AFCRL-62-662 THE U N I VE R S I T Y OF M I C H I G AN COLLEGE OF ENGINEERING Department of Aeronautical and Astronautical Engineering High Altitude Engineering Laboratory Final Report FALLING-SPHERE EXPERIMENT FOR UPPER-AIR-DENSITY: INSTRUMENTATION DEVELOPMENTS Prepared for the Project by: Ho Fo Schulte D. A, Robinson J. Lo Wagener Approved by: Lo M. Jones Project 6690 Task 66900 ORA Project 03558 under contract with:. UNITED STATES AIR FORCE OFFICE OF AEROSPACE RESEARCH GEOPHYSICS RESEARCH DIRECTORATE Lo G. HANSCOM FIELD BEDFORD, MASSACHUSETTS CONTRACT NO. AF 19(604)-6185 administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR April 1962

TABLE OF CONTENTS Page LIST OF TABLES v LIST OF FIGURES vii ABSTRACT ix THE UNIVERSITY OF MICHIGAN PROJECT PERSONNEL xi 1. INTRODUCTION 1 2. ACCELEROMETER DEVELOPMENTS 2 2.1 First Multiple-Cavity Developments 2 2.2 Development of Flight Model 2 2.3 Caging and Retraction 6 2.4 Start Signal 8 2.5 Rebound 13 3. ACCELEROMETER PRE-FLIGHT TESTING 16 3.1 Tests at 1 g 16 3.2 Low-Acceleration Investigations 19 4. SPHERE MECHANICAL DESIGN 24 4.1 Accelerometer Mounting 24 4.2 Balance 25 4.3 Flight Configuration 26 5. SPHERE ELECTRONICS 27 5.1 Contact Detector 33 5.2 Intervalometer 34 5.3 Modulator 40 5.4 Accelerometer Drive 42 5.5 Pull-Off Plug 43 5.6 Power Supplies 43 5.7 Switching 44 5.8 Components 45 6. ACKNOWLEDGMENT 50 7. REFERENCES 51 iii

LIST OF TABLES Table Page I Break-Free Time in Milliseconds II Accelerometer No. 105 17 II Accelerometer No. 104 18 IV Component Function 46 v

LIST OF FIGURES Figure Page 1 Assembly drawing of accelerometer 3 2 Photograph of accelerometer. 5 3 Setup for measuring caging force. 6 4 Caging force vs. displacement. 7 5 Vane and photocell for measuring caging finger motion. 9 6 Transistor amplifier used with vane and photocell. 9 7 Caging-finger displacement vs. time. 10 8 Caging-finger displacement vs. time. 11 9 Caging-finger displacement vs. time. 12 10 Caging-finger displacement vs, time, showing rebound. 14 11 Caging-finger displacement vs. time, showing rebound. 15 12 Transistor pulse amplifier used in drop test. 16 13 Photocell and amplifier used to control accelerometer in drop tests. 21 14 Bullet release device for drop tests. 21 15 Drop bullet, control circuit, accelerometer. 22 16 Accelerometer mounted in control circuit rack. 22 17 Catch basket. 2 18 Drop test setup. 2 19 Sphere on balance testing machine. 26 20 Accelerometer mounted in sphere; adjustable balancing weights. 28 21 Sphere, showing function ports. 28 vii

LIST OF FIGURES (Concluded) Figure Page 22 Sphere circuit block diagram. 29 23 Sphere with top bell removed. 30 24 Sphere with bottom bell removed. 30 25 Sphere with top bell removed, side view. 31 26 Bottom view of sphere with accelerometer removed, output pulse transformer exposed. 31 27 Sphere with top electronics rings exposed. 32 28 Accelerometer bobbin-cavity capacitances. 33 29 Sphere-circuit schematic. 35 30 Ten-section delay line. 49 viii

ABSTRACT The continuing development of the instrumentation for the falling-sphere experiment for upper-air density and temperature is described. A multiplecircuit cavity for the accelerometer, which permitted the elimination of the wire bobbin contact, was developed. A moving iron solenoid caging device resulted in a smaller accelerometer requiring less power than the previous model. A new Attwoodts machine permitting laboratory tests at 10 3 g was constructed. Electronic circuits were developed to replace the intervalometer motoro Circuits were also devised to permit long transit times and to provide discrimination between start and stop pulses. Four spheres and accelerometers for flight test were constructed. ix

THE UNIVERSITY OF MICHIGAN PROJECT PERSONNEL (Both Full Time and Part Time) Estry, Hal W., B.S., Assistant Research Engineer Filsinger, Edward A., Instrument Maker Gleason, Kermit L., Instrument Maker Jones, Leslie M., BoSo, Laboratory Director Mosakewicz, Mary C., Secretary Peterson, John W., M.S., Associate Research Engineer Robinson, Douglas Ao, Technician Schulte, Hal F., Jr., M.S., Associate Research Engineer Sporzynski, George Ao, B.S., Technician Wagener, Jerrold Lo, BS., Assistant Research Engineer Work, Edgar A,, BS., Assistant Research Engineer xi

