E MARIN ye x WOODS HOLE OCEANOGRAPHIC INSTITUTION Reference Rao — a - a point oor OE ati e. BOOK COLLECTION Ore tr Kg Soy —_—_— _——— SS — — — — — _———— —_—_—_— ——= ——=> ———— =y m | aed [-8) ——e —— = = = W WiC Pas a ‘ =: BO, AB pee? 5 meen =f Doc Rar we) u¥ ye oe S= ii / = COLLECTIO! y, J SS — A publication of INSTRUMENT SOCIETY of AMERICA MARINE SCIENCES INSTRUMENTATION Volume 2 Distributed by PLENUM PRESS A publication of INSTRUMENT SOCIETY of AMERICA MARINE SCIENCES INSTRUMENTATION Volume 2 PROCEEDINGS of the SYMPOSIUM ON TRANSDUCERS FOR OCEANIC RESEARCH Held November 8-9, 1962, at San Diego, California EDITOR R. D. Gaul Agricultural and Mechanical College of Texas Associate Editors ’ D. D. Ketchum F. E. Snodgrass Woods Hole Oceanographic Institution University of California R. G. Paquette M. P. Wennekens General Motors Corporation Naval Ordnance Test Station A. J. Carsola W. F. Brisch Lockheed Aircraft Corporation Marine Advisers, Inc. Distributed by 2 PLENUM PRESS NEW YORK F joe? Copyright © 1963 INSTRUMENT SOCIETY of AMERICA Penn-Sheraton Hotel 530 William Penn Place Pittsburgh 19, Pa. PREFACE The scope of the "Symposium on Transducers for Oceanic Research" was expressly limited to measurement devices because it seemed that transducers present the greatest technological obstacles to the advance of research in and over the ocean. It is appropriate that the second major meeting staged by the Marine Sciences Division of the Instrument Society of America should focus on the basic sensing devices in spite of the fact that national atten- tion in oceanography, undersea warfare and marine engineering seems to be concerned with large systems and elaborate schemes. Ocean science and technology, under the pressure of practical demands for knowledge, is placing considerable stress on mental and physical resources at hand. In the final analysis scientific goals and operational concepts must recognize the limitations of tools and technology available for measurement if their planning is not to lead to frustration and failure. Oceanography, like all other experimental sciences, requires the services of scien- tists and engineers from fields far removed from its basic areas of inquiry. Until fairly recently this requirement has been met by a small number of extremely versatile men who managed to design, and frequently hand-build, the apparatus they needed for their investi- gations. Most of the successful instruments in use today derive from this source. Present needs for instrumentation exceed the capacity of those rarely talented individ- uals who can bridge the gap between science and engineering with proficiency. As the ocean research effort gains momentum organizations and professional people are moving with their talents and experience from related fields in an attempt to fill the stated needs. These same demands are pressing specialization and consequently widening the gap between people who design instruments and those who use them. Success must necessarily depend on the degree of mixing achieved. The science will not be served by conversations held exclusively among instrumentation engineers, nor are better instru- ments likely to appear unless scientists are continuously aware of advances in technology. Statements of needs and appreciation for the limits within which they may be satisfied are necessary before the applied scientist and engineer can offer the new measurement techniques that may make feasible previously impossible experiments. We believe that the collection of papers in this volume shows a high degree of mixing and offers a sig- nificant contribution to the fund of knowledge so essential to the advance of marine science. During the course of the Symposium from which this volume has been drawn, one of the most successful practitioners of the art of oceanographic instrumentation remarked that he was disappointed in the number of papers describing instruments that had not yet emerged from the laboratory. His criterion for a successful paper required actual scientific data gathered with the instrument in question and a report based only on the results of prolonged field use. No one can argue the merit of a thorough testing in the environment as a means for evaluating an instrument; neither can an instrument be considered useful until it has contributed to a successful investigation. However, the rapid expansion of effort in the marine sciences demands an early transfer of informa- tion on research and development in progress if duplication and overextension are not to cause ineffectiveness through dilution of our resources. A concerted attempt has been made in the editing of this proceedings volume to strike a compromise between publication lag time and maintenance of literary standards. Each of the associate editors served in the dual role of session chairman at the Symposium and technical reviewer of the papers presented in his session. The topical organization of the meeting has been preserved in the proceedings. R. D. GAUL College Station, Texas D. D. KETCHUM Vineyard Haven, Mass. 20 December 1962 : ‘ Me . | ¢ ; weary y ‘a Te laa , by rae a ne Aris : ‘ . } id bi “ Drebir yer) : Looe ih: () ( ci dal arch i eRe hata ss bid ns o nh mien oi “ae : ae wf imo ah ’ Bes: 1 es Ming vi Hs OEE jenna eat ; 4 - As ee I Se es | paint ‘ehaes genta ’ ons aia ARO, PCa ena Nie iy i Syoiakire Gt mallet | ia Re Bed iat oa dh go ey _aainaee f t A ae eeaitl Mis 2 6 UR Spit oy mame, 4 i ane =H Kiva Dae i Etec ee i ira wh theta ) iF, teaciwamien WAT wide toe. obtain Maem for ook heagtlbas tga a LgORmE? : LLP a ame tani Smee civ" aS ake PNR seit cs Rapa ta mT a : » 5 bans hi Sue AD hats eae nme hast Sa MS ia: sf t a0 COLA . é au cad vr oak Se (hikes “ei: 46082% . core, nity MX, 1 ty Seed. hn, ptey Cs 4 af oe: tale. “Sh aaa agwrl er : i id “iy f ’ ; I? MN pe —s Aa i st peered Blatt ah : i ‘babs w ene 4 i $5 ; ‘ EAE a. ¢- B70. em “aes sae e eee dir ; ’ 4 jah le sept ae a 2 nd voy ‘4 see Toe dee Sie DRO Eee a aioe ert & rg roe aiep, ates wid ah oy a is34 7m “hd een Sh mmiperite ts iis 2 -ralde edd, die bon eh W ree esc, ee praae ote a i putea e. % wees chord. tepan edd ' ee rw sah Sym if MEA be otto Lae ¥ aad Vs wey: ha few Li iw. Soomro ve ‘AA nde’ eo, eaMpah, wit prsara wil tests ail 4 leat. etied & et " SNe Pwo, hee ole (iw (al ORE Abed’ ene ity ASA et Go isy “enege ckaatoibadtin 4 Dit ie BRitentos. opel oe ‘ ene Me eo Get tattoo) ae wee j \ fe hit i eh ae omard, |e ial 8 ee he f Baa Mint jer eo 1 HUT aye “yaoi eee) 1 i ih we Ma Se qa nT dvi ot) , 9 pedro bewntante: Aileen rid earn , ii hoe Dey ; cE Hd » ethers oe 4 based = abe a m sac bu = oF 1 cote sehiowa /t: cut / i r ' (ne O2E haa : ‘thie > ites Fav . # i. vy hal an! yi 7 ft t 50g i” ‘a 7 4 7 y t ; ae acid a ak ate ie owt oohe ae ie Heh “tO Denies . r walt ¢ * nd f CR Mie) wine toy Rally ana ee CONTENTS TEIMIVNGID|. oi lowoman somata Lob a 0! Co guno) bo Rocce ets ecto Keynote Address TRANSDUCERS FOR OCEANIC RESEARCH--A WIDE SCOPE HUG sto, Welisyent G fold o oo Go a ; eLitlih cater die Persmme ne tele Session I - Electrical Conductivity, Salinity and Density INTRODUCTORY REMARKS Reply S. Ce UCC Aaiomiai mein RMLs cfurstarol¥roinpiek ores: «ey Yee bre) i alll “ai Gelq te othe tires Renmin Stabe A PROBE TYPE INDUCTION CONDUCTIVITY CELL Wig Wig) AMcierlehtl hail ING Isla syehaleetersals Hila oO kao ue Gono Golo) a alos A PROPOSED IN SITU SALINITY SENSING SYSTEM Nips cape 2 OW itnarc wtcumtwe ci emnzenLe Peioumt sas subheciL. s Vibise, inj waweneed ety Weds eteNien Gey palbeeeh GMO eed AN IN SITU CONDUCTIVITY METER Di Demo ksinmerssy Wamraisr teMewels tiles) cris 2c) co eisea Gitte) tect wine eevee ULTRASONICALLY CLEANED ELECTRODE Ito dale, \aleieKeyersyay Giavel Ws lio) Weeeerale og GoM SS) 6 65 Gees 4 6 5 Session II - Temperature AN IN SITU TEMPERATURE SENSOR Ps Go Gaal eine! da lilo WaCMWNESOMs 5 oo bo Ooooh A RAPID RESPONSE HIGH ACCURACY THERMAL PROBE als Wily WeWeYONAEIO 5 oh 6 6 Q Sach eo OMNES chemi ae hac ko A MOBILE INSTRUMENT FOR STUDY OF OCEAN TEMPERATURE IN THE THERMOCLINE REGION We Me WsivisliiG@, To Is Ollsoyai eye] Ws 5 Brenly 6 6 5 5 6 6 oo fo TOWED SEA TEMPERATURE STRUCTURE PROFILER Dig Oop AbcHMoyatel GMS to) G6. G5 dio loc ca Oe DRMCHED IAS eaeO Ean mEPEEAE Ene A A ATRBORNE INSTRUMENT FOR PRECISION MEASUREMENT OF SEA SURFACE TEMPERATURE USING INFRARED RADIATION EMITTED BY THE SEA Io IeAleeunta eyael IM, WHS 4G o 5 0 6 8b do oo oo 5b bob oo Go oo Session III - Systems CABLE ASSEMBLY WITH INTEGRAL HYDROPHONES AND INSTRUMENTATION 5 IRs fNalelenesyoyal 5 Go a 5 ci OO SiReG DlacSe dt ong TONe Mie Bar ERLOn COMPUTERS AND TRANSDUCERS Py 1G, yak lelalenavels ids Miwaterelisatvay 6 igh ay Ou Ou 5 aot oglu Gacmon Cele UNTENDED DIGITAL DATA ACQUISITION SYSTEM PHILOSOPHY A. R. IMGGZA STR Wen eubtatital kote. colulcimb ener! we ahaa Matereghhas (0We A mieies . Ptah Gs Ooi vii 10 11 19 25 29 33) 43 ho 93 61 1 81 87 Session IV - Wave Measurement A RESISTANCE WIRE WATER LEVEL MEASUREMENT SYSTEM Ro Ae Agrees Giagl Do Uo Gastallere 5 o go be Goo ooo Oo SH SHIPBOARD ULTRASONIC WAVE HEIGHT SENSOR eta Ish SNH Ge. Wee rain Neg eo, Rebo\es thc so eaumes: fo. ig. & 4 int : , hs ‘ ‘ i i 4 } Mega) Aer" a any - ore, if eames Bl iy Lae.» int 7 ” : i Ae Lriggabte btw’ 6 ¢ (Lat ‘ “yt pe Wil “Oh ae @ iy a) af oy Mh i tt i Aas f * ey y f : 1 Bu ed ah mse " ) Hay ae af 5: 2 Salad, Teoh Maher * " a Ce | Xe \ ' ‘ pepe 3 : ie { 4 a } ris Ab suis mp he bg : Piso on te ot RD e 7 ah ’ any gy abe sey me Pay utheny Fail, Mae “ee? mebat) our bra Ae a.) - - ~~ Teese. Wek ve ed fed fa is sp tte) Zk _npgeubitasy aie. ae ay aah at ae hartig tis, oapo-n er. i " ff re iy =f Dap. aa ee ee 4 ‘ a ; tnt ‘ Ps ‘ ve “s T* 3 ‘> TF Dy hia ee Os Yan? = i a gs ae 7 Bory ‘h + , nie Sah i sadbaial “ * 1 ATO MUR Se. Wd & nese wae ¢ ‘ye oi) (ay ORS : ‘ NC a Fg 0 aoe wah a his 3 oh ene ee “o i ee A 72 we ‘ f ite AE, y yy i th ee 2 J ° SESSION I ELECTRICAL CONDUCTIVITY, SALINITY AND DENSITY Chairman: R. G. PAQUETTE Defense Research Laboratories General Motors Corporation Santa Barbara, California INTRODUCTORY REMARKS FOR PAPERS ON ELECTRICAL CONDUCTIVITY, SALINITY AND DENSITY R. G. PAQUETTE, Session Chairman Defense Research Laboratories General Motors Corporation Santa Barbara, California As an introduction to the papers presented on electrical conductivity, salinity and density, it seems appropriate to dwell briefly on accuracy required of transducers for measuring these parameters. The term "accuracy" is used in its scientifically accepted sense as associated with a 95% probability. SALINITY When asked about his requirements for accuracy in oceanic measurements of salinity, the oceanog- rapher usually will give almost a stock answer of 10.01 parts per thousand. Since the salinity of the ocean is about 35 parts per thousand this represents an accuracy of about 1 part in 3,500. Actually, for many purposes, a lesser accuracy is permissible; in fact, until the advent of the salinity bridge in 1956 the accuracy was about 0.03 parts per thousand or less although few oceanographers clearly recognized that it was so poor. For other purposes, especially any studies in the deep ocean where gradients and temporal changes are extremely small, 0.01 parts per thousand accuracy may be marginal. In large deep estuaries accuracies of about 40.1 parts per thousand are desirable and in estuaries strongly influenced throughout most of their depths by fresh water, accuracies of =0.5 parts per thousand may be permissible. DENSITY It is sometimes implied that the oceanographer measures salinity only to obtain density. This is not true. If he could measure density to the required precision he could deduce circulation patterns by the geostrophic method, but there are other oceanic processes which would still require salinity measurement for their descrip- tion. Consequently, there is not as much inducement to produce an in situ density probe as a salinity probe. If density is to be regarded as an indirect route to the measurement of salinity, note must be taken of the fact that an accuracy of 0.01 parts per thousand in salinity corresponds to an accuracy of 7 x 10-6 em/em3 in density. Such accuracies probably ean only be obtained by differential measure- ments. Since the density dependence on ale) by the working scientist at sea. temperature is equivalent to salinity changes of about 0.4 parts per thousand salinity per degree C, precise control or measurement of temperature would be required. ELECTRICAL CONDUCTIVITY The electrical conductivity of sea water is of little interest in itself; it is normally used to determine the salinity. The electrical conduc- tivity of sea water in millimhos is roughly the same as the salinity in parts per thousand, nominally 30 with a marked dependence on tempera- ture. The conductivity increases with tempera- ture by about 2% to 4% per degree C. To obtain an accuracy of 40.01 parts per thousand in salinity requires a temperature measurement accuracy, or at least a long term reproducibility, of about 0.01°C if all the error is due to tem- perature. The existence of other errors forces distinctly higher requirements on the temperature accuracy. Again, it should be evident that there is not much hope in making absolute measurements and that differential measurements referred to a stable temperature-compensating reference stan- dard will be required. The electrical conduc- tivity also has a pressure coefficient of the order of 0.0005% per psi so that at any great depth provision must be made for a pressure eorrection. CONCLUSION These are difficult requirements, all the more so because they must be met under rugged ship- board conditions and in the hands of technicians who are not experts in the art of measurement. I should like to remind you that in demonstrating the utility of a measuring instrument there is a great deal of difference between 20 measurements in a laboratory and a year of carefully docu- mented testing at sea. There is also an impor- tant difference between the instability and incon- venience which will be tolerated by the enthusi- astie inventor and that which will be accepted I hope we can begin to see more of thoroughly demonstrated instruments in distinetion from those which are so inadequately tested that they are merely interesting. A PROBE TYPE INDUCTION CONDUCTIVITY CELL E. E. AAGAARD and R. H. vanHAAGEN Oceanic Instruments, Inc. Houghton, Washington ABSTRACT The salinity of sea water may be calculated from measurements of water electrical conductivity and temperature in situ. A compact pressure- protected induction conductivity cell and a ther- mistor temperature sensor were mounted in a probe to form a salinity transducer. Constructed for installation in the nose of a deep-sea remote controlled underwater vehicle, the probe is read- ily adapted to other applications such as mounting on or towing behind a research ship. The con- ductivity cell is of the induction type, using two toroidal transformer cores to avoid the polar- ization and fouling problems of exposed electrodes. The toroidal cores are connected in a transformer bridge circuit which is adaptable either to direct or to servo actuated null balance readout. The probe has been proof-tested at pressures up to 3,000 psi. INTRODUCTION One measure of the salinity of sea water is its electrical conductivity. A conductivity cell for measurements of salinity may be calibrated in a sample of sea water of known salinity. The temperature of the sample must be carefully con- trolled since conductivity varies considerably with temperature. For in situ measurements of salinity the temperature of the water as well as the electrical conductivity must be accurately measured. The salinity transducer probe, as shown in Fig. 1, uses an induction conductivity cell and a thermistor temperature sensor. Con- structed for use in the nose of a deep-sea remote controlled vehicle, the transducer posed a major design problem because of the crushing pressure at great depths. In the past, accurate determination of con- ductivity has been a serious problem because fouling and polarization of exposed electrodes in a sea water environment can introduce serious errors in the results. To eliminate this inherent problem with exposed electrodes, the conductivity probe design was based on the induc- tion technique described by Gupta and Hills. The measure of conductivity is the coupling pro- vided by a sea water path between two toroidal transformers. OPERATING PRINCIPLE The transformer cores are encapsulated on a common axis in a toroidal housing which allows the free flow of sea water through the common hole in the two cores. The arrangement is shown in Fig. 2. A primary winding on the exciting transformer core is connected to an alternating current power supply. A single turn secondary winding is formed by the conducting sea water threading through the hole in the toroidal housing. The return current path is the infinite liquid medium in which the unit is immersed. The current through the liquid path will be proportional to the electrical conductivity of the medium. The sea water current path is also the single turn primary winding for the pickup transformer core. The output signal is taken from a secondary winding on this core. By placing on each core an additional winding connected in a series circuit with a variable resistance, the induction conductivity cell becomes a transformer bridge. The setting of the variable resistance required to give a null signal in the secondary winding of the pickup core is a measure of the electrical conductivity of the liquid path. An outstanding advantage of the transformer bridge circuit is that the null adjust- ment is essentially independent of variations in the transformer core losses due to temperature or pressure. TRANSFORMER BRIDGE Calvert et al.° have described the operation of the transformer bridge in some detail. Only a brief discussion of its application in the induction conductivity probe will be given here. Fig. 3 is a schematic diagram of the circuit. Ry represents the unknown conductivity of the sea water path. The circuit is adjusted to null by the variable resistance, R,. The capacitor tunes the null signal to the power supply frequency. Tuning is desirable since it filters out noise above and below the power supply frequency and eliminates null signal phase shift with respect to the exciting voltage. Phase shift is undesir- able where the null signal is compared with the exciting voltage as in applications using a phase sensitive detector or servo actuated null bal- ancing system. Superior numbers refer to similarly numbered references at the end of this paper. ql Fig. 1. Salinity transducer probe. It will be shown that for a discussion of the bridge null adjustment the transformers may be considered to be ideal. Core losses and magne- tizing reactance in the exciting and pickup transformers shunt the power supply and the nuJl signal, and as such tend to increase the required input power and reduce the circuit gain but do not change the null setting. In a like manner the copper losses (resistance) in the exciting and pickup windings, Nj and No may produce small 12 TUNING @aueinen OUTPUT SIGNAL obs ei Mec \ o See She a 25S ae) Vv -O0. 1 a () 5 10 15 20 25 30 TEMPERATURE (°C) Fig. 2. Thermistor compensation network and graph of salinity correction as a func- tion of temperature for water of salinity of 35.3 parts per thousand. which was designed to have a resistance tempera- ture relationship similar to sea water.2) 1; In Fig. 2 a typical circuit and its compensation curve are shown. The errors between 5°¢ and 20°C are usually quite small but outside these limits increase quite rapidly as shown. In some eases, depending on the characteristics of indi- vidual thermistors, errors can be as low as #0.05 parts per thousand in salinity from 0°C to 25°C. However, thermistors have a number of serious disadvantages. They are non-uniform in characteristics and are sometimes unstable. The compensation accuracy is generally inadequate as shown in Fig. 2. Compensating Cell A theoretically ideal method of temperature compensation using sea water itself as the com- pensating element was experimented with by the author and others at the Woods Hole Oceanographic Institution. This technique utilized Copenhagen standard sea water in a small sealed platinum- electrode glass conductivity cell. The cell was fitted with a flexible membrane so that the sealed sample of standard sea water in the cell was at the same pressure as well as temperature as measured sea water sample. The earlier experimental units had a thermal response time of 0.8 seconds. However, their long term sta- bility is poor, at least in the cells made to this date. They are also fragile and difficult to fabricate. Platinum Resistance Thermometer Bridge Circuit A compensation circuit consisting of a double bridge circuit incorporating two precision platinum resistance thermometers has been studied. Two variations of this scheme are shown in Fig. 3. Even though the temperature coefficient of a platinum resistance thermometer is only about nl Ral =Ra2 = Rei —RB2 Rei =Re2 Rr1 =Rr2 Ry] &Rt2 ARE THE RESISTANCE THERMOMETERS Rea RT2 E2 =r; N2 Iy- 12 Rr1 & Ry2 ARE THE PLATINUM RESISTANCE THERMOMETERS Double Wheatstone bridge compensation cireuit (above) and double transformer compensation circuit (below). one-sixth that of sea water at 15°C and of Opposite sign, it can be shown that the circuits shown in Fig. 3 can be made to have a tempera- ture coefficient closely matching that of sea water from 0 to 25°C. An examination of a simple Wheatstone bridge with a resistance thermometer in one arm will show that the relative temperature coefficient of the ratio of the output short circuit current to input voltage, (1/I)) (dIg/aT), becomes pro- gressively larger as the bridge approaches balance and reverses sign on the other side of the balance point. However, an analysis of the simple bridge shows that if the bridge is adjusted to give accurate compensation at 15°C, the temperature coefficient of the bridge rapidly deviates from that of sea water at higher and lower temperatures. However, a detailed analysis of the double transformer bridge cir- cuits shown in Fig. 3 leads to the results shown in Table I. For the various values of R, the bridge resistors, Rp, and Rpo, were adjusted so that the temperature coefficient of the bridge exactly equalled that of sea water at 15°C! Where R, is large the temperature coefficient of the bridge circuit closely matches that of sea water over a wide range. In Fig. 4 a plot is shown of the salinity errors (for S = 35 parts per thousand) with R, at a very large value. The salinity errors are quite small for tempera- tures below 22.5°C and due to the stable charac- teristics of well designed platinum resistance thermometers these errors are quite stable and can be allowed for in the final analysis. PRESSURE COMPENSATION The effect of hydrostatic pressure on the conductivity of sea water) is very considerable as shown in Fig. 5. The figure also shows that the relationship between pressure and conduc- tivity is not linear and that at pressures as Table I. Double Bridge Temperature Temperature Coefficient (4/2C) Sea Water Temperature (°c) Rg=0 Rg=120 Rg=2000 Coefficient (%/°C) 0 3.069 3.020 2.986 2.999 5 2.763 2.734 2.714 2.708 10 2.505 2.492 2.48) 2.481 15 2.286 2.286 2.286 2.286 200 2.099 2.108 2.114 2.118 25 1.935 1.952 1.963 1.976 + .050 4 8 -15° SALINITY i a ate ERROR - P.P.T. 7 T=20 C -.050 0 10 20 30 6 TEMPERATURE “C 5 PERCENT INCREASE 4 IN CONDUCTIVITY 3 2 i+ 0.08 1 soos) (0) DIFFERENCE 1 0-04 0 2 4 6 8 10 12 14 16 %o/°C 40.02 HYDROSTATIC PRESSURE PSIG oo ana (x 1000 ) Rs=INF 4 eae = 0.02 Rs=120 a Fig. 5. Change of conductivity of sea water. = 0.04 Rs=0 0 5 10 15 20 25 SEAWATER TEMPERATURE °C 1.6 Fig. 4. Compensation errors for circuit shown in Fig. 3 and difference in temperature 1.4 coefficient between circuit in Fig. 3 r and sea water. 1.2 small as 50 psi an error equivalent to 0.01 parts PRESSURE per thousand in salinity results. Also, the COEFFICIENT pressure effect is quite dependent on tempera- OF ELECTRICAL 1.0|- tures!® as shown in Fig. 6. The effect of pres- sure at O°C is twice as great as at 30°C. Con- sequently, any pressure compensating device must take into account the temperature. Fig. 7 shows a circuit that has been designed to provide an input-output relationship that closely simulates the changes in conductivity due to changes in pressure. The loading effect of R2 on the pressure potentiometer is adjusted so that the relation- ship between E, and the pressure applied to the pressure potentiometer closely simulates the con- ductivity pressure relationship of sea water. The reduction in the pressure coefficient with increasing temperature is closely simulated by 22 CONDUCTIVITY (10> DECIBAR ') 9.8 0.6 (0) 5 10 15 20. TEMPERATURE °C 25 Fig. 6. Effect of temperature on pressure coefficient at 1,500 psig. the change in resistance of a thermistor resistor combination, Rpqy and Rj. By adjusting Rj, Ro and R3, the effect of pressure on the conductivity of sea water can be simulated with an overall accu- racy of 1% at temperatures below 10°C. At R2>>Rp R2>> RTH THERMISTOR INGR6 Tifa PRESSURE POT Pressure compensating circuit. BRIDGE OUTPUT TOROID BRIDGE INPUT TOROID “TEMPERATURE COMPENSATION CIRCUIT ] | | | i Rp | R3 | (SSS SSS 1% ! FE) = Eg Rey = Reo Ni= N2- R11 = R12 Fig. 8. Complete salinity bridge. 20,000-foot depth, i.e., 8,800 psi and O°C, the inerease in conductivity due to pressure is approximately 5.5%. Therefore, at this depth and pressure, the uncertainty due to pressure compensation errors is 0.02 parts per thousand. At shallower depths the uncertainty is reduced in proportion. COMPLETE SALINITY SENSING BRIDGE The salinity bridge circuit complete with tem- perature and pressure compensation circuits is shown in Fig. 8. A current, Ij, is induced to flow in the sea water loop by the application of EH; to the toroidal transformer, Tj. I,, which is proportional to the conductivity of the sea water loop, sets up a magneto-motive force on the magnetic circuit of To. A counter mmf pro- portional to the difference between IjN] and Ip5N5 is set up by the combined outputs of the pressure and temperature compensating circuits. At one particular value of salinity (depending on Rs) these mmf's will be in balance and Eo will be zero. Any change in I, at constant salinity due to temperature or pressure changes is bal- anced by a similar change in the output of the temperature or pressure compensating circuits, thus maintaining a balance dependent only on salinity. PHASE SHIFTING NETWORK Ra Rp ER=Eo+Eq | + Fig. 9. Block diagram of salinity oscillator. SALINITY OSCILLATOR In a simple system the salinity bridge shown in Fig. 8 could be balanced by a servo system acting on Rg. Rg could be a precision potenti- ometer coupled to a shaft encoder for digital readout. However, this system would require con- siderable amounts of power in an underwater package and would have the disadvantage of larger size and lower reliability due to the number of moving parts. An improved system might utilize the output voltage to input voltage relationship of the salinity bridge to control the frequency of a special phase shift oscillator. A-block diagram is shown in Fig. 9. It can be shown that with suitable design the error voltage, Eo, from the salinity bridge is either in phase or 180° out of phase with the bridge input voltage, Ei. If the error voltage, Eo, is added to a voltage, E,, which is 90° out of phase with Ej, then the phase of the resultant Eo + Eq = Ey, will shift as the bridge balance changes with changes in salinity. The resultant, E,, is amplified and then applied to a phase shifting network consisting of Ry, Rp, Ca and Cg. The output of the phase shifting net- work is amplified in Aj and applied to the input of the bridge, thus closing a complete loop. The loop will oscillate at a frequency at which the sum of the phase shift between EH, and E; plus the phase shift in the phase shifting network amounts to 180°. Experimental oscillators of this type have shown that the salinity uncertainty due to supply voltage and temperature variations on the electronics is not worse than +0.003 parts per thousand in salinity for an oscillator covering a range of 5 parts per thousand. The quadrature voltage circuit shown in Fig. 9 has to be temperature compensated because when the bridge is off balance, EH, will not be zero and will vary with temperature at a given salinity. This means that E, should vary with temperature by the same proportional amount as Eo. However, the accuracy with which E, is compensated becomes less critical as the bridge approaches balance and is completely unimportant at the balance point. Consequently, the accuracy with which Eq is compensated is only important when the salinity value is a long way from the value at which the bridge is balanced. To insure good stability in the basic oscillator, the amplifiers A, and Apo have very good phase shift stability and the level of oscillation is controlled by the auto- matic gain control circuit to a level well below overload. CONCLUSIONS It is estimated that the combined stability of the conductivity sensor, the temperature and pres- sure compensating circuits and the electronics will yield a repeatability of 40.005 parts per thousand and an accuracy of 40.05 parts per thousand in salinity including both temperature and pressure compensation errors from O to 25°C and O to 20,000-foot depth. However, the com- pensation errors can be accurately determined and allowed for in the final analysis with a resulting accuracy approaching +0.01 parts per thousand salinity. REFERENCES 1. JONES, G. and G. M. BOLLINGER, The measure- ment of the conductance of electrolytes TIT. The design of cells, J. Amer. Chem. Soc., 53, 411-451, 1931. 2. HAMON, B. V., A portable-temperature chlor- inity bridge for estuarine investigations and sea water analysis, J. Sci. Instru., 33, 329, 1956. 3. BRADSHAW, A. L. and K. E. SCHLEICHER, A con- ductivity bridge for the measurement of salinity of sea water, Tech. Rept. 56-20, Woods Hole Oceanographic Inst., 1956. UNPUBLISHED. 4. COX, R. A., The thermostat salinity meter, Internal Rept. C2, National Inst. of Oceanog- raphy, 1958. UNPUBLISHED. 5. WENNER, F., E. H. SMITH and F. M. SOULE, Apparatus for the determination aboard ship of the salinity of sea water by the elec- trical conductivity method, National Bureau of Standards J. Res., 5, 711-732, 1930. 6. JACOBSEN, A. W., An instrument for recording continuously the salinity, temperature and depth of sea water, Trans. Amer. Inst. Electrical Eng., 67, 714, 19h8. 7. HAMON, B. V. and N. L. BROWN, A temperature- chlorinity-depth recorder for use at sea, J. Sei. Instruments, 35, 452-458, 1958. 8. HARWELL, K. E., Radio frequency salinity instrument, model E, Tech. Rept. 4, Depart- ment of Oceanography and Meteorology, A. & M. College of Texas, 1954. UNPUBLISHED. ah 9. 10. iuL; 125 335 a. Ic 16. HUEBNER, G. L., Notes on radio frequency salinity measuring equipment at Texas A. & M. College, Publ. 600, National Academy of Sci., National Res. Council, 1958. GUPTA, S. R. and G. J. HILLS, A precision electrode-less conductance cell for use at audio frequencies, J. Sci. Instruments, 33, 313-314, 1956. ESTERSON, G. L. and D. W. PRITCHARD, C.B.1. salinity-temperature meters, Proc., Conf. on Coastal Eng. Instruments, Berkeley, Cais? 5 1955, University of California Council on Wave Research of the Engineering Foundation, Berkeley, 260-271, 1956. ESTERSON, G. L., The induction conductivity indicator, Tech. Rept. 14, Ref. 57-3, Chesa- peake Bay Inst., Johns Hopkins University, 1957. UNPUBLISHED. BROWN, N. L. and B. V. HAMON, An inductive salinometer, Deep-Sea Res., 8(1), 65-75, 1961. BROWN, N. L., A. L. BRADSHAW and K. E. SCHLEICHER, A recorder for in situ measure- ment of salinity by the inductive method, temperature and depth, Unpublished work under Office of Naval Research Contract 2196-7 at Woods Hole Oceanographic Inst. BRADSHAW, A. L. and K. E. SCHLEICHER (private communication) unpublished work in progress on the relationships between conductivity, salinity, temperature and pressure of sea water under Office of Naval Research Contract 2196-7 at Woods Hole Oceanographic Inst. HAMON, B. V., The effect of pressure on the electrical conductivity of sea water, J. Mar. Res., 16, 83-89, 1958. AN IN SITU CONDUCTIVITY METER D. D. SKINNER Westinghouse Electric Corporation Pennsylvania Pittsburgh, INTRODUCTION An indirect technique commonly used for salinity determination is the measurement of the electrical conductivity of the water. Conduc- tivity cells using platinum or similar metallic electrodes have been used for conductivity mea- surements. However, electrode polarization and contamination has limited the long term accuracy of calibration of such devices. Some of the more recently developed conduc- tivity meters have used induced currents in the water to measure the conductivity. This tech- nique eliminates the problems of electrode polar- ization and contamination and appears to lend itself to use as an in situ instrument. THE WESTINGHOUSE CONDUCTIVITY METER An induction conductivity meter suitable for use as an in situ instrument is being developed by the Westinghouse Electric Corporation. This meter utilizes two toroidal inductors in a balanced bridge circuit. A simplified circuit diagram is shown in Fig. l. The inductors, Ly and Lo, are wound on mag- netic toroids made of a ferrite material and are constructed as nearly identical as possible. The resistors, Ry and Ro, are identical and their resistance is approximately equal to the induc- tive reactance of ly and Lo at the frequency used. As a result the bridge circuit should be balanced there being no signal output to the null detector. To make measurements of conductivity, the inductor, L,, is submerged in the fluid whose conductivity is to be measured. The fluid around and through toroidal inductor L, com- prises a one turn secondary winding on Ly with a resistive load inversely proportional to the fluid conductivity. When placed in a fluid, the fluid induces a resistive component into 1), unbalancing the bridge circuit. A secondary winding is placed on Ly and is loaded with a variable resistor. This resistor and secondary winding induce a resistive component into Lo and the variable resistance can be adjusted until the bridge is again balanced. The value of this resistance is thus a measure of the conductivity of the fluid surrounding 1). i) Fig. 1. A simplified circuit diagram of the Westinghouse induction conductivity meter. The inductor 1) is exposed to the surrounding fluid in a manner similar to 1; and in close proximity to lL), but the medium is prevented from loading Lj by the presence of a thin, compliant, electrically insulating membrane closing the hole through the center of the toroid. Thus inductor Io is exposed to the same thermal and pressure environment as Lj. Inductors 1) and Lo are physically made as identical as possible. the thermal and pressure effects on the two inductors should be the same and as a result the bridge balance should be inherently independent of pressure and temperature effects. Thus Because Ly and Lp are identical as are R, and Ro, the bridge balance is also inherently inde- pendent of the frequency, amplitude and waveform of the signal source. This is of considerable value when the meter is used as an in situ instru- ment where it might be difficult to maintain accurate frequency and amplitude control on the signal source. ACCURACY CONSIDERATIONS The accuracy of this circuit is fundamentally limited to measurement of the smallest change in conductivity that will produce a bridge unbalance signal larger than the input noise of the null detector. Therefore, electrical and physical eircuit conditions, which will produce the greatest percentage change in the impedance of Ly for a given change in fluid conductivity, are the conditions for greatest accuracy. The impedance of 1) is essentially equal to the reactance of the inductor paralleled by the induced loss resistance due to the conductive medium. The circuit conditions which will make the value of this parallel resistance smallest and the value of the parallel inductance rela- tively greatest will produce the greatest accuracy until the resistance is small compared to the inductive reactance. When this condition is reached the effect of the parallel inductance will be small and the bridge will essentially consist of 4 resistive elements. Under these conditions the percentage change in the impedance of the Lj arm of the bridge will be determined directly by the percentage change in the conduc- tivity of the surrounding medium. Until this condition is reached, the accuracy of the meter is determined by the following considerations. The value of the induced loss resistance con- sidered in parallel with the inductive reactance of Lj is numerically equal to the resistance of the fluid path through and around the toroidal inductor multiplied by the square of the number of turns on the inductance. The value of the induced resistance can then be decreased by decreasing the resistance of the fluid path and by decreasing the number of turns on the inductor. Decreasing the number of turns of the inductor decreases the parallel inductance by the same ratio that the induced resistance is decreased so that the sensitivity is independent of the number of turns comprising the inductor. The resistance of the fluid path is determined by the physical geometry of the inductance. Increasing the area of the hole through the toroid and decreasing the length of the hole as well as decreasing the cross-section of the core, windings and covering material will decrease the resistance of the fluid path and hence increase the sensitivity of the device. The sensitivity of the device can likewise be increased by increasing the relative value of the parallel inductive reactance of 1, and Lo. The inductance of Lj and Lo can be inereased without increasing the value of the induced resistance by increasing the permeability of the magnetic core material used. The inductance can also be increased by increasing the fre- quency of operation. Thus it is seen that the accuracy of the device is proportional to the permeability of the core material and the fre- quency of operation and is dependent on the physical geometry of the cores. PRACTICAL CIRCUIT CONSIDERATIONS In practice the two inductors, Lj and Io, cannot be made exactly identical. They will be found to differ both in inductance and loss. As a result, some means must be provided to 26 compensate for differences of inductance and loss so that an initial balance of the bridge can be obtained. Assuming that the loss component of the impedance of the inductors is small compared to the reactive component, the effect of an induc- tance unbalance can be compensated by changing the values of Rj and Ro. The effect of a loss component unbalance can be compensated by shunt resistors across the appropriate inductor or across both inductors. In practice R, and Rp have been made variable in different degrees to provide coarse and fine inductance balance con- trols. Likewise, variable resistors shunted across lL; and Lo have been provided to obtain coarse and fine loss balance controls. As was mentioned earlier it is desirable to wind the inductors on toroidal cores having as high a magnetic permeability as is possible. As it is also desirable to operate at as high a fre- quency as is practical, many high permeability core materials cannot be used because of the great amount of loss at high frequency. As a result a compromise must be reached between these con- flicting factors. At present, toroidal cores made of a ferrite material having a permeability in the region of 800 to 1,000 are used at a fre- quency of approximately 500 Keps. The impedance of the elements of the bridge eircuit is determined by the inductive reactance of L, and Lo. If the impedance is made too high, stray capacity will alter the desired circuit eonditions and will make the circuit balance fre- quency sensitive. If this impedance is made too low, stray lead inductance will have the same effect. This latter consideration has been a problem particularly in designing the balance resistance connected to the secondary winding on Lo. If stray inductance is present here. it introduces a reactive component into the null voltage, thereby reducing the accuracy with which the null can be detected. In general the imped- ance level of bridge components has been kept in the region of 100 to 1,000 ohms. As it has been found possible to balance the bridge circuit so that the amplitude of the output to the null detector is of the order of 120 db less than the input signal level, it is necessary to design the apparatus physically so as to pre- vent disturbances from outside influences. The electrical circuitry is all enclosed in shielded containers and the toroidal windings isolated with Faraday shields. The mechanical supports for the inductors are made quite rigid so as to isolate the toroidal cores from mechanical stress. This is necessary since the permeability of the ferrite core is a function of the mechanical stresses in the material. Care must be taken to prevent variation of the circuit characteristics by the use of good quality components and rugged physical construction. An alternate method of balancing the loss com- ponent induced in Ly; by the conducting medium is by the use of a variable shunt capacitor across Rj. In this case, the secondary winding and balance resistance connected with Lo are elimi- nated. It can be shown in this case that the circuit is balanced when the balance capacity is equal to the inductance of 1, divided by the product of the resistance R, and the induced loss resistance. Again the balance is found to be independent of frequency. As in the other balance system, 1) is assumed equal to L,, and Ry is assumed equal to R5. No loss is assumed for Ly) and Lo and, as this is not the case in actual practice, the circuit balance is slightly fre- quency dependent. An advantage of this circuit is that it eliminates the problem of a sliding contact in the balance resistance which had been found to introduce an uncertainty into the value of the balance resistor. THE IN SITU INSTRUMENT For convenient use as an in situ instrument, the bridge circuit is supplied with a self- balancing system to reduce or eliminate the inter- connecting cable requirements between the instru- ment and the survey ship. The balance system used with this bridge utilizes the fact that the Signal out of the bridge to the null detector reverses phase as the bridge is adjusted through the null point. Therefore, the null point is characterized by both an amplitude minimum and a phase reversal. A block diagram of the balance system used is shown in Fig. 2. The bridge output is amplified and fed to a phase sensitive detector. A reference signal for the phase sensitive detector is obtained from the oscillator which is used to drive the bridge. The output of the detector is a DC voltage whose amplitude is proportional to the amplitude of the input signal and whose polarity is a function of the relative phases of the input and reference Signals. The phase of the reference signal is adjusted so that the output of the detector is positive on one side of the null and negative on the other side. This signal is amplified and used to drive a reversible DC motor which in turn adjusts the balance resistance in the bridge in such a direction as to effect a null. A data telemetering system transmits the shaft Reference Signal Phase Sensitive Detector Oscillator Balance Resistance Fig. 2. A block diagram of the self-balancing scheme for the in situ instrument. ll position, which is now a function of conductivity, to the survey ship. As was mentioned previously, stray reactances in components of the bridge circuit introduce a reactive component into the null voltage which reduces the accuracy with which the null can be determined. The phase sensitive detector however responds only to signals which are in phase or 180° out of phase with the reference signal. Since the reactive voltage component mentioned above is made to be in quadrature with the refer- ence signal it does not affect the output of the detector. In the case of a laboratory or manually balanced instrument, this type of null detector has been found to give considerably more accurate results than a simple amplitude null detector. SYSTEM ACCURACY Calibration of a conductivity meter in absolute values is a difficult process when a high degree of accuracy is desired or required. Solutions of known salinity must be used and the temperature must be known to a degree of accuracy similar to the required conductivity accuracy. It is con- siderably easier to make an estimate of the expected accuracy by measuring the repeatability of the device. To measure the repeatability, it is only necessary to know the absolute salinity and temperature roughly and to be able to measure relative temperature to a high degree of accuracy. A sample of artificial sea water was made up with an estimated salinity of 35 parts per thou- sand. As a considerable period of time is involved in making repeatability measurements, evaporation from the sample was inhibited by floating a film of oil on the surface of the sample. The absolute temperature of the sample was measured with a mercury thermometer and a thermistor thermometer was used to measure rela- tive temperature to a repeatability of 0.01°C. The sample was stirred continuously to maintain a uniform temperature. One preliminary laboratory model of the con- ductivity meter has been tested for repeatability under these conditions over a test period of several days. The repeatability was found to be tO.1 millimhos for a solution whose conductivity was approximately 55 millimhos. A large part of this error could be attributed to contact noise in the adjustable balance resistor as was men- tioned earlier. It is estimated that elimination of the contact noise problem would improve the repeatability of this particular device to +0.05 millimhos. As this device did not use the best known physical configuration, it is esti- mated that an improved model now under construc- tion can improve this repeatability figure by a factor of 10. CONCLUSIONS It is believed that the circuit described will give an estimated accuracy and repeatability of 0.005 millimhos or better in its final form. Furthermore, the circuit is inherently insensi- tive to external influences, thereby enabling it to be adopted to use as an in situ instrument. 