ENGINEERING RESEARCH INSTITUTEE UJNIVERSITY OF MC HIGAN ANTI ARBOR TECHNICAL NOTE NO. CT-i TELEMET ING LIMITER FOR AFCRC AEROBEE BEACON SYSTEM By D.G. DOW. Projett Supervisor'N.W. SPENCER The research reported. in this document has been macie possible through support and sponsorship extended by the Geophysics Research Division of the Air Force Camxbridge Research Center, under Contract No. AF 19(604)5)45. It is published for technical information only, and does not necessarily represent recommendations-or conclusions of the sponsoring agency. Project 2096

(A NCI>>

ENGINEERING RESEARCH INSTITUTE* UNIVERSITY OF MICHIGAN TIEETIING Lfl4ITER FOR AFCRC AER-OBEE BEAOONSYSTEM' I. INTRODCTION This technical note will describe a limiter circuit which has been designed for use in connection with the standard U.S. Air Force Cambridge Research Laboratory beacon telemetering system, developed for Aerobee rocket use. Limiter action is necessary to avoid a malfunction of the telemeter when an input voltage greater than 5 volts or less than 0 volts is applied. II. STATEMENT OF THE PROBLEM The information system under consideration is diagrammed in Fig.l. Information Commutator Limiter Telemeter signals Fig. 1

- ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN The commutator permits presentation of many (typically 20) information and calibrating voltages on one telemeter channel. Thus, the output signal of the commutator consists of a sequence of flat-topped pulses, at a rate of approximately 80 per second. Following the commutator is the limiter, which should (a) clip the peaks of any pulses which are over 5 volts and (b) suppress any negative pulses, and in so doing, produce minimum insertion loss. The use of vacuum-tube amplifiers could conceivably result in very sharp limiting, but their use has been avoided due to problems of linearity and drift. Similarly, the use of relays and vacuum-tube switching circuits has been avoided due to the pulsed nature of the signal, which would require very rapid response in the switching loop. The normal problems of missile instrumentation must be met, including reliabilty, compactness, minimum weight, and resistance to vibration, acceleration, and environmental changes. The specific requirements are as follows: a. Linearity of transfer characteristic in the range from 0 to 5 volts must be within 0.01 volt, and preferably 0.005 volt. b. No signals greater than 6 volts or less than -1 volt should appear on the output. c. Input signals of 25 volts should be considered typical, and an input of 60 volts must not damage the limiter. Negative inputs should be expected. d. Insertion loss must be very low. EarL investigation fixed 2200 ohms as the maximum permissable series resistance. e. It is considered that the output impedance of the information source will be less than 5000 ohms; in most cases it will be below 100 ohms. f. The limiters must accomodate the input impedance of the telemeter, which has been specified as greater than 50,000 ohms. III. BASIC FORM The ideal limiter would have a transfer characteristic like that shown in Fig. 2. 2

ENGINEERING RESEARCH INSTITUTE* UNIVERSITY OF MICHIGAN E0-t Emi5 Fig. 2 If perfect diodes and a 0-impedance 5 volt source were available,, this curve could be realized by the circuit of Fig. 5. (<.001 R~L) + 5v p~~~~~~~~~~~~~~~~:Fig. 3 The principal impedimentae to the realization of an ideal circuit are:

IVENGINEERING RESEARCH INSTITUTE UNIVERSITY OF COMPONENTS with voltage as in a semiconductor diode. This current develops an error vo~ltage across the series input resistor (and. the source ouThe firsput imitpedance, which is verye and flow in early missiles uswhere extreme subminiatccure vacuum diodes Sylvania type 5647, which have the advantage). of zero reverse current under any condition. However, the random thermionic 2.energy of the electrons as emitted fr the dicathode with zero anode voltage that inor even with a sm negative voltage, causes forward current. Experimen-vel. ation of the.limiter. The use of a slightly reduced filament voltage logically.lowers this figure to 0.3 or 0.4 volts. There is significant variation from one tube to another in this respect.. InThe large-signal forward resistance anof temperature 5647variations of bximatteries whichly can150 to 200 ohmsed according to the manufacturer's specificationurces. TV. CHOICE OF COMPONENTS Theirmionic Diod.os The complete circuit of this earlier limiters built here and flown in early missiles used4 subminiature vacuum d~iod~es, Sylvania type 5647, which have the ad~vantage andof zero reverse current under ansfy condition. However chae randcom theristonic in Fig 9 2200 energy of the electrons as emitted. from the cathod~e with zero anod~e voltage, or even with a small negative voltage, causes forward current. oExperimenX 5647 @~~ 5647 tal data show that a negative bias of O.4 to 0.6 volt is required. to red~uce the current below a few microamperes, which is necessary for proper oper.ation of the limnliter. The use of a slightly reduced filament voltage logically lowers this figure to 0.5 or O.4 volts. There is significant variation from one tube to another in this respect. The large-signal forward. resistance of the 5647 is, approximately ~~~15 ~g20 to 200 ohms, according to the manufacturers specifications. The complete circuit of this earlier limiter'is shown in Fig. 4 and. its transfer characteristic in Fig. 9. 2200 5647 ~~~~~~~~~~~~564i7 Fig. 4'4