I. INTRODUCTION This is the final report of a project carried on in the Department of Aeronautical and Astronautical Engineering of The University of Michigan, the purpose of which was to complete certain developments of the fallingsphere method for measuring upper-air density and temperatureo The immediate predecessor of this contract was Contract Noo AF 19(604)-2415 under which the developments reported herein were initiatedo For a report of the preliminary work, see University of Michigan Engineering Research Institute Report 2649-9-Fo The sphere development program of the current contract is a continuation of experimental work with small spheres started in 1954. Background information is contained in Refs. 1 through 9. The tasks of the contract were as follows: "Item 1: Develop, construct and test three (3) each completely operational transit-time accelerometers of improved design for use in 7" Falling Sphere experimento The following improvements in the new design shall be attempted: "a. Improved release and pick-up mechanism that eliminates, or reduces to an acceptable minimum, the error due to initial velocity of the bobbino "b. A more positive and less fragile detection system for determining the bobbin time of flighto "Co An increase in overall sensitivity and accuracy over the present accelerometer versiono "do Positive alignment techniques, where possible, to facilitate static and dynamic balance. "e. Lighter construction, where feasible, to reduce the mass. "fo Power drain as low as engineering expediency will permit. "go Compatibility for use with power transistors currently availableo "Item 2: Develop, construct and test 7" sphere electronic instrumentation compatible with new accelerometer. "Item 3: Fabricate three (3) each operational sphere instrumentations complete with accelerometer(s) andpower supplieso "Item 4: Modify ground-station equipment to provide compatibility with 1

new instrumentation." In summary, a new accelerometer was designed and tested in the laboratory. It was smaller and lighter and consumed less power than the previous version. Further, the design successfully dispensed with the "cats-whisker" bobbin contactor. Laboratory tests (including tests on a new Attwood's machine designed and constructed for this development) indicated good performanceo Improved sphere balance was achieved with the aid of a static balance machine. Transistorized sphere circuitry was also designed, built, and tested. It incorporated ant electronic intervalometer (instead of motor-driven) to operate the accelerometer. It also employed a discretionary caging circuit to permit measuring long transit times and it generated start and stop pulses of different widths to permit pulse discrimination in the playback system. Because the performance of the accelerometer cannot be adequately evaluated in the laboratory, four spheres and accelerometers for flight test were built. Three of these were flown at Wallops Island in the summer of 1961. All three spheres operated successfully. However, because of rocket failures in two cases, the spheres were not subjected to sufficiently low accelerations to evaluate the low limit of measurable accelerationo Data from the third flight are being redhcedo The fourth sphere and accelerometer were delivered to G.RoDo in December, 1961. 2. ACCELEROMETER DEVELOPMENTS 2.1 FIRST MULTIPLE-CAVITY MODEL DEVELOPMENTS In the previous report the concept of a multiple-circuit cavity which would permit detection of bobbin contact without using the "cat's whisker" contactor was described. The model constructed to test this concept met with some initial successes but, because of the complexity of the caging-finger system, it proved difficult to release the bobbin with an acceptably small initial velocity, and this particular model was abandoned. The multiple-circuit cavity, however, was retained and incorporated with improvements, including capacitative detection of impact, in the final design. 2.2 DEVELOPMENT OF FLIGHT MODEL On the basis of experience with the original accelerometer and the model described in the preceding paragraph, a third design was undertaken which incorporated sufficient improvements to justify flight tests. This model is the one flown at Wallops Island on June 1 and 6 and July 26, 1961. The new design, shown in Figs. 1 and 2, employs a bobbin and caging-finger combination similar to that in the original accelerometero The chief advantages 2

0IV. W S VI Z m r 0,.-. ~,P (D 14 iy fot *1 0 C8 3 > m I -4 m > I 05 i G)........I —-.... r::> <. fXw > W< o,' 5 G: z m m PIt.....I. G0;n~~~ r, n~'~ol p^ 8 > --. > 0 ) Xar z

Fig. 2, Photograph of accelerometer, which affect primarily the initial velocity imparted at release are simplicity and rigidity. Two metal cavity sections surround each lobe of the bobbin through an angle of 180o The transit distance is 0.050 in. They are separated from each other by an insulating ring. Caging force is provided by two moving iron solenoids on opposite sides of the cavity-bobbin center section. The solenoids are separated from the cavities by rings dimensionally identical to the center ring. One of the cavities is enlarged and is grounded. The ring on that side is of steel. Bobbin release is accomplished by a nonlinear spring system which retracts the caging finger with sufficient speed to escape the pursuing bobbYin at t10 g's acceleration. Solenoids were selected for caging because of their relatively small siz., low power consumption, and large force exerted in the fully caged position, Disadvantages overcome we re: the somewht greater mass (relative to a moving coil system), a short caging stroke due to rapid diminution of force away from the core, and the presence of residual magnetisnm. It was originally thought that tolerances in constructing the various parts could be held close enough so that the bobbin could be used to complete a conductive circuit between the two cavities at any orientation. This proved not to be the case, and we resorted to a capacitive indication of contact between bobbin and cavity. The mul]tiple-circuit cavity was, of course, a necessity for the latter approach so that its conception and development were well used. The capacitive technique eased the very stringent requirements of shop tolerance and in fact allowed actual enlargemenat (by 0,005 in.) of one cavity so that precision bobbin centering (and thus impact time) was required only at the "close" end of the bobbin and its surrounding cavity. Nevertheless, close tolerances were held in any case to insure an accurate distance in the "close" end, and precision performance in general. Further, although the initial contact time is indeed determined by the "close" end, the wave shape of the contact signal is much more constant with respect to the direction of bobbin travel when one cavity is enlarged, 5.

2.3 CAGING AND RETRACTION The caging solenoids must pick up the bobbin and various ing system (depending on orientation) in a 10 g environment. compress the retraction springs. The weights in grams of one parts are: parts of the cagThey must also set of the moving Sliding bearing and slug Caging finger Adjusting nut Coil spring Locking nut Bobbin 11.5 4.4 3.5 1.2 0.2 25.0 Given an iron circuit as in Fig. 1, whose dimensions were to some extent dictated by nonrelated design requirements of the sphere and, given a desirable voltage supply of 8 volts d-c, the following coils were found to be best: wire turns resistance current ampere turns power No. 26 AWG 680 each coil 5.2 ohms each coil 1.54 amp each coil 1047 each coil 24.6 watts total for two coils The latter figure compares with 54 watts (9 amp at 6 volts) for the original accelerometer. The equilibrium operating temperature of the solenoids was measured as 45.50C with an iron-constantan thermocouple and Leeds and Northrup potentiometer. A Hunter force gage Model L-10 was used to measure the retraction force of the caging system. See Fig. 3. Adjustable stops were used to position the slug at various stations of the stroke. The force necessary to pull the slug free from the stops was measured to give a force vs. distance characteristic as seen in Fig. 4. It can be seen that sufficient caging force is available to operate the device over the entire range of displacement, provided that the retracting springs are not too stiff. Fig. 3. Set-up for measuring caging force. 6