28 ULTRASONICALLY CLEANED ELECTRODE R. H. vanHAAGEN and E. E. AAGAARD Oceanic Instruments, Inc. Houghton, Washington ABSTRACT Electrodes, such as used in pH, Redox dis- solved oxygen and electrical conductivity measure- ments of solutions are frequently subject to fouling when used in biological laboratories or marine environments. A probe electrode has been devised which can be subjected to high intensity ultrasonic vibrations to eliminate the accumula- tion of fouling materials on the electrode sur- face. The electrode is driven by a tube of a magnetostrictive material rigidly attached to the electrode. An alternating magnetic field is supplied by a solenoid wound around the probe. A compact transistorized power supply has been constructed to energize the solenoid. The ultra- sonic electrode was originally developed for Redox measurements. Cells for electrical con- ductivity measurements in two and three electrode configurations are currently under test and results will be described. INTRODUCTION AND BACKGROUND All electrode reactions are either oxidation reactions or reduction reactions and these are responsible for typical marine pitting and gal- vanie corrosion which are so costly to the industrialized parts of the world. Oxidation is de-electronation and reduction is electronation, to use the terminology of Professor E. C. Frank- lin of Stanford University.1 Corrosion protec- tion is therefore obtained by cancelling these effects against each other with sacrificial materials or by supplying a counter electron cur- rent. These reactions are further useful since titration end-points may be obtained for certain reactions by measurement of the oxidation- reduction potential.2 Servo-electric control of a reaction can also be affected by this mea- surement but the metallic electrode must be in electrical contact with the solution in which the reaction is taking place. The susceptibility to biological fouling of electrodes in aqueous solutions is well known and several recent ISA papers describe efforts directed toward relieving this difficulty in measurement of A conductivity of solutions.3,+,9; A single metal electrode was recently required to be used for control feedback measurement in a Waters ala solution of human body wastes in which photo- synthesizing algae would flourish to provide oxygen and protein food for mock space flight. The prototype system (Fig. 1) involved a large resin kettle with the electrode inserted through one e@ork hole. Prototype scrubbed electrode system. Superior numbers refer to similarly numbered references at the end of this paper. 29 THEORY OF OPERATION Analysis indicated that comparatively low level excursions of a disk electrode at a high rate should provide sufficient flow, due to non- linearity of the hydraulic coupling, that a strong scrubbing action would occur, even in the absence of cavitation, and that this scrubbing might prevent or remove the normal biological fouling. DESIGN A half-wavelength magnetostrictive tube was rolled of sheet nickel and supported at the nodal midpoint by three setscrews (Fig. A) Ebwas excited at its resonant frequency of 32 Keps by an efficient solid-state driver of simple design. electrical noise pickup in the cable or electrode. This precaution proved unnecessary, apparently due to shielding precautions and effects of symmetry. ‘POTENTIAL FUTURE APPLICATIONS Success of this design has lead to several pro- grams for further development. Design of a simi- larly rugged pH cell using the antimony, antimony- trioxide electrode is being evaluated. This cell is typically erratic in its behavior, but a sur- face contamination may be the reason. The appli- cation of the self-scrubbing principle for improving the thermal contact of thermistors, platinum resistance thermometers and thermo- couples in solutions and powders is underway. Evidence shows that a considerable fraction of Dae 2c A 0.5-inch diameter brass endplate was rigidly fastened to the tube and the exposed surface heavily gold plated. The assembly was seated with silicone rubber in a tube of Nylatron GS, a molybdenum disulphide-filled nylon, which also formed the bobbin for the exciting winding. The winding was potted with an epoxy resin that bonded the conductors mechanically and simul- taneously offered chemical protection. Electri- cal connection was made directly to the brass plate by soft-soldering the copper center- conductor of an unterminated miniature low-noise coaxial cable (Microdot). The complete unit is shown in Fig. 3. Power connections were to ordinary 115 volt 60 eps AC although any source could be utilized by the semiconductor driver which requires a total of 15 watts. The electronic design is perhaps most notable because of the fact that only twenty components are required, of which twelve are in the power supply circuits. This has allowed six months operation without a com- ponent failure and should meet all requirements of both inner and outer space. A pi-section rejection filter was designed and included to reduce anticipated ultrasonic 30 Half-wavelength magnetostrictive tube. the boundary layer insulation effect can be eliminated by sufficiently strong circulation. Heat flow from the magnetostrictive driver losses, as well as from the energetics of fluid reaction, must be considered in the design, although the acoustic power is in the milliwatt region and is of lesser importance. Electrode measurement of conductivity in natu- ral environments, as opposed to the electrodeless transformer methods discussed earlier3»+)7,© has the advantage of utilizing basic measures of length and current with greatly reduced complexity and with higher accuracy than is possible with electrodeless methods. Fig. 3. Self-scrubbing REFERENCES il. 2. PAULING, L., College Chemistry, 1951. MACINNES, D. A., Principles of Electrochemistry, Dover Publ., New York, N. Y., 1-478. WILLIAMS, J., A small portable unit for making in situ salinity and temperature mea- surements, Preprint 43-NY60, Instrument Soc. Amer., September 1960. AAGAARD, E. E. and R. H. vanHAAGEN, A probe type induction conductivity cell, Marine Sciences Instrumentation, 2, Instrument Soc. Amer., Plenum Press, New York, N. Y., 1963. BROWN, N. L., A proposed in situ salinity sensing system, Marine Sciences Instrumenta- tion, 2, Instrument Soc. Amer., Plenum Press, New York, N. Y., 1963. SKINNER, D. D., An in situ conductivity meter, Marine Sciences Instrumentation, 2, Instrument Soc. Amer., Plenum Press, New York, N. Y., 1963. Syl electrode assembly. 7 ia y y Fie qinee y i a ae SESSION II TEMPERATURE Chairman: M. P. WENNEKENS Naval Ordnance Test Station China Lake, California ati) eee vk tae Cay . 0 ¥ \ j Ape ih By Kes le Sa ae ER a) N°) SR AD ay Ln Rea RRM A unm wait ha miata er epiiare tasT otannlinth ines: x: ; Bn Mika eee AL ; eo aT My maytty iy AN IN SITU TEMPERATURE SENSOR F, G. GEIL and J. H. THOMPSON Westinghouse Electric Corporation Research and Development Center Pittsburgh, Pennsylvania ABSTRACT Discussed is a unit for temperature measure- ment at a remote point (accurate to +0.02°C) with a high Q mechanical resonator whose fre- quency is temperature dependent. The resonator is novel in that a mechanical Q near 80, 000 is attainable at 30 Keps with an aluminum resonator whose support posts may be one-third the size of the resonator itself. The posts allow for rapid thermal stabilization. INTRODUCTION An in situ temperature sensor has been devel- oped for use in oceanographic measurements and is capable of measuring the temperature of the ocean accurately to to.02°c. This instrument would also be used in conjunction with an induc- tion conductivity meter which, it is hoped, will determine salinity to an accuracy of 0.02 parts per thousand. meter package will ultimately be capable of opera- tion to depths of 4,000 fathoms. PRESENT METHODS At present the most popular method for mea- suring temperatures to this accuracy employs a thermistor in a Wheatstone bridge circuit, the thermistor being connected to the bridge by long wires. While the thermistor is capable of this accuracy, there are some shortcomings to its con- venient use, the most significant being the heating of the thermistor due to the current in the bridge circuit. A newer method is to use a thermistor in an RC network in a phase shift oscillator, but this requires an extremely stable eircuit or one that is calibrated often. DESCRIPTION OF DEVICE A more satisfactory way to measure tempera- ture employs the device shown in Fig. 1. Inside the brass housing is an aluminum resonator which acts as a high-Q resonant filter whose frequency is temperature dependent. The resonator and amplifier shown compromise an oscillator circuit, and the temperature is converted to a frequency which is transmitted through a pair of conductors and registered on a counter. A two wire cable 35 The temperature sensor-conductivity of any length connects the sensor to the counter, the DC power and temperature signal being super- imposed on the same line. This sensor is especially suited to unattended operation for long periods of time without losing calibration and the frequency can be easily used to modulate a transmitter if needed. The frequency for 25°C is about 36 Keps for the resonator shown, and the frequency change is approximately 10 cps for each degree Centigrade. The resonator housing is evacuated to eliminate any mechanical loading on the resonator because the accuracy of tempera- ture measurement is related to the mechanical Q of the resonator. A more detailed view of the temperature sensi- tive resonator is shown in Fig. 2. The disc between the two posts vibrates in a flexural mode with two nodal diameters, two piezoceramic ele- ments being fastened at the pickup and drive points. The motion is best depicted in the lower left view; in the top view it may be imagined as Occurring in and out of the paper, the top and bottom quadrants moving opposite to the left and right. The nodal diameters are stationary. The two piezoceramic elements are bonded to the disc with conducting epoxy. The elements are made of a lead zirconate titanate and the disc's resonant frequency is far below their natural resonance. Wires are attached to the elements with conducting epoxy using small phos- phor bronze springs for compliance as shown in Fig. 2. It is important to recognize that the dise responds to the longitudinal or left-right motion of the element and not to the thickness motion. The aluminum is connected to the ground potential of the oscillator and the circuit is thus completed to the bottom plates of the elements. Heat is conducted to or from the disc by way of one of the two posts in the resonator of Fig. 1. Ultimately, both posts will be exposed in order to reduce the temperature time constant. Large posts may be used without affecting the Q because the load presented to the disc by the post is reactive. The equal and opposite forces on the post cancel, leaving the post out of the resonant system. The response of this disc to a temperature step function results in a time con- stant of about one minute, or one minute to com- plete 63% of the frequency change. This time constant can be reduced by decreasing the ratio Fig. 1. of the mass of the disc to the area of the post. The practical minimum time constant is not deter- mined by the resonator, however, but by the fre- quency counter. A counter requires a minimum of 10 seconds to count a 36 Keps frequency to the required 6 places; it is probable that the temperature time constant of the resonator can be reduced to match this. MATHEMATICAL EXPRESSION The expression for a thin disc with no post and vibrating flexurally with two nodal diameters is _ 0.238 +t Y Ht Wann? e(1-02) (2) 36 Photo of temperature sensor prototype. where f is resonator frequency in cps, R the radius of the disc in em, t the thickness in cm, Y is Young's modulus in dynes/em@, Gis Poisson's ratio and e the mass density in em/cm3. The change in Young's modulus with temperature accounts for most of the frequency changes, Y changes about 0.04% for each degree Centigrade change. The frequency changes due to expansion alone are of an order of magnitude less. Varying the thickness to diameter ratio changes the disc frequency, as the expression indicates, but the mode of operation is lost as the thickness- diameter ratio approaches one. The expression becomes inaccurate as the dise becomes thicker and with the addition of posts. MOTION OF ELEMENT CONDUCTING EPOXY SMALL SPRING ELECTRODES BOTTOM ELECTRODE ATTACHED WITH CONDUCTING EDGE OF DISC POLARIZING DIRECTION Fig. 2. ACCURACY CONSIDERATIONS The accuracy with which the frequency of this type of resonator follows temperature changes depends upon the Q of the resonator, for the fre- quency must remain constant and be repeatable for any one temperature. The higher the Q, the less the frequency will be permitted to wander. An approximate calculation (Fig. 3) shows that if a maximum random phase variation of 45 degrees is assumed in the amplifier circuit, an accuracy of t0.02°C requires a resonator Q of 45,000. Most experimental resonators have had Q's of 45,000 and above when operating in a vacuum; in fact, one was as high as 80,000. A wide band amplifier in the oscillator cir- cuit with a resulting random phase shift much less than 45 degrees would permit a lower Q resonator to be used. However, this is not feasible due to the presence of harmonics of the described fundamental and many other modes that can be excited and hence must be suppressed, either by tuning the amplifier or filtering. Hither method will introduce some phase insta- bility. A typical spectrum is shown in Fig. }. This spectrum is shown from a slightly different type of disc, having only one smaller post, but the spectrum is typical. Most of these fre- quencies belong to other modes of vibration. PIEZO-CERAMIC ELEMENT NODAL DIAMETERS <——_ FORCES EXERTED ON Sih POST BY DISC Detailed schematic of mechanical resonator. Only the second line at about 60 Keps can be con- sidered a legitimate overtone to the first line with the characteristically high mechanical Q. The fundamental frequency is used in the oscil- lator circuit. The small circuit of Fig.1 con- tains two tuned circuits--a tank circuit tuned to the fundamental frequency and a trap at the first overtone. For laboratory measurements, however, a low pass, sharp cut-off filter is used, following an untuned amplifier (see Fig. 5). This method results in less change of phase with different oscillation frequencies and greater stability of phase at a given frequency. This is shown by the fact that if all the inductors in the filter circuit decrease their inductance by 20%, the resulting phase shift through the filter at a given frequency is only about 20 degrees. Not shown in Fig. 5 is a phase shift circuit following the filter which is necessary to obtain oscillations and will also be used in the final unit to check periodically the resonator Q. This will be done by introducing a known phase shift and noting the resulting frequency change. LABORATORY TESTS Laboratory tests have been conducted to check the repeatability of the temperature sensor CALCULATION OF NECESSARY RESONATOR Q RES = <> | Al IS ee | ce | aS mn P | | = ee ee eee eee ee INCREASING FREQUENCY—> UNIVERSAL RESONANCE CURVE Fig. 3. measurements because this repeatability or fre- quency-temperature stability determines directly the accuracy. To test this repeatability a special double resonator, shown in Fig. 6, has been fabricated with two built-in Veco 51Al ther- mistors. It is housed in a brass container and evacuated. This arrangement permits the greatest flexibility for measurements and comparisons between the four temperature measuring devices. The double resonator was checked for repeata- bility by submerging the entire housing in a water bath. The bath was alternately heated and cooled over a one degree temperature range near 25°C. The resonators were incorporated one at a time in an oscillator circuit consisting of a 60 db amplifier with automatic gain control and the previously mentioned sharp cut-off, low pass filter. The frequencies were monitored by a counter which read to 6 places. The best indication of repeatability is the plot of Fig. 7 which is one resonator's frequency against the other. Ideally the points should lie in a single straight 45° line. Actually, the points lie close to a line and the width of the cluster determines the repeatability. The points on this plot are numbered in the order of their 3 DB POINTS 38 PERMISS/BLE TEMR READING ERROR RANGE —— .04°C FREQUENCY- TEMPERATURE COEFFICIENT LO (CNCMES/ ME »» ALLOWABLE FREQUENCY DEVIATION ——.O4 X 1O=.4 CYCLES ASSUME AMPLIFIER PHASE VARIATION OF 45°: THEN .4 CYCLES ~ 4f WHERE Af = FREQ. RANGE BETWEEN 90° PHASE ANGLE CHANGE Af = .8 CYCLES f —— Re GOCORs Q= KF OO en 8 Q=45000 Calculation relating error, amplifier characteristics and resonator Q. being taken and they show a repeatability, and therefore a potential accuracy, of about O.01°C. This indication of repeatability must be verified by a plot of one of the sensors against a ther- mistor (Fig. 8) to indicate any frequency- temperature hysteresis in the aluminum or tempera- ture gradients due to insufficient time for tem- perature equalization. Hither effect would show as increasing temperature points lying con- sistently to one side of the line and decreasing points to the other. This plot shows no such error. The final plot (Fig. 9) is of one ther- mistor against the other to show the thermistor repeatability. Of course, these tests fail to indicate the long term repeatability that might be expected of the temperature sensor. Barring any loss of vacuum, the only change in calibration that can occur is perhaps through some aging of the alumi- num; the resonator's frequency is otherwise inde- pendent of the external circuit, changes in the ceramic elements or any environmental factor except temperature. DRIVE VOLTAGE T VOLTAGE OUTPU OUTPUT —-DB BELOW DRIVE VOLTAGE 4 S © MY % i} a rs = i = Q < ty 50 100 150 200 250 300 FREQUENCY - KC Fig. 4. Spectrum of similar resonator. MATERTAL CONSIDERATIONS Some factors that are important in order to produce a good resonator should be discussed. The material must have high heat conductivity. Aluminum has more than twice the conductivity of brass and four times that of steel. Copper and silver have about twice the conductivity of aluminum but are not hard enough to permit a high Q. Aluminum has another advantage in that its heat capacity is low due to its low density. Two aluminum alloys have been used, designated 6061-T6 and 7075-16. The latter is the harder and results in a higher mechanical Q. Also, two other materials have been used, Invar and Ni Span C. Invar does not expand with temperature changes; its applications make use of this fact. An Invar resonator has a positive temperature coefficient of 200 ppm instead of the usual negative one. Ni Span C is used in resonant magnetostrictive applications and its frequency is relatively stable with change in temperature. A high Q resonator can be made with this material for temperature independent applications. 39 350 PICKUP DRIVE 9 DB -20 -90 -30 FILTER OUTPUT -50 -60 FILTER -70 OSCILLATOR CIRCUIT BQ ey Se) ED eo) 200 FREQUENCY — KC RESPONSE OF FILTER Fig. 5. Basic oscillator system. Fig. 6. Experimental double resonator. ho PITAE — \ \ i] { 1 | ‘ ‘ l 1 1 | | | | | | | i | | | ! | | | | Seal \ a} | 8 8 S i} all | i St | 9 8 S t | iS 1 < an & 3 3 | g n ed ee Oo H™ S&S \ = Sy \ 5 ° 3 a =e Sf 1 | ae N| Je x 8 VA is i W 9 i ' x 0 8 g ie oO Sl RE 1 N 1 | Mot 8 | i} No a} i ! wy Rt j 5] 1 i + + + + 2 ees Se + == ++ 9S | 3670 67/2 2367/4 2616 367/8 —_6720| g 26°C 25°C © SENSOR S, Fig. 7. Short term repeatability of sensors. 8500 8600 8700 8800 8900 THERMISTOR 7, Fig. 9. Short term repeatability of thermistors. 8600 8700 8600 THERMISTOR 7, Fig. 8. Frequency-temperature hysteresis and thermal gradient test. ya ‘ if ¢ ‘ j r *) ‘ F ! u, \ ; j oe y : \ SA GR aT mE, PNET ‘aoe a ' % i | . hae , wok ' Wy, { i t if ; ‘ P| \ a. ji 1 ré " { DAL) } Ve hi acd i Li ' node i rar estes |, i i, eran = : 1:0 ; * > = a Fl ternary MS he yore } ; i) 1) Fn My sian ¥ ‘sna, my A j : - * wT tule: ‘ ; a cme oe WF» avg i by hy WE oe aaSH / urereuas aed FO —- Pinder eas oo alg A RAPID RESPONSE HIGH ACCURACY THERMAL PROBE H. M. HOOVER Airpax Electronics, Inc. Fort Lauderdale, Florida INTRODUCTION The temperature measuring system described here was evolved in an effort to furnish the oceanographer with an accurate, rapid response measuring system, approaching the precision of the reversing thermometer but being a continuous reading device and one not requiring manipulation by highly skilled personnel. Realized is an arrangement (Fig. 1) with a raw accuracy of +0.05°C over a 25°C span which may be corrected to t0.02°C. The system will maintain this accu- racy, when properly used, for periods of months and it is not affected by normal handling. The time constant is better than one second in flowing water. PROBE SELECTION In approaching such a device, the basic sensor choice was almost immediately limited to resistance type elements by considerations of accuracy. Thermocouples and filled system ther- mometers are not sufficiently precise. The same is now true of relatively novel methods such as paramagnetic susceptibility or capacitance change. In the area of resistance thermometers, the thermistor is outstanding for its high out- put and rapid response. However, our concern over long term thermistor stability, even after aging, made us abandon the device and employ metallic sensors for the very high accuracies required. Here the stability and ease of repro- duction, inherent in platinum, dictated its choice in spite of its low output. Fortunately the linearity of platinum is quite good. Using platinum, stability and reproducibility of resis- tance change is several orders better than the accuracy required, and in fact, a thermometer not dissimilar in construction_defines temperature between -182°C and +603°c.1/@ The probe output follows the Callendar rela- tion of temperature to resistance Fig. 1. Probe assembly. n= 2(= 1) (a) (1) a \Ro 100 100 Superior numbers refer to similarly numbered references at the end of this paper. 43 (2) 1(R, 56 = Ws -@ == iL ————E———Ee (s- 1) (ye) where T is temperature in °C, Rp and Ro are resis tances at T°C and 0°C respectively, Ry09 and R, are resistances at 100° and melting point of sulfur respectively and T, is temperature of melting point of sulfur. Representative values for @ and 5 are 0.00392 and 1.49, respectively, giving a nonlinearity over a 25° span of approxi- mately 0.1%. (3) In order to obtain a reasonably dimensioned package, together with rapid thermal response, a small diameter tube was utilized for the element housing. This combines a relatively low mass with a short thermal path and high structural strength. In practice, a flexible mandrel is wound spirally with reference grade platinum resistance wire. Over this spiral, aluminum oxide or beryllium oxide beads are threaded, leads affixed, the whole slipped into a tube and the assembly then swaged down to final form. Total tube diameter approximates 0.060 inch and in one configuration the outer tube wall is 0.006 inch thick. TIME CONSTANT The thermal response of a probe can be calcu- lated as the algebraic sum of the separate time constants. The thermal situation is analogous to an electrical network of resistance and capaci tance where the time constant, T, is equal to RC (Fig. 2). For a cylinder of unit length, 1? ila = ic) 1 = 27K ) C = 7PCp (x? =) (5) eCp ~2)( 2 2) NOS Ge Core rm - Ty (6) The values of constants in the above equations for the particular probe under consideration are given in Table I. yy (GQ) THERMAL CONFIGURATION (b) ELECTRICAL EQUIVALENT (e; ———©) Fig. 2. Schematic diagram of (a) thermal con- figuration and (b) electrical equivalent for calculation of thermal response. Table I. Thermal probe constants. Stainless Aluminum Parameter Steel Oxide Yo is outside radius of eylinder in feet 0.00233 0.00175 Y; is inside radius of eylinder in feet 0.00175 0.00092 © is density #/£t3 198. 230. Cp is specific heat Btu/#/OF 0.175 0.125 K is conductivity Btu/hr/ft/°F 9.5 1.0 In the probe, which consists essentially of a metal cylinder surrounding a ceramic cylinder with the element beneath, the time constant of the sheath equals RC, =( £498) (0.125) ) (;.0-00833 ) 0.002332 (9.5) (2) 0.00175 : 0.001752 ) 0.322 x 107? hrs 0.0166 sec o—- —o Fig. 3. Complete probe equivalent circuit. and the time constant of the insulation equals (230) (0.175) ) (3 eae) 0. noLOoU)) (c. soars? 0.00092 = 0.000922 ) = 2.87 x 107? hrs = 0.104 sec. The volume of the 0.0017-inch diameter platinum wire is small and is neglected. Solving for the series circuit shown in Fig. 3 yields iS x. GL a2? Dae Pa (7) 7 148 (C]R]+CoRo+CoR] )+8°(CoRoC7R} ) Letting T, = 1, as a convention, and S as the transfer function equal to the negative recipro- eal of RC or the reciprocal of the time constant, T, gives T+(C,Ry+CoRotCoRy )T+(CyR)(CoRo) = 0. (8) Substituting C,R, = 0.012, CoRo = 0.104 and CoR] = 0.005 gives a characteristic time con- stant polynomial T2 + 0.121T + 0.001248 = which yields T = 0.1118 second. This varies from the experimentally determined value by a factor of 3 which is presumably due to boundary layer effects. No effort has been made to solve with the complex physical shapes involved. PRESSURE EFFECT The pressure resistance of the probe may be calculated using Lane's empirical approximation for the crushing strength of long tubes 45 Fig. 4. 2 P= 60,000 x| # -( £) | D\5 Bridge circuit. (9) where t is tube wall thickness in inches, D is tube diameter in inches and P is crushing pres- sure in psi. The constant in the foregoing gives a result reasonably valid for material with a yield strength of approximately 40,000 psi. The configuration described withstands about 8,000 psi. It should be noted that some care is necessary in manufacture to avoid making the element suscep- tible to changing pressure as in a strain gauge. In units tested, strain effect was not measurable (less than 0.001 C) at an external pressure of 1,000 psi. THE BRIDGE To avoid unpredictable sources of error due to resistance in conductors and from resistance changes in similar but not entirely equal lead conductors, the measuring bridge (Fig. 4) is installed in the bulb head where it is protected by the relatively uniform water environment from the effect of gross thermal change. Since the bridge is unbalanced a zener voltage regulator is also mounted at the probe. The unbalanced bridge has an output computed as follows from Thevenin's principle: E(AS- Eo ~ Rp (A+B+S+X)+(A+B) (X+S) (10) where E, is open circuit voltage across detector, E is source voltage, A, B and S are fixed bridge resistors, X is the platinum resistor and Rp is the source impedance. The current, Ig, passing through is E Ig= OSE (11) Gl Ry ae ha where Rp is the bridge resistance seen by the detector and Rg the detector impedance. An addi- tional nonlinearity accrues from the unbalanced bridge which is minimized by keeping the per- centage of resistance change in the bridge rela- tively low and the amplifier impedance high. Bridge output changes proportionally in the unbalanced bridge with bridge voltage supply. However, the load is relatively fixed and tem- perature changes are small so there is little difficulty in zener regulating the bridge supply to 0.015%. Manganin bridge resistors are used, suitably aged. Care must be exercised in avoiding thermocouples at junctions. (Manganin has an output of 3 microvolts/°C with respect to copper.) The bridge is mounted in a cylindrical pres- sure housing approximately 1.75 inches in diameter by 3 inches long, made of steel or high strength bronze. One end houses a conventional }-conduc- tor underwater electrical connector. Various mounting arrangements are furnished for attaching the probes to cables or structures. Sealing is accomplished with "0" rings. A mechanical shield for the sensitive element is insulated from the housing. FOULING Where thermal probes are to be immersed for long periods in relatively shallow water, marine fouling of the element will increase the time response sharply. As an example, a coat of paint on an element of the type described delays response time about 25%. We have noted weed for- mations several inches long and barnacles 3/8 inch in diameter after 3 weeks of unprotected immersion in tropical waters. The element shield therefore is made of nonfouling sintered material impregnated with anti-fouling agents. We find that this arrangement, with the relatively high and constant toxic dispersion mechanism, holds an area several inches adjacent to it clear of marine life except for some of the bacterial slimes. These vary in thickness and tenacity and the anti- fouling agents have some effect on them but at this time we have no definitive information as to their thermal conductivity. It is believed that slime acts as an additional boundary layer. SIGNAL CONDITIONING To convert the relatively low output of the probe to a signal suitable for feeding a data system a multiple stage second harmonic magnetic amplifier (Fig. 5) has been designed. This has the advantage of being hermetically sealed and not subject to maintenance so that its precision 46 Fig. 5. Amplifier - power supply assemblies. is not a matter of careful adjustment. These second harmonic amplifiers have a null shift into the bridge impedance of the order of 10 micro- volts referred to the input over a 40° to 10°F temperature span and a +10% supply voltage varia- tion. A feedback factor of several hundred sta- bilizes the gain and uae the output, usually O to 5 volts DC. The same amplifier housing furnishes an individual DC bridge supply, an on-off switch and circuit protection. A SPECIAL APPLICATION An interesting application of the probe sug- gests itself for dynamic height measurements. If one probe uses the pressure protective sheath and another a compliant sheath, we have the equiv- alent of protected and unprotected reversing ther- mometers of the same or perhaps improved accuracy. It is practical, in this instance, to use a 2- eonductor cable carrying DC power to the assembly and 2 frequencies proportional to temperature to the surface. All of the hardware is available and could be packaged in a cavity about 2 inches in diameter and 4 inches long. The addition of an in situ salinometer to this package would take over some Nansen bottle functions and the com- bination could make rather rapid deep stations possible. REFERENCES ats MEYERS, C H., Coiled filament resistance thermometers, Bureau of Standards J. Res., 9, 1932. STIMSON, H. F., The international tempera- ture scale of 1948, Bureau of Standards J. Res., 42(209), 1949. 7 MARKS, L. E., Editor, Mechanical Engineers Handbook, 4th Edition, McGraw-Hill Book Co., New York, N. Y., 1-2274, 1941. MAZZEO, B. A., A low level high accuracy DC magnetic amplifier, Electrical Manufacturing, November 1958. ‘7 A MOBILE INSTRUMENT FOR STUDY OF OCEAN TEMPERATURE IN THE THERMOCLINE REGION A. A. HUDIMAC, J. R. OLSON and D. F. BRUMLEY U. S. Navy Electronics Laboratory San Diego, California INTRODUCTION A heavy fluid with a free surface, having a density variation only in the vertical direction (for the undisturbed medium), is subject to two basically different types of gravity wave excita- tion: (1) surface waves and (2) internal waves. Internal waves are characterized by having, for every mode, the maximum vertical particle dis- placement below the free surface. For many analytical purposes most of the ocean for most of the year can be taken to consist of three strata. The first stratum is the surface layer consisting of well mixed, nearly isothermal water. The second part is the lower, more dense water in which the temperature and density vary slowly. Separating these two parts of the ocean is the (seasonal) thermocline region in which the vertical temperature gradient is relatively darge. Since the horizontal salinity gradient is very slight in this region, isothermal sur- faces nearly coincide with isopyenal surfaces. Thus, an obvious way to observe internal waves is to measure the vertical temperature structure in the vicinity of the thermocline. A vertical string of thermistor temperature sensors for detection of the passage of internal waves was first used by Ufford;1,2 such arrays are commonly used today. This technique is widely applicable in shallow water and can even be used in deep water when attached to deep-sea mooring devices. When it is necessary to survey an area, an instrument on a moving platform is desirable. One such scheme is a towed chain designed for measurement of thermal microstruc- ture in the upper 500 to 1,000 feet.3 To mea- sure the vertical displacement of isotherms in the thermocline where steep temperature gradients are encountered, fairly close spacing of sensor elements, precisely located, is required. Close, precise spacing can be achieved with the moving strut internal wave recorder described herein. APPARATUS The vertical sensing array consisted of normalized thermistors spaced equally in the leading edge of a vertical strut mounted on the bow of a submarine. The first strut (Fig. 1) Fig. 1. Low drag strut array mounted on USS BAYA. was 18 feet long and made of wood with a stream- lined cross-section. Its base was mounted in a steel boot and the main member was held in place by a number of smaller struts. The temperature sensors were placed 2.5 feet apart. The shoulder of the sleeve of the sensor element was flush with the leading edge and the thermistor bead extended about 1/8 inch into the fluid stream. The strut current in use (Fig. 2) is made of three 10-foot sections of a commercial antenna tower. Each section is triangular in cross- section. The tower is guyed laterally as well as fore and aft. Operation is satisfactory up to Superior numbers refer to similarly numbered references at the end of this paper. Fig. 2. 5 knots and guy wire vibration is excessive at 6 knots. There are 34 thermistor beads in the vertical array. Of these, 26 are spaced at 1-foot intervals and used with a digital data system. Five are used as in the original array except that they are spaced at 6-foot intervals. The remaining 3 beads all have fast thermal response and are used for continuous temperature measurements. Two of these beads are located near the middle of the tower while the third is located at the top. In addition there is a fixed resistor which is mounted at the foot of the tower. It is clear that the motion of the platform through the water causes an essentially time- independent distortion of the streamlines about the bow. This leads to a distortion of the time- dependent vertical temperature structure and the shape of the internal wave is modified. Since the distortion is most severe near the hull, the lowest beads are mounted some distance above it 50 Tower array mounted on USS SEA DEVIL. and data from the upper part of the strut are preferred. TEMPERATURE INDICATING SYSTEM Thermistors were chosen as temperature sensor elements because of their high sensitivity and simplicity. Thermistors have a large negative temperature coefficient of resistance and the resistance is an exponential function of tempera- ture. Commercially available thermistors do not have the same resistance vs. temperature charac- teristics (typical variations are +20% in nominal resistance at 25°C) and hence are not directly interchangeable. Since the feature of inter- changeability in sensor units is very desirable, a method was worked out for determining the optimum values of 2 resistors, one in series and one in parallel with the thermistor, that would successfully match the sensor outputs. The ther- mistors used were Veco 32Al1 (Victory Engineering RESISTORS Fig. 3. Corporation) which is the bead type with a glass probe covering. They are 2 inches in length, have adjacent leads and have a nominal resistance of 2,000 ohms at 25°C. The construction of the thermistor assembly is shown in Fig. 3. The thermal time constant of the thermistor thus encapsulated was approximately 0.5 second. Since the temperature readings were to be taken at intervals of from 10 seconds to 1 minute, "aliasing" of the temperature signal could occur unless a thermal lag was introduced. For this reason, the thermal time constant was increased to about 16 seconds by applying several coats of an epoxy resin. PRESSURE INDICATING SYSTEM It was necessary to record depth variations concurrently with the temperature records for subsequent compensation during analysis. Since the depth desired was that of the temperature sensors, the ideal location of the pressure transducer inside the hull would be directly under the strut or tower containing the tempera- ture sensors. The transducer was simply con- nected to the nearest sea chest by means of a high pressure line within the hull. The pres- sure at the sea chest has a quasi-static com- ponent and a fluctuating component due to the surface wave action. The latter pressure drops off exponentially with distance from the free 51 LER A RR ROR RTE RE NN ARETE SERCH Ln Aten aarenmee tnt ak mg PLASTIC CYLINDER 7 EPOXY ei FILLING Mele GUARD FIBER SPACERS THERMAL LAG COAT ING Thermistor assembly. surface and in usual operation is attenuated below the system detection level. Near the sur- face these fluctuating components would have to be filtered out to prevent aliasing when the pressure reading is sampled at intervals equal to or longer than one-half the period of surface waves present. A Statham Laboratories, Inc. Model PA2O08TC unbonded strain gage pressure transducer was used. This was an absolute gage with a range of 0-500 psia and a combined nonlinearity and hysteresis of less than 0.75% of full scale. MONITOR AND RECORDING SYSTEMS The main purpose of this paper is to describe the transducers so the following discussion of recording techniques is limited to features that are pertinent to transducer design or operation. An independent and relatively simple visual moni- tor and data recording system (Fig. 4+) employs strip chart recorders to handle temperatures from 5 thermally lagged thermistors, a single pressure transducer and the relatively fast temperature fluctuations of one uncoated thermistor. All data analysis is done on a CDC 1604 (Control Data Corporation) digital computer. A block diagram of the digital data recording system designed to match the CDC 1604 is shown in Fig. 5. The bridges were built at USNEL; the associated digi- tal system was built according to our specifications by Electro-Instruments, Inc. An input scanner, with a maximum switching speed of 50 channels per second, selects sequentially 26 temperature out- puts, one pressure output, 3 calibration outputs and 2 time signals. The signals are amplified in a DC amplifier of adequate bandwidth. The analog to digital conversion is accomplished by means of a digital voltmeter with a maximum con- version rate of 1,000 readings per second. The time base for logging is a digital clock. The logging control allows for selection of scan eycles with intervals of from one second to 30 minutes (also continuous). Recording is by means of a printed tape recorder and/or a tape punch. With the current set of thermistors, the log- ging rate is normally set at 10 seconds. Im this mode both the tape punch and the printed tape recorder are used at all times in recording data. FIVE FIVE COATED TEMP THERMISTORS BRIDGE SIX CHANNEL M-H RECORDER PRESSURE TRANS-— DUCER DEPTH BRIDGE SINGLE ONE ONE UNCOATED TEMP CIAL THERMISTOR BRNDGE M-H RECORDER Fig. 4. Visual monitor and recording system. By cutting the printed recorder out of the system and using only the tape punch, the sampling rate may be increased from 3 to 4 channels per second to 9 to 10 channels per second. With a faster tape punch the sampling rate may be increased to approximately 17 channels per second and addi- tional changes can be made for further increases. OVERALL ACCURACY AND PRECISION The accuracy of the temperature measurement depends on the accuracy of the calibration, linearity of the bridges and the accuracy of the digital system. The maximum error is +0.005°C. The overall precision of the temperature measure- ments, using probable error as the criterion, is the square root of the sum of the squares of the precision of the calibration and of the precision of recording which has been evaluated at to.0032°C. The accuracy and precision of the depth indicating system are not well known but are probably a small fraction of a foot. REFERENCES 1. UFFORD, C. W., Internal waves in the ocean, Amer. Geophys. Union Trans., 28, 79-86, February SCNT. 2. UFFORD, C. W., Internal waves measured at three stations, Amer. Geophys. Union Trans., 28, 87-95, February 1947. 3. RICHARDSON, W. S. and C. J. HUBBARD, The con- touring temperature recorder, Deep-Sea Res., 6, 239-244, 1960. on | ao GO 2 je l= Oar 29 Os NPUT o2 TEMP. oul BRIDGES SCANNER Nr | = i} 26 fo o PRESSURE TRANS DUCER DEPTH BRIDGE A-D CONVERTER DIGITAL DIGITAL CLOCK OPERATION FEN, CONTACT VOLT METER BEN CONTA FOR STR IP. CHART RECORDERS PRINTED LOGGING WARE CONTROL RECORDER OUTPUT CONTROL Fig. 5. Digital data recording system. 52 TOWED SEA TEMPERATURE STRUCTURE PROFILER E. C. LaFOND U. S. Navy Electronics Laboratory San Diego, California ABSTRACT Vertical strings of temperature transducers have been deployed by the U. S Navy Electronics Laboratory from fixed platforms buoyed from floats or the sea floor and towed horizontally. The technique of towing from surface ships has proved most valuable in the study of internal waves, fronts and other thermal structures in the program to achieve vertical profiles of the sea. The USNEL thermistor chain is a string of 34 temperature sensors that operates in essentially a subsurface vertical line as the ship moves on the surface. The signals from the sensors are seanned electronically and interpolated for all whole degree Centigrade isotherms, which are printed on a continuous chart. In addition to the analog presentation of 2 dimensional tempera- ture structure (depth and distance) the water temperatures at the 34 levels are planned to be punched on computer tape for later machine analysis. The depth of the array is monitored by means of a pressure transducer at its sub- merged end. The thermistor chain has been used for a year and a half in the Pacific Ocean from Canada to Central Mexico and as far afield as Honolulu. The main difficulty experienced, now eliminated, was lack of watertight integrity in the underwater temperature sensors, electrical harness and cable connectors. The advantages and disadvantages of the equip- ment for oceanographic research as well as some of the results are discussed. INTRODUCTION The USNEL toyed sea temperature structure pro- filer is essentially a string of thermal sensors held suspended in a near vertical attitude in the sea while the ship is moving forward. The sensing elements are located at selected inter- vals along the faired link chain from the sur- face to a depth of 800 feet and cause associated electronic recording equipment to provide iso- therm profiles that are recorded with reference to depth and distance while the ship cruises ahead .