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN A somewhat involved, bias system was necessary to obtain biases of approximately 5.5 and. -0.5 volt (Fig. )4) to overcome the "contact potential" current d~iscussed. above. It was based. on the 6-volt source available in the missile, and. -led. to serious problems to be d~iscussed. in the section on bias sources. Semicond~uctor Diod~es If germanium d~iod~es are consid~ered., the question of reverse current becomes important; it has been found. that with presently available d~iod~es., the back current, rather than the insertion.loss, d~etermines the maximum aLlowable value of series resistor R1. In the range of 0to 5 volts the d~iod~es are in a low-current cond~ition; the equivalent network is shown in Fig., 5. rb Fig. 5 In this case rb1 and. rb2 are very large, and. nonlinear. Since an error is caused. by the current through Ri, and.rb and.rb are much greater than Rj, it can be seen that: E. Eero (i1 + i2) Ri in + 5-E (1 rbl(E) rb2(5-) 5 E( 1 + 11(2) rb2(5) rbi rb2(5E The subsubscript on rb ind~icates the voltage at which the nonlinear resistance has been measured.. The current. drawn by the load. resistance RL has been neglected.. For simplicity of thought., the problem will be consid~ered.

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN large compared with Ri, and therefore can be neglected. Accordingly, the error is given by simple voltage-divider action as shown in Fig. 6. Ri S 2(5) 5V Fig. 6 Data-reading considerations allow a maximum error of 0.005 volt Typical diodes of good quality (Hughes type lNlOO is an example) can be obtained in which the reverse current at room temperature is specified to be less than 5 microamperes at 5 volts. Thus for the specified allowable error, a 1000-ohm series resistor is indicated. Consideration of the forward characteristics then shows the limiting features of the device. The newer high-conductance diodes offer the highest forward-toback current ratio available (see Table I); thus the type NlON100, which appears to be the best at low voltages, was selected for use in the limiter. Three of these diodes have been tested and show conformity to the manufacturer's specifications, as tabulated. TABLE I DIODE CHARACTERISTICS AT 1 VOLT FORWARD AND 5 VOLTS REVERSE Reverse Current Maximum Forward Diode Number at -5 volts, current at 1 volt, Temperature, microamperes milliamperes ~ C lN48 10.5 (exp) 4.0 (mfr. spec) 25 1N48 25.5 (exp) 41 1N69 5s~7 (exp) 4.0 (mfr. spec) 25 1N69 12.7 (exp) 41 1N34 8.9 (exp) 25 1N63 15.2 (exp) 3.0 (mfr. spec) 25 1N63 30.8 (exp) 41 lNlOO No. 1 1.4 (exp) 26 (exp) 25 I~O (mfr. spec)n 5.0 (max.) 20 (mm.)r 251