0.74 AMP 0.92 AMP. 1.10 AMP 662 AM 822 Ah 983 AJ 1001' 66c I ul -1 3 - 0Cmm 4000 ~CE - G^ C000 C=P~0 &Lstlcee^t -.at force Vs- Fi. Ca S fo F7i8- \ooo

In the course of choosing the springs it was found necessary and desirable to be able to trace the motion of the caging system. To do this a small vane was attached to the end of the caging finger. The vane intercepted a collimated beam of light directed to a photocell, the output of which was amplified and viewed on an oscilloscope. See Figs. 5 and 6. With this set-up the following significant parameters were measured: (a) elapsed time for caging and retraction (b) retraction acceleration (c) magnitude and duration of rebound (d) effectiveness of rebound damping (e) lag time from field collapse to break-free A variety of springs was tried before deciding on a cascaded combination of two which gave satisfactory initial acceleration together with sustained accelerationo One spring is a helix of ten turns of Oe032-ino-diameter steel wire, the other a flat diaphragm of 0.020-in.-thick beryllium copper, The helical spring operates throughout the entire caging and retraction while the diaphragm is involved in only the last 0.015 in, of the fully caged position. The relatively strong diaphragm was used to achieve the high initial acceleration required by the fingers in the sharp break from the bobbin under 10 g's of acceleration. The acceleration of the caging finger with various spring combinations is shown in Figs. 7, 8, and 9. The displacement of the bobbin under an acceleration of 10 g is shown in each case, Figure 9 is for the combination used in the flight models, 2.2 START SIGNAL It was at first intended to use the break-free from the stops to generate the start signal, that is, the beginning of the transit time interval, However, the repeated severe hammering of the stops, together with oxidation, caused deterioration of the contact surfaces. Misses in the signal occurred often enough to warrant changing to collapse of the magnetic field as the source of the start signal and then incorporating the constant lag time in the data, The magnitude of the lag time is determined by the residual magnetic field, the distance the slug is stopped from the core, and the forces exerted by the dual spring. The lag time is regulated by varying the deflection of the heavy spring and the po — sitioning of the slug by the adjustable stops. A distance of 0o010 ino between the core and the stops was finally selected after an investigation of the residual magnetic field effects at various distance. This gives control over the too large forces against the core but does not seriously diminish the effective caging distance. Slight changes in this spacing are sometimes made in making the final close balance between the two solenoids0 The adjusted lag times in the various accelerometers were about 0,001 second. The fact that the retraction speed changes under varying acceleration environments is relatively unimportant in the motion of the bobbin as long as ti.e 8

LIGHT PHOTO < P SCOPE ~_ CELL CAGING FINGER. _ VANE ADJUST NUT - - - -DIAPHRAGM SPRING nDJUSTABLE SLUG STOPS.u,~~~~~~~~~~~~~~ i ~ OELICVELICAL SPRING WLnJ-U 9 vr "W rqwv www vl Fig. 5. Vane and photocell for measuring caging finger motion. GND SCOPE I I I SI I L _.......-_ __..J Fig. 6. Transistor amplifer used with vane and photocell. 9

0 wI o / DISPLACEMENT OF BOBBIN' Z _,UNDER Ig ACCELERA TION z | DISPLACEMENT OF BOBBIN 0 |UNDER lOg ACCELERATION 0 a. 4 o 0 o 6 CAGING FINGER- \ 0 Q. w Cl) (0 D 10 I I 0.0005.0010.0015.002 TIME - SECONDS Fig. 7. Caging-finger displacement vs. time. Displacement spring 3, beryllium copper 0.010 in. thick.;0

n0 I 0 U) w 0 z I z (.9 UJ 0 cn U. z Q. 0. U) zP D/SPL ACEMENT OF BOBBIN 0*0r / UNDER /Og ACCELERA TION I NS, NS. NlN." CAGING 0.0005.0010.0015.0020 TIME - SECONDS Fig. 8. Caging-finger displacement vs. time. Displacement beryllium copper 0.015 in. thick. spring 7,

2 \ DISPLACEMENT OF lz |W~l\^~~~~~ ~ \. - UNDER /Og ACCELi I \ 06 w DO o |CAGING FINGER \ 0[ r 8 — U. 2\ 10 _-j (0\ 12 —-- 0.0005.0010.0015 TIME - SECONDS Fig. 9. Caging-finger displacement vs. time. Displacement spring 11, beryllium copper 0.020 in. thick, micro-switch mode. BOBBIN ERA T/ON 44 N%. NS" NNS. N.0020

acceleration is greater than that of the bobbin. However, to promote constant lag time and low initial velocity the time of break-free from the stops should be kept within measured acceptable limits. A setup to measure this parameter was made. The accelerometer was positioned so that the caging stroke was on a vertical axiso A variety of weights) simulating larger-than-one-g conditions, was screwed on the caging finger. The accelerometer was then cycled. The time from field collapse to break-free from the stops was measured. The results were as follows: TABLE I BREAK-FREE TIME IN MILLISECONDS Voltage Axis Added Weight in Grams 0 62.2 169.8 294 3 6 Horiz,.8 6 Vert..8.8.8.8 7.5 Horiz. o75,*8 7.5 Vert..8 -.9 o8 o75-o8.8 -.9 No weights were added in the horizontal position because of the unduly large frictional force imposed by the cantilevered load. This test also demonstrated the ability of the accelerometer to pick up the weights with applied voltages somewhat less than the planned figure of 8 volts, 2,4 REBOUND As anticipated, the caging fingers were found to rebound into the cavity volume after the termination of the retraction stroke. A lead seat behind the caging head had been provided to damp the rebound but proved to be inadequate. Room for an additional damping pad existed at the other end of the fingers, and various devices and materials were tried in this space: springs and pads of Bakelite, Lucite, sponge rubber) Teflon, and polyureathaneo The latter material proved to be satisfactory under long periods of operation, It has the one disadvantage of remaining compressed after long storage in the compressed state but this is easily handled by installing new pads on occasion, Figures 10 and 11 show the undamped and damped rebound of the fingers compared with the bobbin displacement) both as a function of time,