+ The USNEL thermistor chain thus makes feasible a worldwide acquisition of data on the vertical temperature structure of sea water since the ocean- ographic research vessel USS MARYSVILLE (Fig. 1), which deploys the chain, is capable of sailing anywhere in the oceans. The hardware, consisting of hoist and chain, was designed and manufactured by the Commercial Engineering Corporation of Houston, Texas. The contouring temperature recorder was manufactured by the Scientific Services Laboratory, Inc. of Dallas, Texas, and is based on a design by Dr. W. S. Richardson of the Woods Hole Oceano- graphic Institution.© Three previous units have been constructed. The thermistor beads were pro- duced by S. P. Fenwal Electronics, Inc. of Framing- ham, Massachusetts, from specifications by the Scientific Services Laboratory, who encapsulated the beads. The harness (underwater electrical leads) were manufactured by Spectra Strip Wire and Cable Corporation of Garden Grove, California. DECK AND SEA UNITS Oceanographic Chain Hoist The oceanographic chain hoist3 is a self- powered winch designed for oceanographic work requiring measurements to depths as great as 840 feet when the ship is underway at very slow speed and to depths of approximately 540 feet at a speed of 13 knots. The hoist that raises and lowers the chain is powered by a diesel motor and controlled by a hydraulic drive, all on a single foundation measuring 11.5 by 18 feet. The drum is 7 feet in diameter and its total height with chain is 9.5 feet from the deck. The drum stores 900 feet of articulated chain. The 900-foot chain, the drum on which it is wound and the hoist weighs a total of 37,500 pounds. The towing device connecting the surface vessel to the weight, or "fish," is a special articulated tow chain (Fig. 2) which is composed of fairly flat streamlined links about a foot long. 9 inches wide and 1 inch thick. Between the two steel fairing cheeks on each link is a channel to house the electric cable Superior numbers refer to similarly numbered references at the end of this paper. 93 Fig. 1. USNEL sea temperature profiler (thermistor chain) hoist on USS MARYSVILIE. LOCATION FOR THERMISTOR BEAD Fig. 2. Chain links, harness and fairings. Fig. 3. Encapsulated thermistor bead: (a) thermistor bead, (b) phenolic resin, (c) silicone rubber, @)) “or ring seal, (e) silicone rubber cap and (f) leads. oh harness. The plastic fairings that form the trailing edge of each link serve to reduce drag and hold the electrical conductors in position. When the harness is inserted in the channel, the plastic fairings are pushed into place between the cheek plates and are held in position by springs which snap into holes in the cheek plates. To prevent tension in the harness when the chain is bending sharply over the towing sheave, the harness is looped into two rounded depressions in the fairings. The 34 fairings that hold tempera- ture sensors have a drilled hole for insertion of the thermistor bead and a drilled diagonal channel that permits water to flow over the bead as the chain is towed through the water. At the lower end of the chain a streamlined, 2,300 pound weight holds the chain in a nearly vertical posi- tion while it is being towed. Beads and Connectors The thermal sensing beads used are type GB 32 P168 thermistors whose resistance varies as a function of temperature, yielding ambient tempera- ture measurements in the form of electrical cur- rent amplitude. The beads are carefully matched at a fixed temperature to an accuracy of 1 part in 2,000, equivalent to 0.02°C, at the matching temperature. Each bead and its two connecting leads are encased in silicone rubber (Fig. 3). A B | = a sl ome) Fig. 4. A watertight connection between the leads and the harness is important because this is the most probable location of water leakage. The connectors, consisting of brass plugs crimped on the end of each lead and then soldered, are inserted in special neoprene connecting plugs that fit together without air space as the brass connections are made. A watertight neoprene sleeve slides over the wire leads and the con- nected plug. Any defective thermistor can easily be replaced by unplugging the unit and inserting a new one. The leads extending from the encapsulated beads (F, Fig. 3) developed leaks after prolonged use. The addition of a silicone rubber cap (E, Fig. 3) provided a satisfactory seal. Another important improvement was made in the connectors which now consist of tightfitting neoprene sleeves so constructed that there is no air space to permit leakage (Fig. 4). These con- nectors, manufactured by Electro-Oceanics, have proved watertight through long immersion. Harness The present harness, manufactured by the Spectra Strip Wire and Cable Corporation is made of No. 22 19-strand copper wire. It is first covered with a 0.008-inch layer of polyethylene to prevent water from getting between copper and insulation. Then a 0.008-inch layer of poly- vinyl chloride is applied to reduce abrasion in the links. Wine color-coded wires are cemented together with polyvinyl to form a ribbon. Thir- teen of these ribbons make up the harness, a total of 117 leads, which is preformed in a zig- zag fashion, taped at intervals and laid in the channel of the chain (Fig. 5). All 117 leads do not go the full length of the chain but are sealed off at different distances along the har- ness so that only 27 leads reach the chain end. The 34 thermistor beads use 2 leads each and 3 leads service the pressure element. The ther- mistor chain thus requires a total of 71 leads, leaving a number of spares. The inner leads of the harness are normally used as conductors since they are less susceptible to chafing and leaks. oy) Detachable connector from bead to harness: (a) leads, (b) neoprene cap, (c) neoprene body, (d) gold-plate brass male connector, (e) gold-plate brass female connector, (f) neoprene body, (g) neoprene cap and (h) lead to harness. An initial difficulty was the lack of water- tight integrity in the electrical harness. Points of vibration and friction in the bundle of wires. connections of the thermistors to leads, connec- tion of leads to the detachable plugs and the plugs themselves all proved susceptible to leakage. Connections at the greatest depth were the most prone to leak. The first attempt to prevent leakage utilized a heavy rubber jacket over leads and vulcanized connectors. This proved too stiff to spread properly as the chain passed over the sheave on the fantail and to loop back into the channels in the links when the chain hung vertically. ELECTRONIC COMPONENTS Contouring Temperature Recorder The contouring temperature mecoracrs built by the Scientific Services Laboratories,* is the heart of the unit, taking information from the thermistors in the chain towed behind the ship and plotting the vertical distribution of tempera- ture as a continuous record (Fig. 6). Each iso- therm (line of constant temperature) is displayed as a depth profile similar to that made by a depth recorder and the complete record gives a 2 dimensional representation of the thermal struc- ture of the sea. Since the thermistors are scanned from top to bottom the vertical scale of the record represents the length of the chain in the water. The thermistors are placed at even intervals on the chain. Usually 34 measuring thermistors, spaced at 2/-foot intervals, are programmed. As these are towed through the water the recorder interpolates and prints on the paper roll the con- tours of the various isothermal surfaces where the horizontal scale represents either time or dis- tance and the vertical scale depth. Since the ship speed at 6 knots is believed to be several times faster than internal waves, the effects depicted are primarily spatial. Normally tempera- ture increases toward the surface. However, in case of temperature inversions, where warm water underlies cold, the positive temperature gradient area may be shaded on the record for identification. Fig. 5. Preformed harness: The lower quarter of the chart (Fig. 6) is marked every 6 minutes with a vertical row of 16 dots by means of timing pips. For tempera- ture readings the dots represent temperatures at @-degree intervals from 0° to 30°C. A con- tinuous line marks the temperature of the selected sensor. When the temperature of the bead is known the upper isotherm on the record and all other isotherms are easily identified. The depth of the pressure transducer near the end of the thermistor chain varies with ship speed and with subsurface currents. For recording this depth on the lower part of the chart the 16 vertical dots represent depth at 60-foot inter- vals from the surface. There is no possibility of confusing the depth record with the tempera- ture record for the values are simulataneously shown on 2 of the dial indicators at the base of the instrument (Fig. 6). Moreover, the depth reading is likely to decrease as the speed of the ship increases. A third dial indicator at the base of the instrument is connected with the scanning mechanism which prints temperature con- tours on the upper part of the chart and shows the temperature currently being plotted. The recorder uses the helical-drum-and-blade principle and writes on electrosensitive paper such as Westrex Timemark No. 118 or Westrex No. 44. It can accommodate rolls of paper 19 inches wide and }00 feet long. The paper speed may be varied from 2 to 12 inches per hour but 56 (a) preformed bead and (b) lead for bead. the most satisfactory speed for presenting internal waves is approximately 6 inches per hour. At this speed a 400-foot roll should last approx- imately 33 days. Several difficulties were experienced with the temperature contours due to shorts or open cir- cuits in the harness and connecting leads. Other troubles in the recorder were caused by skewed paper feed, burning of paper, shorts in the cir- cuit, smudging of records and improper scanning of the input resulting in flat isotherms. Most of the difficulties were eliminated by proper tuning, adjustments and replacement of parts. One improvement was in the change from Westrex No. 44 paper to Westrex Timemark No. 118 paper which gave a better quality of trace and reduced handling and recording smudges. Malfunc- tions still occur but sufficient spare parts now insure a successful cruise. Recorder Schematics The thermistors are arranged in an array. Each thermistor forms one arm of a voltage divider similar to those used at USNEL on other tempera- ture measuring devices. The voltage at the junction of any thermistor with its load resistor is the analog function of the temperature of that thermistor. Several such thermistors equally spaced on the chain produce analogs of temperature F G H Contouring temperature recorder: (a) isotherms, (b) surface temperature plots, (c) depth of end of chain plot, (a) patch panel for incoming leads, (e) recorder paper case, (f) scan temperature dial, (g) station tempera- ture dial and (h) transducer depth dial. aleeor at those stations. A block diagram of the pro- filer system is shown in Fig. 7. A commutating switch connects these stations sequentially to the interpolating potentiometer which assumes a linear temperature gradient between any two successive stations. As the wiper of this interpolating potentiometer travels along its resistance path the wiper picks off a uniformly graduated voltage between successive thermistors. Since the water temperature at the thermistors will not fall at whole degrees Centi- grade (nor even at whole tenths of degrees), it is necessary for the potentiometer to interpolate aT between degrees and fractions of degrees to obtain whole degrees Centigrade or tenths of degrees. The interpolated voltage produces a continuous gradient that can be followed by an amplifier and servo-mechanical components. This scan voltage is fed into a scan servo-amplifier. The servo-amplifier and its associated follow- up mechanism comprises a DC amplifier loop. The feedback from the potentiometer on the servo- mechanism is a nonlinear function adjustable to the nonlinearity in the thermistor output voltage. The nonlinearities cancel, producing an output shaft rotation that is linear with reference to temperature variations. Since the output of this servo-mechanism is linear with temperature, it is only necessary to provide a pick-off suitable for printing on a record, such as slotted or digitizing drum driven by a scan servomotor through gearing so arranged that one turn of the motor equals 1°C. The drum is slotted for 1/10° and 1/20°C marks. These slots allow light to actuate photoelectric cell pick-offs whose outputs operate the print coder, print amplifier and recorder at the moment of passing through a given temperature. As the linear interpolation progresses down the record temperature marks appear at the same depth as long as the temperature at the thermistor beads remains constant. The measuring circuits cover a range from -20°C to 32°C and 1° isotherms can be plotted for this entire range. When gradients allow finer plotting, 0.1° and 0.05° isotherms can also be recorded. While the rate of scan can be varied, 12 seconds is the optimum setting for the overall eyclic process of taking temperature information from the nearly vertical column of water, the surface bead and the pressure sensor. At this setting the isotherms are contoured in 8 seconds, the remaining 4 seconds being used for plotting depth and surface. Depth Sensing Element The depth sensing element is a Bourdon tube driving a potentiometer. The DC output signal from this potentiometer is balanced against the DC feedback from the feedback potentiometer in the depth servo-mechanism. The balanced signal is then amplified by a conventional 400-cycle servo-amplifier. The depth of the pressure transducer is recorded on the lower portion of the paper record which is interrupted every 6 minutes while the scale marks are printed. This interruption dis- tinguishes the depth record from the temperature line. The vertical scale marks represent depth at 60-foot intervals from the surface and they are 6 minutes apart horizontally. D9 COMMUTATING | INTERPOLATING SCAN SERVO SWITCH POTENTIOMETER AMPLIFIER § & LON BEADS PRINT , DIGITAL AMPLIFIER DRUM FEED BACK POTENTIOMETER Fig. 7- Schematic diagram of profiler. Fig. 8. Profile of temperature structure. TYPICAL TEMPERATURE STRUCTURE PROFILE each printed line represents whole degree iso- therms. The vertical scale represents depth and An example of vertical temperature structure descends from the surface to 800 feet. The hori- of sea water recorded with a thermistor chain zontal scale represents both time and distance; off the coast of Mexico is shown in Fig. 8. Here in this example either 24 miles or 4 hours. Normally this is taken as a distance plot. 58 The example indicates the nature of vertical oscillations in the thermal structure which are commonly called internal waves. It is apparent that the cycles in the vertical displacement of the isotherms contain a wide spectrum of wave- lengths. The longer and higher waves are found where the vertical gradients are weaker, in this case near the surface and at the greater depths of 600 to 800 feet. In the main thermocline where the isotherms are more compact, around 200 to 300 feet, wavelengths of 0.2 to 0.5 mile are common but not uniform. The heights of the larger waves are about 20 feet in the thermocline, whereas at the other depths they may be several hundred feet. Most of these small waves on different iso- therms in the thermocline are in phase with each other; however, this is not necessarily true of the larger waves above and below the thermocline. In fact, around the 6 mile range the 14 degree and 18 degree isotherms are almost out of phase. Another feature is the temperature inversions that occur below the main thermocline in the 11 and 12 degree isotherms. This example clearly shows the nature of sea water temperature struc- ture recordings that can be obtained with the thermistor chain system. CONCLUSIONS The USNEL towed sea temperature structure pro- filer is a satisfactory device for measuring and recording the thermal structure of the ocean. This profiler has been very successful in record- ing, for the first time, the details of the thermal structure of the Pacific Ocean and the nature of fronts and internal waves which cannot be obtained by any other means at the present time. REFERENCES 1. lLaFOND, E. C., Two-dimensional oceanography, Bureau of Ships J., 10, 3-5, December 1961. 2. RICHARDSON, W. S. and C. J. HUBBARD, The contouring temperature recorder, Deep-Sea Res., 6, 239-2h4, 1960. 3. COMMERCIAL ENGINEERING CORPORATION, Manual for 900 foot oceanographic chain hoist no. Mk II-F chain, Publ. 24, 1961. UNPUBLISHED. 4. SCIENTIFIC SERVICE LABORATORIES, INC., Preliminary instruction and operation manual for the contouring temperature recorder - model 1, 30 June 1961. UNPUBLISHED. 5. IaFOND, BE. C., Slicks and temperature struc- ture in the sea, Rept. 937, U. S. Navy Electronics Lab., 2 November 1959. UNPUBLISHED. 29 ‘ Vie eee «Dy eR nar = ; eis 4 =! 1 ry « i” - J i "ef 4, t , : yw * } Prey i ¥ ¥ Ms arr Ur doneesali #4 RY Bee 2 wes i 4 + te * > Eee Daa) ( A 4 j . diy on pe a ru ihe be ye ‘ b bib feed ance dase 4 Se ad LU he c ti ; a , a) - i i ‘4 * ye ‘ t + i ¥ Ke ) fi 1 ‘meow a i v b : . i at 1 ppd A n a i » | i ‘ i yp i ’ ras! x ’ 5 \ i 1 / eae ‘ i fi a4 pA bi bu § 4 ; i Ly , : ’ “ vh \ fi A % ‘ ; j \ : i yee Ae in =f SS agement F ¥ i 4" t; re Bite x J { 6 a" oo ae ' ene BOLSA 0 ” nie set HT i) a ye ee bee) .* aihy om, h, Path bre AIRBORNE INSTRUMENT FOR PRECISION MEASUREMENT OF SEA SURFACE TEMPERATURE USING INFRARED RADIATION EMITTED BY THE SEA R. PELOQUIN U. S. Navy Oceanographic Office Suitland, Maryland M. WEISS Barnes Engineering Company Stamford, Connecticut INTRODUCTION The combination of a precision radiation col- lecting and detecting system with highly stable electronics has resulted in an infrared radio- meter capable of measuring absolutely the tempera- ture of the sea surface to an accuracy of 10.2°C. The instrument uses an in-line black body refer- ence radiation cavity temperature controlled to better than 40.05°C. Integral with the cavity are the detector, germanium optics and chopper system, comprising a highly stable optical unit independent of temperature. The electronics system is largely transistorized and uses pre- cision components to provide the overall stabil- ity to maintain the 0.2 C performance. Operation of the instrument has been reduced to the essential steps, requiring only selection of temperature range and selection of the mode of operation. No other operating adjustments are required. The output is a direct indication of sea surface temperature and it is presented in degrees Centigrade on a meter and on a strip chart recorder. A second highly compact instrument, known as the infrared thermometer, of a less complex design has also been developed. It is capable of a measuring accuracy of 0.5 to 1.0°C. Its output is displayed directly in degrees C on a panel meter. THEORY OF OPERATION Radiation Characteristics of the Ocean Surface Over different parts of the earth the tempera- ture of the ocean surface ranges from 4 minimum of -2°C (271°K) to a maximum of +35°C (308°K). Only a thin layer of water is required to absorb infrared radiation completely. In the infrared region of from 4 to 12.5 microns the emissivity of the ocean surface is 0.98 for radiation normal to the surface. The reflection for normal radia- tion is 2% and increases to 4% for radiation incident at 60° from the normal. The ocean surface has essentially the radia- tion characteristics of a black body at 300°K. 61 Such a surface emits maximum radiation at a wave- length of 9.6 microns. The radiation emitted for the extremes of ocean surface temperature are shown in Fig. 1 as radiation emission curves for black bodies at +35°C and -2°C. Examination of these curves shows that the major portions of both peaks are included in the region between 6 and 20 microns. Atmospheric Attenuation Since the instrument is to be used aboard air- eraft it is important that atmospheric attenua- tion effects be considered. Fig. 1 also shows a curve of the spectral attenuation through 1,000 feet of atmosphere in the 6 to 20 micron region. The curve shows that there is a good transmission window between 7.5 and 12.5 microns and that the infrared radiation outside of this window will be highly attenuated by the atmosphere. The effects of atmospheric attenuation, therefore, can be lJargely eliminated by use of a 7.5 to 12.5 micron band-pass optical filter. Approximately 29% of the radiation emitted by the surface of the ocean falls within this band-pass and would be available for detection. In this application an optical filter con- sisting of arsenic trisulfide glass and a thin coated slab of indium antimonide was used. The transmission characteristic is shown in Fig. 2. There is still some attenuation present within the pass band. It is possible to account for this effect and correct the temperature measure- ment by introducing a quantity know as "optical thickness" which is dependent on specific humidity, pressure and layer structure of the atmosphere. Another approach to eliminating atmospheric effects is to use a narrower window, 9.2 to 10.9 microns. Fig. 1 shows the relatively com- plete transparency of the atmosphere in this region. WATTS/CM2 MICRON INTO A HEMISPHERE Fig. 1. PERCENT ATTENUATION WAVELENGTH (Microns) Blackbody distribution (Planck Law) for -2°C and +35°C and atmospheric attenuation of 1,000 feet of air path at sea level. PERCENT TRANSMISSION WAVELENGTH IN MICRONS Fig. 2. Filter percent transmission (upper trace shows 100% transmission). 62 Energy Considerations The least amount of radiation available for detection is from the ocean at a water tempera- ture of -2°C (271°K). It is further required to detect a 0.2°C change at this temperature. The change in radiation emission for a small change in temperature can be found from the differen- tiated form of Stefan-Boltzmann Law Aw = 4oet3ar . (1) Thus at T = 271°K, AT = 0.2° C or 0.2K, AW = 8.5 x 107? w/em® (total into a hemisphere). Experience with detectors in previous appli- cations indicates that it is quite practical to detect radiation differentials of this order. In fact, in the spectral region of interest (7.5 to 12.5 microns) a thermistor bolometer has been found to be quite capable of detecting con- siderably smaller radiation differentials than that calculated. DESCRIPTION OF AIRBORNE RADIATION THERMOMETER General The airborne radiation thermometer equipment consists Of 3 basic parts: a radiometer optical head, an electronic processing system and an indicating-recording system. The radiometer optical head collects radiation from the sea sur- face and generates an electrical signal propor- tional to the difference between this radiation and the radiation from a precisely controlled internal black body reference source. This sig- nal is processed by the electronic circuits to produce a precise DC signal. This output is linearized and broken up into } overlapping temperature ranges, each spanning a 10 C interval with 1°C overlap. Finally the measured signal output is monitored on a panel meter while a continuously operating strip chart recorder simul- taneously produces an accurate and permanent record of the sea surface temperature data. As shown in Fig. 3 the airborne radiation thermometer is integrated into a single cabinet. Controls, panel meters and the recorder are arranged on the sloping control panel of the cabinet for maximum operating convenience. radiation collecting and detecting unit is located at the base of the cabinet behind the forward-bottom access plate and faces out through an opening in the bottom of the cabinet. Its line of sight is directed vertically downward and a shutter mechanism in the base of the cabi- net covers the entrance aperture when the radi- ometer is not in use. The Printed circuit and component boards, mounted primarily at the rear of the cabinet, contain the electronic circuitry. Access plates are provided and some panels are hinged so they 63 Model 14-320 airborne radiation thermometer. Fig. 3. may be serviced on either side without removing them from the cabinet. When the unit is in operation these panels are secured into position by spring lock fasteners. Special provisions have been made to keep a record of the selected temperature range while the instrument is in use and a marking system is incorporated into the recorder to mark the range setting of the instrument in code along the edge of the strip chart. Radiometer Optical Head A layout of the optical system is shown in Fig. 4. Radiation enters the system through a germanium doublet which focuses the energy onto a thermistor detector. The detector is mounted at the apex of a temperature controlled black body cavity operating at 50°¢ and assumes the temperature of the black body. A chopper blade and mask, each consisting of two opposed 90° sectors, are placed at the front end of the cavity. The chopper blade rotates, and as it rotates it either completely blocks the incoming radiation by closing the 90° sector openings in the mask or allows entering radiation to fall on the detector when the chopper blade sectors are aligned with the mask sectors. The inner surface of the chopper blade is gold plated and highly PHOTOTRANSISTOR REFERENCE (PRP) REFLECTIVE OPTICAL CHOPPER BLADE TIMING BELT TEMPERATURE CONTROLLED BLACK- BODY CAVITY GERMANIUM LENSES PRE AMPLIFIER FILTER BOLOMETER CHOPPER BLADE SHAPE jaa MASK Optical system layout. reflecting in the infrared. When the cavity is closed to incoming radiation the detector receives the cavity black body radiation as reflected by the gold plated surfaces. Thus, as the blade rotates, the detector alternately receives the target and the black body radiation. The detector consists of a thermistor bridge network which produces an output signal propor- tional to the difference between the incoming radiation and reference black body radiation. The radiation is chopped at a rate producing a 20 cps signal which enters a high gain, highly stable hybrid preamplifier which amplifies the signal approximately 2,000 times. A second chopper blade, external to the opti- cal system, is driven by the chopper blade motor through a timing belt to provide a square wave signal in phase with the detector signal. The signal is generated as the second chopper blade interrupts a light beam directed on a photo- transistor. The reference generator is called a "phototransistor reference pickup" (PRP). The PRP signal is used later on in the electronic circuits for synchronous rectification of the signal generated by the thermistor detector. An electronic temperature offsetting signal (E.T.0.) for zero suppression in range selection is also derived from the reference signal. The black body reference cavity is of the conical type and its temperature is precisely controlled and monitored with thermistor beads embedded in the cavity and used as resistance 64, thermometers. The control circuit for the cavity is designed to maintain its temperature at 50°C t0.02°C. Considerable care has been taken with radi- ometer design to provide a thermally-stabilized optical structure. This has been accomplished by placing the detector and signal chopping elements within an enclosure consisting of the cavity and germanium lenses in a compact in-line thermal structure, thereby achieving, as far as possible, an isothermal entity. This is done to minimize the effects on the system of thermal variations dueto changes in lens emission and chopper mask emission as well as changing lens transparency. Changes in detector responsivity also occur with varying detector temperature. A point of particular interest is that in order to achieve the in-line structure the chopper blade drive shaft actually passes through the germanium lenses through an on-axis hole ground in each lens. Electronic Processing System Details of the electronic processing system are shown on the right hand half of Fig. 5. The output of the preamplifier is a 20 cps signal composed of the thermistor generated signal and an out-of-phase offsetting signal. This com- posite signal is fed to a synchronous rectifier where it is converted to a DC signal. Synchron- ous rectification, as used here, provides the equivalent of a very narrow band-pass circuit; hence, it affords a great reduction in noise signal while operating at low signal levels. As mentioned earlier, the signal used for both offsetting and as a reference for the synchronous rectifier is obtained from a second chopper in the radiometer structure. Linearizing and range determination occur automatically during range selection by simultaneously switching in the proper amount of E.T.0. and appropriately changing the system gain. The resultant output signal is fed to the panel meter through a meter amplifier and to the strip chart recorder. Since black body radiation normally varies as the fourth power of its temperature, linearizing is necessary if the total temperature range is to be presented on a linear scale. The linearizing, and range determination and selection, are accom- plished in the following way. Referring to Fig. 6 the output of the synchronous rectifier is fed to an adjustable attenuator controlled by the range selector switch. A second deck on the range switch receives the E.T.0O. signal and feeds a selected portion of it to the preamplifier. The E.T.0. signal reduces the output signal to zero at the beginning of each range (-2, +7, +16, +25) and each section of the gain attenuator reduces the slope of the output function from that of the fourth power curve to that of the linearized curve, providing a signal to the WINDING RESISTANCE THERMOMETER HEATER THERMISTER BOLOMETER SYNC SIGNAL PHOTOTRANSISTOR RADIOMETER) HEAD REF PICKUP | Fig. 5. RANGE GAIN ATTENUATOR} SEA WATER SURFACE TEMP METER RANGE SWITCH RECORDER ETO. SIGNAL Fig. 6. lLinearizing and range changing block diagram. recorder such that a 10°C change in sea surface temperature produces a full span deflection of the recorder. As indicated in Fig. 5, the temperature con- troller precisely maintains the reference cavity at 50°C to.02°C. The cavity temperature is sensed by a thermistor bead embedded in the eavity and used as one arm of a Wheatstone bridge. The error signal from the bridge con- trols a thyratron which proportions the current through the heating element of the cavity. A second thermistor bead, located within the cavity, is used to monitor the cavity temperature. Changes in resistance of this bead are monitored by the reference bridge circuit. The bridge output is displayed on the reference temperature 65 RECTIFIER AMPLIFIER OFFSET SIGNAL SOURCE ELECTRONICS & RECORDER REFERENCE TEMP BRIDGE TEMPERATURE CONTROLLER REFERENCE ELECTRONIC REGULATED SEMBEBACURE BS POWER METER SUPPLY. SUPPLY SYNC. RANGE GAIN ATENUATOR SEA WATER SURFACE TEMPERATURE METER METER RECTIFIER AMPLIFIER RANGE SWITCH RECORDER System block diagram. meter which is calibrated to read cavity tempera- ture directly in degrees Centigrade. Thermistor operation requires a DC bias which is developed by the electronic bias supply. It originates in a 5 Keps transistorized oscillator. The output voltage is stepped up and then recti- fied to produce an output of 7300 volts. The DC output voltage is additionally filtered and highly stabilized using RC networks and zener diodes. The DC power ous circuits is required for operating the vari- obtained from a transistorized regulated power supply operating from the 117 volt 60 cps line. The supply also provides a highly regulated output at 26.5 volts. In an effort to produce a precision instrument of high reliability, considerable attention has been given to design fundamentals. These include temperature stabilization of the detector, the cavity and optical elements, compensation of the preamplifier and detector at 20 cps to provide a nearly perfect flat response to eliminate the effects of chopper speed variation and reduction of the dynamic range over which circuits are required to operate to insure a high degree of. linearity. In addition, extreme care has been taken in the choice of circuits, selection of components and method of fabrication to achieve a highly stable electronic system, e.g., large amounts of feedback are used in the preamplifier; silicon transistors are employed to avoid temperature dependency; wire wound resistors are selected wherever dividers or calibration adjust- ments and networks are used; very highly regu- lated power supplies are employed using either closed loop regulation or precise zener diode regulation; highest quality components are util- ized and all leads that might be a source of extraneous or varying pickup are shielded. Presentation Data output is provided in two forms. An approximate value of the sea surface temperature is indicated continuously on the sea surface temperature panel meter. Temperature precise to +0.2°C is presented as an instantaneous indica- tion in the form of a continuously printed recorder output on the strip chart. Both the panel meter and strip chart recorder present the data in four 10°C ranges; =20) 160) 58°C), 720) tol 417°C, 416°C to +26°C) and! 425° to +35°C. These ranges overlap one degree, with the upper end of each range overlapping the lower end of the next higher range. Operating Characteristics The following operating characteristics have been determined from laboratory tests. Calibra- tion runs show an accuracy of absolute tempera- ture measurement of better than +0.2°C. System electronic noise is equivalent to 0.05°C and it ean be reduced further. Reference USMIDEE ELE (cavity) control drift is within 0.05°C. INFRARED RADIATION THERMOMETER The infrared radiation thermometer (IRT) was developed to provide a small, compact and por- table radiometer where the size and precision performance requirements of the ART were not needed. Basically, the two instruments are identical in principles of operation. The IRT differs from the ART in the following respects: 1. Smaller entrance aperture. 2. On-off cavity controller rather than pro- portional controller. 3. More rudimentary reference cavity structure. 4. AC signal panel readout instead of DC synchronous detection. 5. Calibrated to temperature range as desired. These basic changes have resulted in a highly compact, simply operated unit having an accuracy of the order of 1°C and a sensitivity of 0.1 to 0.2°C. 66 A photograph of the IRT is shown in Fig. 7. As in the ART it consists of three units-- radiometer, electronic unit and optionally sup- plied recorder. Operation consists of setting up the radiometer to view in the desired direc- tion, turning on main power and reading panel meter (and/or recorder) after the reference cavity is up to operating temperature. The IRT can be supplied with a filter similar to the ART (to pass only energy in the 8-13 region). The unit shown in Fig. 7 was calibrated on a black body and adjusted for a full scale panel meter readout of 15°F to 100°F. FIELD TEST OF ATRBORNE RADIATION THERMOMETER Field tests have been carried out by the Navy Oceanographic Office for which the instru- ment was developed. Test data have been obtained from 3 types of platforms under the conditions described below. Static Tower Tests In June 1962 the ART was mounted on ARGUS ISLAND, an oceanographic tower in 194 feet of water 22 miles southwest of Bermuda. The tower platform was easily adapted for mounting the ART. The instrument had an unobstructed view of the sea surface from a height of 65 feet and was operated continuously during daylight hours. Bucket temperatures were taken periodically. Comparison of ART temperatures and bucket tempera- tures are shown in Figs. 8 and 9. The trend of the curves shows a greater change in ART tempera- tures than in bucket temperatures with the maxi- mum change occurring at 1400 hours. On 26 and 27 June skies were 0.8 obscured by clouds and air temperature reached a maximum value at about 1300 hours. The ART appears to be recording the tem- perature of the surface of the water accurately. This has been shown in the laboratory; however, methods remain to be devised for relatinc these values to water temperatures at depths of 5 to 10 feet. Future tests are being planned to repeat the ARGUS ISLAND tests on a more compre- hensive scale. The ART measurement will be related to humidity, air temperature, wave height and albedo. A thermistor chain will make con- tinuous recordings of water temperature at various levels in the upper foot of water. Helicopter Flight Tests Prior to acceptance of the prototype ART several flight tests of the instrument were made over Chesapeake Bay on 23 and 24 June 1960. The ART was mounted in a HUP-2 helicopter. A series of passes were made across the track of a small boat which towed a thermistor probe clear of the boat's wake. These instruments were calibrated under identical conditions; equal accuracies were indicated. TEMPERATURE (° F) 0800 1000 1200 1400 1600 1800 TIME (LOCAL) Fig. 8. ART sea surface temperature comparison tests at ARGUS ISLAND, 26 June 1962. 67 TEMPERATURE (° F) TIME (LOCAL) Fig. 9. Data from the 23 June test are shown in Figs. 10 and 11. Temperatures were recorded by both the ship and the aircraft. Sharp spikes appeared in the ART trace each time the heli- copter crossed the ship's wake. Although the temperatures generally agree, an exact comparison is difficult because of the variability in sur- face thermal conditions shown by the traces. Meteorological conditions offer a possible explan- ation for the surface conditions; surface winds were light and variable and skies were clear. On the second day, light winds were present at the surface, the water appeared to be more thoroughly mixed and the thermistor trace was steady. ART temperatures were lower than ther- mistor temperatures (Fig. 11). Analysis of the data shows a probable ART deviation of +0.31°C from the reference sea surface temperature. WV-2 Flights The ART was shock mounted in the baggage compartment of a WV-2 aircraft. It views the water surface through an opening in the fuselage bottom. The aircraft is normally pressurized to allow air to exhaust around the sensing unit. This was found necessary to prevent instrument noise resulting from excessive turbulence around the lens. The recorder is separated from the console and mounted at one of the radiomen's stations. Continuous temperature records are manually noted at one minute intervals. During normal operations the aircraft was flown at an altitude of 1,500 feet at 190 knots. All flights to date have been conducted during daylight hours; however, night flights are being scheduled. Flight tracks do not usually exceed 1,400 miles; average flight duration is usually 6 to 8 hours. Aircraft surveillance with the ART provides a quick method of obtaining sea surface tempera- tures over large areas. Accumulated data aid in the construction of sea surface temperature 68 ART sea surface temperature comparison tests at ARGUS ISLAND, 27 June 1962. charts. A sample of surface isotherms for 2 October 1962 is shown in Fig. 12. The data were collected off the North Carolina coast in connection with sea surface temperature studies in the Gulf Stream. During the flights skies were clear, the sea was calm and the air tem- perature was 23°C (73°F). A series of flights was completed between 15 and 25 July 1962 over a triangular track centered at 39°N 70°W. The exercise was designed to determine time-space variability or persis- tency of sea surface temperature patterns. The aircraft was flown at an altjtude of 1,500 feet at 190 knots. The range of the observed tempera- tures was 18° to BG. Temperatures compared over 2, 48 and 72-hour periods are shown in Figs. 13, 14 and 15 respectively. CONCLUSIONS The ART is becoming a valuable means of obtaining sea surface temperatures. Aircraft surveillance with the ART makes possible the col- lection of large quantities of data in relatively short periods over broad areas. When first put into operation, the instrument was subject to frequent failure of electronic components; however, this problem has been resolved and the instrument now has good operational reliability. Instrument accuracy of 0.2°C has been shown in the laboratory. The ability of the instru- ment to detect horizontal temperature gradients at the surface has been of particular value. In field use the ART has not produced results com- parable to those obtained in the laboratory. On several occasions ART temperatures have concurred with immersion temperatures; on other occasions significant differences have occurred. At pre- sent, these differences remain unexplained. Future temperature measurements with the ART are expected to increase in accuracy and become a valuable source of information. 23 JUNE 1960 23 JUNE 1960 "COPTER : PASS, aoaieen 15 20 21 22 23 (24) 25 22 TEMP CALIBRATION— °C TEMP. CALIBRATION- °C WATER IMMERSED THERMISTOR (YP BOAT) AIRBORNE RADIATION THERMOMETER (AIRCRAFT) Fig. 10. ART helicopter flight test data (normal sea), 23 June 1960. 24 JUNE 1960 24 JUNE 1960 TEMP CALIBRATION- °C TEMP CALIBRATION-°C WATER IMMERSED THERMISTOR (YP BOAT) AIRBORNE RADIATION THERMOMETER (AIRCRAFT) Fig. 11. ART helicopter flight test data (mixed sea), 2 June 1960. 69 35° Boas 7@e 73° Fig. 12. Sea surface temperature pattern derived from ART flight test, 2 October 1962. eae TEMPERATURE (° C) 1400 1410 1420 1430 1440 1450 1500 1510 1520 1530 1540 1550 1600 1610 Ist TRIP TIMESCG My) 1610 1620 1630 1640 i650 1700 1I710 1720 1730 1740 1750 1800 I810 1820 2nd TRIP Fig. 13. Comparison of 2-hour temperature changes, 20 July 1962. 70 27 26 24b 2l TEMPERATURE (° C) Sig ses tes Sood goons fusst oat "TIME (GMT) 1400 1410 410 1420 1420 1430 Fig. 14. 1430 1440 1440 1450 Comparison es aoEsa| 1450 1500 1810 1520 1530 1540 1550 1600 1500 I510 1t§2Q 1530 1540 1550 1600 i610 of 48-hour temperature changes, 20-22 July 1962. 