ENGINEERING RESEARCH INSTITUTE* UNIVERSITY OF MICHIGAN Generally, the diode which has the highest ratio of rb5 to rf 1.should be selected. Ri then can be determined from error and insertionloss considerations. The ef fect of high temperature on the back resistance is rather large. Experimental data show that at 4i20C, few diodes can be expected to have currents less than 5 microamperes at -5 volts. (See Table I) A typical instrumentation provides periodic calibration to permit the desired accuracy of data interpretation. In this calibration system, four of the commutator segments are devoted to fixed-voltage signals, of 0.00,0 1.55, 2.70, and 4.05 volts respectively. Reference to Fig. 7b shows that at elevated temperatures., assuming a diode of minimum allowable quality, this system employing four reference levels will limit the determinable error caused by the -limiter circuit'-to less than a millivolt except between 0 and 1.35 volts. The sharp curvature of the diode characteristic (diode No. 1) near 0 volts causes the largest error. If a highquality diode is used here,. the error can be reduced to a maximum of 5 millivolts at 420C and 0.8 millivolt at 25CC as shown in Fig. 7a. At the ~5-volt end of the range, the hump in the error curve (with calibration) is much less pronounced., since the bias voltage is normally a f ew tenths of a volt over 5 volts., and most of the rapid change in the diode characteristic curve is very close to 0 volt. At low temperatures, the bias voltage may drop to 5 volts. However., at low temperatures, the diode-back current, and thus the error., is greatly reduced., so the accuracy is still high over this part of the scale. This is indicated in the data of Fig. 7. If there is a selection of diodes available, the units with the lowest reverse current at -5 volts will give the least error., and thus should be selected. This characteristic is especially valuable in regard to the diode which'is biased at 0 volt. V. BIAS SOURCE Of the several ways of providing a reference bias voltage of 5 volts, one possibility mentioned earlier is the use of a voltage divider in conjunction with a N-T-6, (6-volt) battery. This arrangement would have a very low apparent impedance without significant current drain from the battery. For example, if 10- and 50-ohm arms were used across63 volts, the bias would be 5.25 volts at an 8.5-ohm impedance. Unfortunately,, experience has shown that the available voltage under load of the N-T-6 battry dcreaes rpidl with,, —44, 1 — derasn temeraureLt a value1 as low -1

-0 250CENTIGRADE - -WITHOUT CALIBRATION ~~ ~WITH 3 CALIB3RATION VOLTAGES X40CENTIGRADE -WITHOUT CALIBRATION IC___ - - WITH 3 CALIBRATION VOLTAGES 4 0~~~~~~~~~~~~~~~~~~~~~~~. 1.0 2.03040 -2'0___ ___VOLTS INPUT __ -4_____CALIBRATING VOLTAGES -( - - ~~~~~~~~~~1. 33, 2. 70,4.0 5 FIG. 7a - - ___ - ~~MAXIMUM ERROR IN LIMITER DUE TO ____ ~DIODE BACK CURRENT

FIG. 7b MAXiMUM LIMITER ERROR DUE TO BACK CURRENT IN DIODES AT 420C (POOR DIODE) IC_ _ _ _ _ ERROR WITHOUT CALIBRATION ERROR WITH 3CALIBRATIONt 6 71 7 ~~~~~~~~~~~~VOLTAGES VALUES COMPUTED FROM E-1I CURVES OF THE DIODESI 4~~~~~~~~~~~~~~~~~~~~~~ cc ~~~~~~~~POOREST ALLOWABLE DIODE IN ZERO O BIAS POSITION w 0 1.___I0 2.0 3.0 4.0 5.0 -2 ____ / ~VOLTS INPUT _ __ _ _I I.4L CALIBRATING VOLTAGES1.35, 2.70,4.05 -6 -8 __ ____ __ -IC _ _ __ __ _ _

ENGINEERING RESEARCH INSTITUTE* UNIVERSITY OF MICHIGAN Correspond~ingly, if the battery voltage is 6.5 volts, the bias voltage will be 5.7 volts. Since the resultant bias voltage d~etermines the point at which limiting commences, it is important to select a stable supply source. For these reasons, the use of a d~ifferent type of bias battery was consid~ered.. Four series-connected. Mallory RM-l mercury cells, with an open-circuit voltage of 5.3 volts were selected. after preliminary investigation. RM-l's were chosen in preference to other mercury cells because they appear to hold. their voltage above 1.25 volts for a longer time und~er load.. Und~er anticipated. circuit use an overload. (limiting action) wil~l force current through the battery'in the reverse d~irection. Some tests were mad~e and. it was found. that the battery will polarize und~er the influence of a reverse current., allowing the terminal voltage to rise to as much as 1.6 volts per cell. This would. be the case in the limiter if one or two of the information channels became overload~ed. to 20 or 25 volts. To overcome this effect, a constant "tpreload." is put on the battery to insure that the current will not, in general, reverse. With a d~rain of 15 milliamperes., the bias is very close to 5.2 volts with an internal ipd~ance of about 10 ohms for the four cells at room temperature. If several limiters are to be used., they can be supplied. from a common battery source, in which case the use of the RM-12 would. be ind~icated.. This cell is roughly the same electrically as three RM-l's in parallel; thus it should. be peload~ed. to approximately 45 ma. With this system, the bias source would. then be capable of handling simultaneous overload~s on three channels. The lower internal imped~ance of the larger cells likewise enhances the d~esired. -limiting function. The foregoing d~iscussion has been on the basis of using this system at room or higher temperatures. In actual use of the limiter,. at the Holloman range it is possible that temperatures may go to as low as 000. This will have three effects on the limiter. 1. An increase in the back resistance of the d~iod~es will d~ecrease the alread~y small error d~ue to back current. 2. An increase in the forward. resistance of the d~iod~es will make the limiting a little less sharp. However, this effect is much smaller than that on the back resistance, and. will raise the output voltage only a few tenths of a volt for input voltages as high as 30 volts.