BOBBIN DISPLACEMENT UNDER lOg ACCELERATION U) w z 0 I C 0 aw CL J cL) Ln L M T OF BOBBIN DISPL ACEMENT VELOCITY (a INITIAL IMPACT = APPROX 60 IN./SEC \ 0 4 7 8 9 10 II 12 13 14.1I TIME - MILLISECONDS Fig. 10. Caging-finger Displacement spring 11. displacement vs. time, showing rebound.

BOBBIN DISPLACEMENT UNDER /Og ACCELERA TION.0 \ \ 0 \<_CAG/NG FINGER \ o.04- DISPLACEMENT \ < |\ \ L IMIT OF BOBBIN.H \ \'DISPLACEMENT 0.05 —---- LL z \ \ Z.06F \ \ o a..07 -.08 0 1 2 3 4 5 6 7 8 9 10 11 TIME - MILLISECONDS Fig. 11. Caging-finger displacement vs. time, showing rebound. Displacement spring 11.

3. ACCELEROMETER PRE-FLIGHT TESTING 3.1 TESTS AT 1 g The majority of tests of the accelerometers was made on the bench at 1 g. In addition to representing an important region of the accelerations encountered in flight and of course being the easiest to perform, tests at 1 g are most important in discovering errors in centering, under and over-caging, errors in dimensions, and operational malfunctions. Past experience, reaffirmed by the work on these accelerometers, has indicated that an accelerometer performing properly at 1 g will do so at lower and higher accelerations as well. The transit times may be measured from the oscilloscope (Tektronic 545) trace obtained with the moving vane described above, or more conveniently, with an electronic counter (Hewlett-Packard 522B). In the latter case the oscillatordetector circuit used in the low g drop test described below is used to trigger the counter on the sharp rise time of actual contact. The necessary gain was achieved with the transistor amplifier of Fig. 12. The circuit will trigger on a pulse of 0.2 volt with a fall time of 20 pjsecs/volt. The performance of two of the accelerometers which is typical of all four is given in Tables II and III for various orientations. The standard deviations in 10 trials are shown. The average transit times in the various positions may be compared with the predicted transit time of 16.09 milliseconds. -6.0v -6.O~v -6.0v -6.0 v -6.0v 47K I r -2.7 K 42.7 2.7K - 4II1I 2N1382 47 0 K K O UPUT ~~~0 —11 ~~~I la laIIl a I AA AOUIPUT 1090 pf Fig. 12. Transistor pulse amplifier used in drop test.

TABLE II ACCELEROMETER NO. 103 Transit Times at 1-g Acceleration Transit Distance: a1IZ1 16.99 16.91 16.93 16.89 16.91 16.85 16.89 16.90 16.95 16.99 169.22.09.01.02.03.01.07.03.02.03.07.oo49.0001.000ooo4.0009.0001.oo49.0009.01804.0184.050 in. 16.89 16.95 16.86 16.90 16.87 16.92 16.84 16.93 16.89 16.90 168.95 Calculated Time: Lag Time: 16.09 millisec 0.80 millisec 16.89 millisec.00.06.03.01.02.03.05.04.00.01.0000.0036.0009.0001.0004.0009.0025.0016.0000.0001.0101 16.76 16.73 16.70 16.80 16.95 16.79 16.87 16.88 16.79 16.85 168.09.05.08.11.01.14.02.06.07.02.04.0025.0064.0121.0001.0196.0004.0036.oo49.0004.0016.0516 /.0018 =.043 /.o01 =.033 /.0052 =.072.043 + 16.92 =.25%.033 + 16.89 =.191.072 + 16.81 =.42% 16.77 16.83 16.71 16.76 16.72 16.82 16.81 16.75 16.70 16.73 167.59.01.07.05.00.04.o6.05.01.06.05.0001.0049.0025.0000.0016.0036.0025.0001.0036.0009.0198 16.69 16.77 16.76 16.77 16.86 16.71 16.91 16.82 16.88 16.89 168.06.12.04.05.04.04.10.10.01.07.08 (1=,* — zIII.0144.0016.0025.0016.0016.0100.0100.0001.0049.0064.0531 16.78 16.84 16.85 16.81 16.88 16.89 16.85 16.91 16.89 16.87 168.57.08.02.01.05.02.03.01.05.03.01.0064.0004.0001.0025.0004.00ooo9.0001.0025.ooo0009.0001.0145 /.002 =.045 /.0055 =.075 /.0014 =.038.045 + 16.76 =.29%.073 + 16.81 =.43%.038 t 16.86 =.235 *900 17