71 IGIO—JULY 20 JULY 22 27 TEMPERATURE (° C) 44 f E 1 coed sewed eae Gage Bed seaes Gaus i TIME (GMT) 1400 I410 1420 1430 1440 1450 1500 {510 1520 1530 1350 1400 {1410 1420 1430 1440 1450 {500 {510 1520 7a eos T i ing lene 1540 1550 {1600 1610-—— JULY 20 1530 I540 1550 1!600— JULY 23 Fig. 15. Comparison of 72-hour temperature changes, 20-23 July 1962. U2 SESSION HI SYSTEMS Chairman: W. F. BRISCH Marine Advisers, Inc. La Jolla, California } Th iy re ‘ CABLE ASSEMBLY WITH INTEGRAL HYDROPHONES AND INSTRUMENTATION P. R. ANDERSON Westinghouse Electric Corporation Baltimore, Maryland INTRODUCTION The cable assembly described in this paper was developed as a result of the need for a long multi-element vertical acoustic array that could be raised and lowered in and out of deep water rapidly by remote control without human super- vision. The location of the equipment limited the space available and required that a minimum of space and weight be added, that storage be automatic and that the entire array length change direction at least 90°. It was also important that the attitude and shape of the array be known at all times. Many sophisticated storage and handling arrangements were examined including methods for Flemishing, spinning reel type storage, linear cable hauler, capstans, etc. All methods had serious faults for this application--too large, too expensive, too complex, too slow, etc. A simple arrangement with a large power driven storage drum and an overboarding sheave to accom- plish the change of direction, appeared to be desirable if an array could be designed that could be stored under tension on the drum and flexed sufficiently to pass over a reasonable size sheave. A requirement that operation be remotely controlled eliminated the possibility of storing array sections in short lengths and assembling them as they passed over the side even if a system could be developed that per- mitted the rapid assembly necessary to meet the raising and lowering speeds specified. DESIGN CRITERTA The final approach decided wpon required the development of an array in which the hydrophone elements, preamplifiers if required and atti- tude sensing elements although larger than the cable in diameter, became an integral part of the cable, without restricting the flexibility required for changing direction and storage on adrum. This approach itself immediately initiated many problem areas. The strain member and electrical conductors of the cable would have to be concentric with the instrument pack- ages. The packages themselves must maintain as small a diameter and length as possible. An array of this length would be difficult to manu- facture in one continuous operation. The break- out of individual leads to each array element 75 must not weaken or disturb the continuity of the strain member but the leads themselves must sus- tain frequent flexings without failure. Cable diameter, even with 96 leads, must be kept to a minimum but the strain member must provide a satisfactory safety factor for reliability. The above problems required various solutions. To maintain the cable leads and armour concentric with the instrument packages, all units were designed with a hole around their center axis through which the cable could pass. A cylinder with a hole through its center does not present the most desirable volume in which to package a hydrophone, preamplifier, tilt sensor or magnetic bearing detector. The necessary components were fitted into packages with an external diameter of 2.5 inches and a maximum length of 3.5 inches with a hole 0.8 inch in diameter for the cable. An array of this type necessitated a hydrophone with a high degree of insensitivity to accelera- tion effects and the ability to operate at high pressures in addition to the necessary acoustic sensitivity and impedance. A stabilized transistorized preamplifier with 40 db gain was fitted into the desired package configuration. Linear accelerometers, used in pairs to measure cable deviation from the verti- cal, were available that would fit the package configuration. Magnetometer probes measuring field strength by the even harmonic method were found small enough to be packaged as an integral part of the cable. A system was conceived by which the array could be manufactured in short sections although the final assembly would have the appearance of one integral section. Although construction of the array was facilitated in many respects by assembling the entire length from short sections, the multitude of connections necessary to pro- vide continuity of the leads from section to section presented a problem. A quick disconnect type connector was considered but none was avail- able with the necessary small size, sufficient electrical connections and suitable strain member connection. The lack of a quick disconnect con- nector plus the other considerations resulted in the method described here. Wafer, Ib, Fig. 2. Cone and collar on cable. THE CABLE SYSTEM The cable to be used contained 48 twisted pairs in 7 bundles enclosed in a watertight jacket of neoprene. The conductors are solid copper to reduce cable diameter and facilitate design of an anti-hosing cable. The armour braid of flat stainless steel ribbons woven external to the neoprene jacket provided opti- mum flexibility with a minimum of added diameter and weight. Flat stainless steel ribbons readily adapt themselves to a terminating scheme which produces reliable strain member connection easily and 76 Terminal assembly. quickly. Hach terminal assembly consists of a cone and collar held together by three bolts (Fig. 1). The collar is slipped over the armour braid to the position the assembly is to occupy on the cable. The cone is slipped over the neoprene jacket with the partially unbraided armour ribbons ascending the cone. Hach ribbon is inserted through the proper hole in the base of the cone as the cone is pushed toward the collar, clamping the armour between the two identically inclined surfaces (Fig. 2). The three bolts are tightened sufficiently to lightly clamp the ribbons in place. A slight pressure is applied to each ribbon by pulling easily where it exits from the back of the cone. After each ribbon has been positioned, the three self-locking bolts are torqued to 30 foot points to firmly entrap the armour in the terminal assembly. Numerous tests of this assembly on a tension testing machine produced breaking strengths equal to that of the uninterrupted cable. Two terminal assemblies after being clamped to sections of cable, are joined by three stainless steel rope assemblies so that a flexible connec- tion exists between the strain members of two cable sub-assembly sections. The separation of } inches between the two terminal assemblies provides space for the elec- trical connections which present another problem. To keep the array diameter at a minimum and still connect the multitude of leads, each individual connection must occupy little more space than that required for a single lead. Connector TLE Sho inserts, crimp type pins and other methods of joining two leads were investigated and rejected for various reasons of size, reliability or com- plexity. After many tests, it was decided that the connections would be made by soldering two leads together side by side without twisting and covering the joint with heat shrinkable plastic tubing. The tubing, in addition to adding mechanical strength, provided a watertight bond with the insulation of the cable leads. Numerous tensile tests proved that the above method resulted in a junction stronger than the copper lead itself. During array manufacture the terminal assem- blies of two adjacent array sections are mounted about 4 inches apart. All the necessary elec- trical connections are made, tubing shrunk on and then three wire rope assemblies are installed joining the two sections mechanically (Fig. 3)n Each end of the rope assembly is terminated in a threaded steel fitting swaged to the wire. Self- locking nuts hold the assemblies in position after they are inserted through holes in the terminal assemblies. The volume between the terminal assemblies and around the cable leads and wire rope is filled with polyurethane which bonds to the neoprene jacket on the cable pro- viding a watertight joint that will stand the pressures expected for this application. SYSTEM TEST AND OPERATION A serious problem was discovered, however, when the array junction was cycled repeatedly under simulated load conditions over a large diameter sheave. The individual leads in the cable did not stretch as a unit with the outer armour and watertight jacket. During repeated flexing under tension, the leads appeared to walk up the cable and eventually the leads that walked the fastest failed in tension. This type of failure did not show up during straight ten- sion test. Repeated flexing under simulated load conditions was necessary to allow the cable parts to work sufficiently until differences in length of the various cable parts caused a break in a lead. Tt Assembled instrument section before potting. A number of different cable designs were manu- factured and tested with varying success. Part of the slippage problem was traced to the insula- tion on each lead. Because so many leads had to be soldered in such a small volume Teflon insula- tion was originally used. Teflon, however, is very slippery and cannot be bonded easily, so it was replaced by polyethylene. Test evidence also indicated that the relationships of the lays of each twisted pair and the cabling lay of each bundle contributed to the amount the leads appeared to walk in the cable. No cable tested eliminated the problem but the effect was mini- mized by proper design of cable insulation, direction and amount of lay and jacketing thick- ness and hardness. To provide protection against lead failure due to non-uniform cable stretch, a service loop was included at each electrical connection. The loop provides approximately 2 inches of slack to allow for non-uniform cable and lead stretch. To prevent the encapsulating compound from filling the interstices of the connector and capturing the leads thus negating the action of the service loop, the connector area before encapsulation is wrapped with tape. The polyurethane moulds around the wrapped area leaving a small void internally in which the lead joints may slide freely past one another as the elements in the cable are worked over the sheave. As pressure increases on the cable in deep water, the connector area is compressed and the internal void volume is decreased tending to freeze the leads in place. This is not serious because slippage occurs primarily during the time the cable curves around the sheave when pressures are negligible. It is necessary to bring leads out of the cable at every location of a hydrophone or instrument package. To eliminate breakouts through the cable armour all elements are located next to a ter- minal assembly and the required leads are broken out at the electrical joint section between the terminal assemblies. When the electrical con- nections are being made between two array sec- tions the cable leads associated with the hydro- phone located adjacent to the terminal assembly are selected and brought to the surface of the Fig. }. Instrument section after potting. Tne, 5c group of connections being made. To each cable lead to be used at this point a special single conductor wire about 8 inches long is soldered and the joint covered with shrinkable tubing. This special wire is non-hosing and is insulated with material compatible with polyurethane. The diameter is sufficient to allow passage through a hole in the cone and collar to the area immedi- ately adjacent to the terminal assembly where the preamplifier and hydrophone packages have already been positioned on the cable. This lead and a lead from the preamplifier package are soldered together. Two such leads are required for each preamplifier. Two leads from the oppo- site end of the preamplifier are soldered to two shielded leads from one end of the hydrophone. A service loop to provide stretch length during flexing is provided by rotating each lead 180° around the center cable. The volume between each component is filled with polyurethane. This encapsulation positions the units on the cable, provides a watertight seal and mechanical pro- tection for the necessary connections and pro- vides a flexible cushion between the units to prevent interference during the flexing that occurs as the assembly passes over a sheave or is wound around a storage drum. To facilitate the passage of the assembly over curved surfaces, a tapered section is moulded at each end of the package (Fig. 4). 78 Instrument section bent around 44-inch diameter. Polyurethane was used for all moulding appli- cations. Of the large number of materials tested it possessed, in addition to the desired physical and electrical properties, the greatest ability to bond to metal, rubber, fiberglass and PVC. is easy to handle, particularly if the material is used in the frozen cartridge form, moulds can be simply made and low temperature or room tem- perature cures are reliable and easy to make. Tests on the instrument bundle over a sheave under load simulating the worst actual operating condi- tions have shown that these compounds have very good abrasion resistance in addition to their flexibility. It To provide array position information, tilt indicators and magnetometer probes were included in the array assembly. As many as five leads were required for these instruments so that as many as five holes were needed through the cone and collar. Assembly procedures, however, were similar to that used for the hydrophones and preamplifiers. Instrument assemblies manufactured according to the above procedures have been tested under tensions of 1,200 pounds while flexing over a sheave 44 inches in diameter (Fig. 5). Minimum time to failure has been about 200 cycles with some sections surviving over 500 cycles before lead failure. ACKNOWLEDGMENTS The work reported herein was supported under a contract with the Office of Naval Research. 19 D fae 1 x a Ly : my) is ‘ ) xt a { } hb ys \ ‘ a Ld r 4 é ; ; ¥ " ‘ ii wae ij ‘a ita ao de p Bid it oo dipshit Liem Cheah sgpoeanninl id to ami tee iia Mi wee " 5 tne 1 pret henabents ay eis aT oe be Wi hunt ire beni it ' potsrpetaen i wh * au alive | ‘ he ny >> ‘Oe 4 ay ou . 1 ake a ; fal, naaiaLian? Teal oe ee A ) vn, Rake fren ee w a fess phe i ‘i Ae ee oe is Page? iNet ay Vila, eee : a PG > hy: en i eu nied et 1 mat, * woe pel, Ved Re iat camry ies) a poder Y iv Bre vst 4 fa desweres ot Ue eg meee! Pee tal im Rae eee oes. i. Ge Aghywltiaacdiay wg OF vile a Si ) * Te wa, : vw OT vide a ha Srahow. is rectified, filtered and applied to the control diode, CRl, by means of resistor R,. Since the DC bias current through a diode also determines its AC impedance the magnitude of the feedback AUTOMATIC GAIN CONTROL CIRCUIT CRI Fig. 6. Twin T oscillator Be STAFF Eos 10,000 v 200 0 O=10AV o——_—_ —__—_———_ Fig. 7. Series resistance constant current source. Eog Rot Ry [| IF: BA260db Foal Paths cen = ' aaa = E> R, eee =|zse80 fo! & 1, PAG 1+>— PA, Fig. 8. Simplified diagram of operational amplifier. voltage transferred through resistor Rj will be determined by this bias current. The effect is such that if the oscillator output level increases the rectified level of the automatic gain control increases, thereby driving more bias current through the control diode and lowering its dynamic impedance. Less positive feedback voltage is applied to the front end of the oscillator restoring the oscillator output level to very nearly its original value. A thermistor- resistor network is used to compensate for the temperature shifts in the control and rectifier diodes. The oscillator maintains the required amplitude stability of less than +0.1% over an ambient range of 50°F. Constant Current Amplifier This amplifier is used to drive a constant current through the staff independent of resis- tance. The excitation level of 10 volts rms results in a good signal to noise ratio. In Fig. 7 is shown a circuit configuration in which a resistance is interposed between the oscillator and the staff to approximate a constant current source. Full scale voltage of 10 volts rms will generate a current of 5 milliamps through the 2,000 ohm staff. Holding 5 milliamps constant to within to.1% as the staff resistance changes oii { 4 2000-1 Eos 0-8 V { 5001 Eoa 3-llV 100K 250 ' ° o- Fig. 9. Actual constant current amplifier. from zero to 2,000 ohms requires a series of resis- tance, ne of 2 megohms or greater. Under these conditions the oscillator output would have to be at least 10,000 volts rms which is not practical. By utilizing a high gain amplifier in the opera- tional configuration, the needed current accuracy can be attained and the oscillator output level of 1 volt rms is multiplied by 10 which is needed to drive the staff. A simplified diagram of the operational ampli- fier is shown in Fig. 8. If the raw gain, Ay) is at least 80 db and the closed loop gain is 20 db or less, there will be a net negative feed- back of 60 db or more. Under these conditions as R3 is varied over the 2,000 ohm range the output voltage of the amplifier, ee changes from 2 volts to 10 volts as the current, Ty, is held constant. This follows from I, being equal to In within better than 0.1% as a result of the 60 db of negative feedback around the amplifier. As Ra is varied downward In tends to increase causing a slight decrease in amplifier sensing current, q,. The amplifier output voltage, E,., is in turn lowered and In very closely restored to its initial value. It follows that the vol- tage, E,,, developed across R3 is proportional to the value of R3 since I, is maintained constant by the action of the operational amplifier. Again it may be important to state that the preceding is true only if E, is constant to within 0.1%. Fig. 9 shows the scheme used in the actual circuit to arrive at a practical value of amplifier input impedance, the principle of operation being the same as for the simplified case in Fig. 8. SIGNAL CONDITIONING AC-DC Converter The AC-DC converter consists of a high gain amplifier again connected in the operational con- figuration but this time including silicon switching diodes in the feedback path. This tech- nique develops a proportional DC output voltage with relation to the rms value of the input vol- tage. With a pure sinusoid the average value is equal to 0.45 of the rms value. This clearly shows why the waveform distortion must be kept low in this instrument. With total harmonic distortion below 0.5%, a conversion accuracy of better than 0.1% can be obtained related to the rms value of the input voltage. A diagram of a simplified AC-DC converter is given in Fig. 10. The open loop gain, A,, of the amplifier is approximately 66 db insuring over 60 db of net feedback around the loop including the silicon diodes. The diodes are connected such that on the positive half of the output wave, 6 Eoc, diode CRl conducts and diode CR2 is biased off. The reverse occurs on the negative portion of the wave. The sum of currents In and I) are z equal to current I, since sensing current Iz is oc reduced to below 0.1% of I, as a result of the | 60 db of negative feedback. Operation with diodes in the feedback is as Fig. 10. Simplified AC-DC converter. follows. If there is no input signal to the amplifier and the amplifier output noise is below 1 volt peak-to-peak, the rectifier diodes are for all practical purposes open circuits allowing the amplifier to have its full raw gain of 66 db. The waveform appearing at the output of the ampli- fier will be that of white noise. If a positive voltage is applied to the input of the amplifier, a current I) will flow into the amplifier as sensing current I2 since the diodes are still open. Sensing current I3 instantaneously drives the output voltage, Eqc, negative causing CR2 to conduct. Input current will now flow through R32 as Io. I3,; the amplifier sensing current, will drop to a negligible value. At this time the forward gain of the amplifier has dropped to slightly greater than unity. Therefore, the vol- tage developed across Ra represents one-half of I, and the other half is developed across Ro. Further, the rectified voltage, Eg., developed across Ro is proportional to the input voltage, Eos, to better than 0.1%. In Fig. 11 is shown the actual AC-DC converter diagram used to obtain L L the 2 megohms input impedance required for the ~~VO> 7 amplifier. This requirement is necessary since the amplifier input impedance loads the staff. Fig. 11. Actual AC-DC converter. Low Pass Filter The output voltage, Ege, from the AC-DC con- verter, though being a precise conversion of the AC voltage across the staff, is still not ina Fig. 12. Low pass filter. form usable for most applications due to its pul- sating nature. It is necessary to smooth this voltage to keep the ripple to less than 5 milli- volts. The 2-stage LC low pass shown in Fig. 12 is employed for this purpose. The two inductors ae UNFILTERED, Ep, have a series resistance of approximately 1,000 WU pe ohms. This resistance limits the load resistance FILTERED. E! that may be connected to the instrument without d DC . 4 ! Sv PZ appreciable loss of filtered output voltage, Epc. aXe At a carrier frequency of 1 Keps the filter has an attenuation of over 70 db. The filter has less than 0.2% attenuation at 5 eps giving additional useful data up to 20 eps. Fig. 13 shows both the unfiltered converter output, Enq, and the filtered output, Epc. The final output voltage is in a Fig. 13. Converter output. form useful for most analog recorders, telemetry inputs, digitizing devices, etc. CONCLUSIONS The instrument described herein is capable of directly measuring variations in sea water level Over a nominal band from 1 cps to 1 cycle per day. Accuracy may be held to 0.2% of the length of the staff which can be readily manufactured in lengths from 5 to 100 feet. These specifica- tions apply only so long as the surface of the staff is clean; in situ testing is in progress to determine degradation caused by marine fouling. The system is readily portable and the transducer is rugged enough to withstand rough handling and adverse environmental conditions common to oceanography. ACKNOWLEDGMENTS The authors wish to acknowledge the assistance and suggestions rendered by J. I. McQuilken and R. D. Gaul during development of this instrument as well as the help of Mr. Gaul with preparation of the manuscript. REFERENCES 1. DeLEONIBUS, P. S,, Resistance wire wave staff p. VII-14 in Oceanographic Instrumentation, 2nd Edition, Special Publ. }1, U. S. Navy Hydrographic Office, 1960. 2. FARMER, H. G. and D. D. KETCHUM, An instru- mentation system for wave measurements, recording and analysis, Proc., 7th Conf. on Coastal Eng., 1, Chapt. 5, August 1960. ) sp) CLI SARIS Win ti | q M4 + bs “ y b “ihe pay 3 i get A A eh f bh Sieh vl Sy wh Ehe Siw eed vigemioasioog OF a pe LL) A = an wail i. yany a Ta " oe sh oe eS ie Bie te SHIPBOARD ULTRASONIC WAVE HEIGHT SENSOR R. B. MARK Radio Corporation of America Burlington, Massachusetts ABSTRACT This paper describes a self-contained, elec- tronic wave measuring and recording system developed for the U. S. Navy Oceanographic Office. In waves up to 40 feet it measures relative dis- placement of the sea surface from mean sea level with an accuracy of better than 5%, while compen- sating for all significant ship motions. Verti- eal distance is measured by a box-mounted, pulsed ultrasonie echo ranging unit. Ship motion is measured by a doubly integrated, accelerometer- vertical gyro system. The two signals are com- bined with a fixed offset to produce true wave height. GENERAL SYSTEM DESCRIPTION A new approach to wave measurement is embodied in the shipboard wave height sensor developed by BOW MOUNTED HOUSING VERTICAL GYRO ACCELER- OMETER ROLL STABILIZATION DRIVE PITCH CRADLE DRIVE TRANSDUCERS Fig. 1. 101 RCA for the Navy Oceanographic Office. This instrument is capable of long-term, unattended operation in the normal shipboard environment of vibration, temperature change and poorly regu- lated primary power supply. Data in the form of displacement of the local sea surface from mean sea level is recorded on a strip chart at a sampling rate of 10 per second. Height resolu- tion is 2 inches with maximum wave height of 40 feet or greater and accuracy of better than 5%. Long-term, stable integration of ship accel- erations permits accurate recording of waves having periods of 2 to 25 seconds. The mode of operation of the wave height sensor is shown in the block diagram of Fig. 1. The rugged, watertight, stabilized housing shown in Fig. 2 is mounted at the ship's bow projecting over the surface of the water. This housing con- tains and protects the vertical gyro, acceler- ometer and roll and pitch stabilization drives. INTERIOR ELECTRONICS DOUBLE INTEGRATOR STRIP CHART RECORDER SUMMING AMPLIFIER | WAVE HEIGHT SENSOR MEASURING CIRCUITS AIR TEMPERATURE COMPENSATION Block diagram of wave height sensor. Fig. 2. Stabilized transducer housing. The ultrasonic transducers are mounted on the bottom of the housing. The vertical gyro, which has a low erection rate after spinup, acts as a long-period pendu- lum and provides a vertical reference accurate to 1/8 degree in spite of normal motions of the ship's bow and independent of the accuracy of the mounting surface. The housing is roll stabi- lized and the inertial quality, temperature regu- lated accelerometer is additionally stabilized in pitch. It is mounted on a small servo driven pitch cradle that is erected to synchro signals received from the vertical gyro. The output of the accelerometer is a direct current proportional to the true vertical accel- eration including gravity. The current is con- verted to voltage by means of a precision pick- off resistor located in the interior electronic package shown in Fig. 3. The gravity component is removed and the signal doubly integrated to produce a voltage proportional to the vertical displacement of the instrument housing. Distance from the housing to the local sea surface is obtained by means of a pulsed, ultrasonic, echo- ranging height sensor previously developed by RCA for hydrofoil craft control applications. The two signals are subtracted and combined with an adjustable bias to give the instantaneous dis- placement of the local sea surface from mean sea level. Correction for the variation of the velocity of sound with temperature is accom- plished by means of a single dial setting. This could be made completely automatic, if desired, by addition of a thermistor temperature sensing eircuit. TRANSDUCER BEAMWIDTH Most of the problems encountered in developing the wave height sensor arose, naturally enough, from such properties of the ocean surface as its local slope, reflectivity, displacement and velocity. The first two properties primarily affect the design of the echo-ranging part of 102 Fig. 3. Interior electronic package. the wave height sensor. As shown in Fig. 4, the apparent profile of a wave is distorted by an excessively wide beamwidth, Fig. 5 shows quan- titatively the error due to beamwidth in the presence of sloping wave surfaces. The lower group of curves corresponds to the shortest range within the beamwidth and the upper group to the longest range. For pulsed operation the lower group represents the first return. The extreme curve of the lower group shows the return for specular reflection, which is necessarily from perpendicular incidence. It implies operation =— TRANSDUCER NOMINAL =— ~ oa NGA =. SURFACE MEASURED BY EARLIEST RETURN BEAMWIDTH s —~ SEA SURFACE SURFACE MEASURED BY AVERAGE RETURN ime, Mh, BEAM a@=30° INCLINATION DIFFUSE REFLECTION LAST RETURNS ae PGES 35 He FIRST RETURNS SPECULAR REFLECTION fo} 5 iKe) 15 20 25 30 INCLINATION IN DEGREES Fig. 5. Error in height due to beamwidth and wave slopes. on sidelobe reflections or, alternatively, loss of signal when the wave slope is greater than one-half the beamwidth. It is apparent from Fig. 5 that very narrow beamwidths would minimize the errors due to wave slopes but a compromise must be made if loss of signal is to be kept within allowable bounds. Such loss occurs when the slope of the entire surface in the area encompassed by the beam exceeds one-half the beamwidth. Furtunately, large slopes are statistically improbable under conditions of light wind and smooth surfaces. Moreover, the presence of wind roughens the sur- face and diffuses the return reflection thereby causing an effective increase in beamwidth without an attendant loss of accuracy. Statistical measurements of the slopes of waves have been made.~? The distributions are approximately Gaussian with parameters shown in Table I. Since the measurements were made from photographs taken at substantial altitude com- pared with wavelength, the slopes represent the superposition of ripples and chop on the larger Effect of excessive beamwidth on height error. Table I. Distribution of inclinations of reflecting facets. Direction Crosswind Upwind-Downwind Wind velocity, knots 10 20 10 20 Bias, degrees (0) ) -0.5 -1.0 Standard deviation, degrees 5 Tod i 10 waves. From Table I it is apparent that a slope of *20°, corresponding to sinusoidal waves of length-to-height ratio of 9:1 or greater, would include all the statistically significant cases. However, a half beamwidth of 20° would produce a range error of about 6%. In practice it is feasible to operate at a frequency of 38 Keps and reduce the beamwidth to 8° between nulls, a value which can be obtained with commercially available transducers of 5 inches diameter. The RCA hydrofoil ultrasonic height sensor has undergone many hours of successful testing with such transducers. Sea conditions ranged from glassy calm to white caps with waves up to 5 feet. Relative winds from zero to 50 knots were encountered. Although receipt of 80% of return pulses is sufficient for proper opera- tion, usually more than 90% and often more than 98% were received. SELECTION OF TYPE OF MODULATION Of the types of modulation applicable to dis- tance measurement two of the simplest are short pulse and sinusoidally modulated CW. The mea- sures of range are respectively the time interval between transmission and reception and phase shift of the sinusoidal envelope. For a maximum range of 40 feet the total transit time is about Superior numbers refer to similarly numbered references at the end of this paper. 0.1 second and the highest non-ambiguous envelope frequency is about 10 cps. A carrier frequency in the neighborhood of 35 to 40 Keps is suitable for either type of modulation since it is high enough to avoid machinery noise interference and low enough to avoid the effect of the rapidly increasing attenuation in the region of 50 Keps and above. For a relative horizontal velocity of 2 feet per second, the 38 Keps transmitted frequency is shifted by 419 eps for a half beamwidth of 8 degrees. A continuous spectrum is generated by echoes from various points within the beam. The envelope of such a signal approaches 100% modulation at noise frequencies up to 19 eps and interferes with the necessary signal modulation of 10 eps. Pulse modulation was chosen for the RCA height sensor to avoid this source of noise and to obviate the necessity for heavy filtering with attendant low time response. Operation on the leading edge of the return pulse eliminates the effects of doppler shift. MINIMIZATION OF THE EFFECTS OF SPRAY The effect of spray on the distance measure- ment is minimized by the action of a slow auto- matic gain control loop. Spray droplets are generally small compared to the transmitted wave - length and, therefore, tend to scatter the energy incident upon them. If a sheet of spray is suf- ficiently dense the true echo may be so reduced in intensity that it will not exceed the auto- matic gain control threshold and hence will be 60 50 40 SPECTRAL DENSITY [A(w)]* 30 IN FT.2-SEC 4 6 Fig. 6. lost. To provide for this case the previous cor- rect range must be held. Since the height sensor measures range by counting clock pulses, the pre- vious count can be held exactly for any desired period of time. If the spray persists, the auto- matic gain control loop gain is automatically adjusted, within limits to pass the true signal. It is unlikely that a persistent wind-borne spray would produce a sharply defined echo at suffi- ecient intensity to exceed that of the true echo because the particles of such a spray are neces- sarily finer and more uniformly distributed than those which occur momentarily in dense sheets near the surface of the water. SHIP MOTION COMPENSATION Ocean waves further affect the design and per- formance of the wave height sensor by causing motions of the ship's bow upon which the wave height sensor is mounted. The principal motions are pitch, roll and the attendant translations of the housing. It is required that the ultra- sonic transducers and the input axis of the accel- erometer be maintained approximately vertical but the necessary accuracy for the accelerometer is much greater than for the transducers. It is for this reason that the accelerometer is pitch sta- bilized while the transducers are not. Surge and sway accelerations are eliminated by keeping the accelerometer vertical and heave acceleration is used to derive the heave displacement height com- ponent which is subtracted from the echo ranging height. Fig. 6 shows the Neumann spectrum for a fully developed state 6 sea and also the rough relative WAVE AMPLITUDES RELATIVE SHIP RESPONSES 8 1.0 : 1.2 FREQUENCY IN RAD./SEC. 1.4 Neumann spectrum of state 6 sea. 104 response curves for pitch and roll of a typical oceanographic ship. To the right of the appro- priate natural frequency the ship response is 180° out of phase with the slope of the waves and to the left it is in phase. Maximum amplitude of pitch response is expected to be less than 8 and roll response will not ordinarily be of much greater magnitude because the ship is required to be hove to and headed into the sea during wave measurements. Since the major portion of the pitch response is in phase with the wave slopes, the incidence of the echo ranging beam will usu- ally be near perpendicular and signal loss will be minimized. For higher frequencies, corres- ponding to pitching that is out of phase with the wave slope, the pitch amplitudes and wave slopes are smaller. To provide the necessary accuracy of roll sta- bilization at the frequencies near the peak of the spectrum, the servo bandwidth has been set considerably higher than the frequency of peak wave response. The accelerometer pitch cradle servo bandwidth is several times that of the roll servo. : CONCLUSIONS The shipboard wave height sensor has been designed for accurate wave measurement at sea. It is believed that this design provides capa- bilities far exceeding those of existing systems, particularly with respect to measurement of large waves in deep water. ACKNOWLEDGMENT Development and construction of the equipment described herein was supported by the U. S. Navy Oceanographic Office under contract number N62306-932. REFERENCES 1. COX, C. and W. MUNK, Measurements of the sea surface from photographs of the sun's glitter, J. Optical Soc. Amer., 44, 37-h0, January 195). oe 2. COX, C. and W. MUNK, Statistics of the sea surface derived from sun glitter, J. Mar. Res., 13, 198-227, November 1954. 105 Bay sl a 4 Me «Ww see keh 8 oh » PT HN ey Pa i fi ie n y ' if t mH eps eet, i yes: a" ena “ee 5 Ay rt) DI) ay i vet Sit 1 mehy, aa Pil eae ane ¢ ihe f y ‘ ‘ i i ah % 7 aR deg a ae Porn 22 vps td gs AL EQUIPMENT FOR TELEMETERING OCEAN-WAVE INFORMATION W. A. VON WALD, JR. U. S. Naval Research Laboratory Washington 25, D. C. INTRODUCTION Wave measurements in mid-ocean have been plagued by three main obstacles: (1) observers generally do not agree on wave height (and also tend to ignore wave period), (2) in heavy seas the main emphasis of a ship captain is on sur- vival rather than on measurements and (3) instru- mentation for shipborne wave measurements is not generally available. Further, survey vessels that are equipped with wave-measuring instruments are not likely to provide sufficient data to describe the wave climate. An unmanned buoy that could measure ocean waves, particularly if it eould do this on a currently-reporting basis, would seem to be a valuable asset to oceanographic instrumentation. THE WEATHER BUOY The United States Naval Research Laboratory. has developed a weather-telemetering buoy (shown in Fig. 1) designed to transmit wind speed, wind direction, barometric pressure, air temperature and water temperature at 6-hour intervals. The messages are sent in Morse Code with numerous "repeats." Provision has been made to add up to 5 information channels without modifying the equipment and an additional 5 channels after minor modification. These added channels have been pro- vided for telemetering outputs of oceanographic sensors that may be available in the future. The buoy case is 36 inches in diameter and 30 inches high. A telescoping ballast structure extends 15 feet below the buoy. A telescoping superstructure with an 18-foot whip antenna ex- tends above the buoy. All external instrumenta- tion is mounted in a single assembly on top of the antenna. A UHF beacon is provided to permit aircraft to locate the buoy or, if the buoy is moored, permit its use as a navigational aid. The buoy weights 600 pounds and carries batteries for 12 months of unattended operation. Inside the buoy case a lower compartment houses the batteries. The instrumentation for coding, programming and measuring barometric pressure, station direction and average wind speed are located in the upper portion. Also in the upper compartment are a de-de convertor, 2 transmit- ters, 2 command receivers for remote turn-on. On clock-control one frequency is used in the Fig. 1. The weather buoy. 107 Fig. 2. daytime and the other at night. If the station is operated on demand it will respond on the fre- quency used for interrogation. The two methods of control are completely compatible. The main instrument package is shown in. Fig. 2. MEASUREMENT OF SEA CONDITIONS The first additional oceanographic sensor to be investigated is a sensor to obtain some measure of the sea condition. One description of the sea surface is that of an energy spectrum which gives estimates of wave energy as a func- tion of frequency. Energy spectra are usually obtained by digital computer processing of a time series of data of the surface elevations taken over a time interval of the order of 20 minutes. The area under such a spectrum is 4 measure of the total energy of all the waves. Since the weather buoy cannot be equipped with a complex computer, nor is it practical to supply the power required to transmit 20 minutes of wave- height data to a shore station for analysis, the possibility of obtaining the integrated value of the mean square of the surface deviation over a 20-minute period using a pressure transducer suspended below the buoy and working into an 108 Weather buoy instrument package. integrating device was explored. The fact that this integrated value is a measure of the total energy of the measured waves added to the desira- bility of this arrangement. Further study indicated that a vertical array of sensors could provide a certain amount of wave period information and this will be discussed. Wave Measurements Trochoidal wave theory depicts ocean waves as eircular motions of water particles which decrease exponentially with the depth expressed in wave length. If one prefers to consider the ocean surface as an infinitely broad distribution of sinusoidal waves, the variations in pressure due to the waves decrease with depth, the rate of decrease being a function of the wave length. For water depths of the order of a few wave lengths either concept results in virtually the same rate of decay of wave effects with depth. Because the wave effects diminish with depth as a function of wave length and because wave length is directly related to the square of the wave period, there is available a means of sensing and telemetering some estimates of the energy spectrum. A vertical array of pressure sensors suspended at various depths below a buoy on an effectively rigid cable would follow the vertical motion of the buoy provided the weight of the array is sufficient to insure that (a) the array remains essentially vertical in the presence of current shear and (b) the weight is greater than the hydrodynamical drag so as to force the sensors to descend as rapidly as the buoy. Consider the suspended vertical array of pres- sure sensors indicated in Fig. 3. A 5 unit array is shown with the buoy at the trough (a) and at the crest (b) of a surface wave. A sinusoidal wave is used for simplicity. The constant- pressure surfaces shown at the various depths in the figure were computed by using the exponen- tial decay in wave activity derived from tro- choidal wave theory and also by using the hyper- bolic cosine ratio resulting from the concept of a sinusoidal wave on the surface which is infin- ite in extent. Both methods resulted in essen- tially the same configuration of constant- pressure surfaces. (b) Buoy @ GAGE | 5, — {GAGE 2 40 GAGE 3 GAGE4 60 140 Wave sensor configuration schematically displayed relative to constant pressure contours for a wave period of 9.34 seconds. The following equations were used: meyers! (2) K = cosh [2ra/L(1-Z/a)] (2) where d is the water depth, Z is the mean depth of the pressure surface, L is the wave length and K is the factor by which the amplitude of the con- stant-pressure surface at depth Z is reduced com- pared to the constant-pressure surface at the water surface. One can see from Fig. 3 that the shallow sensors cross fewer constant-pressure surfaces as the wave passes than do the deeper sensors. The pressure change to which each sensor in the array is exposed is proportional to the difference in amplitude between the constant-pressure surface at the water surface and the constant-pressure surface at the depth of the sensor. Therefore, the response of each sensor is proportional to 1-K. Integration of Wave Measurements To obtain an integration of the output of a wave sensor over a period of time (for example, 20 minutes) the concept of an electrical calo- rimeter is used.! If a voltage which is at all times proportional to the deviation of the pres- sure from a mean value as measured by any one of the pressure transducers in the vertical array (Fig. 3) is applied across an appropriate resis- tance, whose heat capacity is known and which is highly insulated, the rise in temperature of the resistance over a given period of time is a measure of the energy introduced into the resis- tance during that time. This temperature change can be telemetered as a measure of the mean square of the amplitudes (also the total energy) of the pressure variations to which the wave sensor responds. Such an "electrical calorimeter" has been con- structed and is illustrated in Fig. 4. The unit consists of a copper bobbin on which are wound a heater winding and a resistance-thermometer winding. The calorimeter is insulated from its surroundings by foam plastic. A 30-gram copper bobbin in the center of a 3-inch cube of foam plastic has a thermal time constant of about 20 minutes and averages over a 20-minute period with good accuracy. With the full scale output of the wave sensor described above, the calorim- eter would show a temperature rise, above ambient, of 70°F. The coding system of the weather buoy has a resolution that corresponds to about 0.4°Rr so that overall system accuracy should be adequate. Superior numbers refer to similarly numbered references at the end of this paper. 109 HEATER ELEMENT —120 OHM KARMA a ELEMENT - 9500 OHM NICKEL ax 2; PLASTIC MOUNTING PLATE 05055028 “0054 x SSIHSSSS eH KC XXX RR TERMINALS x KR COPPER SLEEVE WASHERS PLASTIC WASHER FOAM PLASTIC INSULATION - 3" CUBE Fig. 4. Typical electrical calorimeter. Calculation of Energy Spectrum Estimates by Use of a Vertical Array Assuming that the sea surface can be approxi- mated by a combination of sine waves having 5 different periods, one can compute (1-K) for each of the 5 sensors as illustrated in Fig. 3 for each of the 5 different wave periods. Each sensor will respond to the 5 waves but with different sensitivities due to the different depths of the sensors below the buoy. The squared and inte- grated output over an interval of time of each sensor in the array is a linear combination of such response to the 5 waves having different periods. Thus one can form a set of linear equa- tions describing the outputs of the 5 sensors as follows: R, = (1-K,,)"B, (3) where R; is the output energy over a given inter- val from the ith sensor, K;; is the depth attenu- ation for the i sensor for the jth wave period and EB; is igrsteretentestigiae la to the energy associated with the J A wave period. These equations can then be inverted to obtain the E. in terms of the Rj which is the data telemetered. The inverted set of equations for an array with the gages located at the 95% response depths for periods of 4, 8, 12, 16 and 20 seconds are: E59 = +0 -393R39-2-80R157+12.39R 355 (a) -28. DR6p6t18. 56Ro78 E16 = -1.000R39+7 .12R157-29.8R355 6) +53 .OR6p6-30.0Ro7g 110 Bio = 4+1.46R39-10.03R)57+28.12R355 (6) 6 -38 .61R626+19 .20R978 le, = -2.67R39+7-11R)57-12.19R365 (7) +11. 26R¢56-6-79Ro7g By = +1.84R39-1.56R1 5742. 26R365 ra -2. TAREo6t1 3 PRo78 A The subscripts on the E's indicate period; those on the R's indicate depth. Since a short appendage to the buoy might be much more convenient in field operations than a long one, an array of gages located at the 10%- response depths was also considered. The inverted set of equations for the short array are: E59 = 418, 271-Ry 9-5,102.Ryg gtl, TOOR,), 6 (9) -405..9R7g 1,431 -4R) 5), +8,814.R Ei6 = -32,870.R -2,689.R 4.9 +400. ORZ9 3 437 ARI ok 19.8 Wh.6 (10) Eyp = +14,474.R) 9-3, 783-Ry9 gt1,064.Ryn 6 (11) -111.5R79,4+2-86R1.), Eg = -1,721-R), ot30.5R,9 gtl09.aR 4.9 19.8 kh.6 (12) +9) 5 1879 a y-2 . 