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN Experiments conducted near freezing temperatures'indicate that a load of 240 ohms per cell (5 ma) instead of 80 ohms (.15 ma) will insure that the bias stays above 5 volts for about 5 to 6 hours of battery life, at this temperature. This was determined on an intermittent-use basis, using 50-minute "on" periods. Experiments have also determined that if forward and reverse currents are passed alternately through the cell, so that the net charge passed is approximately zero, the cel~l will not change noticeably in potential., even at low temperatures. This means that the battery loading resist ance can be made higher without distortion due to polarization. The use of standard carbon-zinc "penlight" (or larger) cells was also considered. However, preliminary tests at 00C indicated that the internal impedance of the cells became so large as to preclude their use in the limiter circuit. Thus,' the choice of mercury cells for a given system is a function of the number of overloaded channels and the extent of the overload. The significant quantity is the current that must be delivered by the cell to equalize the reverse current. If this amount is 5 ma or less,. the RM-l may be used. If it is greater than 5 ma, the FuM-12 batteries must be used. In our application, many of the input signals are calibrating voltages, and other signals which cannot overload; thus the use of the RM-l is feasible. On the basis of these considerations, a limiter was built (for low-temperature use) with 1NlOO0 diodes and four RM-l mercury cells for bias. The series resistor of 1000 ohms was selected as a compromise between optimum limiting and minimum back-current errors. The final circuit is shown in Fig. 8 and its transfer characteristic is shown in Fig. 9, compared with that of the earlier limiter of Fig. 4. This circuit meets all the specifications except that of 6-volt maximum output at 25 volts input. It was decided that this was acceptable 1000 02 i1NlOG 1NlOO Signal Input Signal Output Anode 4x~m-1 Fig. 8 ~ Ti

3.3 - - - -X 10 X - LIMITER TYPE 2 WITH IN10OIOES 0 0. I AND MALLORY CELL FOR BIAS X LIMITER TYPE I WITH SLYVANIA TYPE 5647 DIODES AND VOLTAGE.- l -!-/- 10 — IIE YP IHNO IDDIVIDER l3!_SE FUNCTYPE I AND I2 IT

ENGINEERING RESEARCH INSTITUTE* UNIVERSITY OF MICHIGAN for the sake of higher accuracy. -If a more stable and exact 5-volt supp~ly were available, the limiting could be made sharp enough to meet this requirement. A 60-volt input pulse will seriously polarize the battery, resulting in an increased, bias and poorer limiting unti~l the cells can recover; however it will not damage the limiter, thus fulfilling the specifications in this respect. In addition, it wil~l not interfere with the transmission of information, which is unaffected by moderate changes in the bias voltage. The use of RM-.12 cells f or a bias supply is recommended if two or more limiters are to be flown'in a mis'sile. One set of four such cells is adequate for three limiters when operated as discussed above. The diodes selected, Hughes type lNlOO, are hermetically sealed and thus should be subject to a minimum of environmental change. Their extremely small size is advantageous, and their mechanical ruggedness is high compared to that of vacuum tubes. With the present instrumentation it is not possible to exceed their maximum current ratings of 50 ma dc and 500-ma surge. VI. CONCLUSION This technical note has presented the steps in the design of a'limiter-for use with a specified telemetry system, capable of limiting signals in excess of 5 volts positive and greater than 0 volts negative. A circuit has been designed which is superior to that originally used in this application. The merits of new limiter as compared with its predeces-. sor are as follows: 1. Much sharper limiting, especially at high values of input voltage. 2. Freedom from effects due to interaction through common batteries from other instrumentation in the missile. It should be noted that even if several limiters are operated from one bias battery, interaction wil~l only occur between signals greater than 5 volts, where accuracy is not a consideration. 5. Avoidance of vacuum tubes with their higher failure probabilities and need for filament supply., h. Improved lo-teprtrepromncb sngsprt

UNIVERSITY OF MICHIGAN mutated with widely varying input voltages.

UNIVERSITY OF MICHIGAN 3 9011111111111111115 02656 6466 3 9015 02656 6466