TABLE III ACCELEROMETER NO. 104 Transit Times.050 in. Transit Distance: at 1-g Acceleration Calculated Time: Lag Time: 16.09 millisec 0.90 millisec 16.99 millisec Wz 16.93 17.11 17.02 17.09 17.07 17.20 17.02 16.97 17.01 16.97 170.39.11.07.02.05.03.16.02.07.03.06.0121.0049.0004.0025.0009.0256.0004.0049.0009.0036.0562 17.07 17.01 17.07 17.01 17.07 17.04 16.99 17.08 17.15 17.05 170.54.02.04.02.04.02.01.06.03.10.00.0004.0016.0004.0016.0004.0001.0036.0009.0100.0000.0190 17.21 17.24 17.29 17.22 17.23 17.17 17.19 17.19 17.17 17.21 172.12.00.03.08.01.02.04.02.02.04.00.0000.0009.0064.0001.0004.0016.0004.0004.0016.0000.0118 /.0056 =.07483.074 + 17.04 =.435 /.0019 =.04359.o44 + 17.05 =.26% ~ /.0o12 =.035.035 + 17.21 =.20% * 17.01 16.93 16.94 16.99 16.95 17.04 17.02 17.02 17.02 17.05 169.97.01.07.06.01.05.04.02.02 A.02.05.0001.0049.0056.0001.0025.0016.0004.ooo4.ooo4.0004.0004.0025.0165 16.94 16.94 16.97 16.94 16.93 16.95 16.95 16.99 16.97 17.01 169.59.02.02.01.02.03.01.01.03.01.05.ooo4.0004.0001.0004.0009.0001.0001.0009.0001.0025.0059 17.03 16.98 17.05 17.01 17.00 17.03 17.03 16.98 16.93 16.96 170.00.03.02.05.01.00.03.03.02.07.04.0009.0004.0025.0001.0000.0009.0009.0004.0049.0016.0126 /.00165 =.041 /. 00059 =..024 /.00126 =.036.041 + 17.00 =.24%.024 + 16.96 =.14%.036 + 17.00 =.21% *90~

3.2 LOW-ACCELERATION INVESTIGATIONS Because the low-acceleration performance of the accelerometer determines the maximum altitude of density measurement of the sphere experiment, considerable effort in the program was expended on this factoro Initial velocity imparted to the bobbin by the release mechanism and initial velocity imparted to the bobbin as a result of its center not coinciding with the center of spin of the sphere are the two important sources of difficulty at low accelerationso The latter problem is approached by precision construction and careful balancing of the sphere, a process discussed below Initial velocity imparted by the caging mechanism is a function of the design and quality of construction, In a completed accelerometer the initial velocity is most satisfactorily tested for in an implicit way by an over-all performance test in nearly free fall, ioeo, in an Attwood's machine. In the previous report a different approach to checking for initial velocity was described but was not applied to this design, one reason being the impracticality of removing the center section. Consideration was given also to the possibility of testing the accelerometers at zero g in aircraft flying special trajectories. The zero-g condition lasts long enough so that eventual contact of the bobbin may be used to calculate the initial velocity and hence the error to be expected from such an initial velocity when using the accelerometer at a small acceleration which is not zeroo The zero-g aircraft tests were beyond the scope of this contract and were not attempted. The most promising approach appeared to be to construct a new Attwood's machine with extended range. The range was extended toward the low-acceleration end by using a new pulley of lower moment of inertia and a more massive bullet than in the previous machine. The moment of inertia of the new pulley was measured by the method of Refo 35 that is, by measuring the time of fall of a small mass through a measured distance, the mass being attached to the pulley by a thread wrapped around the rim0 The moment of inertia is given by M1 r2T2(g - 2) I- = ---- 2S where MI = mass of small mass r = radius of pulley T = time of fall S = drop distance 19

I for the new pulley was found to be 662.64 gm cm2 compared with 1002.02 gm cm2 for the old. Now, using the new large bullet weighing 24,381 grams (vs. 7,171 for the old), a transit time for the accelerometer was found to be 446.6 milliseconds from / 2s(M r2 + I) gI where s = transit distance of accelerometer M = mass of large bullet This corresponds to an acceleration of a = 2s/t2 = 1.283 cm/sec2 or 0e00138 g. To avoid errors in the acceleration due to heavy leads to the accelerometer coils, these were eliminated by incorporating the batteries in the bullet. The solenoids were controlled by means of a photocell circuit operated by light passing through a small opening in the bullet casing. The circuit is shown in Fig. 13. The photo-cell A is a self-generating type. When it is energized, the 2N456A conducts and energizes the solenoid coils. Shortly after the bullet is released by a trigger, it passes out of the beam of light, thus de-energizing the coils and releasing the bobbin. The bobbin transit time signals are carried in three small (O.004-in.-diam) wires leading from the bullets The bullet is 33 in. long and made of standard 4-in. pipe and couplings. The center section, made of brass, houses the accelerometer and auxiliary circuitso The accelerometer cavity is located on the cago of the entire unit to minimize centrifugal acceleration due to rotation of the bullet in fall. A special fork drop latch was devised to keep horizontal motion to a minimum. See Fig. 14. The bullet, accelerometer, and control circuit are shown in Fig. 15. Figure 16 shows the accelerometer mounted in the control circuit rack. A catch basket was built to prevent damage to the accelerometer and circuits in the bullet. It was designed to keep impact accelerations below 80 go The construction is a tripod arrangement consisting of a shaped and padded ring free to slide vertically on three posts* The base of the ring is supported by three auto springs with shock absorbers inside. A large inverted pyramidal chute guides the bullet into the ring and prevents "fall over" damageo The catch pad is seen in Fig. 17. Impact accelerations were measured with a Massa Laboratories crystal accelerometer Model Ml91 having the following characteristics: sensitivity 42 20

-6.0 v - 6.0 v -6.0v 70 K loo0 S -6.0v -6.0 v It 1_...~~if a 10. 1 AI W NE 1090 %V Ie.sa A N414' -' n j —i -N 414 --— ^ T11 ^ I TI-^AA^ —'2N 456 A LIGHT 22 Fig. 13. Photocell and amplifier used to control accelerometer in drop tests. )1 tvl*t I FA - r////////l x X XXNXW///AINX XXXX XXX XXXXXX XX XX Fig. 14. Bullet release device for drop tests. 21