86Ry 5) By = 443.3Ry9-75-SRyg_ gtl5-TRyy 6 a 13 -0.15R79,4-3-O6Ry 5), Obviously the coding-system resolution required to use the short array would be far greater than that required to use the long array. Once the equations for 4, 8, 12, 16 and 20- second periods have been solved to obtain 5 estimates of the energy spectrum the location of any maximum might or might not be indicated. If a maximum is indicated another set of equations for a more narrow range of periods can be deter- mined and solved to locate the energy maximum in more detail. FIELD TRIALS Preliminary field trials of suspended pres- sure transducer arrays were carried out during September and October of 1962. Long period waves were encountered in only one test when a 2-unit array was in operation with sensors at a depth of 88 and 245 feet respectively. Connec- tions to the pressure transducers were by means of cables to recorders mounted in a small boat. The water depth was well over 2,000 feet. The wave records indicate that the sensors did not follow the downward motion of the buoy as accurately as they did the upward motion. For this experimental setup the output of the pressure transducers was recorded on strip recorders rather than feeding into the electrical calorimeter as would be done for telemetering. Digital data taken from the 2 gages in the array were subsequently processed to obtain a power spectrum of the energy as measured at the par- ticular depth of each gage (not corrected for depth). The areas under each of the plots of these spectra (as shown in Fig. 5) are equal and the respective normalizing factors required to reduce the area to the particular value are shown. Since the area under a non-normalized plot would be proportional to the total energy, the normalizing factor is proportional to the energy and would therefore be directly related to the rise in temperature which would occur in the electrical calorimeter. The ratio of the NORMALIZING FACTOR 10:52177 x 107° NORMALIZING FACTOR 61-:15405 x 10° © RELATIVE ENERGY normalizing factors is 5.8. The single wave period which would give this particular ratio for two gages at the depth used is about 23 seconds. The actual spectra have maxima around 10 seconds, thus giving a rather large discrepancy. These preliminary tests have pointed up a number of changes which must be made in the experimental setup for future experiments to evaluate the overall feasibility of this approach to sea-state measurement. REFERENCES 1. VON WALD, W. A. Jr. and J. E. DINGER, The electrical calorimeter as an integrating device, Rept. 5796, U. S. Naval Research Laboratory, July 1962. UNPUBLISHED. GAGE A 88' GAGE B 245' 2018 16 14 |I2 11 10 9 8 tf 6 5 PERIOD (SECONDS) Fig. 5. Energy spectra from suspended wave sensors. dat ; ; ) ; Ns ‘ ‘ poe oleic bee © Saale stn ee Ta Qa abwoae! tag itl om ee I herd yy ,ow . » >. va a a he Pera «| ‘ i A Oe iy ae se thee FON t Mitihe nS ote pokey ey RWerrk oro, be Hy avivin Matt 2 Se "ho 1 ae Se ae cra ’ dy BAL hy ee ae ake Batts re is ‘weds PS Saha Sai xd onre aie queasy ae Git a if ‘ aa tm it ah yy Seo ee papi a rns he: tayo ae G1dHIDs, LD We Seep” ie eae BMRB E> dit TA gua ye aR beiete oi: On gal heee Several strane | ood g+ aie waeet SAape TR aes Fee ih ears P Bi aR yu QA) fete Sota 2 ' ee: : makes aie Pyitias al? al panies See ate mae leat Coty 3 bi pb fee L4zlau ia qeassouny (i JOR e ur @oR) es iv os Danae Qe Yyteds ole 2D me ries ae CORES Tye, toe Tia Soe) ee, aa oe lg ie Lia rie! ». 1 GT ett orc kak ampere Tiago am ” ‘ { iY aie DE Bae y i vated owe 4 o) ‘aesaeye 60 zene -0.247 11.8 43 3.65 -0.680 3.4 12.7 2.68 123 90% response are also given. In a non-inertial system these values would be 2.3 times that for 63% response. Another feature of the deceleration curves of Fig. ll is of interest. Immediately after stop- ping the tow carriage it is noted that the response factor goes above unity before falling off as it should. This is due to the wake that has been generated by the meter as it travels by the stopped rotor, thus giving an additive com- ponent to the omnidirectional sensor. The greater the flow speed before the negative step the more rapidly the wake moves past the meter, consequently shortening the time during which the rotation rate is significantly influenced, as seen in Fig. Ji. Fig. 12 gives the flow speed indicated by the rotor as its speed through the water is varied in a somewhat irregular fashion. Some of the characteristic response features described above are readily discernible. Note that when the speed varied almost sinusoidally with a period of roughly 4 seconds (between time of 160 and 170 seconds in Fig. 12) the indicated speed fol- lowed the mean trend but failed to indicate the variations. The plotted points represent single rotor revolution averages; somewhat better response might have been obtained by shorter averages, but at low speeds the previously dis- cussed angular rotor speed variability due to non-uniform torque would introduce considerable error. a ACTUAL TOWED SPEED INDICATED BY ROTOR SPEED E (ONE REV, AVG) CO 0.4 2 00S 0 0F PS, tN \ = re \ | a ; ‘ \ Ww Ww / * \ ao03 \ : o | \ ‘ \ a a | \ / \ a ! ‘ \ = 0.2 fs 2 \ 2 ° = \ N a = Ne \ z ° 4 L. Sa 5 Oe a > = (3) cr ¢ oO 100 110 120 130 140 150 160 170 180 TIME (SECONDS) Fig. 12. Response of rotor CS-2 to irregular changes in tow carriage speed. CONCLUSIONS an anti-fouling aerosol similar to petroleum We consider it presumptuous to assume that there is such a thing as the "best" current meter. A current meter, just as any other instrument, must be evaluated on the basis of the user's requirements. It is a good instru- ment if it satisfies the particular needs with high reliability and within the user's allowable limits of accuracy, cost, time and manpower. It is certain that the Savonius rotor current meter is a good instrument for many applications. In practically all cases its characteristics and limitations must be recognized if not accounted for at the data analysis stage. With reasonable manufacturing control the standard Savonius rotor current meter should operate reliably in the range of 0.05 to 3 or knots. Turbulent flow encountered in the natural regime may be expected to bias rotor performance; much yet remains to be done before this bias can be quantitatively estimated. Normally, measurement of highly turbulent or rapidly varying flow with the rotor probably should be avoided. Rotor output is likely to be greater in the natural environment than in a tow tank so the meter will register a higher mean current speed than actually present. Marine fouling has a marked effect on rotor output even when not very severe. The rotors should either be kept clean or a correction factor greater than unity applied. Presence of 12 jelly has little effect on performance causing a slight increase of output. Above about 0.2 knots steady state performance is essentially unaltered by changing the length or number of tiers of vanes in the rotor. At speeds approaching the threshold output will be altered inversely proportional to surface area of the rotor plates. Reduction of rotor diameter increases efficiency, presumably due to reduc- tion of surface area, particularly of the tier separators. The Savonius rotor is a highly inertial device. Response to acceleration can be several times faster than to deceleration (when the flow past the meter is stopped ) and response factors are strongly dependent on the magnitude of speed changes. ACKNOWLEDGMENTS Support for these studies was jointly fur- nished by the Hytech Division of Bissett-Berman Corporation and the Office of Naval Research under Contracts Nonr2119(4) at the A. & M. College of Texas and Nonr2116(1) at the Scripps Institu- tion of Oceanography. George Barlow of SLO assisted with the testing and N. HE. J. Boston of Texas A. & M. directed much of the data analysis. REFERENCES al SNODGRASS, J. M., Prototype telemetering current sensor, Section in Progress Rept. to Bureau of Ships, SIO Ref. 55-10, Scripps Inst. of Oceanography, January 1955. UNPUBLISHED. SAVONIUS, S. J., The Wing-Rotor in Theory and Practice, Published by Savonius and Company, Printed by Nordblad and Pettersson, Helsingfors, Finland, 1925. SAVONIUS, S. J., The S rotor and its appli- cations, Mechanical Eng., 53(5), 333-338, 1931. ia, GAUL, R. D., The Savonius rotor current meter, Tech. Rept. 62-271, Dept. of Oceanography and Meteorology, A. & M. College of Texas, February 1962. UNPUBLISHED. GAUL, R. D., Evaluation of the Hytech Corpora- tion current meter calibration system, Tech. Rept., Hytech Corp., San Diego, Calif., October 1961. UNPUBLISHED. MARINE ADVISERS, Calibration of Savonius rotor current meter, Tech. Rept. to Naval Ordnance Test Station, Prepared under Con- tract N123(60539)21697A(FBM), July 1960. UNPUBLISHED. STEVENS, R. G. and L. F. SHODIN, A fast response cup anemometer for measurement of turbulent wind over the ocean, Marine Seiences Instrumentation, 2, Instrument Soc. Amer., Plenum Press, New York, N. Y., 1963. U2) A i } 4 i » hy % i ie a | Hil BA ON : us Whe ad ; ,, ~ | ( mene i ¥ ; : \ a Laake Aha h Ma ies ‘e ie EOS VN RES mb a ee ‘ ‘ 2 mR pe Juries em ; tah ih s iy at Vena i | u tA r : d 4 wi¥ ; ; Wl ey 5 L v * at Pf ot = a ee . " my, : ei 7 i y yi i i i f Tae) ‘A j DUR I a aa Ne Sg, \ a i ij int : er a" a ey vas mead ic hea yee nye pay eat 7 Se Toatt : Of i Mhdy? Tan: Ay ppp Oh a 1 eh, ‘ A DOPPLER CURRENT METER F. F. KOCZY, M. KRONENGOLD and J. M. LOEWENSTEIN University of Miami Miami, Florida ABSTRACT A Doppler current meter designed and assembled at the Institute of Marine Science, University of Miami, was tested in model tanks and in the ocean. The output signal under representative oceanographic conditions was recorded and analyzed. INTRODUCTION An acoustie current meter based on the Doppler shift principle was designed and assembled early in 1961 and tested in order to study character- istics of the output signal and evaluate the limi- tations of this meter. Doppler shift meters are of interest because they have high sensitivity, are inherently self-calibrating, have good tran- sient response and have no moving parts. During a period from June 1961 to June 1962 tests were conducted in a towing tank and in representative oceanographic environments. The instrument consists of a transmitting and receiving transducer operating on a 5 Mcps acoustical signal. The receiver is a single con- version superheterodyne type which provides a good signal to noise ratio. The intermediate frequency passband was shifted upwards to uniquely select the upper sideband signals derived from flow towards the receiver. Addi- tion of double conversion and double sideband detection could provide instantaneous data on flow direction and magnitude. THEORY The Doppler shift or frequency change that results when a transmitted wave is reflected from a moving object may be expressed as: where ec is velocity of sound, f is transmitted frequency, f; is transmitted frequency shifted by the Doppler effect, c is positive when movement is towards the receiver and K is unity if the transmitter and receiver heads are identically located and the reflecting object moves in direc- tions parallel to the transmitted wave. For other angles, the angular relationship of the trans- mitting and receiving heads to the direction of the reflecting object must be recognized in the formula. In the interest of simplicity, K com- bines these angular functions since our objective was not to prove the Doppler theory but study the meter characteristics. The instrument is self-calibrating if f is held constant, K is accurately determined and c is known. Frequency is held to a 0.002% tolerance using a precision quartz crystal and instability is not a significant source of error. If a vane arrangement is used to align the instrument into the direction of current flow and the relative head angles are accurately measured, K becomes a known constant. At a current speed of 2 fps an error in sound velocity determination of 5 fps results in an error of less than 0.002 fps in speed indication. When sound velocity is determined indirectl the resulting errors introduced are shown in Table I. Table I. Measurement Error C Error Temperature: 0.036°F 0.18 fps Salinity: 0.02 parts per thousand 0.086 fps Depth: 5 feet 0.091 fps The error magnitudes are in excess of those generally incurred in oceanographic tests. Contribution No. 421 from the Marine Laboratory, University of Miami. Superior numbers refer to similarly numbered references at the end of this paper. 127 OSC DRIVER VARIABLE REGULATED POWER SUPPLY TRANSMITTER RECEIVER Fig. 1. System block diagram. INSTRUMENT DESIGN The transmitter section consists of a crystal controlled oscillator driver whose output is matched to a barium titanate transducer. The receiving system consists of a broadband crystal controlled superheterodyne receiver, transducer, frequency meter (Model 41-7991, Airpax Electronics, Inc.) and tape recorder (Fig. 1). Both the transmitting and receiving transducers were mounted on a head fixture which permitted variation of the "interocular'' distance between the 2 units and experimental changes in the cross- over point. The rear surface of the l-inch diameter transducer discs are air-loaded to elimn- inate back radiation. Theoretical beamwidth at 1/2 power points at the operating frequency is given as VOLUME OF REVERBRATION X @ = 1.22 5 radians (2) where ) = c/F and D is transmitter diameter. For the test instrument ’ is 3.08 x 107© and a is 0.836°. The beamwidth measured in a tank has been found to be somewhat less than theoretical and minor lobes are at least 60 db down in amplitude. A frequency of 5 Meps was selected as a best compromise between achieving narrow beamwidth, ability to resolve small reflecting particles and high Doppler sensitivity vs. increased attenu- R ' ras SMUT ation by the medium with high frequency. The 2 units of channellite 100 (barium titanate, Channel 8" Industries) were found to have resonance peaks 30 Keps apart and more careful frequency matching Fig. 2. Transducer positioning. could undoubtedly improve the signal to noise ratio and further reduce the transmitter power requirement. A transducer spacing of 8 inches with a crossover point of 10 inches (Fig. 2) was RECEIVER 128 20kc + 3DB 10 DB BANDPASS FOR +V 435 Fig. 3. | REGION OF -V 455 ke CARRIER Receiver bandpass characteristics. found to optimize the signal to noise ratio and minimize the effects of head turbulence. Adequate frequency stability was obtained by use of 0.002% tolerance quartz crystals in both the transmitter and receiver. A drift of the receiver's local oscillator causes a shift in the received signals with respect to the IF band- pass but does not affect the instrument accuracy. Transmitter frequency drift effect is discussed in the section on theory. The receiver is a {-transistor circuit employing an RF stage, mixer, oscillator (erystal- controlled), 2-stage IF strip, diode detector and 2 audio stages. AVC was not incorporated in an effort to maintain system linearity. The IF amplifier was tuned as shown in Fig. 3 so that the carrier and upper IF sideband only were passed. This limited the instrument response to read current flow toward the transducer faces and reduced response to turbulent flow in one- half of the possible directions. A receiver sensitivity of 1 microvolt at 0.15 VPP output for a 12 db (S+N)/N ratio was achieved by the use of 70 Meps cut-off PADT transistors (post alloy diffused. Amperex, Inc.) in the critical circuits. The receiver was powered by a 12 volt mercury cell. The transmitter utilized a conventional crystal controlled vacuum tube oscillator. A variable voltage regulated power supply provided control of transmitter power input from 15 watts to less than 10 milliwatts. In practice, a power input of 45 volts at 6 milliamps to plate and screen of the oscillator tube was found to be optimum. The low impedance of the barium titanate trans- ducer (-J = 2 ohms, R<2 ohms) made it necessary to place it in series with the final tank circuit in order to transfer sufficient driving power. No environmental housings were fabricated and the heads were fed remotely by means of coaxial cable. Under average sea reflectivity condi- tions, audio outputs of 0.1 to 0.35 VPP were obtained with a signal to noise ratio of 10:1. Higher signal to noise ratios can be obtained by closer transducer frequency matching and/or higher sensitivity crystals. 129 THE EXPERIMENT Several experiments during a one year period were conducted at the Institute of Marine Science laboratory dock in 8 feet of sea water. The head assembly was tripod-mounted 4 feet off the bottom and aligned parallel with the current flow. The heads faced toward the Bear Cut bridge 1,000 feet away and were 4 feet to one side of the dock pilings. Various head angles and crossover points were tried as well as positions parallel and normal to the current flow. A series of tests was run in the towing tank at Stevens Institute of Technology and the output was tape recorded. The tank was 80 feet square with damping beaches and was filled to a depth of 4 feet with still fresh water in an air condi- tioned room. Further tests were conducted off the west shore of Bimini, Bahamas, in 20 feet of water during slack tide when wind and current were nil. The transducer head assembly was tripod mounted 4 feet from the bottom and heads were successively positioned in all 4 quadrants, starting normal to the shoreline facing seaward. Various head spacings and angles were tried. The meter output was monitored and tape recorded. Tests in the bay at Bimini off the Lerner Marine Laboratory dock were conducted with the transducers 8 inches from the surface in 2 feet of water. The heads faced into the current well forward of the dock pilings but 2 feet from an anchored barge. The water was clear and the sur- face was smooth. A small drifting wood chip was earefully timed over the length of the barge to get an approximate flow velocity. Tapes were made of the meter output. The head spacing was 8 inches with a crossover of 10 inches. ANALYSIS Three principal types of equipment were used for readout. An Airpax magmeter read frequency of the Doppler shift. Integrating time constants were varied from 0.25 second to 4 seconds in 0.25 second steps. A 4-second integration gave the most stable readings. A Panoramic sonic analyzer, Model LP1A, was used for spectrum analysis using one second duration and O to 20 Keps log frequency sweeps at linear amplitude settings. Relative amplitude relationships were studied. A Tektronix oscilloscope, Model 535, was used to display output wave shape and fre- quency. Tests were recorded on magnetic tape and analyzed. A 4-second tape loop was made of a representative portion of each test. Succes- sively 1, 20 and 40 sweeps were recorded on the analyzer. Single sweeps at rates of land 2 cmper second were recorded on the scope and magmeter readings at 4 seconds integration time were noted. In addition, 15 and 60 consecutive one second sweeps were taken of continuous segments of the * 400 600 FREQUENCY Tes lh Tests at Institute of Marine Science dock May 1961 to February 1962: (a) oscilloscope, single sweep, 1 ms/em, (b) Panoramic, 20 sweeps, 4 second tape loop, (c) Panoramic, 16 sweeps, continuous segment and (d) Panoramic, 60 sweeps, continuous segment. original tapes. In general, the audio output of the Doppler meter fluctuates considerably due to flow velocity variation and the number of scatterers in the reverberation volume. Under violently turbulent conditions the frequency meter reading was fairly stable. Examples of extremely turbulent flow can be observed in the records taken from the Institute of Marine Science dock. Many local eddies were observed visually around the pilings and large swirling patterns from the Bear Cut bridge extended to the dock. The Doppler shift fre- quency, converted to an audible signal, sounded complex and resembled wind in blizzard conditions. When the transducer heads were positioned normal to the general current flow appreciable 130 Doppler shift was still produced, signifying turbulent conditions. With head parallel to current flow a complex wave structure was gen- erated (Fig. ha) and the frequency spectrum was broadband with no well defined "peak" (Figs. 4b, ke and 4d). The current meter output was recorded during tests at the Stevens Institute tow tank. No Doppler shift was produced until the transducers were mounted 2 inches from the surface and the towing arm crossed its own wake on its second and subsequent revolutions. Low frequency fluc- tuations modulating the sinusoidal output were caused by the motion produced wavelets (Fig. 5a). The frequency spectrum of the signal is narrow after a 1-second signal sweep (Fig. 5b). There Oe tanh ha FULL ATA TE eae a ee FREQUENCY th CP: Gs Ine, 5)5 Tests at Stevens Institute towing tank on 3 June 1962: _EREQUE NCYIN.CPS. b. (a) oscilloscope, single sweep, 0.2 ms/cm, (b) Panoramic, single sweep, 4-second tape loop, (c) Panoramic, 16 sweeps, continuous segment and (d) Panoramic, 60 sweeps, continuous segment. is no appreciable broadening after 16 consecu- tive sweeps (Fig. 5c) and there is still a sharp frequency spike after 60 consecutive sweeps (Fig. 5d). An interesting phenomenon was noted in the offshore Bimini tests where the sea was essen- tially still. The meter output was zero except when long gentle swells produced a Doppler shift output coincident with the below surface orbital motion. There was a poor signal to noise ratio (Fig. 6a). A broad plateau contains the entire forward range of the orbital water motion caused by the swell (Fig. 6b, 6c and 6d) except for the frequencies below 50 cps which were cut off by the combined tape recorder and spectrum analyzer response. dtsjal Frequency meter indications were generally lower than those of the Panoramic analyzer because the instantaneous "peak" frequency is not as well defined or densely distributed in time as appears on a well integrated record. Tests at the Lerner Marine Laboratory dock, where near zero wind speed and seemingly low turbulence conditions prevailed, show a more complicated wave structure than produced in the tank (Fig. 7a). Note the broader structure and some sidebands of Fig. 7b in contrast to Fig. 6. The sidebands were caused by turbulence and/or instantaneous variation in flow velocity (Fig. Tc). Additional integration time produced a somewhat denser version (Fig. 7d) and the low frequency peak is due to reflections from a barge riding at $e ii ais tae a REQUENCY IN CPS Co Fig. 6. anchor a few feet from the current meter. Very little Doppler was noted when the heads were placed normal to the current flow. Magmeter readings were reasonable compared to the current velocity measured by timing a floating chip. CONCLUSIONS The Doppler current meter described will pro- vide a sensitive and accurate sinusoidal output for measurements of a unidirectional, turbulence- free current containing particulate matter such as would exist in a properly designed flow tube. A meter based on the Doppler shift principle was described in a paper by Chapulnik and Green.3 Their experiments indicated that this instrument 132 Tests offshore from Bimini on 27 June 1962: (a) oscilloscope, single sweep, 2 ms/em, (b) Panoramic, 20 sweeps, 4-second tape loop, (c) Panoramic, 16 sweeps, continuous segment and (a) Panoramic, 60 sweeps, continuous segment. was well suited for measuring unidirectional, constant-velocity current flow. Since this meter is non-inertial, it responds to rapidly changing irregularities or turbulence in the medium. Most mechanical meters read the average or integrated gross water movement past them and smooth out these irregularities. It is possible by analytical or electronic means to effect integration of the output Doppler meter; however, to do so suppresses the capability of this device to resolve instantaneous turbulent flow. In a highly turbulent flow where complex sidebands are produced, flow determination becomes difficult, but there is reason to believe that with a proper choice of integration time constants ele i akaatthanht alld FEMI VV YE C. BIS> Yo 60 FREQUENCY qIN cPs or b. - oe 2500 Tests at Lerner Marine Laboratory dock, Bimini, on 28 June 1962: (a) oscilloscope, single sweep, 1 ms/em, (b) Panoramic, 20 sweeps, h4-second tape loop, (c) Panoramic, 16 sweeps, continuous segment and (d) Panoramic, 60 sweeps, continuous segment. this instrument will produce no greater error than do other integrating types. Analysis of the frequency spectrum exhibits a predominant frequency which corresponds to the primary or average flow component and other side- band frequencies that are proportional to the product of the direction cosines and velocity of turbulent flow. The major contributing component is comparable to a carrier frequency and the turbulence to modulation sidebands. In order to exploit the instrument's sensitivity to turbu- lence it could be designed to operate in three planes with simultaneous velocity and direction sensing sections, suitable for tape recording, that would give the mean and turbulent components of motion in each coordinate direction. High 133 sampling rates can be used for obtaining an extremely fine structure. Tests in the towing tank, which represented an apparently homogeneous particle-free environment, confirmed the requirement for the presence of particulate matter for reflection of the acoustic beam. Tests in the ocean off Bimini demonstrated the ability of this instrument to respond to water particle motion induced by swells in an otherwise currentless environment. Another appli- cation to be explored is a 2-plane instrument mounted on a free drifting neutrally buoyant float. It would read only turbulence components because current flow would be cancelled by the float drift. Future investigations will include considera- tion of the characteristics of the reflecting particles and a rigorous mathematical analysis of the output signal. ACKNOWLEDGMENTS The authors wish to acknowledge the valuable assistance of Mr. Herbert Cook of Airpax Elec- tronics, Inc. who furnished the magmeter fre- quency meter, Mr. Hall Kaighin who designed and fabricated the transducer fixture head and Mr. Harvey Sachs who participated in the experi- ments. Work on this project was made possible by support of the Office of Naval Research, Contract Nonr840(01). REFERENCES 1. HORTON, J. W., Fundamentals of sonar, U. S. Naval Institute, Annapolis, Maryland, 1-18, 1959. UNPUBLISHED. 2. JENKINS, F A. and H. E. WHITE, Resolving Power of a Circular Aperture, Optics, McGraw-Hill Book Co., 2nd Edition, 1-316, 1950. 3. CHAPULNIK, J. D. and P. S. GREEN, A Doppler- shift ocean-current meter, Marine Sciences Instrumentation, 1, Instrument Soc. Amer., Plenum Press, New York, N. Y. 1962. 134 PRACTICAL PROBLEMS IN THE DIRECT MEASUREMENT OF OCEAN CURRENTS R. G. PAQUETTE Defense Research Laboratories General Motors Corporation Santa Barbara, California ABSTRACT The motion of water in the sea is a complex of turbulent, oscillatory, sporadic and steady movements with wide scales of size, velocity and period. All of these vary with wind, season and year and generally decrease in magnitude from the surface downward. Attention may be concentrated on various aspects of the motion by taking mea- surements which are time-integrated with appro- priate time constants. In many situations the integrated mean vector is much smaller than peaks of the instantaneous velocities and may have little correlation with the instantaneous direc- tion. Attempts to deduce mean velocities from measurements having too short a duration may lead to large errors. The current meter of most general utility would be one which can be kept in place continuously to record continually on ‘as short a time scale as is feasible. Current meters are subject to many errors due to undesired motions of the flexible supporting cable and the supporting platform. These errors are described and estimated. Existing current meters are of severely limited value for any but the grossest measurements when the suspensions must be long, relatively flexible or disturbed by waves. Periodic stray motions could be removed by vector-integration either in the meter or at a later date on the record, but with con- sequent serious deterioration in the representa- tion of real changes of similar period. Small Systematic errors accumulate to such important magnitudes in long integrations that meters hav- ing accuracy, rapid response and faithful inte- grating characteristics in a high degree are required. INTRODUCTION The direct measurement of ocean currents appears simple, perhaps because current meters are relatively simpie and because current meters are described in text books and in the litera- ture well dissociated from any mention of prac- tical field problems. It is to dispel this fallacious idea of simplicity in the minds of the many new instrument designers entering the field of oceanography that this paper is written. It will be shown that the measurements of cur- rents by means of conventional meters suspended from ships, buoys or other platforms are 135 seriously distorted in many ways due to the stray motions induced by the platform and supporting cable as well as the imperfect response of the meter to the resulting transients and to the real transients which exist in the sea. Some of the information to be presented has appeared inci- dentally and in a qualitative way mostly in scat- tered papers describing successful series of current measurements at sea. It is hoped that the present more unified and semi-quantitative discussion will be a stimulus to the design of equipment and techniques which will be useful in the era of unattended instruments which is now beginning. The discussion will treat specifically those current meters whose velocity element is a rotor but it may be applied with little modification to other types, such as those having pressure plates, tilting bodies or acoustic paths as sensors. The emphasis will be on depths and velocities typical of the open sea but will apply also to estuarine motions in which the problem has not become grossly simplified by the shallowness of the water. The problem of current measurement cannot be attacked in the abstract. Instead it must be faced in relation to conditions which exist in the sea and to that aspect of the complex motion which it is desired to study. For this reason the discussion will begin with a general descrip- tion of the types of motion in the sea in relation to the portion of the motion which the experi- menter may wish to extract. DISCUSSION Types of Motion in the Sea The net motion of water in the sea may be considered as a system of average established currents upon which are superimposed transients that may be slowly changing and with a size scale of thousands of miles. At the other extreme is the random motion of turbulence of a size scale smaller than we care to deal with now and a time scale correspondingly short. There are motions which are periodic, such as those of tidal period, inertial rotations and currents associ- ated with surface and internal waves. Others are more or less random, such as the wind driven transients and the various scales of turbulence. There are regions of strong currents, such as the Gulf Stream, in which mean surface velocities rise to several feet per second (fps). The super- imposed macro-turbulence has velocities with mag- nitudes a large fraction of the mean and direc- tions vary through the full rotation of the compass. In much of the ocean, established cur- rents have mean velocities of the order of a few tenths of a foot per second and there are large areas in which the mean may be measured in hun- dredths of a foot per second. In areas of weaker surface current the winds are often the chief disturbing influence and, since wind-driven water velocities may approach 2% or more of the wind velocity, a situation can exist in which the instantaneous velocities are many times larger than the mean and the direction of flow has little relation to the mean direction. Surface waves commonly have periods in the region of 4 to 25 seconds and heights reaching 50 feet or more. Heights of 6 to 12 feet are common. The particle velocity in the wave is quasi-sinusoidal and, at the surface, has a peak velocity equal to the velocity of a particle rotating uniformly in a circle of diameter equal to the wave height and making one complete rota- tion during one wave period. In a 12-foot wave of 10 seconds period, the particle velocity is then 1.27, or 3.77 ft/sec. This motion may affect near-surface meters directly or may intro- duce stray motion into the surface-floated sup- porting system. The particle motion decreases rapidly with depth in proportion to e-eT 2/L where Z is the depth and L the wavelength.1 Thus, for a decrease in depth equal to 1/9 of a wavelength, the orbital diameter decreases by one-half. For the 10-second wave in deep water the computed wavelength is 512 feet; the particle velocity would have decreased to one-half at 57 feet of depth and to 1/512 at 512 feet of depth. Under these conditions a meter or buoy even at a depth as shallow as 228 feet (4/9 of the wavelength) would experience an orbital velocity of only 0.23 knots. In the presence of a long high swell this would not be true. Water velocities are generally considered to decrease with depth, at least down to some depth at which velocities are relatively small. This is a plausible conclusion where the driving forces are the winds at the surface. However, there are circumstances in which the decrease with depth is followed by a reversal and an increase to velocities comparable to those at the surface. Two such cases are the Cromwell Current in the Equatorial Pacific and its counterpart the Atlantic Equatorial Undercurrent. Some driving forces, such as the tides, the moving atmospheric pressure disturbances or the long waves of tsunamis, make their relatively small contribu- tions nearly equally at all depths with magni- tudes of the order of 0.1 fps. There is always the possibility that such deep-acting forces may cause more important velocities by convergence into deep narrow channels between islands or between seamounts. Sporadically, in some areas having sufficient bottom slope, turbidity currents will flow along the bottom with velocities reach- ing 15 fps or more. Then there are the massive, but extremely slow, thermohaline circulations of the deep ocean water generated in the Antarctic and Arctic by the sinking of surface water made more dense by cool- ing and partial freezing. The velocities of these flows average at most a few hundredths and perhaps a few thousandths of a foot per second, but the few direct measurements which have been made in deep water have shown transient velocities of as much as 0.7 fps and directions having little rela- tion to the indirectly deduced mean. The Need for Direct Current Measurements For many years the velocities and patterns of average ocean currents have been deduced by two principal methods: (1) the statistical treatment of the reported deviations of naval vessels from their courses and (2) indirect computation by application of the geostrophic equation to the experimentally (and indirectly) measured density field. The first method is inaccurate and yields only surface currents. The second gives some information about mean velocities at all depths in deep water but fails near shore. The determin- ation of flows at great depth is strongly depen- dent on the choice of a depth of no net motion which still must be determined by methods which are theoretical and not well supported by experi- ment. Both methods produce means with a summing period of one or more weeks and give little or no information about transients of shorter period. A determination of mean velocities is useful and the geostrophic method has advantages but it is now necessary to check its validity by experi- ment and supplement the findings where the method is known to be weak. In the vast regions where transient velocities in the sea are many times greater than the mean it can be seen that many processes probably have little relation to mean velocity. Examples of such processes are the transport of sediments, the mixing of water masses and the transport of plankton and fish. The understanding of these processes awaits an ade- quate description of the transients. The Experimental Problem With the above outline of the complexities of motion in the sea it is now possible to look at the practical problems in current measurement. These will be considered in two senses: (1) the design of the experiment and (2) the design of measuring equipment and suspensions. The two elements in the design of a current measuring experiment which are most often poorly done are: (1) design of the experiment to Superior numbers refer to similarly numbered references at the end of this paper. 136 extract from the complexities of motion those com- ponents which are of interest for the purpose at hand and (2) sufficient distribution of the mea- surements in time and space. Attention often must be concentrated upon a particular component of motion because the entire field of motion is too complex for present understanding or because a current meter designed for one range of veloci- ties may not function in a range which is grossly greater or smaller. A current meter designed for measuring the particle velocities of waves may be of little use for general purposes and one which might measure the low velocities in deep water may not serve at higher velocities. Furthermore, One must understand what he wants. When near- surface measurements are made, either by drift bottles or with the GEK, the result obtained is sensitive to a few feet difference in the depth of immersion of the measuring device. There has been a tendency to dismiss as inaccurate those measurements made quite near to the surface, explaining the errors as due to "windage." writer believes that such differences likely represent real gradients in velocity with which we are not yet prepared to deal. We are more comfortable describing currents which represent greater volumes of flow or which are representa- tive of the forces acting on more deeply immersed objects like ships or buoys. The If a mean is desired then the period of mea- surement must be long enough to obtain a good mean and if the result is to represent a volume rate of flow over a large section, the complete section must be investigated in width and depth with due regard to the variations caused by weather and other variables. Too much wishful thinking has been done of the type which purports to describe a sine curve by means of one point measured at random. Exploration, of course, is legitimate and necessary. What is objectionable is any tendency to assign wide validity to isolated measurements. THE ERRORS IN CURRENT METERS HUNG FROM A SHIP OR BUOY To again clarify the limitations placed on this discussion it is emphasized that attention is being concentrated on current meters which can be suspended from a ship, buoy or other plat- form. For convenience, considerations are limited to meters whose velocity sensors are rotors but the same principles apply, almost without exception, to other types of meters Similarly suspended. Other devices such. as the geomagnetic electrokinetograph, parachute drogue and Swallow float are omitted. These still have their spheres of usefulness but at present they appear too slow, inaccurate or expensive (in terms of ship time) to solve the problem of mea- suring transient flows in the detail which will be necessary in the future. No attempt will be made to catalog the various types of current meters as this has been done by a variety of investigators.2)3, 137 It is scarcely necessary to point out that the current meter, even in ideal behavior, measures only the water motion relative to it so relative motion of the meter introduces an error. If the direction element fails to respond precisely another error results. These errors are discussed under two somewhat overlapping headings: (1) those errors which can occur without any motion of the supporting platform and (2) those due to motions of the support in company with failures of the meter and its suspension when exposed to rapidly changing flows. The latter have been called "dynamic errors." ERRORS WHICH CAN OCCUR WITHOUT MOTION OF THE PLATFORM The errors in current meters hung from a ship or buoy arise mostly from the motions of the platform bitthere are at least 5 sources of error even if the platform is fixed. These are: (1) distortion of the near-surface flow by the platform, (2) deviations of a magnetic compass in the meter by iron on the platform, (3) elasti- city and hence distorted response of the long suspension, (4) dynamic errors of the current meter itself and (5) indirect error which results from the error in depth due to wire-deflection when the meter carries no depth element. The first error may readily be demonstrated by simultaneous comparison of current meters near the bow and near the waist of a vessel, or of meters On opposite sides of a vessel, at depths less than perhaps 2 or 3 times the ship's draft. It requires little imagination to picture a variety of distortions due to the large body of the ship at various angles relative to the cur- rent. Such effects may extend laterally for distances of the order of one ship's length as one quickly discovers when using a drift-pole paid out on a measured line from the stern. The effects are particularly serious in small currents (less than one knot) when wind and the yawing of the vessel may make the vessel lie at a sizable angle to the current. In measurements made from a freely drifting vessel the distortions may be less serious unless there are important velocity gradients in the sea between the surface and twice the depth of the keel or the vessel is drifting rapidly with respect to the water. The deviations of a magnetic compass by iron on the ship are well known. Yet, meters con- taining magnetic compasses have been used blithely, close to the vessel, with no determination of the errors. Errors up to 12° in the indications of a current meter hung near the surface at the stern of the wooden vessel BROWN BEAR have been demonstrated. On steel ships the errors are much more serious. The writer has found errors of the order of 100° at a depth of one-quarter ship's length and 2 keel depths below the surface on a 275-foot ice-breaker. The effects of a long suspending cable are a time lag in the response to changes in current and an attenuation in the peaks of the recorded flow. A more important contribution to the same effect is introduced by the elasticity of the moorage. This subject as well as the dynamic behaviors of current meters will be discussed below under "Dynamic Errors." The problem of error in depth is more or less self explanatory. Its importance depends upon the gradient of velocity with depth and would be especially serious in situations where there are relatively sharp changes, as in the region of the Cromwell Current. DYNAMIC ERRORS The dynamic errors associated with current meters have been classified into 5 types: (1) those due to slow, more or less random movements of the platform, (2) those due to the elasticity of long suspensions and the elasticity and slack in moorings, (3) those due to dynamic failures in the meter itself which prevents the accurate following of rapid transients, (4) those due to pendulous or elastic-cord types of oscillation of the suspension excited by the rolling and heaving of the platform or by turbulence and (5) those due to vertical motions of the meter. Effects of Slow Movements of the Platform The effects of slow movements of the platform will be discussed first. Platforms anchored on the surface on relatively long and elastic anchor lines undergo a complicated cycle of motions even in constant currents. There is the oscillation about a center near the bow, called yaw, oscil- lation about a center near the anchor, called swinging, both superimposed upon a fore-and-aft (riding) movement due to cyclic tightening and relaxing of the anchor cable. The platform moves with each change of current and with each gust of wind. The current meter dangling below on a wire experiences these motions with some time delay and attenuation due to the elasticity of the sus- pension. In depths of 100 meters or less, on small vessels the writer has found yaw, swinging and riding nearly small enough to ignore at velocities of one knot. In deep water with larger ships and different weights and scopes of anchor wire this conclusion might well require modification. A long series of observations has been made on the ARMAUER HANSEN in depths of 1,800 to 4,000 meters. Cursory examination of these data shows fluctuations in the bearing of the anchor cable of the order of one or two degrees at water velocities approaching one knot, whereas in more common conditions when velocities were about 0.5 knot the scatter is about 75°. More extreme fluctuations occurred at lower velocities. Consider the spurious error due to yaw alone calculated on the assumption of a swing of 15° during a period of about 2 minutes at a point 100 feet from the bow of a 140-foot vessel and 138 assuming a sinusoidal variation in the velocity of yaw. It is found that the amplitude of yaw is t9 feet and the spurious velocity at the middle of the yaw is 0.25 fps. The use of a buoy as a platform will reduce the yaw greatly. Swinging and riding still remain, however, and it is likely that their combined effects are at least com- parable to the velocity calculated above. A strong and gusty wind blowing across the direction of current flow can severely accentuate the motion of the ship at low water velocities. Reduction in Dynamic Response Due to the Elasticity of Long Suspensions The discussions of this section apply equally well to the effects of rapid stray motion at the top of the current meter suspension and to the dynamic response of the meter and its long sus- pension to real transients in the flow. It will be shown that long suspensions buffer the current meter against short rapid movements of the plat- form but diminish the ability of the meter to record transients in the velocity at depth. The response of a current meter to a step-change in velocity is derived first. Table I shows the time required to attain 90% response to a sudden 20% increase in velocity for a typical current meter suspended on a light wire under a number of conditions. The initial wire angle also is shown to give some feeling for the depth error involved. The velocity of 3.0 fps at 10,000 feet is admittedly unrealistic, but the reciprocal situation of currents being measured from a ship drifting at 3.0 fps is quite a real possibility. Table I has been calculated by a gross simpli- fication of a complex problem. The meter area and weight have been lumped together with the area and weight of the lower third of the wire and assumed concentrated at the resulting center of area, at which point the water is presumed to act. The remainder of the wire is assumed weight- less and without drag. The waterforce has been taken proportional to the square of the slip velocity and the drag coefficient as unity. The meter and terminal weight together are taken to have a frontal area of one square foot and a weight in water of 150 pounds. The supporting wire is assumed to have a diameter of 0.1875 inch. Inertial forces are neglected and the velocity of the meter at any time is taken as the difference between the new water velocity, Vo, and the slip necessary to maintain the wire angle, 8 , at that instant. The slip is given by aL ia ® cos 9 (1) Table I. increase in velocity of 20%. Initial Velocity ft/sec 0.3 Wire Length V6 ity ft Degrees Sec 100 0.05 1.6 300 0.09 9. 