Fig. 15. Drop bullet, control circuit, accelerometer. Fig. 16. Accelerometer mounted in control circuit rack.

r iK# 1.1-7. Catch;~ basket., Fig 1-8. Drop test setup. 23

millivolts/g; dynamic range 0.001 to 1000 g; resonant frequency 56 kc; frequency range 10 to 50,000 cpso First drops indicated accelerations of 122 g's, but this was reduced to 45 g's by the addition of padding in the catch device. The bobbin transit time of 446.6 milliseconds required a drop distance of 4 feet which, together with the dimensions of the apparatus, resulted in mounting the release mechanism about 10 feet above the floor. A heavy steel shelf holding the drop mechanism was attached to a vertical steel beam imbedded in the wall of the buildingo The over-all drop test setup is illustrated in Fig. 18. In making successful drops, attention to all the details noted in describing the use of the previous Attwood's machine was of even greater importance. It is reasonable to state that the adjustments and sources of error in the Attwood's machine were more troublesome than in the accelerometer. Since good agreement with the predicted transit times was finally achieved, the technique must be considered successful, although it is not a "calibration" in the sense that the errors may be ascribed to the accelerometer and corrected for. The technique is really a check of proper operation, after which the accelerometer is taken to be an absolute device whose performance is predicted from calculations based on dimensions* A typical population of five measures of a predicted -transit time of 446,6 milliseconds (reading all digits presented by the counter) is: 444.27 448.87 45011l 447.92 455-53 Avg. 449.3 4o SPHERE MECHANICAL DESIGN 4.1 ACCELEROMETER MOUNTING Special consideration was given to mounting the accelerometer in the sphere because of the effect of the mount on bobbin centering and because the accelerometer acts as a diametral bolt which holds the sphere end bells tight to the center section against pressurizationo The exterior length of the accelerometer is about 0.020 in. shorter than the interior sphere diameter (cap to cap)o This dimension varies from sphere to sphere because of the accumulated effects of tolerances and strains, Centering of the accelerometer was accomplished as followso The accelerometer was fully seated in one sphere cap with the cap screw tightened a specified amount as measured by a torque wrench, The second cap was then tightened with equalntorqueo The first-eap screw is then loosened and the second one tightened with the same torque, the number of turns of the second screw required to do this being carefully noted. 24

The second cap screw is then backed off by one-half the measured turnso Finally the first cap screw is tightened with the same torque used throughout. With one atmosphere pressure inside the sphere and vacuum outside the total force carried by the accelerometer as a tension member is about 500 lb0 The accelerometer was designed in the first place to accommodate the load and did so reasonably well. However, small changes in transit time were sometimes noted between the pressurized and unpressurized conditionso The trouble was eliminated by separating the end caps of the accelerometer from the working parts and connecting the former with four 10-32 boltso These carried the tension load leaving the accelerometer working parts unstressedo 4.2 BALANCE The necessity for sphere balance has been discussed previouslyo In checking sphere balance it is assumed that the center of the bobbin is on the geometrical center of the sphereo Considering the tolerances used in machining the various parts, this is probably true to plus or minus one or two thousandths of an incho The balancing operation then becomes one of getting the center of gravity of the sphere as close to the geometrical center as possibleo In Table I of Refo 7 the balance of previous spheres is showno Distinction is made between the displacement of the cogo from the center and from the spin axis because, if the sphere preserved its angular momentum about the spin axis (i.e., the longitudinal axes of the rocket and accelerometer), it would be more important to keep the latter displacement small. The range of these displacements in 14 spheres were, respectively, 0003 ino to 0o050 ino and Oo001 ino to 0.025 in.; the later spheres were the better oneso In the spheres of this report several techniques and changes were resorted to for improved balance. First, the intervalometer motor which may have had a gyroscopic effect was replaced with an electronic circuito The only significant moving parts in the new spheres were the pick-up fingerso These are inherently symmetrical. However, the balance with the fingers advanced and retracted was checked and found unchanged. Another variable, the possible sloshing of electrolyte in the batteries was eliminated by having no excess electrolyteo Balance was measured with a Taylor Universal Static Balancing Machine seen in Fig. 19. It may be noted here that static balance of the sphere is all that applies because a freely falling body can rotate only about its c.go. i.oe. it has no constraining bearingso Rough balance of the sphere was obtained by properly located lead weights in the form of plates and washers placed in both the body and caps of the sphere. Three trimming weights, adjustable from the outside with the sphere "buttoned up" were incorporated in one of the end bells of the spheres. These were used for final balance just before flighto The Taylor machine is calibrated to read unbalance in ounce incheso This reading, divided by the weight of the sphere in ounces (about 175) gives the c.go displacement in thousandths of an inch. Corrections for the different masses of 25

Fig. 19. Sphere on balance testing machine. the individual spheres and for the mwass of the ada pting fixture must be mace. Errors due to out-of-roundness of the spheres and air drafts in the room were noted at the point of best balance. Everything considered the e.g. in all the spheres was found to be within about.:t: 0.002 in. of the geometrical center. This compares with about the same figure for the tolerance of construction of the sphere and also is about t; o as lose a the balance machine could be practicably read. It should be noted Ithat uith different fixtures, the sensitivity of the balance machine can be improved. In checking balance the sphere was merely rested in the cuplike adapter fixture of the balance machine so that balance about any axes (two at a time) could be read on the spirit level with great ease. The balance machine saved many hours of lab.or over the?revious system of suspending the sphere on a wire and observing it with a theodolite. 4, FLIGHT CONFIGURATION The four spheres in their final f-lighit configuration were nearly the same in mechanical construction as the earlier ones. The outside diameter remained at 7 inches. The antenna slot ad components were unchanged. Internally, the end bells had minor changes to aceommodate the new accelerometer end caps. One end bell was fitted with the externally adjustable balancing weights. The center sece tion had some webbing removed. Four function outlets were used as before but only the rotary switch was unchanged. The ball starting switch was redesigned, the pressure port modified and the monitor plug was wired for battery charging.