1000 0.2 59 3000 0.4 450 10000 0.8 2600 and the horizontal velocity of movement of the meter by 9x -1.2 fi Oa Oy fire See. (2) de cos 8 cos 9 te) where k = 2a PCDA W = weight of meter + terminal weight + lower third of cable in water, lbs, p = density of sea water, 64 1b/ft3, Cp = drag coefficient, assumed 1.0, A = area of meter + weight + lower third of cable, ft", 8 = angle wire makes with vertical, radians, and oy =initial wire angle. The current meter is able to sense only the slip, which at time zero is the initial velocity, Vo, and approaches 1.2 v, asymptotically. To solve the above equation for time, t, dx is replaced by its equivalent L cos 9 d8 where L is the wire length. The time to attain a given wire angle is me ah cos 9 cols) (3) VEU% 1,2 an Q tan9 cos 9, cos9 Wire angle and times for 90% response in current meter subjected to a step 1.0 3.0 0.6 6 5.0 17 i1.O 28 8.5 83 2.0 193 16.5 420 4.8 1420 32.3 1340 8.6 8500 Yas 3120 139 For 90% response, 6 is chosen as the value cor- responding to v iL 18} Vo. This equation is integrated numerically for angles greater than Oe but at smaller angles it may be approxi- mated by the equation 2 ae cay Gage we) which, on numerical integration, gives 5 Doli Oe Vk (5) The times for 50% response are about 0.27 as great and for 95% response are 1.3 times greater. From Table I it may be concluded that there are severe restrictions on the ability to detect short term transients when using long suspensions. For placement of deep current meters one would prefer to fasten the meter in the span of a taut- wire mooring with a stiffness much greater than that corresponding to 150 pounds of buoyancy. The buffering offered to deep current meters against short term fluctuations of the surface platform is evident. It is of interest that the most important contribution to the area in the preceding calcu- lation comes from the wire itself. Table II shows the areas and weights of wire used in the pre- ceding problem. Since the times increase in pro- portion to the wire diameter, every effort should be made to keep the suspending wire as small as possible if excessive lag times and wire angles are to be avoided. Table If. current meter suspended on Length, ft 100 Weight lower 1/3 of cable, lbs 1.6 Area lower 1/3 of cable, ft@ 0.5 Cable weight and area contributing to the time constant of a 3/16-inch wire rope. 300 1,000 3,000 10, 000 4.9 16.3 kg 163 1.6 ha 15).6) 46.8 FIGURE 1 (b). PLATFORM ANCHORED _ IN DEEP WATER, SCOPE 1.1:1 REGION OF ELASTIC MOTION, 600 ft. RESTRAINT ON PLATFORM 600 + 252 ft. FIGURE 1 (a). IN SHALLOW WATER, SCOPE 3:1 PLATFORM ANCHORED RADIUS OF UNRESTRICTED Wale Abo Effect of Elasticity and Slack in the Mooring The mooring inevitably has elasticity; often by design in order to absorb shock loadings. When the mooring line is heavier than water the elasticity is in the catenary. If the line is synthetic fiber rope it is mostly in the elas- ticity of the material. Typical situations are shown in Figs. 1, 2 and 3 for 4 types of mooring: (1) a ship or buoy anchored in 300 feet at a scope (ratio of line length to depth) of 3:1 shown in Fig. 1(a), (2) a surface platform in 6,000 feet of water at a scope of 1.1/1 shown in Fig. 1(b), (3) conventional taut-wire moor in the same depth (Fig. 2) and (4) the taut- rope moor now being used by W. S. Richardson at Woods Hole Oceanographic Institution (Fig. 3). The diagrams for the first two are self- explanatory. In the taut-wire moor, the deviation of the submerged buoy will vary with the constants of the system. The wire angle has been estimated at 1° for a buoy of 500 pounds net lift and 140 REGION OF aes RESTRAINT, 600 + 2750 ft. Conventional mooring features of surface vessel platform. RADIUS OF UNRESTRICTED MOTION, 600 ft 12 square feet cross-section on 3/16-inch wire rope at an assumed velocity of O.5 fps when negli- gible drag is contributed by the surface float. This is a conservative assumption since it is often difficult to keep the drag of the surface float as small as that of the submerged float. Actual knowledge of the instantaneous velocities at 150 feet depth is sparse indeed. Frequently at some locations, and perhaps occasionally in many, the velocity certainly will exceed 0.5 fps. The degree of restraint on the motion of the sur- face float with respect to the buoy depends upon whether or not the upper mooring line is heavier or lighter than water. If distinctly heavier, a ecatenary forms and there is a corresponding restraint shortly after the surface platform moves away from vertical alignment with the submerged buoy. If the line is nearly neutrally buoyant there is little restraint until it is stretched taut. To obtain a feeling for the kind of distortion in the measured current due to slack and elasti- city, the simple case of a purely sinusoidal 100 ft. MOVEMENT (ELASTIC) Fig. 2. Water Sha ~ MORE OR LESS UNRESTRAINED RADIUS OF PLATFORM MOVEMENT 426 ft. Platform moored to submerged buoy at 150 feet in 6,000-foot water depth with upper mooring scope of 3:1. Richardson toroidal buoy on taut elastic propylene rope. ya WATER VELOCITY DEEP-WATER MOOR FRACTION OF PEAK VELOCITY 0 3 6 9 12 TIME, HOURS Fig. 4. Distortion of surface tidal current by elasticity and slack in the moorage. reversing current of semi-diurnal period has been considered. To suit the shallow water case the peak velocity has been taken as 1.0 fps while for deep water the more reasonable value of 0.5 fps is taken. Fig. 4 shows the kind of distortion to be observed for the two simple moorings. In the shallow water case the analysis is simple. Shortly after slack water the elastic extension of the mooring has been relaxed and the platform is free to move with the current for a distance equal to twice the slack in the mooring. In the simplified case there is no relative motion between platform and water during this period and the current meter registers zero. Rather abruptly the mooring line begins to tauten and the apparent velocity rises sharply. The rise is sharp because the elastic extensibility of the moorage is small compared to the total excursion of the water during one half- cycle (252 feet versus 13,770 feet). The measured velocity never does reach the peak velocity because some elastic extension of the moorage is continuing at this point. When peak velocity is past, the moorage begins to relax and moves the platform against the current to maintain the apparent velocity higher than it should be until after the true velocity has reached zero. The cycle then repeats in the opposite direction. In the deep water moorage the total water trans- port in one half-cycle is 6,885 feet and the elasticity amounts to 2,750 feet in addition to the 1,200 feet of unrestrained motion. The result is to attenuate the apparent peaks quite severely. The 1,200 feet of slack contributes its period of apparently zero velocity as before. The degree of attenuation and phase lag depend greatly on the stiffness of the mooring so the diagram in this case must be regarded as illustrative rather than in any degree quantitative. No attempt has been made to derive the more complex response of the taut-wire mooring since there has been so little standardization that it is more difficult to put plausible values on the constants of the system. However, the response will differ only in degree from those illustrated. If the submerged buoy has a large buoyancy com- pared to the drag forces on the lines and surface float, it is likely that it may be the most rigid system, even for a meter near the surface. For a meter at the submerged buoy or in the span of the taut wire it definitely has an advantage. No analysis of Richardson's system is offered since the extensive experimental results now forth- coming from current meters on this moorage will form a better basis for an analysis of errors. This system may compare favorably with the taut- wire mooring. It can be seen from the above that phenomena occurring when the moorage is slack are completely lost. Those velocities which occur during the elastic stretching of the moorage are attenuated and shifted in phase to a degree directly related to the ratio between the effective extensibility of the moorage and the linear transport of water during a half-cycle. Rotary motions also are subject to similar distortions. If the radius of the streamline of rotary motion is less than that of the stretched moorage the motion is lost, except for any effect of the drag of the cable on the bottom. If the radius of the true motion is somewhat larger only a fraction of the velocity and transport will be observed. A more hopeful situation exists if there is a relatively steady current, involving large trans- ports, upon which the fluctuations are superim- posed. If the moorage is stretched so as to remove most of its elasticity it will yield rela- tively little to fluctuations along the line of the moorage and these fluctuations may be fairly faithfully represented. Lateral components involving small transports, however, will be severely attenuated or absorbed. It is easily demonstrated that a moorage stretched to a radius of 1,000 feet under a current of 1.0 fps will respond to a small change of direction at a rate which will attain 90% of the change in 2,300 seconds or 38 minutes, during which time the direction record in the meter changes corres- pondingly slowly. All of the stray motions which have been described in this section are slow enough for the meter to accurately register the relative velocity with the consequent advantage that the integral of all cyclic transports is zero over one cycle so that the distorted velocities at least can be removed from the results. In the next sections effects will be treated which do not sum to zero and which can contribute to spurious determinations of long term transports or average velocities. ie Meter Response to Rapid Changes in Horizontal Motion Rapid transient horizontal motions, real and artificial, are presented directly to the current meter by wave motion or indirectly by wave induced platform movement or natural turbulence of the water. Current meters are subject to a number of spurious responses and response failures which can lead to some confusion. Again the effects are the most serious when the real currents or their means are small compared to the real or artificial transients. When a surface-floated platform rides in the waves it undergoes complex motions due to the waves. It is subject to the particle motion of the waves which moves the platform back and forth with wave period. This motion, combined with the restraint of the moorage and the rolling of the platform moves the current meter suspension back and forth or, more generally, in a small highly irregular loop, usually elongated. The effect is amplified if the meter is suspended from a boom extending far from the metacenter of a ship. This motion, insofar as it is transmitted to the current meter, is undesirable. If the meter responded accurately the motion could be recorded and integrated to zero. For various reasons the current meter does not respond accu- rately. The errors may be classified as due either to failure in directional response or non- ideal velocity response. Both will be treated as though the suspension were fixed and the water fluctuating but the treatment applies equally well to the reciprocal situation. Failure in directional response will be illustrated by reference to two general types of current meter. One has a propellor-like rotor intended to be exposed to the current from the front only and it carries a tail fin or other orienting device to turn the entire meter to face the current. The other has a rotor presumed equally sensitive to flow from all orientations, e.g., the Savonius rotor. Direction, if obtained, is derived from a light vane, of short time-constant, rotated with reference to an internal compass. In the first type of meter there is a definite limitation in directional response. Take for example a meter with a flat tail fin and the center of the fin at a radius, r, from its center of rotation. This meter is subjected to a step- wise reversal of velocity at magnitude, v, and lies with an angle between the source direction of the current and the projection of the meter axis through the tail fin. On a flat fin it may be assumed that the component of the velocity normal to the fin turns the current meter with no slip and no inertial forces. The normal com- ponent, of course, is zero when the meter is at 0° and 180° incidence, which is unreal because it leads to an infinite period of rotation, whereas the tail fin is certainly started on its way in these indeterminate regions by stray motion or turbulence. A more realistic estimate may be obtained by integrating the equation between limits of 3° and are cos 0.632; the latter a not NTT Fig. 5. The Ekman current meter. strictly justifiable simulation of a condition of 63.2% response. Then 1299 18 -o) ae = [me (6) Wet sing fo) and 129° a eee log, tan g = 3.6 aie (7) ae If we use a velocity of 1.0 fps and a radius, r, of 20 inches, corresponding to the Ekman current meter shown in Fig. 5, we obtain a time constant of 6.3 seconds. During part of the period water flows through the meter backward. In the Ekman meter reverse turns are subtracted but in most electrically registering meters the rotor has no directional discrimination and the registration of velocity is always positive. Current meters with small radii of rotation and those with hydrofoil sections for tail fins will have bet- ter characteristics. Particularly good is the Von Arx current meter® which uses two Garbell fins on either side of a cylindrical body and probably the most satisfactory device is a small sensitive direction vane such as used in the Snodgrass meter. A practical problem is to inquire what happens to a meter like the Ekman meter in the continu- ally oscillating currents in waves near the sur- face, which is similar to the case of a rapidly fluctuating suspension. The calculation requires only minor modification to that above but is sensitive to the boundary conditions. Here it is assumed that the meter will carry out a syn- metrical rotary oscillation reaching a maximum conformity to the current direction at the end of each half-cycle. The results of this calculation are shown in Fig. 6 for the Ekman current meter in a recipro- cating current corresponding to the particle velocity of 5-foot sinusoidal waves at the 143 WATER VELOCITY ORIENTATION OF METER FRACTION OF PEAK VELOCITY ANGLE BETWEEN METER AND CURRENT APPARENT VELOCITY 37 /9 7 15 2n TIME ANGLE Fig. 6. Response of current meter with tail fin to rapidly reciprocating current. surface. It is interesting that the result is independent of the period of the wave and depends only on the ratio of wave height to the radius of rotation of the meter fin. In Fig. 6 the apparent velocity is shown as negative when the water flows backward through the rotor. In this idealized case the Ekman meter would sum current flow to zero in one cycle whereas electrically registering meters would count all the flows as positive although the independently registered direction, if it were registered continuously, would give indication of a spurious result. Discrete registrations of direction could lead to confusion, depending on the frequency of sampling. Unfortunately, the real situation is worse since most meters are not equally sensitive to forward and reverse flows and the time-integral of vel- ocity will not be exactly zero even if direction could be properly incorporated. Rapidly respond- ing meters will behave much better. The negative excursions of the apparent velocity will be smaller and the orientation of the meter will be near 0° and 180° during a larger fraction of the eyele. Such meters, however, will not be immune to the accumulation of apparent flow registration due to their front-to-back asymmetry and the many times repeated oscillations. Integration of such cyclic fluctuations to zero over an integral number of complete cycles theoretically is possible if the velocity sensor is equally sensitive from all directions, the direction sensor extremely rapid in response, and the sampling sufficiently frequent. In the general case, in which fluctuations are rotary, it is necessary to carry out vector-integrations continually, either in the current meter itself or from the record at a later date. This requires that the velocity and direction be associated as vectors. If they are dissociated, it generally is impossible to integrate. Of existing current meters, the Snodgrass meter (Fig. 7) in its continuously recording form is closest to having the desired characteristics Fig. 7. The Snodgrass current meter. although the actual integration from the velocity and direction records would be impossibly tedious in practice. Richardson's modification of the Snodgrass meter, with its digital recording sys- tem, may provide a solution but it remains to be proved that the discrete sampling is being done at a sufficiently high frequency. No suitable internally-integrating meters are known. Errors Due to Pendulous, Elastic-Cord and Rotary Types of Oscillation Pendulum action is an effect which is not well documented. One worker reported observing unexpectedly high apparent currents at depths of the order of 100 to 200 meters which he believed due to such resonances. A calculation of reson- ant lengths shows a suspension 100 meters in length resonant to a 20-second period. However, the period of ship's roll is in the region of apn 5 to 10 seconds, corresponding to lengths of 20 to 80 feet, and one would anticipate a greater likeli- hood of resonances in this region. The increased damping of greater wire lengths also diminishes the likelihood of long-period resonances. Meters mounted in the span of a taut wire generally will be more stiffly supported and responsive to shorter periods. Experimental work would be desirable to detect such phenomena. Another form of error due to a short rapid pendulous motion is "jitter," a situation occur- ring when an electrically registering rotor is stalled or almost stalled on the contact. The pendulous motion causes the rotor to oscillate back and forth across the contact, producing rapid contact closures which are seen as rapid forward motion by the registering mechanism. Richardson has pointed out that a related but not so serious behavior occurs with current meters in his mooring systen. Apparently there is a rotary oscillation of the cylindrical current meter body about the rotor caused by surges which induce torque gradi- ents in the helically-laid supporting rope. For- tunately, the register that counts turns is in a form which permits the excess counts to be detected. There is a suspicion that such rota- tions may affect the reference compass and cause errors in direction. Richardson also points out that the Russian practice of mounting current meters on an arm projecting from the supporting cable must exaggerate errors due to cable rotation. Errors Due to Vertical Motion The errors due to up-and-down motion of the current meter arise from 4 causes: (1) asym- metrical water flow about the rotor generated by the body of the current meter, (2) direct sensitivity of unhoused rotors to vertical motion due either to front-to-back asymmetry in the pro- peller blades or to a form of turbine action which oceurs in horizontally oriented bucket wheels, (3) porpoising and (4) constant tilting of the current meter due to water drag which exposes a projection of the face of the meter to vertical motion. The first cause is difficult to avoid. In the Ekman current meter the propeller has been housed in a horizontal tube which undoubtedly removes much of the effect. However, the writer has directed tests in which meters of the Ekman type were cycled up and down through a 5-foot motion every 5 seconds to simulate the rolling of the BROWN BEAR. Under these circumstances the rotor ran backward at a rate corresponding to about 0.25 fps. The Ekman-Merz meter similarly treated ran forward at the equivalent of 0.61 fps. The Price current meter (Fig. 8) which has an unhoused bucket-wheel ran forward at about 2.2 fps in a similar test. A fraction of the effect prob- ably was due to porpoising. One would expect the Snodgrass current meter to be relatively insensi- tive to vertical motion since all aspects of the rotor and housing are nicely symmetrical with respect to such motion. Fig. 8. The Price current meter. The term "porpoising" refers to the tendency of the meter to nose upward when rising and down- ward when falling. This action occurs most severely on current meters which are mounted in trunnions and carry improperly designed horizon- tal tail fin surfaces. One meter that porpoises is the Price meter but it already is so sensi- tive to vertical motion that the additional effect is unimportant. Another in which the effect can be most serious is the modified Roberts current meter, one model of which both horizontal and vertical fins have several times the area of those in the standard meter of Fig. 9. Some horizontal fin is necessary in this instrument to balance the pressure moments of the large rotor during vertical motion but the writer has observed the meter to turn its nose nearly ver- tically downward and upward during moderate heaving of the vessel. The standard Roberts meter also exhibits porpoising to a significant degree. It is important to recognize that heaving motions may be transmitted along the supporting cable to much greater depths than are horizontal disturbances. , 145 The standard Roberts current meter. Fig. 9. CONCLUSIONS A Seaeiivicidiy pleas has been painted of our ability to measure ocean currents on a continuous basis with suspended current meters and existing techniques. The problems of stray motion and limited dynamic response are extremely serious. Generally speaking, small short period fluctua- tions will be difficult to find. The large, long period transients may be measured with fair accu- racy but probably not well enough to determine mean currents in those areas where the transients far exceed the means. There is a great need for a current meter of rapid response in both direc- tion and velocity which is insensitive to stray motion and will integrate accurately to zero all of the undesired cyclic motions. Of existing current meters the most promising is the Snodgrass current meter, in its continu- ously recording form, although it is far from perfect. Something will have to be done to con- vert the form of the record into one which either internally integrates or is easy to integrate later from the record. Richardson's modification of the Snodgrass meter appears to be producing useful information and solving many of the problems. Of existing mooring systems the stiff taut- wire moor with a small surface float serves fairly well the region from the submerged buoy downward, but not the upper section. A close competitor may well be the Richardson moor. It may be that early work with continuously moored systems in the more difficult areas may have to be done with bottom mounted meters. The present discussion has tended to emphasize situations which are serious because they are easier to illustrate. It is readily granted that in particular experiments much less serious con- ditions may exist. In such cases the experimenter must accept the burden of proving that his results are real and meaningful. REFERENCES 1. PIERSON, W. J., G. NEUMANN and R. W. JAMES, Practical methods for observing and fore- casting ocean waves by means of wave spectra and statistics, Publ. 603, U. S. Navy Hydro- graphic Office, 1-284, 1960. 2. BOHNECKE, G., The principles of measuring currents, Assoc. d'Oceanog. Physique, Union Geodesique et Geophysique Internationale, Pujol, Sen. Wh, U2, iG5 5). 3. JOHNSON, J. W. and R. L. WIEGEL, Investiga- tion of current measurement in estuarine and coastal waters, Publ. 19, California State Water Pollution Control Board, State Printing Division, Sacramento, Calif., 1-233, 1958. UNPUBLISHED. 4. THORADE, H., Methoden zum Studium der Meeresstromungen, Abderhaldens Handbuch der biologischen Arbeitsmethoden, Abt. ITI, Teil 3, 2965-3095, 1933. 5. FLEMING, R. H., The effect of ship's heading On current measurements, Paper presented at First Western National Meeting, Amer. Geophys. Union, Los Angeles, Calif., December 27-29, 1961. UNPUBLISHED. 6. VON ARX, W. B., Some current meters designed for suspension from an anchored ship, J. Mar. Res., 9, 93-99, 1950. 7- SNODGRASS, J. M., Prototype telemetering current sensor, Section in Progress Rept. to Bureau of Ships, SIO Ref. 55-10, Scripps Inst. of Oceanography, January 1955. UNPUBLISHED. 8. RICHARDSON, W. S., Personal communication, 1962. 146 A FAST RESPONSE CUP ANEMOMETER FOR MEASUREMENT OF TURBULENT WIND OVER THE OCEAN R. G. STEVENS and L. F. SHODIN Woods Hole Oceanographic Institution Woods Hole, Massachusetts ABSTRACT A fast response cup anemometer suitable for measuring turbulent winds over the ocean is described. The electrical characteristics of this instrument are suitable for use at remote observing stations where low power drain and telemetry are desirable. Dynamic response charac- teristics of the cup anemometer are discussed together with an ingenious method for dynamic calibration. INTRODUCTION It should be remarked at the outset that mea- surement of wind is a proper concern of the oceanographer since the wind serves to transmit considerable mechanical energy into the ocean and strongly influences other processes such as evapo- ration and mixing. Im fact, wind driven phenomena such as large current systems and wind generated waves are certainly among the most spectacular features of the ocean. Before discussing the details of the cup ane- mometer it is worthwhile to examine some aspects of the measurement of dynamic variables in ocean- ography, particularly since it is becoming increasingly fashionable to deal with dynamic measurements by use of digital computers. It should be apparent that meaningful measurement of a variable implies a rather long term observa- tion of its time history. If the time history is interpreted in terms of its corresponding fre- quency spectrum it is found that oceanographic variables rarely contain frequencies above 10 cps. A more usual upper frequency limit would be 2 cps for phenomena near the surface, while the prac- tical low frequency limits for many phenomena may be measured in cycles per day or month. Obviously, low frequency data of this sort may be processed conveniently and economically on a general purpose digital computer as contrasted with, say, acoustic data which may be analyzed with so-called analog devices. On the other hand, new technological capabilities have so vastly expanded the capabilities for acquiring large quantities of oceanographic data that the digital computer becomes an essential element in the interpretation, cataloging and analysis of data. Now, once the commitment is made to use a digital computer in data analysis the computer becomes, essentially, a part of the "instrument" itself. This fact can be important in the design and evaluation of the transducer and data acquisition system since it is frequently possible to incor- porate instrument corrections into the computer program. Thus it is possible to apply corrections for nonlinearity, temperature compensation, etc., to the data in the digital computer. This pro- cedure may serve to vastly simplify the trans- ducer design or (because extremely complex correc- tions may be made using a digital computer) it may be possible to use transducers that would otherwise be unsuitable. Cup anemometers have been a standard device for measurement of wind speed for many years. In recent years, however, their use has been extended to the measurement of turbulent wind in micro- meteorological research. While the cup anemometer is a rugged and reliable device, serious ques- tions have been raised regarding its suitability for accurate dynamic measurements in turbulent wind regimes. In particular, the linearity of response to increasing and decreasing wind speed has been questioned. The anemometer described here was designed for use in conjunction with a study of wind generated ocean waves. Since it was intended to subject the wind speed data to power spectrum analysis, the dynamic characteristics, particularly in regard to linearity of response, had to be known. This is particularly important since any instru- ment nonlinearity must be removed before calcu- lating the covariance function and the power spectrum. However, as will be shown, the actual performance of the cup anemometer did not exhibit the anticipated difficulties. Nevertheless, it seems worthwhile to point out the necessity for eareful evaluation of transducers intended for dynamic measurements and the possible means of compensating for undesirable transducer charac- teristics. DESIGN SPECIFICATIONS The following criteria and specifications were used in the design and ccastruction of the anemometer : Woods Hole Oceanographic Institution Contribution No. 1330. 1h7 1. Wind speed range of 1 to 30 mph. 2. Fast dynamic response which requires minimum inertia and friction of all moving parts. Frequency response to 1.5 eps. 3. Design should be rugged, weatherproof and capable of withstanding winds up to 60 mph in the presence of rain and salt spray. The electronic circuitry in particular should be stable over a wide range of environmental CSMIDSE ULE with an extreme range of -20° to +60~C. 4, Electrical characteristics: (a) Minimum current drain since anemometers may be battery operated. (b) Output should be compatible with standard TRIG FM subcarrier oscillators, prefer- ably 0 to 5 volts DC. (c) Provision should be made for automatic field calibration of all electronic circuitry. Consideration of these specifications led to the construction of a compact anemometer, com- pletely self-contained except for a 2 volt power source and field calibration equipment. In order to minimize inertia and friction the moving parts were reduced to the shaft which sup- ports the cups and a slotted cylinder which acts as a light beam interrupter. Several slots are cut in the light "chopper" to give better resolu- tion at low wind speeds and to enhance the dynamic resolution. Mechanical Construction The mechanical layout of the anemometer is shown in Fig. 1. The entire assembly is 13 inches overall and the main instrument case is 2 inches in diameter. The anemometer cups are press fit into a hole in the upper cap. No retaining screw is needed. The upper cap is locked onto a stain- less steel shaft which is supported by two minia- ture precision ball bearings. The upper cap, which rotates with the shaft, forms the outer portion of a double weather seal. There is an inner cap, fixed to the case, inside of which is a spinner which in turn rests on the inner race of the upper bearing and which makes a loose con- tact with the shaft. A second cap is locked with set screws to the shaft at the lower bearing, just inside the instrument case. This is the light chopper which projects downward and over the lamp housing. It has 30 slots 0.01 inches wide milled parallel to its axis. The lamp housing consists of an aluminum tube surrounding a low voltage incandescent bulb and having a single slot 0.02 inches wide. A plastic ferrule fastened to the lamp housing serves to mount 2 solid state photoelectric switch. The photo- switch is placed in line with the slot in the lamp housing about 3/32 inch distant, while the chopper rotates betwem the lamp housing and the photodetector mount. Thus as the shaft rotates, light passing the lamp housing slit is alternately 148 interrupted and passed by the slots in the chopper causing the photoelectric switch to alternately open and close an electric circuit. The instrument case is fabricated from aluminum tubing. The lower end plug and the case extension which supports the shaft and bearing assembly, are machined from aluminum and are fitted with "0" ring seals, being secured to the tubular case with small machine screws. The lamp housing, photo- detector mount and the electronic circuit boards are mounted on 3 small rods which are in turn fastened to the lower end plug. A 4-pin her- metically sealed MS type electrical connector is threaded into the lower end plug. The assembly is finished with 2 coats of white epoxy paint which not only provides corrosion protection but also acts as a reflective coating to prevent the electronic components from becoming overheated when exposed to sunlight. ELECTRONIC CIRCUITRY The essential features of the electronic cir- cuitry are shown in the functional diagram of Fig. 2 and Fig. 3 illustrates one of many workable circuits capable of performing the basic func- tions. The photo-sensitive switch manufactured by Solid State Products, Inc. of Salem, Mass., and called a Photran, is of special interest since it functions much like a switch with high conductivity when exposed to light but with very high resistance when dark. Once the device con- ducts, however, it continues to do so until the source of current is nearly spent. This accounts for the high value of 1 megohm in the Photran circuit. The pulses derived from the Photran charging the 0.001 mfd capacitor appear on the one-shot multivibrator as differentiated pulses of a couple of microseconds duration. The one shot shapes these into pulses of constant width and amplitude, the amplitude being determined by the zener diode. Since the pulses are of con- stant amplitude and width and only change in fre- quency, they may be integrated to produce a voltage proportional to the incident wind velocity. Field calibration of the electronics is made by extinguishing the lamp and introducing standard frequency square waves which correspond to the light chopper frequency for a given wind velocity. The temperature environment limits of the instru- ment are essentially those of normal transistor tolerances when used in a one-shot multivibrator and the Photran, being a silicon device, is capable of equal or better performance over a wide temperature range. Power consumption of the instrument is approximately 1/70 milliamperes, the rated current of the No. 313 lamp. The rest of the circuitry adds very little to this drain and depends somewhat on the frequency of the pulses generated. IG asi Son NRT “rt SVN RV Se RR apples STATIC CALIBRATION Wind tunnel tests were conducted at the Round Hill Field Station of the Department of Meteor- ology, Massachusetts Institute of Technology, to evaluate the performance of the Beckman- Whitley cups and the mechanical portions of the anemometer. The results that were used in the circuit design are shown in tabular form. They are averages for the } anemometers tested. 