Figure 20 shows the accelerometer in the center section of the sphere (antenna facing camera)o The adjustable balance weights are in the right-hand bell with rough balance lead washers mounted underneath. In Fig. 21 the four function ports are shown. 5- SPHERE ELECTRONICS The complete falling-sphere system is shown in block-diagram form in Figo 22. Figures 23 through 27 show various sphere-construction detailso This system can conveniently be broken down into six sections as follows: 1. Accelerometer 2. Contact detector 35 Intervalometer 4. Modulator 5 Accelerometer drive 60 Transmitter The accelerometer with its two electrically isolated cavity sections has been completely described previously in this report. Along with the new accelerometer, is a newly devised capacity-actuated contact detector which provides a highly precise measurement of the time of accelerometer bobbin contact with the cavity wall, This is accomplished without the need for any physical connection between the bobbin and the rest of the accelerometero The intervalometer is the basic controller for the system. It initiates and controls the following series of events: lo Generates a periodic series of signals (called "clock pulses") which set the maximum rate at which the accelerometer can be recycled. These clock impulses are transmitted as 5-microsecond pulseso They serve to assist groundstation personnel in tracking the sphere during its trajectory with the groundstation telemetry antennas. The clock pulses are also useful as timing markers and should be particularly helpful if noise pulses are present on the telemetry recordso 2. Generates a signal in synchronism with the release of the bobbin by the accelerometer caging fingers. This signal is transmitted as a lO-microsecond "start" (i.e., "release") pulse. 3o Prevents the system from recycling the accelerometer until after a "stop" (ieeo, "contact") pulse has been received. If 2.5 seconds after bobbin release a stop pulse has not appeared a- stop pulse is simulated within the intervalometer and the accelerometer is recycled. The stop pulse is transmitted as a 20-microsecond pulse which completes the series of pulses trans 27

Fig. 20 Aceelromoeter mtounted -In er nhe dj;,- tdl &:.l;abI.e butlanc Ang weights. T t _1, 1~~,~~ C: I unc. o.::~~'~~

INTERVALOMETER CLOCK —-— 1 INPUT "AND" I I I F I I I IM.BINARYS RESETE I-AMP-. GAT__ OVER DELAY RIDE GENERAT I CONTACT DETECTOR r —-. —--. r — __ _ _ T — -11 -- I R.F TUNED DETECTOR ---- CIRCUIT AMPLIFIER AMPLIFIER TOI MC I I 20 MCROMICROSE XTAL SE — CR I SCILLATOR I PFN PFN MODULATOR r —--- r\3 L_ __....___1 _,_.. ACCELEROMETER ACCELEROMETER DRIVE.. ___... —--—, ACCELER-_.P _ I POWER_ OMETER 2cK SWIrTCHk d —MLIFIEagr L____d L__________r l Fig. 22. Sphere circuit block diagram. TRANSMITTER ^ I 0_,r - - - AI 3LUI r7- NTENNA I NI | I-TRANSMIT- 1 400 MC. I TER I I I I l

Fig. 25. Sphere with top bell removed. Fig. 24i. Sphere with bottom bell removed. 50

Fig. 25. Sphere wi'th to) bell. remomve4d, s:ifde view. Fig. 26. Bottom()~ view of:' sphele o wx:.th acceleromalcbet removed, output pu.lse'tr:ansf~:O o ntoer ex]p:osed. 51I

Fig. 27. Sphere with top electronics rings exposed. mitted and received at the ground station. In addition to the above, the intevalometer ig nore r min r detector output pulses which omentime1s occur hlen the accelerometer cagingr fingers release the bobbin* It also neglects all outputs due to bobbin bouncing after initial contact has toccurred. This systr has be d des cbd s c retionatry because the acelerometer does not normally recycle until. after a stop pulso is generated by the bobbin. The disc retion s not unlimited, however. After a stop pulse has been generated the andi,'gate xremains closed until the succeeding clock pulse appears. Thus, ac elerormeter ecycling is sychronized with the basic clock rarte o' integrI tervals.s thereof * The above also applies whenever a mandatory aeel^eromeTter recyce ocurs af te a stop pul s does not appear within: 2*5 seconds from bobbin release. It will be noted from th is description that in contrast to eaxrlier falling spheres, a clock pulse has been added and both start iand stop pulses are also "width-coded.*" Povision hias also been mande:for longer bobbin transit times and faster accelerometeor recycling tlan was feasible in the past. These features all assist in obtaininxg addi tional yupper-atmosphere-density data over a ider range of altitudes, an2d facilitate data handling rhen either manual or automatic data reduction is undertaken. All sections of Fig. 22 except the transmitter are described in detail below. The transmitter, operating at 400 megacycles With a peak pulse power output of approximately 15 vattst, is identical ith th ose used in earlier falling spheres. The reader is referred to Ref. 6 for additional details. 52

Ci Ct (a) (b) Fig. 28. Accelerometer bobbin-cavity capacitances. 5.1 CONTACT DETECTOR The contact detector utilizes the change in capacitance between the electrically conductive bobbin and its two surrounding and electrically isolated cavities. Figures 28a and 28b illustrate the bobbin-cavity relationship and the equivalent electrical circuit. For simplicity, all fixed and stray C that does not vary with bobbin position within the cavities is not shown. Consider now the variation of capacitance between terminals A and B of Fig. 28 as the bobbin moves in a direction which tends to increase C1 while C2 remains approximately constant. The capacity between A and B is given by: 1 1 1 CAB C C2 As the bobbin approaches the "close" cavity, Cl increases, asymptotically approaching infinity in the limit. This means that at the instant of contact: CAB = C2 A similar result occurs even when the cavities are the same size and C2 is allowed to vary (as it generally does) in response to bobbin movement. Since C1 varies approximately linearly with respect to the reciprocal of the distance between the bobbin and the cavity wall, it seems that CAB would never be subJect to discrete changes in value, even at bobbin contact. Consequently, if the capacitance variation is transformed into a varying voltage by appropriate associated circuitry, the rate of change of voltage with respect to bobbin position would be a function of bobbin velocity, and thus acceleration as well. 33