149 Anemometer assembly. Miles/hr Meters/sec Pulses/sec Rev/min 30 13.411 260.00 520.00 10 4.470 86.67 173.34 iL, 0.447 8.66 17.30 No departure from linearity could be detected on wind tunnel tests. However, allowance had to be made for a certain amount of background noise eee) FUNCTIONAL COMPONENTS PHOTO SWITCH > romry | LIGHT CHOPPER GME SHOT INTEGRATOR TYPICAL PULSE TRAIN PULSES a cA BURR Babee ul 002 MSEC >| be | MSEC >| 270 MSEC baie STS ive TRANSMISSION LINE DEMODULATOR 6 CPFiLTER Fig. 2. in the wind tunnel which required taking rather long term averages to determine mean pulse rates. Average pulse rate is 19.38 pulses/sec per meter/ sec or 0.646 rev/sec per meter/sec. Since the distance from the center of rotation to the out- side edge of the cups is 9.0 cm, the ratio of cup tip speed to wind speed is 0.366. This means that the outside edge of a cup will travel 0.366 meters while the air moving past the cup travels 1 meter. The threshold velocity for four anemometers varied from 0.5 to 0.9 meter/sec. The lowest threshold velocity was observed on an anemometer which had been in outdoor service for over a year. Assuming that the cup rotation rate is linearly proportional to the wind speed, as indicated above, the remainder of the electronic circuitry may be evaluated by substituting a square wave generator for the photoswitch output. The integrating circuit (Fig. 2) is an essential 150 Bes Funetional diagram of anemometer circuitry. part of the anemometer circuitry since the pulse output of the one-shot multivibrator must be smoothed somewhat before it can be used to modu- late the subcarrier oscillator (VCO). Both the integrating circuit and the VCO are nonlinear circuit elements in this application. The VCO output frequency is normally a linear function of the input voltage. However, since there is considerable ripple present on the integrator output voltage and because the magnitude and fre- quency of the ripple also vary with the pulse rate, the output frequency of the VCO is not strictly proportional to the average input voltage. The RC-integrating circuit is, of course, inherently nonlinear. Fortunately, the two effects tend to oppose one another so that the frequency output of the subcarrier oscillator is a linear function of the pulse repetition rate with a maximum deviation from the least squares straight line fit of 40.5% of full scale. The voltage output across the integrating circuit as measured with a DC voltmeter has a maximum deviation of 42% of full scale from the least squares straight line fit. +24 PHOT RAN D I8V F ZENER ateuee 1/4 WATT TANTALUM ANEMOMETER CONVERTER SQUARE WAVE GENERATOR Fig. 3. An electrical circuit design for converting anemometer output. DYNAMIC CALIBRATION One way to evaluate the overall response of any transducer is to subject it to a step change When observing a cup anemometer in the labora- of the variable to be measured while observing tory it is immediately noticeable that the cups the output from the transducer. The technique is will accelerate rapidly in response to a sudden highly developed for purely electrical measure- draft of air but will continue to coast for a ments and much of the existing technique may be relatively long time in still air when the applied directly in evaluating transducers of driving force has been removed. This raises the various kinds. The apparatus for simulating a possibility that the angular acceleration of the step function in the wind tunnel is shown sche- cups in response to an increasing wind velocity matically in Fig. 4. A variable speed motor differs from the deceleration in response to a mounted on a hinged support drives a circular decrease in wind velocity of the same magnitude. puck on the end of a long shaft. The puck may Furthermore, it can be expected that the angular be engaged against the anemometer upper cap by acceleration of the cups will be larger; that is, forcing the motor to swing upward by means of a the cups will respond faster to the same magni- push bar extending through the floor of the wind tude of velocity change at higher mean wind tunnel. The angular velocity of the cups may velocities. Thus, it is at least possible that now be regulated by adjusting the motor speed to the dynamic response of the cups is dependent be either greater or less than the wind speed. on both the mean wind velocity and the sense of Upon release of the push bar the motor and puck velocity change. assembly is pulled rapidly back to its retracted position. Thus, so far as the anemometer is con- cerned, it experiences a realistic step change WSL a CA / UA / i / / i | | ----+f-+ + [heges Poe ene 1 ie a i a ud | Fig. 4. Apparatus for producing simulated step function response in a wind tunnel. in wind speed as soon as the puck loses contact with the anemometer cap. This procedure was carried out for several values of mean wind, for increasing and decreasing step functions of various amplitudes. The sur- prising result was that the response character- istics were uniform for all wind speeds from 2 m/sec to 10 m/sec for both increasing and decreasing step functions. The observed response was, in fact, an exponential function with a time constant identical to the RC time constant of the integrating circuit, which is 0.27 sec. This result holds true, even at the extreme con- dition, for a mean wind of 2 m/sec with a te m/sec step function. This result is both encouraging and disap- pointing. If the pulse output had been recorded as well as the integrated output much more could have been learned about the cups themselves. The requirement for a smoothed voltage into the telemetry system forced the inclusion of the integrating circuit in the anemometer. While the integrator has masked the response of the cups themselves this is not entirely without virtue since the observed dynamic response of the anemometer is uniform over a wide range of mean wind speeds. Hence, within the limits imposed by this relatively slow response, the instrument has a single and simple dynamic response. The only question remaining is whether 152 or not the response is adequate for investigation of turbulent wind. FREQUENCY RESPONSE As mentioned earlier, one of the more useful means of presenting turbulent wind data is the power spectrum. The reason for desiring to use a frequency scale in preference to a time scale for data presentation is that certain manipula- tions of the data are easier to perform in the frequency domain and frequency data are almost always easier to interpret. The time constant when transformed into the frequency domain beeomes the frequency response function. If then, the time constant can be experimentally deter- mined and if the response to a step function is (approximately) exponential the frequency response function W(f) may be written UG wv /2 w(t) = 2 = [as(oree)2] (1) where f is the component frequency in cycles per second of the turbulent wind, U_ is the observed amplitude at frequency f, U is the "true" ampli- tude at frequency f and k is the time constant. This shows that the amplitude, U, is attenuated proportionally to 1/kf so that at higher fre- quencies the observed amplitude, U,, becomes smaller. For power spectra the quantity of interést is we(£) and this function may be used to correct the observed power spectrum for the attenuation due to the instrument's time con- stant by the formula Up (£) we(£) e) U(f) = Fig. 5 shows a plot of W-(f) for k = 0.27 superimposed on a portion of a typical power spectrum Up" (£) calculated from observation of wind over Buzzards Bay, Mass. Wind velocity was measured with one of the anemometers described here over a period of about 10 minutes. The velocity signal was sampled 5 times per second so that the highest frequency computed in the power spectrum is 2.5 eps (only a portion of which is shown in Fig. 5). Since the observed power spec- trum attenuates very much more rapidly than the frequency response curve it can be assumed that the frequency response of the anemometer is high enough so that important high frequency components in the wind are not cut off. However, if the observed spectrum is corrected for the effect of the time constant according to Eqn. (2)some modifi- cation of the spectrum does occur with the possi- bility of a secondary peak occurring near f = 0.4 eps. Whether or not such a peak has real meaning can only be determined by careful con- sideration of the statistical reliability of the power spectrum reinforced by the experimenter's own prejudices and preconceived notions as to what he is looking for. 100 LEGEND @----FREQUENCY RESPONSE FUNCTION O—TYPICAL WIND POWER SPECTRUM ° (@--COMPENSATED POWER SPECTRUM LOG (VEL) (ARBITRARY UNITS) i 0 200 400 600 800 1000 1200 1400 FREGUEINGY (MILEICVYCLES/ SEC) Fig. 5. Typical power spectrum of wind speed measured over Buzzards Bay, Mass. ACKNOWLEDGMENTS this paper were made possible by the Office of Naval Research Contracts Nonr2734(oo0) and The authors wish to acknowledge the contribu- Nonr3351(00). tion of Mr. Harlow G. Farmer, now of the Univer- sity of Washington, Seattle, Washington, who initiated the design and early development of the anemometer. The developments reported in 153 — Ee CSE eee ie NN BAL one i sk re er sie : oo | Ee, i i ieee ‘ ae vik eae On Le Acie eeu as ts “a : ier 0) Gee lee eee j trio, oye at 5 aa rcwey pe tie RE ikea) cay ¢ nea SOLION ELECTROCHEMICAL DEVICES J. L. COLLINS Defense Research Laboratory University of Texas Austin, Texas ABSTRACT The solion is a new class of electrochemical devices which act as current limiting diodes, signal processing elements, pressure or flow detectors, etc. Of particular interest is the solion linear pressure transducer or linear flow detector, with its wide range of possible appli- cation at the lower frequencies normally associ- ated with oceanography and geophysics. A discus- sion of the principles of solion operation will be presented along with some of the practical applications of solions in the field of oceano- graphic instrumentation. INTRODUCTION The science of electrochemistry began around 1800 when Galvani noticed that if two dissimilar conducting materials were placed in contact with a freshly prepared frog's leg, the leg twitched as if alive. Although the solion does not util- ize a freshly prepared frog's leg, it does util- ize some of the principles noticed in Galvani's experiment. The solion is a very low power consumption electrochemical device, commonly known as a redox system. With a redox electrode the reaction occurring at the electrode is completely rever- sible. Furthermore, the electrodes consist of an unattackable metal, usually platinum, which will not enter into the reaction itself. The electrochemical system consists of an electrode set immersed in a solution containing soluble forms of the same chemical in two different oxi- dation states. For the case of the solion this is generally an iodine/iodide electrolyte system. The name solion is derived from the phrase "ions in solution." Electric current is trans- ferred via ion flow in a solution as opposed to ion flow in a gas or solid. The redox electro- chemical system is discussed in this paper, but other solions utilize different electrochemical principles such as electrokinetic transduction. THE SOLION ELECTROCHEMICAL DIODE Consider the case of a solion diode. This will illustrate the basic characteristics of the DC MILLIAMP METER —(1) DC VOLTMETER CATHODE ELECTRODE ANODE ELECTRODE MOLDED AND SEALED PLASTIC HOUSING ELECTROLYTE SOLUTION Fig. 1. The basic electrochemical diode system. solion, not including those effects due to hydro- acoustic flow. Fig. 1 is a schematic of the basic electrochemical-diode system. The diode consists of a pair of platinum electrodes sealed into an airtight chamber that has been filled with an iodine/potassium iodide/water solution. One elec trode, the anode, is chosen with an effective sur- face 10 times that of the other electrode, the eathode. The chamber is constructed of a chemi- cally inert plastic material such as Kel-F. The external electrical circuit consists of a variable low voltage, DC supply, a DC milliamp meter and a high impedance DC voltmeter. With the circuit connected as in Fig. 1, the voltage between the electrodes is increased and the current through the cell is monitored. The voltage-current relationship is shown in the con- eentration polarization curve of Fig. 2. Maximum voltage is about 0.9 volts DC; any substantial inerease above this value will tend to cause hydrogen evolution at the cathode. This curve illustrates one of the basic features of the solion--the output current is independent of the applied voltage over the range 0.1 to 0.9 volts. Typical value of the slope of the plateau is approximately 1 megohm. Over the region of the plateau, the limiting current is given by the following relationship:~’ Superior numbers refer to similarly numbered references at the end of this paper. 1.2 [ TYPICAL SLOPE ~ | MEGOHM Co ie) @ TYPICAL SLOPE #100... 0.6 Co fo) aS T ELECTRODE CURRENT - MILLIAMPS fo) (2) Cael 0.2 1 L n el % 02 04 06 08 1.0 12 ELECTRODE BIAS— VOLTS The concentration polarization response of a simple solion diode. ig = a EN (1) where A is effective area of cathode electrode, N is iodine concentration or normality, 2 is effective thickness of diffusion layer, n is 2, number of electrons involved in reaction and F is Faraday constant, 96,500 coulombs/g-mole. The diffusion coefficient, D, is given by (2) olf where T is absolute temperature in degrees Kelvin, o is viscosity of the solution and K is a con- stant for the particular system. Therefore, for a solion diode the limiting current is primarily determined by the cathode area (a cathodic con- trolled reaction), the iodine concentration and the absolute temperature. The electrochemical reactions occurring at the electrodes consist of reducing iodine at the cathode and oxidizing iodide back to iodine at the anode. The process is illustrated by the following: At the cathode: I, + 2e > 217 (3) At the anode: eT > Ip + 2e (4) The electrode material itself has not entered into the reaction and the completely reversible process can continue indefinitely. The shape of the "knee" of the curve can be adjusted by varying the potassium iodide concentration but 164 EXTERNAL LEADS CATHODE BUTTON CATHODE ELECTRODES Solion flow detector. Gye 3), the shape illustrated makes greatest use of the constant current feature. Another unusual feature of the solion is the "double" source impedance characteristic. Although the "signal" output appears to originate from a 1 megohm source, the cell appears as a 100 ohm source to any 60 cps "pickup" that might appear in the external electrical circuit. "Reverse" voltage characteristics of the diode are very similar to those shown in the curve of Fig. 2. The anode and cathode leads are now interchanged with the cathode becoming an elec- trode of 10 times the previous area. The polari- zation curve will again have a similar shape except that the limiting current will be approxi- mately 10 times the "forward" limiting current, due strictly to the increased area of the cathode electrode. The "front to back” ratio is there- fore primarily determined by the ratio of the electrode areas. Values in excess of 100:1 have been obtained. THE EFFECTS OF HYDROACOUSTIC FLOW The diode has been discussed to illustrate some of the basic principles of the electro- chemical system utilized in the solion. The assumption was made that hydraulic flow was absent. By proper design, the solion can be utilized as a hydraulic flow or pressure detector. Consider the design of a solion, similar to that indicated in Fig. 3. The rigid plastic body of the diode has been replaced with a cell of different design. The cathode element is now mounted in a solid web at the center of the plastic housing so that any fluid flow between chambers is possible only through the cathode electrodes. The electrolyte solution is con- strained by compliant diaphragms which permit limited volume flow between chambers. Since the cathode structure tends to obstruct any flow of fluid between chambers it is con- sidered as an acoustic resistance, R, measured in acoustic ohms. Since the diaphragms are com- pliant, this effect is considered as an acoustic compliance, C, measured in acoustic farads. The product of the acoustic resistance and compliance determines the low frequency response of the transducer. If the flow detector is connected to the external electrical circuit shown in Fig. 3 and if flow through the cathode elements is assumed to be zero, the individual cathodic currents will tend to a quasi background current as the iodine ions in the neighborhood of the cathodes are depleted. The background current is then deter- mined by Eqn. (1). If the two cathode electrodes indicated in Fig. 3 are identical, their back- ground or “no flow" currents will be identical. The differential "output" will then read zero or near zero volts, depending upon how nearly identical the cathodes actually are. The iodine ions contained in the volume in and around the cathodes are reduced to near zero except for the small amount of iodine diffusing into the region. Assume now that, with the cell connected to the external bias, a net differential pressure is developed between the compliant diaphragms. A net hydraulic flow of the electrolyte solution will commence from one chamber to the other chamber. The amount of volume flow is determined by the net differential pressure and the acoustic resistance through the cathodes. This is illus- trated by the classical relationship? ~ dv Ap =R a (5) where Ap is the net differential pressure, R is the acoustic resistance of the cathode and dV is the volume flow rate. The volume flow rate is related to the pressure by R and can be a linear or a nonlinear relationship depending upon whether R is constant or some function of pLR(p)]. At low frequencies (f<1 cps), pressure and volume flow rate are related by R, but at higher frequencies the relationship must become |Z| so as to include the acoustic inertance of the fluid in the flow path. A linear flow detector has a useful upper frequency response in the 30 to 50 cps region. As flow commences through the cathode elec- trodes, electrolyte at the bulk iodine concentra- tion is forced into the region of the cathode electrodes. The output current is not limited to the diffusion currents of Eqn. (1) but increases in relationship to the number of iodine ions per unit time arriving at the cathode. If the cathodes behave as linear detector electrodes, all the iodine arriving at the cathodes is reduced. The linear detector cathode, therefore, furnishes an output current which is linearly proportional to the volume flow rate. This is shown by = dv = TS Bihere 3: ers (6) 165 LOAD LINE SLOPE 750. ELECTRODE CURRENT — MILLIAMPS (0) 2 04 06 08 1.0 1.2 ELECTRODE BIAS — VOLTS Fig. 4. Output characteristics of a typical solion linear pressure detector. where I is the electrical current in the external cathode circuit. From Eqn. (5) a substitution for the volume flow rate gives T= yn ( 22 ) x 1073 , (7) During the flow cycle discussed the electrolyte passing through the "upstream" cathode has been depleted of its iodine ions and only dilute elec- trolyte arrives at the "downstream" cathode. The downstream cathode has its small background cur- rent reduced even further since the dilute solu- tion flowing through this cathode tends to over- come the iodine diffusing in from the bulk solution on the downstream side. The net effect is to cause an increase in the external electrical current associated with the upstream cathode, while the electrical current associated with the downstream cathode remains small or even decreases. The resulting voltage developed across the 2 resistors in the external load is such that one output lead becomes positive with respect to the other and, in the case described, this differen- tial voltage is proportional to the applied dif- ferential pressure and preserves the phase as well as the amplitude. A return to zero differ- ence pressure lets the output difference voltage return to zero volts. A reversal of the pressure causes a reversal in the polarity of the output difference voltage. Output characteristics of a typical linear detector are illustrated in Fig. 4. The load line has been adjusted so that at some hypotheti- cal maximum pressure the voltage between the cathode and anode is always greater than 0.1 volts. In this manner, the solion always operates in the proper region as a constant current device. For Table I. Parameter Cathode acoustic resistance Current sensitivity Pressure threshold Frequency range Dynamic pressure range Background power consumption Maximum signal output power Operating temperature range Maximum temperature coefficient of sensitivity Maximum size - diameter thickness Maximum weight a typical solion linear detector, operating in the linear region, it is interesting to note the volume flow sensitivity. If an iodine concen- trate of 1.0 normal iodine is used, from Eqn. (6) it can be seen that a flow of 10°° ec/sec will produce an output current of 100 microamps. This offers an extremely sensitive flow detection capability. A comment should be made regarding one other characteristic of the linear flow detector. If an excessive differential pressure is applied to the transducer the flow rate exceeds the linear ion reduction capacity of the cathode electrode. This does not cause the output current to "clip," but causes the output current to increase as the square root of the flow above the linear range. The flow signal is not "lost" but simply modi- fied. In this manner pressures far in excess of the linear range can be monitored and even mea- sured with proper corrections of the output signal. Linear flow detector transducers have been built with a wide variety of parameters. These parameters are given in Table 1. Although a wide range of values is indicated, all combina- tions of extremes are not obtainable in a single detector. NONLINEAR TRANSDUCERS The flow detector just discussed is a so- called "linear" flow detector transducer. It operates on two principles: (1) linear hydraulic flow and (2) linear electrochemical response. By proper design of both the acoustic (flow) Range of parameters of solion linear flow detectors. Range of Values Units 1o#-107 acoustic ohms 0-300 microamp/dyne/em= 0.01 to 100 dynes /em= 0.0001 to 30 cps 1:1 to 30,000:1 -- 10 to 1,800 microwatts up to 27 milliwatts -10 to +30 Co 42.5 %/°C 3}.(0) inches 0.75 inches 8.2 ounces 166 system and the electrochemical system, various nonlinear relationships can be obtained. Eqn. (7) indicates an output current that is linearly related to pressure. If the acoustic circuit is modified to become nonlinear with pressure, say by inserting an orifice of proper design, then R is not constant but becomes a funetion of pressure: R = R(p). Eqn. (7) also assumes that the cathode elec- trodes are behaving as linear detectors. By proper design of the cathode electrode structure, various nonlinear ion reduction responses, such as the square root response mentioned, can be obtained. With a combination of both acoustic and electrochemical nonlinearities, a wide range of pressure-current relationships can be obtained such as square root, linear, exponential, square and higher powers (powers up to + have been measured). By removing an anode from one bulk fluid chamber and replacing it with a scavenger cathode electrode, an output current sensitive to flow in only one direction can be obtained. Therefore, it becomes a rectifying pressure or flow detector. Other unusual effects and responses can be obtained although some of these effects are undesirable and difficult to remove from a desired response. APPLICATION OF SOLIONS Because of its extreme sensitivity at low frequencies, along with the stable properties of a redox electrochemical system, the solion is well suited for application to the low frequency problems associated with geophysics and/or ocean- ography. A few examples of how solions have been utilized in these fields are listed here. Microbarograph By choosing a "backup" pressure reference chamber with the proper acoustic "leak," very minute atmospheric pressure changes (of the order of 0.01 microbar) can be detected within the fre- quency limits of the system. If this sensitivity is excessive, a less sensitive transducer should be chosen or an acoustic resistor should be inserted in series with the transducer. A linear operating range of 10,000:1 would place the maxi- mum pressure limit around 100 microbars in the case described. Measurement of Vertical Pressure Gradient in the Atmosphere or Ocean By coupling the solion flow detector to a long rigid wall pipe, mounted in a vertical (or any angle desired) position, dynamic differences in pressure within the frequency and pressure response of the transducer can be measured. Again, 0.01 microbar of pressure change is detect- able. Although changes in pressure (the dynamic pressure) can be detected, static pressures can- not. Oceanographic Bottom Pressure Transducer Again, by choosing a proper backup chamber (pressure changes must be measured with respect to some constant reference pressure), extremely small changes in bottom pressure can be Beuscnee: Pressure changes of the order of 4 x 10 inches of water are theoretically detectable, so long as the changes lie within the limits of the system's frequency response. Housings have been designed which physically protect the solion and allow pressure measurements in the 1.0 eps to 0.001 cps frequency range. A Solion Seismograph If the solion linear flow detector is coupled to a long, horizontal column of fluid, say mercury, and if the column is equally divided about the solion (the same amount of column length on each side of the transducer), there is no net differential pressure across the solion and its output is zero. Output will be obtained when a shock signal, containing acceleration components within the frequency response of the solion system,is directed along the length of the column. The pressure developed (and, by Eqn. (7), the output current) will be propor- tional to the length of the column of fluid, the density of the fluid and the acceleration com- ponent along the length of the column. Detection of accelerations of the order of 10° g is readily possible. 167 Detection of Small Oscillatory Water Currents in the Ocean A device to detect small oscillatory water eurrents has not been constructed but limited investigation indicates the possibility of designing such a transducer. Fluid flow through a Venturi tends to create a pressure effect that is proportional to the velocity of fluid flow. A compound Venturi system coupled to a solion transducer, with one Venturi "pointed" 180° away from the other, could furnish a directional, low velocity water current transducer. The same effect also appears possible with the pitot tube, but a "full wave'' response would be more difficult. CONCLUSIONS These are only a few of the many possible applications of the solion. Solions have been used extensively to measure ocean bottom pressure signals in the study of the harbor seiche problem. These transducers have been in operation for approximately three years. Solions are also being used in the study of low frequency atmospheric noise. Reliability and stability have been excellent. Some of the unusual characteristics of the solion are: (1) low power consumption, (2) remote operation capabilities, (3) high sensitivity, (4) low pressure threshold capabilities, (5) very low frequency response, (6) broad-band response at low frequencies, (7) a reasonable temperature coefficient that can be corrected with thermistors, (8) excellent stability and reliability and (9) surprisingly rugged construction, except for diaphragm puncture. ACKNOWLEDGMENTS This work was sponsored by the Navy Department through the Bureau of Naval Weapons and the Office of Naval Research. REFERENCES 1. WILLARD, H. H., L. L. MERRITT and J. A. DEAN, Instrumental Methods of Analysis, D. Van Nostrand Co., New York, N. Y., 2nd Edition, 1951. 2. KOLTHOFF, I. M. and J. J. LINGANE, Polarography, Interscience Publ., New York, N. Y., Qnd Edition, 1952. 3. OLSON, H. F., Dynamical Analogies, D. Van Nostrand Co., New York, N. Y., nd Edition, 1958. vt" i enue whe wat , haa cant pe er fatty aA ait ine spell N eas Fas dallas - dash Boman: ty Sihen ST Spresi Bec ttt canna! seams iiprloal occreroml Daerah ish eb aah wily Wy Sis Bent ® «ie Lipa ai eee Wake pibvel f yo Crewe cnt Mae ip ie i/ ) fad a Lap yent ase ~ ig aia so ibn ic nde en pain Ene: & ee ee INFLUENCE OF HYDROSTATIC PRESSURE ON COMPONENTS M. FLATO U. S. Naval Research Laboratory Washington, D. C. ABSTRACT After a cursory investigation into the effects of high hydrostatic pressure upon components used in electronie circuitry, a more thorough look has been made in certain areas. Components which are satisfactory for short term tests have succumbed after longer periods under pressure. In the last reporting! it was shown that a test block of thin-walled aluminum cylinders embedded in plastic withstood a short term test. In the present test series the same block was subjected to 6 months duration. One cylinder failed and the second was reduced in diameter along its length. This test indicates the "cold flow" effect upon plastics which in turn can be transmitted to components. The pressure tank used for these investiga- tions has been modified to vary temperature as well as pressure. It is eegeele of lowering tem- perature to approximately OC. LONG TERM PRESSURE TESTS Tn the earlier report of pressure testst it was mentioned that a long term test was in progress. This test was devised to determine the thickness of potting material necessary to pro- tect components such as transistors and capaci- tors which have internal cavities and voids. Fig. 1 shows 4} very thin walled aluminum tubes cast Fh calm a eae . into a plastic block. The specimens were placed at 1/32 inch increments from the surface of the block. The two closest to the surface (1/32 and 1/16 inch from the surface) imploded at about 2,500 psig; the other two, more deeply embedded, successfully withstood the full 10,000 psig pres- sure. The end of the test block containing the two tubes that had not been crushed in the origi- nal pressure test was placed in a small pressure tank where the pressure of 9,000 psig was main- tained for 6 months. The samples were removed from the tank at the completion of the long term test and its condi- tion was observed to be as shown in Fig. 2. The most obvious change was the failure of the cylin- der situated 3/32 inch from the surface. Con- siderable damage occurred, however, at other places in the sample. The remaining cylinder (situated 1/8 inch from the surface) was reduced in diameter over a portion of its length. In addition, a depression immediately over the axis of the cylinder had occurred and was measured to be 0.025 inch deep (averaged). When the sample was first removed from the tank a bubble noticed prior to the long term test was almost invisible and on the underside of the plastic casting was a depression 0.046 inch deep. In the first few hours after removal from the tank, the tiny bubble increased in size and forced the oil out through a crack in the plastic. The photograph shown in Fig. 3 was taken 5 days after the test block was removed from the pressure tank. Fig. 1. pressure of 10,000 psig. Hollow aluminum cylinders in plastic and exposed to short term hydrostatic Superior numbers refer to similarly numbered references at the end of this paper. Hollow aluminum cylinders in plastic as photographed 2 hours after removal from pressure tank following 6 months exposure to 9,000 psig. Hig. 22 From this test we know that long term pressure tests must be integrated into the overall environ- mental simulation tests. It is essential to determine the characteristics of various resins and fillers with respect to cold flow if success- ful operation of potted circuits without heavy metallic pressure housings is to be achieved. In a 2-week test a number of resistors and capacitors were placed in the tank under 10,000 psig and at room temperature. The resistors tested were of a type in which a glass encapsula- tion is fusion sealed around a tin oxide resis- tance element. Sixty percent were unaffected by the test, 20% leaked but still read correct resis- tance, 10% developed encapsulation cracks but still read correct resistance and 10% broke and were completely unusable. The capacitors tested were of a ceramic type and their capacitance, in most cases, had decreased enough to cause concern at the conclu- sion of the 2-week test. These same capacitors were placed into a long term pressure container and the pressure held for 2 months. When the capacitors were removed from the test most of the capacitors had increased their capacitance. What seems most unusual is that the capacitance increased more in the 2-month test than the capacitance decreased in the 2-week test. Mea- surements were again made after the capacitors had 2 days in the atmosphere to recover. The larger valued capacitors returned almost to their original value but the lower values were still reading high although they did show some recovery. An important discovery in this test is that all components must be tested carefully under many conditions. An instrument designed and con- structed with certain of these ceramic type com- ponents would have become unreliable sometime after lowering. 170 Fig. 3. Hollow aluminum cylinders in plastic as photographed 5 days after removal from pressure tank following 6 months exposure to 9,000 psig. HOLLOW VESSEL TESTS A typical test method for determining the external burst pressure of an air filled vessel is to place it in a fluid filled pressure tank and build up the pressure until the test vessel implodes. This technique does give the desired answer but usually cannot show how or where the failure began. A method which is now being used at the Naval Research Laboratory does not take the test vessel to complete destruction. The unit is filled with fluid and a small, say 0.025 inch inside diameter tube is inserted into the vessel (Figs. 4 and 5) and led out of the pressure tank. This tube is led into an upright or inclined manometer tube, open at the top. As pressure is applied on the test vessel, it will shrink slightly, forcing the fluid up the glass tubing. This gives a dynamic, instantaneous and accurate measure of the test item's deflection. If the deflection is measured and plotted against pressure as the test proceeds then material yield points are easily recognized and the test stopped if desirable. On the other hand, if carried to burst pressure, the resulting implosion is stopped far short of complete col- lapse, due to the back pressure occurring within the test vessel, since the thin 0.025 inch tubing will not allow the inside fluid to be squeezed out fast enough. A view of the deflection in an aluminum sphere is shown in Fig. 6. Other samples have been tested using this technique and analyses conducted. Figs. 7 and 8 are pictures of Pa ie, Mh Aluminum sphere with manometer needle attached. Haterer Fig. 6. Aluminum sphere of 0.094 inch wall thickness and 3.0 inch diameter erushed at 5,300 psig (deformation began at 4,900 psig). Ialf25 0 To Fiberglass cylinder with manometer needle attached prior to testing in pressure tank. Fiberglass cylinder crushed at 10,200 psig. 171 Fig. 8. Fiberglass cylinder segmented for inspection after failure at 10,200 psig. segmented sections of fiberglass cylinders which were stopped short of complete implosion. Inspection of these segments have shown where and how the failures occurred and where the designers should give added attention in building a more reliable container. TEMPERATURE CONTROLLED PRESSURE TEST FACILITY To simulate more fully the deep ocean environ- ment one of NRL's pressure facilities has been modified to include the capability of cooling the tank and the enclosed fluid. The tank is cooled by a small unit with a 3500 BTU/hour capacity. The cooling coil is tightly wrapped to the out- side of the pressure tank as is the thermal con- trol bulb. Flexible hoses connect the cooling unit (mounted on the pressure tank enclosure) to the cooling coil. The cooling unit of this system is capable of lowering the temperature of the pressure vessel from 75°F to 32°F in about 1 hour and 40 minutes. The pressure tank is 5 1/16 inch ID and 6 7/8 inch OD with an inside length of 20 inches. The volume of the tank is 410 eubic inches and usually contains paraffin oil. A tank heating unit is also incorporated to bring the pressure tank temperature from its cooled state to room temperature in a reasonably short period of time. The heating unit is a 750 watt surface heating element from a domestic electric range. The element is firmly in con- tact with the base of the pressure tank. The surface heating element can return the tempera- ture of the tank from 35°F to 55°F in 2 hours and 50 minutes. Both the cooling and heating elements are thermally insulated from the sur- rounding atmosphere. 172 The temperature of the contained fluid is sensed with an iron-constanan thermocouple. The thermocouple junction is located on the longi- tudinal axis of the pressure tank 1 inch above the bottom. A thermoelectric device maintains the thermocouple reference junction at 32 F. At present, the fluid used in the 15,000 psig pressure system is paraffin oil. This oil becomes sluggish near 32°F. It is desirable to use an oil whose viscosity is fairly constant from 32°F to 85°F. Some synthetic oils exhibit this desirable trait but their coefficient of compressibility is too great. The fluid which embodies both of the above desirable character- istics is water but some other features are not so advantageous. DISCUSSION AND CONCLUSIONS With every instrument discussed at this Sym- posium there is probably a common unknown; namely its accuracy when operating in the real ocean environment. The program at NRL is aimed at testing and evaluating oceanographic sensors in a simulated deep ocean environment. At this time only the parameters of pressure and temperature are cousidered. Unless there is an organized and concentrated effort to design and produce secondary standard instruments and test instru- ments under controlled simulated parameters it seems apparent that some doubt must be associated with any measurements taken in situ. A wide range of tests have been undertaken to learn more of the influence of hydrostatic pres- sure on components. Many of the tests described deal with areas of the utmost concern to instru- ment designers--criteria for selection of elec- tronic components to be used in long term submer- gence and effectiveness of encapsulating materials used to contain electronic circuits and seal them from hydrostatic pressure. System designs could be improved and production of prototypes accel- erated if a clearing house for this kind of infor- mation was readily available and taken advantage of. This would minimize duplication of experi- ments and component manufacturers could also improve product design through application of this knowledge. REFERENCES 1. BUCHANAN, C. L. and M. FLATO, Influence of a high hydrostatic pressure environment on electronic components, Marine Sciences Instru- mentation, 1, Instrument Soc. Amer., Plenum Press, New York, N. Y., 1962. NUCLEAR DIGITAL TRANSDUCERS J. L. HYDE Charter Laboratories Division of Charter Wire, Inc. Northport, L. I., New York ABSTRACT Measurement transducers whose signal output is a pulse rate, derived by modulation of the radia- tion from a nuclear source, are an unconventional type now being developed in this laboratory. Analog to pulse rate conversion is performed in the sensing head essentially by mechanical means. A digital readout is obtained by means of a pulse counter. These transducers can be designed for measure- ment of any physical variable that is sensed through a mechanical displacement. Two develop- mental models are described and test results are presented. Proposed applications include digital pressure and temperature transducers and a digital bathythermograph. INTRODUCTION This paper describes our development work on measurement transducers based on the inherently discrete or digital properties of radioisotope decay. It had seemed apparent that transducers of this nature would permit physical data to be digitized substantially at the source, thus pre- serving the initial sensing accuracy during telemetry and data processing. It had also appeared that a transducer could be built which, once calibrated during assembly, would retain its accuracy indefinitely. These impressions have been reinforced by our experience to date. It now seems likely that the special features inher- ent in these devices may be especially attrac- tive in the design of measurement transducers for the marine sciences. The basic principle of a nuclear digital trans- ducer is so unsophisticated that it is not entirely clear why somebody did not try to build one long ago. Two of the most plausible reasons are that digital data were not in any great demand and that silicon junction nuclear par- ticle detectors, which are particularly suitable for these transducers, were not previously avail- able. Therefore, it is unlikely that a useful device could have been built until quite recently. Several types of pulse count transducers, for which a time base is generated in order to accumulate a digitized total count, are now available. Among these are: (1) tachometers and flowmeters (including anemometers and ocean cur- rent meters) of the "toothed wheel" or revolution counting type, (2) the vibrating wire type of pressure transducer and (3) DC analog transducers in combination with a voltage controlled oscil- lator which generates a frequency analog. Nuclear digital transducers represent a fourth type. FUNCTIONAL DESCRIPTION A digital transducer has been defined! as a coupling device which passes data along in some form of discrete code to an unlike system or sub- system. Ina nuclear digital transducer, the data are transmitted by means of nuclear particles or by electrical pulses derived from nuclear par- ticles and the coding is accomplished by counting a@ variable but predetermined fraction of the dis- integrations occurring in a radioactive source. Most nuclear transducers, including various types of thickness and density gauges, are analog in nature owing to the fact that a DC analog signal is developed by pulse integration within the detector. When nuclear radiation intensity or radiation flux density is measured by a Geiger counter, either a DC analog or a pulse rate analog can be used to indicate the radiation level. In the latter case, when a pulse counter including a time base is used for readout, the combined instru- ment is similar to a nuclear digital transducer. In other applications where the nuclear digital method might have been used, for example in some of the thickness gauges, the DC analog method seems to have been preferred. A block diagram of a nuclear digital trans- ducer is shown in Fig. 1. Particles from the nuclear source pass through the variable aperture controlled by the shutter and reach the detector where they generate electrical pulses. These pulses are counted by a conventional type of digital counter over a preset time base. The mechanical sensor controls the size of the aperture through which the particles pass. In the case of a pressure transducer this may be 4 bourdon tube or bellows. In the case of a Superior numbers refer to similarly numbered references at the end of this paper. 173 temperature transducer it may be a bimetallic strip. Im either case a displacement is gen- erated by the mechanical sensor which controls the shutter aperture, this displacement in turn being determined by the measurand. The sensing head is a pulse rate transducer; strictly speaking, the instrument is not a com- plete digital transducer unless a counter and time base generator are included. The sensing head is also a displacement transducer in which analog to pulse rate conversion is performed essentially by mechanical means. The output sig- nal is a representation of the physical variable to be measured in which all the information is contained in the pulse rate. NUCLEAR SOURCES The nuclear disintegration rate of a radio- active source is absolutely independent of tem- perature. Physico-mechanical effects inside the source such as thermal expansion which might alter the observed counting rate are negligible. DIGITAL COUNTER NUCLEAR RADIATION DETECTOR NUCLEAR VARIABLE SOURCE SHUTTER MECHANICAL TIME BASE MEASURAND SENSOR GENERATOR Fig. 1. Block diagram of nuclear digital transducer. RATE PULSE PULSE HEIGHT ALPHA SPECTRUM Fig. 2. Therefore, the data are effectively "frequentized" at the source by non-electronic means and subse- quent pulse recovery is almost completely insensi- tive to electronic circuit stability. Alpha particles are preferred to betas for use in this transducer. The reasons are apparent from the source energy spectra shown in Fig. 2. Alpha particles are essentially monoenergetic and their differential spectrum shows a single peak. The tailing off of this peak is due to slowing down of a few alpha particles when they escape from the source. The peak due to electrical noise in the detection equipment is extremely well sep- arated from the alpha peak and the discriminator setting for elimination of the noise pulses is not at all critical. This means that a simple discriminator is sufficient to insure that each alpha particle reaching the detector generates one, and only one, pulse. In the case of the beta spectrum shown in Fig. 2 the discriminator setting would be extremely critical and also would prevent a large fraction of the beta particles from being counted. This arrangement does not allow very good counting stability. Gamma rays also are not preferred because, being very penetrating, their use would require an inconveniently heavy shutter and aperture. The variable aperture arrangement used in our experimental transducers preserves the spectrum of the source so that the alpha pulses are much larger than the noise pulses and pulse discrim- ination is very simple. This would not be true NOISE RATE PULSE PULSE HEIGHT BETA SPECTRUM Source energy spectra. if the alpha particles were to pass through some absorbing material on their way to the detector, thereby distorting the alpha spectrum. Nuclear decay is characterized by randomly spaced pulses having approximately a Gaussian frequency distribution. The statistics eistfelen= ated with nuclear counting are well known. A statistical error in each accumulated count exists such that the,standard deviation of N counts is equal to N°. This means that a trade- off exists between accuracy of measurement and sampling rate. For example, if the measurand is to be sampled once per second and if the required readout CR ey is 0.1% of full scale, then at least 10~ basic counts must correspond to full scegle and the transducer must deliver at least 10~ pulses per second to the counter or scaling eircuit. An interesting corollary of these nuclear counting statistics is that for a fixed sampling time the transducer readout error, expressed as a percentage of full scale, decreases as the scale reading decreases. In other words, more accurate readings are obtained at the low end of the scale. This contrasts with conventional readout devices whose error is practically con- stant at all points on the scale. There should be no hazard problem with respect to nuclear sources for these transducers. Very weak license-free sealed alpha sources have been satisfactory for our experiments to date. If much stronger activities should be needed, the entire source chamber can be sealed. We have used Radium D sources (the actual alpha emitter is Poe40) which have a half-life of 20 years. In any final design a compromise between source life and data rate, consistent with a clean alpha spectrum, would be necessary. If the source chosen for a particular application should be relatively short-lived, the decay can be compensated either by adjustment of the time base or in the course of data processing. SILICON DETECTORS Silicon junction nuclear particle detectors34 are particularly useful in nuclear digital trans- ducers that employ alpha particles. They are relatively insensitive to gammas. The detector-limited pulse rise time, which may range from 2 to 50 nanoseconds? should be approachable by using very fast electronics. At least one manufacturer markets detectors having a 6 nanosecond rise time. If the 6 nano- second figure is accepted as a reasonable limit for the rise time, the minimum practical resolving time of silicon junction detectors is about 20 nanoseconds, leading to a maximum useful random pulse rate of about 107 per second with *As used in the formula no = — a corresponding dead time correction, nt, of approximately 20%*. At the present time a more conservative estimate of the maximum pulse rate would be 10% per second. The nonlinearity cor- responding to the required dead time compensa- tion® can be built either into the counter elec- tronics or into the mechanical modulator as may be appropriate. The transistorized preamplifier (Fig. 3) used with the silicon junction detector is a modifica- tion of a circuit described by Chase, Higinbotham and Miller of Brookhaven National Laboratory, ¢ to which we have added an extra stage of amplifi- cation and a pulse inverter. The input circuit contains a charge integrator so that the voltage height of the output pulses is proportional to the detector and therefore is proportional to the incident particle energy. OTHER DETECTORS Scintillation detectors or gas filled propor- tional detectors theoretically could be used in nuclear digital transducers. However, since either of these types would be bulkier and its resolving time longer than silicon junction detectors, no further consideration will be given to these types. Geiger tubes are the simplest and cheapest detectors available and may be interesting for use when economy is a prime consideration. How- ever, since Geiger tube pulses are exactly the same for all types of radiation, any required discrimination of radiation must be accomplished by variations in shielding and in detector design. Furthermore, the use of Geiger tubes, which typically have resolving times of 20 to 1,000 microseconds, would limit the maximum pulse rate to an undesirably low value for many transducer applications. It is possible that silicon junction detectors may be capable of operating in an avalanche mode similar to a Geiger counter, thus resulting in pulse amplification in the junction. This effect should be most readily observable at cryogenic temperatures .3 If solid state avalanche detectors should ever come into use, and if they should be capable of room temperature operation with reliable performance, they might be preferable to Geiger tubes for use in low performance types of transducers. TELEMETRY Counters with electronic scaling decades com- monly provide a binary coded decimal output suitable for telemetering. Therefore, if the counter is located ahead of the telemetry link, all the advantages of PCM telemetry are available. PCM telemetry is most suitable when several transducers are to be sampled serially and = where no is mean number of events occurring, n is mean number of events recorded and tT is counter recovery time. 2N393 DETECTOR 2NI500 2NI091 -6V 2) OUTPUT Fig. 3. Preamplifier schematic diagram. PREAMP DETECTOR SHUTTER SOURCE Fig. 4. Seetion of sliding shutter transducer. Fig. 5. Photograph of sliding shutter transducer. 176 telemetered over a single channel. If the most efficient use of one channel for a large number of transducers is required, each nuclear digital transducer in such cases will require a separate counter. When the data rate is low and bandwidth is not a limiting factor, the counter may be used in the system following the telemetry link. In this ease the original transducer pulses can be telemetered, in a manner analogous to CW radio- telegraphy, or by frequency shift keying. Although pulse rate telemetry is extravagant of bandwidth, it may offer considerable attraction from a cost standpoint when the transducer is to be expendable. MECHANICAL CONFIGURATION A cross-section of the sliding shutter trans- ducer is shown in Fig. 4. In this transducer the aperture opening is controlled by the setting of a micrometer screw. The range of setting from a fully closed to a fully open aperture is 0.100 inch and the open aperture is square. A photo- graph of this transducer is shown in Fig. 5. A corresponding transducer with rotary shutter is shown in Fig. 6. The full shutter range corresponds to less than 20 degrees rotation of the drum dial which is read by a vernier. A photograph of this apparatus, with the preampli- fier circuit board visible, is shown in Fig. 7. The chamber which contains the source and detec- tor may be evacuated through a valve when desired in order to eliminate air scattering and energy spectrum distortion. 'O RING DETECTOR Be / Fig. 6. SHUTTER Cut-away of rotary shutter transducer. CHARTER LABORATORIES, yoo = aK « 4 J 100, 10% 3 , 3 off 30) sOK Yaa CYCLE wey = PRESET Tose wes s MANUAL J MODE P. ’ POWER HS. SOURCE PEDESTAL NortHPoRt, tt. My. 3 Fig. 7. Photograph of rotary shutter transducer. 4 CHARTER LABORATORIES Fig. 8. Fig. 8 shows a photograph of the rotary shut- ter transducer connected to a digital counter readout. This counter is modified so as to include an infinitely variable time base gen- erator so that direct transducer readings may be obtained in terms of any units desired. For example, the sliding shutter transducer can be made to read the micrometer screw setting directly in thousandths of an inch. EXPERIMENTAL RESULTS Shutter plots obtained with the sliding shut- ter and rotary shutter transducers are shown in Figs. 9 and 10, respectively. The straight lines obtained in these plots demonstrate the highly linear characteristics of nuclear digital transducers. It is also possible, by shaping the shutter and aperture, to produce a nonlinear characteristic. For example, it might be desirable to compensate an otherwise nonlinear system in this manner. 177 Experimental transducer with counter readout. The slight scatter of the experimental points in Figs. 9 and 10 is a result of nuclear counting statistics.2 Use of stronger sources or longer counting times would greatly reduce the scatter. If each accumulated count had been 100 times as large, for example, so that all numbers shown on the ordinate scale were multiplied by 100, then the percentage standard deviation of the trans- ducer readout would be only 1/10 as great and the points would lie still more closely on the line. Tt should be mentioned that the data shown were taken with very weak Radium D-E-F sources con- taining only 1 to 5 microcuries of activity. Weak sources were used in these experiments inas- much as a high data rate was not required. In transducers designed for practical use, con- siderably stronger sources would be desirable. In order to test the immunity of nuclear transducers to temperature changes, the sliding shutter transducer shown in Figs. 4 and 5 was placed in the refrigerator at 40°F. After thorough cooling a plot similar to Fig. 9 was fo) DIGITAL TOTAL PER 30SEC (10° COUNTS) yo fp oO AN ®@ oO 10 20 30 40 50 60 70 80 90 100 SHUTTER OPENING (10 °INCH) Transducer readout graph - sliding shutter. Fig. 9. taken with the door being opened only briefly in order to change the micrometer settings. This plot was completely indistinguishable from Fig. 9 showing that a temperature change of this magnitude had a negligible effect. ACCURACY In considering the accuracy of nuclear digital transducers it is necessary to distinguish the mechanical sensing accuracy from the purely statistical readout accuracy. the readout of shutter position can be obtained to any desired statistical accuracy simply by accumulating a sufficient number of counts. This means that a nuclear digital transducer is an infinite resolution device. The mechanical sensing accuracy of the trans- ducer is limited entirely by the mechanical sensor. For example, although bourdon tubes of good linearity and low hysteresis are available, these sensors still have certain deficiencies which should be capable of measurement by means of a nuclear digital transducer. These trans- ducers, due to their infinite resolution and low friction, may offer an improved means for Stated differently 178 (or) ro) o aS DIGITAL TOTAL PER 60 SEC ( 10° COUNTS ) De) 2 5 (0) Sis azo SHUTTER ROTATION ( DEGREES) Fig. 10. Transducer readout graph - rotary shutter. measuring the performance limitations associated with various types of mechanical sensors. If the transducer is to measure a changing variable, the accumulated count will represent an average over the time interval determined by the time base generator. This automatic averaging feature may be desirable in certain applications for the purpose of eliminating the effect of rapid signal fluctuations. ; APPLICATIONS In discussing applications of nuclear digital transducers, we shall first mention the obvious one consisting of an analog-to-pulse rate or an analog-to-digital converter based on a D'Arsonval meter movement. Such a device is illustrated in Fig. 11, in which the detector, shutter and aperture are visible. The shutter is attached to the moving coil of the meter movement. This type of converter might be advantageous for use when digitizing equipment having long term stability and accuracy is required as, for example, in an untended remote data station. SHUTTER SOURCE DETECTOR APERTURE Fig. 11. Analog-to-digital converter. A pressure transducer configured for under- water use is diagrammed in Fig. 12. tight compartment with "0" ring seals contains the helical bourdon tube which operates the shutter. With this type of bourdon tube there are no extra mechanical linkages; one end of the helix is attached to the pressure port and the other end to the rotary shutter. A corresponding configuration for a tempera- ture transducer, utilizing a bimetallic ther- mometer movement to operate the rotary shutter is shown in Fig. 13. A good bimetallic movement is capable of a 3-second time constant. A some- what shorter time constant may be obtained by using a temperature sensor of the type employed in an ordinary mechanical bathythermograph. The latter sensor consists of a long thin tube full of toluene that expands with rising temperature and operates a bourdon tube. The remaining figures show a proposed appli- eation of nuclear digital transducers in a digital bathythermograph. Fig. 14 shows a block diagram of the underwater unit which contains a coupling unit to permit multiplexing the sig- nals. The single conductor performs 3 functions; it carries power down to the underwater elec- tronics, pressure and temperature signals up from the transducers and DC control pulses down to the underwater unit for in situ calibration. For checking the transducer calibration, a DC control pulse gates a generator which operates an electric switching device in the underwater unit, thus simulating zero and full scale transducer readings. Such simulated readings can be obtained in either of two ways: (a) by use of an auxiliary source and detector in fixed geometrical relationship or (b) by rotating the shutter so as to fully open or fully close the aperture. An underwater unit, containing these components in a housing similar to the standard electronic bathythermograph The pressure- 179 LEGEND: A= SOURCE CAPSULE B=PARTICLE DETECTOR C= SHUTTER D= HELICAL BOURDON TUBE E = OPENING TO SEA PRESSURE F= ELECTRONIC COMPARTMENT G,H= O-RING SEALS |= WATERTIGHT CABLE CONNECTOR Fig. 12. Pressure transducer. LEGEND: A=SOURCE CAPSULE B= PARTICLE DETECTOR E= PRESSURE TUBING '/g 0.D. F= ELECTRONIC COMPARTMENT C= SHUTTER G,H= O-RING D= BIMETALLIC THERMOMETER l= WATERTIGHT CABLE MOVEMENT CONNECTOR Fig. 13. Temperature transducer. specified by the U. S. Navy Oceanographic Office, is illustrated in Fig. 15. A block diagram of the deck unit is illustrated in Fig. 16. The digital equipment includes two counters with variable and programmable time bases to give readouts directly in temperature and depth units, and a digital printer or other recording device. The analog equipment includes two simple integrating circuits and an X-Y recorder. The test control panel shown at the top of Fig. 16 is used to actuate the in situ calibration system described above. The deck unit would be equipped with records for making both analog and digital records. Although the above discussion has been limited to the application of nuclear digital transducers in a bathythermograph, these devices are also being evaluated for other applications. Inasmuch as the cost of the sensing heads is expected to be moderate for pulse rate transducers of excel- lent long term stability and high accuracy, their suitability may depend upon system require- ments as to sampling rate or data rate. It is apparent that increased readout accuracy can be obtained at the expense of sampling rate and vice versa. COUPLING UNIT PRESSURE HC FILTER ZENER REG- |_|LOW PASS TRANSDUCER-{—| ricren | ULATED SUPPLY | | FILTER &PRE-SCALER eo a 2 2 CHANNEL Hosta t Pitas | — ER 3 TEMPERATUREL CY fF PLEXER TRANSDUCER[ [Jy BATTERY &PRE-SCALER [\_J[ -6VDC AND SWITCHING SIGNAL SINGLE CONDUCTOR CABLE SEA WATER RETURN TO DECK UNIT Fig. 14. Block diagram of bathythermograph underwater unit. HARNESS WITH WATERTIGHT CONNECTOR SS SSP SS es | roo = COUPLING PRESSURE TEMPERATURE UNIT TRANSDUCER TRANSDUCER Fig. 15. Bathythermograph underwater unit. 180 TO UNDERWATER UNIT ‘4 AS [4 = A Dc DC eu PAN PULSER FILTER oe ou 2 SYSTEM GROUND o& TO SEA WATER ial ow a 7) P COUNTER & DISPLAY & DISPLAY DIGITAL PRINTER, TAPE PUNCH OR MAGNETIC RECORDER P | recaaren X-Y RECORDER Block diagram of bathythermograph deck unit. Fig. 16. The data rate capabilities of a system based on nuclear digital transducers will be dependent upon the maximum obtasnable pulse rate. Ifa full scale rate of 10° randomly spaced pulses per second is obtainable, as we anticipate, this would permit one data sample per second with readout accuracy of 0.1%. Longer sampling times would, of course, produce greater readout accuracy. Ultimate mechanical accuracies will be limited by the mechanical sensors chosen. The best available bourdon tubes, for example, have an overall accuracy within a few tenths of one percent. CONCLUSIONS On the basis of preliminary tests, the unique characteristics of nuclear digital transducers may lead to specialized digital data systems having outstanding long term accuracy. Nuclear digital transducers appear to offer the following advantages: (1) data digitized substantially at the source, (2) long term calibration stability, (3) overall accuracy limited almost entirely by the mechanical sensor, (4) infinite resolution of readout, (5) excellent linearity of readout or alternatively, prescribed nonlinearity for system compensation, (6) automatic averaging of data over the desired sampling time, (7) unaf- fected by temperature and other ambient condi- tions and (8) suitable for rough service applications. REFERENCES 1. KOMPASS, E. J., What about digital trans- ducers?, Control Eng., 5(7), 94-99, July 1958. 2. KOHL, J., R. D. ZENTNER and H. R. LUKENS, Radioisotope Applications Engineering, D. Van Nostrand Company, New York, N. Y., 146-178, 1961. 181 BROWN, W. L., Introduction to semiconductor particle detectors, Inst. Radio Eng., Trans., NS-8, 2, 1961. DONOVAN, P. F., Nuclear radiation detectors, Control Eng., 8(9), 144, September 1961. MANN, H. M., J. W. HASLETT and G. P. LIETZ, Pulse rise time for charged particles in p-n junctions, Inst. Radio Eng., Trans., NS-8, I5il, LEE, aa SUGARMAN, R. M., Nonsaturating transistor cir- cuitry for nanosecond pulses, Inst. Radio Eng., Trans., NS-7, 23, 1960. CHASE, R. L., W. A. HIGINBOTHAM and G. L. MILLER, Amplifiers for use with p-n junction radiation detectors, Inst. Radio Eng., Trans., NS-8, 147, 1961. Be y Pe rh ; f fi f 4 it teas | - ide iy ; 4 Tl Yn 5 a , 3 wei Pe he 4 t t re At u br 1h itt y a ed Belg a iy i i ais > coed eatatee © 4 POT dw ASS ON) OL ane I te eae aa wae) iy 4 a ¥ \ ELH elton Hae Msdhed oe pee i Fy it ead i \. i Gite! H) , Tals ‘ a) rycen " i for \ : { v hi NIN, : * ie 1 i 4 i on s a - 7 in if re ‘ oh Sip a ¥ F Mleat Lope ‘ou rinse wi iucwrtc i Pad TS Es ‘ ate ser | a; AN IMPROVED SELENIUM PHOTOVOLTAIC CELL T. K LAKSHMANAN Weston Instruments, Division of Daystrom, Inc. Newark, New Jersey ABSTRACT The selenium photovoltaic cell is frequently used in measurements of water transparency and in underwater measurements in oceanographic research. The photovoltaic cell described in the present paper is an improved type of Weston photronic cell. It consists of two semiconduc- tors in intimate contact and the junction between the semiconductors is the source of the photo-enf. The cell has high sensitivity particularly at low light levels. At very low illuminations the new cell has been found to generate an emf sev- eral times higher than the older types. The various cell characteristics, such as dependence of\ output on illumination, spectral response, time constant, fatigue and temperature dependénce are described. A brief description is also given of the cell construction. It is presumed that a photosensitive heterojunction exists between the two types of semiconductors. INTRODUCTION Accurate measurement of light levels under water is important in certain phases of oceano- graphic research. Of the several types of opto- electric transducers available the most commonly used device is the selenium "barrier layer" cell. This is a photovoltaic cell which generates a potential GRE Sie) oe between its terminals when exposed to light. It is self-contained and can operate a meter without the application of an external source of power. On the other hand, photoconductive cells and phototubes can only be used when an external voltage supply is inserted in the circuit. The selenium photovoltaic cell is rugged and compact and can easily be encased in a water- tight housing. Its output is high enough to drive an ordinary meter even at low light levels. Its spectral response is closer to that of the human eye than that of any other photo-device. With the use of a simple filter the resulting response can be made to match that of the eye so that the device can measure illumination as opposed to radiant flux. Furthermore, the cell shows excellent stability over a period of several years. Cells of this type have been used in under- water irradiance meters for attenuation measure- ments of blue-green irradiance. Holmes and Snodgrass have described an underwater irradiance meter or submarine photometer designed to operate in the upper 100 to 150 meters of water. It per- mits direct measurement of the downward blue- green irradiance. The detectors in the multi- detector unit are cosine flux collectors equipped with Weston photron cells. The use of the pho- tronie cell in arctic oceanography has been reported by LaFond.3 Water transparency is measured with a hydrophotometer which consists of a standard light source in a watertight housing which transmits a focused light beam through one meter of water to another housing containing the photocell. The cell is also used in an ambient light meter for measuring the ambient light intensities at various depths under water or ice. CELL CHARACTERISTICS The new cell to be described is an improved type of selenium photovoltaic cell. It differs slightly in construction from the older type but is considerably more sensitive than its predeces- sor at low light levels (below 1 foot-candle). Hence it is particularly suited for underwater light measurements. The fabrication process is more flexible and the cell has been made up in a variety of shapes and sizes (Fig. 1). Standard configurations are given in Fig. 2. Experimental cells have also been made up on flexible substrates. These flexible cells, as well as concave and convex cells, are useful in special applications. The process also permits the fabrication of large area cells. The output characteristics of the new type and the old type cells at moderately light levels are shown in Fig. 3. The output is shown as a family of I-V curves at various light levels ranging from 5 to 200 foot candles at a color temperature of 2870°K. At each light level the external resistance is changed from 3 to 10,000 ohms and the potential difference across the resistor measured with a Superior numbers refer to similarly numbered references at the end of this paper. Ala R5 R4 R3 Se eae OL: alle one pea ee R2 t RI t | 2 ae sl Ege ee fale Fig. 1. The new cell in various sizes and shapes. Fig. 2. Standard circular and rectangular cell shapes (dimensions in inches). EXTERNAL 10 as eS 3 OHMS 10 '! OHMS 102 OHMS ——— site aaa Conese ac000 LATE a PZ a FH He (oI A ea A a aI aA imayee ca ee AS 4 HJ 103 OHMS OUTPUT CURRENT ,MILLIAMPERES EHR EE jeune tities : BnimD ater oat ell ae Sei aeaet=ecs Sie can Hilt mem TH = ba Se) Mill Pililbaan ROY TTT ANAT TERMINAL VOLTAGE MILLIVOLTS Fig. 3. Characteristics of cells at higher illuminations (active area, 11 em@) . 184 EXTERNAL RESISTANCE 102 OHMS 109 OHMS 10° OHMS 108 OHMS OUTPUT CURRENT ,MICROAMPERES TERMINAL VOLTAGE MILLIVOLTS Fig. 4. Characteristics of cells at low illuminations (active area, 11 cm@). potentiometer or a high input resistance milli- voltmeter. On such a plot the external resis- tances are seen as diagonals. The end points of the curves on the current axis are the short cir- cuit currents and the end points on the potential axis are the open circuit voltages. The curves ee) OUTPUT _VS ILLUMINATION 100 . oe ae Eo 1000 give the mean values for 3 cells of each type. aa | 1] 1000 SE aee It is seen that the new type cells are appre- ciably more sensitive than the old type. ne 29199 The improvement is seen to be much more pro- 20h 10,000 nounced at light levels below 1 foot candle. The Li light levels, shown in Fig. 4, range between 0.1 i and 1 foot candle and the external resistances EZ from 100 ohms to 1 megohm. While the short cir- Aaa cuit currents of the new type cell are appre- f= ciably larger, the open circuit potentials “Wie exceed those of the old type by more than an “Te order of magnitude. The difference is more pro- alr nounced the lower the light level. A pronounced increase in performance is present also in the | maximum power capabilities of the cell. The maxi- mum power is obtained from the limiting power diagonal at a given light level. The output current is seen replotted as a function of the illumination of Fig. 5. With low values of external resistance the output is linear with respect to the light level. The open | circuit potential, on the other hand, increases rapidly at first and gradually later and satu- Fig. 5. Current output of new cell as a function rates at higher light levels, as seen in Fig. 6. of illumination (active area, 11 cm@). Troealaeals Tplalioesats rH ILLUMINATION - FOOT CANDLES TUNGSTEN een 2700°K RSLS SS eI at a) ! 10 185 OPEN CIRCUIT VOLTAGE VS ILLUMINATION 500 400 300 MILLIVOLT|S nn re) ° T 100 =i | 10 ILLUMINATION FOOT CANDLES TUNGSTEN LIGHT 2700°K 100 6. Open circuit voltage as a function of illumination. Fig. The improved performance is believed to be due to the difference in the construction of the cells. The photovoltaic junction in the old type cell is between crystalline selenium and a thin, sputtered film of cadmium. A top layer of gold is applied over the cadmium to improve the conductivity of the cells. In the new type cell a thin film of cadmium oxide replaces the cadmium and gold films. Sputtered cadmium oxide is more or less trans- parent and electrically conducting, forming a highly photosensitive barrier with seleniun. Selenium is a p-type semiconductor and cadmium oxide is n-type. A complex junction is believed to exist between the two types of semiconductors The semiconducting properties of selenium and eadmium oxide, and the manner of application of these materials, determine very considerably the photovoltaic properties of the resulting cell. The crystalline nature and the type and amount of impurities in selenium are extremely important as well as the degree of non-stoichiometry and the impurity concentration of cadmium oxide. optical and electrical properties of cadmium oxide which is an oxygen deficient semiconduc- tor, have been shown to be strongly dependent on these parameters. The The relative spectral response of the cell is seen in Fig. 7. The cell sensitivity is moderate in the blue, peaks in the green and falls off rapidly towards the red. By using a visual cor- rection filter with the cell the resultant response is made to match that of the human eye. The time constant of the cell is a function of the illumination and the load resistance. The rise and decay times are usually different from each other. For moderate illuminations and with moderate values of load resistance, the rise and decay times range between O.1 and 5 milliseconds. All selenium cells show a small amount of fatigue which is the gradual change in the out- put for a brief interval immediately following exposure to light. With the present cells such 186 RELATIVE RESPONSE 0.0 i 400 450 500 550 600 650 700 WAVELENGTH MILLIMICRONS Fig. 7. Relative spectral response. EXT RES IN OHMS 1000 Ww (o) i 4 35 Oo io} 2 100 = z ) « tj 1000 a. 4 TIME IN MIN. 100 Fig. 8. Cell fatigue. effects are small and are generally within 5%. Fig. 8 shows this effect. Temperature effects are also a function of the illumination and the load resistance. Fig. 9 shows the temperature dependence between 5° and 45°C at two different light levels. The varia- tion with temperature is due to the fact that the internal resistance of the cell is strongly dependent on temperature. A selenium photovoltaic cell can be repre- sented electrically as a current generator, shunted by a capacitance, C, and an internal resistance, Ba in series with a small resis- tance, R,. If the primary photoelectric current is ip and the load resistance R, the current through the latter is given by at AR, agai Sea Rj +R,+R (1) This picture is a rather naive representation of the cell performance. The actual mechanism of the cell is much more involved and can only be explained in terms of semiconductor junction theory. So far the exact junction parameters have not been worked out. TEMPERATURE EFFECTS ILLUMINATION -200 FOOT CANDLES -2700°K PERCENT DEVIATION IN OUTPUT BASED ON 25°C 5000 ILLUMINATION -20 FOOT CANDLES-2700°k 5 [3000 1000 100 o [52 300- 500 300 1000 100: 3000 -5 5000 () 10 20 30 40 50 TEMPERATURE IN DEGREES CENTIGRADE Fig. 9. Temperature dependence. Exposed cells are subject to attack by je moisture and corrosive gases. For most precision measurements encased cells are used. Both her- metically sealed metal-glass cases and non- hermetic bakelite cases have been developed. Hermetically sealed cells have excellent sta- 6. bility over a period of several years. A model 596 cell, encased in a bakelite case, is shown in Fig. 10. ACKNOWLEDGMENT The author is indebted to Walter Slegesky for his assistance in compiling the data for this paper. REFERENCES 1. ZWORYKIN, V K. and E. G. RAMBERG, Photo- electricity and Its Application, John Wiley & Sons, New York, N Y., 1-494, 1949. 2. HOLMES, R W. and J. M. SNODGRASS, A multiple detector irradiance meter and electronic depth-sensing unit for use in biological oceanography, J. Mar. Res., 19(1), 40, 1961. 3. LaFOND, E. C., Arctic oceanography by submarines, Proc., U. S. Naval Inst., 86, 90, September 1960. 4. PRESTON, J. S., Constitution and mechanism of the selenium rectifier photocell, Proc., Roy. Soc. of London, A, 202, 449-466, 1950. 187 Fig. 10. A model 596 cell. LAKSHMANAN, T. K., P-N junetions between semiconductors having different energy gaps, Proc., Inst. Radio Eng. 48(9), 1646, 1960. = LAKSHMANAN, T. K., Electrochemical Society Meeting, Boston, Mass., September 1962. i : i Ce ay) ry : J ¢ . | wee =) aoe y bs Ns Dy fe es) _ Bir ae y ST Sind Viole AT: Ae oo. r i { as : y } i Aa, : ’ wecleguie THQ Y . 4 - M ' + 7) ietisete< Bee a ? P. f Faye | iy i, } Sa 7 ‘w@ CA eee | Oe 4 ~ x 7 inh = % P : 7 ; {iter Dieta Fie iwt Yi Rt Onis {pokes detwes Se ete Meee ( (ito tc O iy Co 2 t dig Phew orth Bl iu® damume Gate ! pow ; Lrg og heruaee he d BDL BVA Devdas of. coupe) shawl g wea Shee Ei fovihet AY GE Rae Se Laid wiht’ teste So Balan, % F ; b ¥? * ot SORE ahs. “eaiaanecth ot ye emt Cee che teat ApS hu tae “KGS » ~ . Gian ss dll a ; . Wel «ti M 1 j wal a2: "? ear) pens AA | acd. Vere oee ia a | ame ree Pes Rae Py ry mies, are we ol Pr 3 hw st ig ata: “ne QUE ey a oo oe gee mT | Xan eeee € . ‘ "i Pic ” aa ee Ce or ae co a 0 hoe A et Ss) al y ; Aa Pea! fiw Satu th rial a4 Reh nhs sa 4 Sarat MENS) he! oid be! ge a L . ei Moen rank ey i ats et ae , bitte 5 ‘hale i LS ee Wah hers le f ed Iw ripe Wt OC fs Gt my feos t f = eres LS ah i had ee a W ey a6 Se we eggn, % Als Mi a ‘ 4 or { i Rs PvG MG Arh Nip -ernonh | ‘ ‘ 7) DIGITAL PRESSURE TRANSDUCER STUDY A. E. SNYDER and A. D'ONOFRIO Pratt and Whitney Company, Inc. West Hartford, Connecticut ABSTRACT The tunnel diode has a characteristic per- formance curve which can be utilized for under- water transducer applications. One very impor- tant application of this component would be as a pressure transducer providing a direct digital output reading. A tunnel diode and appropriate electronic circuitry has been examined and evalu- ated for this purpose. The design, analysis, test and evaluation results of this study are presented. INTRODUCTION The discovery of the tunneling earece in heavily doped P-N junctions by Esaki- in 1958 has made available a component which exhibits characteristics that are useful in areas ranging from computer logic to communications. A great many papers have been published since 1958 regarding tunnel diode characteristics and appli- cations. This paper will describe an investigation, conducted in the Research Laboratories of the Pratt and Whitney Company to evaluate a digital output tunnel diode transducer for oceanographic pressure measurement. The possibility that future oceanographic instrument systems may be computer oriented establishes the requirement for a digital output pressure transducer capable of measuring pressures to 15,000 psi with an accuracy of 40.5% of full scale. The approach taken during this investigation has been to use the tunnel diode in a unique hybrid oscillator cir- cuit rather than in the conventional amplifier or oscillator modes. TUNNEL DIODE AS A PRESSURE TRANSDUCER The reported works of Mason® and Sikorski and Andreatch> demonstrated clearly that the charac- teristic I-E curve of the tunnel diode is pres- sure sensitive. Curve "A" of Fig. 1 represents the I-E characteristics of a germanium tunnel diode at atmospheric pressure. The peak current and voltage are designated by Tp and ED respec- tively. The negative conductance, -gq, is the slope of the I-E tunnel diode curve between Ih d Ip 3=-= ee ef I i] I I I t I I I-gy | = 1 < i = 1 = i 1 1 i A : B i Cc I i Tey se Essoss Nes i] I i] I I 1 i] t i] i] J 1 ED Ey MILLIVOLTS Fig. 1. Tunnel diode characteristic curve. and I, where I, is the valley current and E, is the valley voltage. Germanium differs from sili- econ tunnel diodes in that the peak current decreases with pressure. The change in the [-E curve, represented by curves "B" and "C", results from the application of hydrostatic pressure. The applied pressure stresses the semiconductor, affecting the energy gap and effective mass (ratio of effective mass of the electron to its mass in free space) which are related to the tun- neling probability.© The tunneling probability is related to the current through the junction mecur une in the change observed in the I-E curve. Mitchell™ has derived the tunnel diode curve utilizing quantum mechanics and the energy level relationships. Superior numbers refer to similarly numbered references at the end of this paper. 189 Depending on the circuit configuration asso- ciated with the tunnel diode, the device can operate in a switching, amplifier or oscillator mode. The amplifier mode, when used as a pres- sure transducer, has a gage factor which can be as high as 30, 000. Considering the bonded wire strain gage with a gage factor of less than 10, the potential of the tunnel diode as a pressure transducer can be appreciated. The 15% decrease in peak current associated with germanium tunnel diodes under 20,000 psi pressure is responsible for the great sensitivity. TUNNEL DIODE PRESSURE TRANSDUCER OSCILLATOR MODE Conventional Oscillator Mode The series tunnel diode oscillator shown in Fig. 2a and the series parellel tunnel diode shown in Fig. 2b are the conventional oscillator circuits used to generate a sine wave. The series parallel oscillator shown in Fig. 2b is consid- erably more stable than the series oscillator because the "tank" circuit in the series parallel configuration is primarily responsible for the output frequency. Changes in frequency with pres- sure would be small except at very high fre- quencies where the tunnel diode shunt capacitance is a large part of the "tank" circuit. The series tunnel diode oscillator shown in Fig. 2a depends on the value of the negative con- ductance for its frequency and stability.? The negative conductance, -gg, is not single valued and the value specified by the manufacturer is an average one. Obviously, the slope of the curve is voltage dependent; therefore, the sta- bility of this configuration is not very good. With pressure changes affecting the negative slope of the I-E curve, stability is difficult to maintain over a wide range. The change in | Ba | due to the variation of the I-E character- istic curve with pressure can be shown to vary with the output frequency. Thus gag ela (1) where Rp is total equivalent circuit resistance, Lis series inductance, C is shunt capacitance and leal is the absolute value of negative con- ductance. The requirement for stable oscilla- tion is that L/C be made equal to Rp/| el p Hybrid Oscillator Mode The tunnel diode switching operation can be understood by reviewing curve "A" in Fig. l. The peak and valley currents are the key switching points. If the current reaches the peak value the diode voltage increases to correspond with the point "d" on the I-E curve. When the cur- rent is reduced to the value of the valley cur- rent the tunnel diode switches back to the low 190 |-9g| R= wer + lay!” (b.) SERIES PARALLEL Fig. 2. Tunnel diode oscillator circuits voltage state. A current change is all that is required to operate the tunnel diode switching mode. This switching characteristic does not depend on the value of &a- The hybrid configuration shown in Fig. 3 is a relaxation oscillator with a square wave output. The tunnel diode, when switching from the low voltage state to the high voltage state, turns the transistor, T-1, ‘on.’ Turning the transistor on discharges the capacitance, C,. When C, has discharged to the point where the current through the tunnel diode switches back to the low voltage state turning T-1 "off," C, then charges up through the circuit provided until the current through the tunnel diode reaches the value of the peak current and switches the tunnel diode back to the high voltage state. This turns the tran- sistor, T-1, "on" and another cycle begins. The tunnel diode characteristic curve shows that the values of the peak and valley currents of the I-E curve depend on pressure. These in turn establish the point of the charging or dis- charging cycle where switching occurs. This characteristic of the tunnel diode results in a Fig. 3. Tunnel diode hybrid oscillator circuit. change of frequency with pressure. Referring to Fig. 3, the characteristic times for the hybrid oscillator circuit can be written as (rj 4175) r2C te = peti e2 Sak log. ry tpt; (2-Euttotra)ie) (2) E Lor2eC I CeeeiacsOml og. Av (3) Yorth3 il tL=tetta (4) where t_ is charging time in seconds, t, is dis- charging time in seconds, ty is the total time per cycle and f is the frequency of square wave output in eps. The capacitor discharge characteristics shown in Fig. 4 result in a problem due to the non- linearity of the curve. The peak and valley 191 CURRENT ATMOSPHERIC PRESSURE PRESSURE Fig. 4. Capacity discharge curve. currents are functions of pressure; therefore, the time required for the valley current to reach the switching point during the discharge cycle varies and produces shifts in frequency which are not linear with pressure. In order to eliminate this problem the charge and discharge characteristics of the capacitor must be linearized. The charging or discharging current must be constant in order to linearize the characteristics of the charging and discharging time. The gen- eral expression for capacitor voltage may be written as (5) where €, is voltage across the capacitor, C, is the capacitor and i is the charging or discharging current. However, if i is constant, Eqn. (5) reduces to (6) where k is i/C, and t is the charging or dis- charging time. The charging and discharging periods, there- fore, can be made linear with time if i is held constant. This is accomplished by making ry very large and E a high voltage, producing a constant current source for charging C,. The transistor, T-2, has the base current set with P-1 to produce a constant discharging current. The resistor, rc, is provided to prevent the voltage from being applied directly to the base while the base drive is being set. The value of rz is selected such that the current reaches the peak current, IL,, before the capacitor is completely charged. The transistor must be saturated when the peak vol- tage of the tunnel diode is applied to the base. Therefore, resistor r), must have a value such that the product of the collector current of transistor T-1 and r) is greater than the value of the voltage applied to the transistor when the base voltage is equal to the peak voltage of the tunnel diode. If the shift in frequency due to pressure is caused to follow some predetermined schedule, the transistor T-2 can also serve another pur- pose. By rectifying the output and using the voltage to drive the base of the transistor, T-2, we can also perturb the frequency pressure characteristics as a function of frequency. +E COUNTER EXPERIMENTAL SETUP AND INSTRUMENTATION A pressure cell was fabricated as shown in Fig. 5. Pressure cell and electronic elements. Fig. 5. A commercially available germanium tun- nel diode was obtained and the top of the hat to measure the output frequency. Fig. 6 is a carefully removed. Silicone grease was used to photograph of the test setup. fill the cap and it was the only element of the circuitry exposed to pressure. The hydrostatic pressure was applied using an air operated EXPERIMENTAL RESULTS hydraulic pump that permitted setting the air pressure to hold a given hydrostatic pressure Using the series amplifier configuration pre- during readings. A pressure gage was carefully viously described the plot of the frequency vs. calibrated and was used to indicate the pressure pressure shown in Fig. 7 was obtained. Two sets in the cell. The voltages applied to the cir- of data are plotted on this curve for comparison. cuit were well regulated and instrumented. A Many additional sets of data were also taken and Hewlett Packard (Model 524c-10) counter was used the results were uniformly the same. A 500 eps Fig. 6. Test setup. 192 a ~ U x > S) z w = fe, ww oe Oo 4 8 12 16 20 PRESSURE (psi x 1000) Fig. 7. Series oscillator frequency vs. pressure. frequency shift per 1,000 psi with a frequency of 90 Keps at atmospheric pressure was observed. The deviation from the best straight line was +50 eps on the average. The minor shifts were due to the natural instability of the circuit under pressure. The data from the tests of the hybrid oscil- lator configuration are shown in Fig. 8. The frequency variation with pressure was plotted at various atmospheric pressure frequencies, f., ranging from 10 Keps to 1 Meps. The shift in frequency per 1,000 psi increased as the atmos- pheric pressure frequency was increased. An output of 20 eps/psi was obtainable when fy was 1.8 Meps; however, the accuracy is better at the low frequencies. In the region from O to 1,000 psi a larger shift in frequency was noted, indi- cating the I-E curve did not vary linearly. However, compensation can be used to correct for this nonlinearity The average shift in fre- quency per 1,000 psi plotted against f, for a 20,000 psi change in pressure is shown in Fig. 9 for the same configuration. The shift in fre- quency is 15% or better, depending on fo: For a 200,000 eps shift in frequency from 1 Meps when 20,000 psi is applied, the average shift is 10 Keps per 1,000 psi representing 20%. In order to obtain the desired accuracy we must measure the frequency to at least 1 Keps. 26 160 1200 QD oA uU 4 = 140 6 z 18 w 1100 ) 4 tra 120 14 f, = 10 f, = 100 f,- 1,000 0 4 8 12,16 20 (a.) (b.) (c.) PRESSURE (psi x 10 ) Fig. 8. Hybrid oscillator curves. 10 i B18 ° fo} Sg S 6 a ES a e~ SS 4 Vv m 2 9 > i) s w rg C) 0 200 400 600 800 1,000 ¥, (Kc/s) Fig. 9. Frequency shift vs. frequency at atmospheric pressure. However, much better resolution can be accom- plished. Common mode rejection can be accomplished by utilizing 2 hybrid tunnel diode oscillators and beating the two frequencies as shown in Fig. 10. The difference frequency output will result because one tunnel diode is exposed to pressure but the other is not, although they are in the same environment. The output frequency is then gated into an accumulator whose output collectors indicate the stored count in binary. The dif- ference frequency, being less than f) or fo in Fig. 10, requires less accumulator capacity. A serial or parallel readout can be accomplished with suitable circuitry. The voltage and tempera- ture problems are now common to both circuits and effectively cancel out. Greater accuracy may be obtained using this technique. The test circuitry did indicate that there is some apparent hysteresis. However, the overall performance characteristics indicate that this is not a serious problem because the hysteresis is well within the accuracy limits. Zero reference shift for the single oscillator configuration is due primarily to bias voltage changes and tempera- ture. Common mode rejection would solve this problen. PRESSURE SENSITIVE HYBRID OSCILLATOR FREQUENCY COMPARATOR ACCUMULATOR COMPENSATING HYBRID OSCILLATOR BINARY OUTPUT Fig. 10. Block diagram of measuring system. The test results demonstrate that a tunnel diode transducer will provide an accuracy of 70.5% of full scale with high sensitivity. The applicable pressure range extends from atmos- pheric to 20,000 psi. It is our opinion that oceanographers can make excellent use of this device in their instrument systems. DISCUSSION AND CONCLUSIONS A transducer system for obtaining a direct digital output proportional to pressure has been described. The highly responsive tunnel diode as the pressure transducer provides steady state and transient pressure readings in digital form with a sensitivity of 3 cps/psi when f, equals 300 Keps and 10 eps/psi when f, equals a Meps. The demonstrated accuracy is 40. 5% of full scale which we believe is satisfactory for many oceano- graphic requirements. Improvements in accuracy can be foreseen in thenear future. The ability of the hybrid configuration to operate stably over a wide range of frequencies and provide means for selective direct compensa- tion for nonlinearities make this configuration most attractive. Common mode rejection will reduce the effects of temperature, supply vol- tage changes, etc. Reliability, life and performance tests under actual field conditions remain to be conducted but because of its favorable size and digital output the tunnel diode pressure transducer may become an important component of computer oriented oceanographic research instrumentation systems. LIST OF SYMBOLS sf 4 peak tunnel diode current. I, valley tunnel diode current. Vp, tunnel diode voltage corresponding to the peak current. Vy tunnel diode voltage corresponding to the valley current. 19h |ea| absolute value of the negative conductance. r resistor. Cc capacitor. L inductor. Epp applied voltage Vv voltage applied to transistor in hybrid configuration. t time 6 frequency at atmospheric pressure. f,51 frequency shift from f, tof). Rp total equivalent circuit resistance. Eo voltage across capacitor. k ratio of current to capacitance. REFERENCES 1. ESAKI, L., New phenomenon in narrow ger- manium P- N junctions, Phys. Review, 109(2). 603-604, 1958. 2. MASON, W. P., Semiconductor devices as pres- sure frensduesre, Electronics, 35(8), 35, 1962. 3. SIKORSKI, M E. and P. ANDREATCH, Tunnel diode hydrostatic pressure transducer, Review of Scientific Instruments, 33(2), 155, QOS Sk ty anor ae ©: ee nammET 4. MITCHELL, F. H. Jr., Deriving the tunnel diode curve, Electronic Industries, 20(10), 96, 1961. 2. LOWRY, H. R., J. GIORGIS and E. GOTTLIEB, Tunnel Diode Manual, General Electric Co., First Edition, 1961. Aagaard, E. E., 11-17, 29-31. Anderson, P. R., 75-79. Ayers, R. A., 93-99. Brown, N. L., 19-2h. Brumley, D. F., 49-52. Buchanan, C. L., 157-161. Collins, J. L., 163-167. Cretzler, D. J., 93-99, 115-125. D'Onofrio, A., 189-194. Flato, M., 169-172. Folsom, T. R., 1-7. Fredkin, E., 81-86. Gaul, R. D., 115-125. Geil, F. G., 35-41. Hoover, H. M., 43-47. Hudimac, A. A., 49-52. Hyde, J. L., 173-181. Koezy, Bo H., Le7-a3h Kronengold, M., 127-134. LaFond, E. C., 53-59. Lakshmanan, T. K., 183-187. Loewenstein, J., 127-134. Mark, R. B., 101-105. Metzler, A. R., 87-90. Olsens Je Re, 49=52. Paquette, R. G., 10, 135-146. Peloquin, R., 61-72. Shodin, L. F., 147-154. Skinner, D. D., 25-28. AUTHOR INDEX 195 Smith, P. F., 81-86. Snodgrass, J. M., 115-125. Snyder, A. E., 189-19}. Stevens, R. G., 147-15). Thompson, J. H.. 35-41. vanHaagen, R. H., 11-17, 29-31. Von Wald, W. A. Jr., 107-111. Weiss, M., 61-72. io = % © ~ a oe sf : bg eae | F 2 bs i » o AL i = a a 4 are tke kar =a i i ' P ‘ ei v ¥ 4 2 ~ \ , Naioty od - ca 'y ie : wei F 4 P a 4 oe Lt ALeb, a ee Sy » eyes ih: seat tess Si yee Ye RRR? Sai OR | ae 8 A saoeratan | A Wirt’ Nae peti ‘ Re re to i “iR-ee ne: perth, j 4 =o = ‘= Avegar ale on) Al , ; eit: ' ; ra id . > : rah Fen 4 3 & ’ 4 Vee are , Mk, f 7 it "did aay try ; we. pi ial’ ae 4 N pa “_ Ae, _ - if gai . Play r é hy tep Pay if a akg et ' . 5 ae ok A sheet n ae aN . “hile y ‘ + iM ; ah ; ee ae! n ~ " y Eh, rr iin , : F Heys i . \ Fi ORL om ’ ~ 2 4k ' Wy = cae! ; : : wal . - OL" Vint? {= : om ow) : , ; see ees ivy Hae Fa | ‘a i r rhe, ; rere) ' ; ? Anes: Eheim i” one 7 é pa, , * 4 % e ; Ss " ims TR papel de ' t = ris ' ' er i Dare: y x Alieex ie Relea ae 7 2 aon Gg eee roy hy 5 _ oe ie py «abe ne ry op sepa ihe H i 4 ene mean ot (pala : et a f fe ; a say Fas i é j oe eee iat ot RAO rene ya hla 4 “Fe Ts . » ar 2, 1 ae ple ru: ou pa Mtn Pale a : J's i) CY Ue Poll Wy CuK \ ! ; : fp NA i AM ‘COE Hine ey ) ’ eee 1 MD iT nn. eras a ] i ; rane at oie Hy i } a / Dk j i] t {; sae acy, ( i 1 Se panes ac Se