In fact, there should not be a pulse output at all, particularly at low acceleration levels. Fortunately, however, the above speculation has proved to be incorrect in actual practice. Extensive testing of the accelerometer and its contact detector at 1 g and at very low acceleration levels indicates that a contact pulse is always generated and its amplitude is independent of bobbin velocity. The contact-detector circuitry shown in Fig. 29 consists of a 7.0-megacycle crystal-controlled transistor oscillator loosely coupled to a parallelresonant tuned circuit. A peak responding diode detector and the accelerometer cavities are also connected across the parallel resonant circuit. C24 is adjusted so that when the bobbin is in its centered position and uncaged. the circuit is tuned somewhat off resonance. Any change in capacitance caused by bobbin movement will now affect the tuning of the resonant circuit and thus cause a change in the d-c output voltage of the peak responding diode detector. The proposed mechanism responsible for the sharp discontinuity always observed upon bobbin to cavity contact is dielectric breakdown of the air when the distance between the bobbin and cavity becomes sufficiently small. In this event, the discontinuity at "contact" really occurs slightly before actual physical contact. However, the error thus involved amounts to less than 0.1% of the total transit distance. Extensive testing to confirm the above hypothesis has not been undertaken, but some experimentation and calculations have definitely indicated the possibility of a breakdown phenomenon. The presence of C25 provides for more uniform contact pulses as a function of accelerometer orientation. When correctly adjusted, the maximum detector output is approximately 7 volts, which results in contact-pulse amplitudes of approximately 3 volts. 5 2 INTERVALOMETER The operation of the intervalometer can be followed by referring to the circuit diagram, Fig. 29. To begin, assume that the circuit has been turned on, several recyclings have occurred, and that the bobbin is in transit. During transit, the binary is so set that the "AND" gate is turned off-io.e., no pulses can pass through the "AND" gate to trigger the RESET circuit. The clock circuit generates a pulse every 350 milliseconds which triggers the "clock pulse" modulator, The modulator generates a 5-Pisec pulse and thus a "clock" pulse is transmittedo When the bobbin makes contact with the cavity wall, a contact pulse is generated. This pulse, in the form of a "step" in d-c voltage, is transformed into an "impulse" (no d.-c level) by differentiation and amplification in the input circuit. This impulse triggers the controlling binary (a bistable multivibrator) into its opposite state. 34

INPUT BINARY "AND" CLOCK RESET OVERRIDE DELAY AMP. AMP - 6 VOLTS (REG.)_.OS005f' F I} * RI 42.7K — I2N138 R4 T'I _? I.001, Tf IOK 1.8KL R2 R3 R5 2.t K R7 N1090 >IOK R6 10 K 10 K 2N414 -- - - - IOK IOK -"Pf C3 Rg RI 2.7K K2.7K o470 K CIOK 1Rl2 R13 "R14 R15 R16 D, D2 C4 5 IN482A IN482A — + - / loaf t 2.ajAr 2N414 2N41 —-- n ri TsS -- 4.7K7IK I 22011 5.6K 220n 17R R18< R19 R20 R24< ^ C6 4.7K - R23 47Of \ /\ ~~2N414y N1231 2N41 315K 4.7K Rp| R22 0; IN482A 2.0K R27 l IOK D5 lOK <R 3 "K 1 —--- C';C N482, A0K' H ^-^- <R32 ClO^^-U^^ R2.7 K 100K 3 R R TitT2; 28!8 f29 22K V ^ Cg~~~~~~~~~ O 22* f' I -I — N-I JF _5l7^150 K;9 K lK 22K 1.8 R3. IN482.1 3.A, R33 R'4 2 R36 R38 __ L~ N1090;IOOK'R40 lOOK R45 4.7 K T T4 R —A/ —% 2NN1090 R46 C16 Ir I 470K 1 1 C 4.7 K T'3 C12.01)lf R41 CI II~ I @470 I,,f -. IK > >22K 4 L R39 R42 R43 1.) 11090 I4OK4 R44 59 I.0olIf 4 4 4.IRf IOK.22K < R47 > R48:470 J' R49 J_ 1 1 SELF RESONANT At 7.0 MC. — t"HT.I..j I 27Cn o100,,. L.26 100 A9f INIOOA D14-ZOf C25 I' -7. OV.(REG.), n %.^4- 625,,f S15K 7.0MC 4-251 TR2 TR I 22 T21 68 CR -. 9 2N370 - -24 f Tz23 C23 I OOjAsf 2J80A Q 3.9K ol R69 R67 IOOQ AAA IN754A L.Olu.f Z,(7v) TC D,3 19 -9 VOLTS < LTS | 47 A IN2071b ACC. R63 D,2 DRIVE N 2N184 1 2N456 0 22 <:OOQ A:OOQ I 64 R65 >IK r 62 -9 VOLTS,IOK [ -85 VOLTS 221 I —-- FISHER D5315U4A.OOl1uIf R B+ C II2A TR59 400 MC. TRANSMITTER Ci e1 IN 4A I R54 -9VOLTS R60 IN2071 - J A1 51 64: PT2)' |-5.68^ J~ PT2J\ DIO -a I I \ I rpT FIL-. 3 (ww) I/2w(ww) -2NI90\ TECHNITROL IN482A I %n TI \03A PT2 73WA PTI "9 4gn ~I - NN O71 - -- I, Tie~ I - 1_1 I I R52 ilon <o n _ PULSE MODULATOR G.) IK 0 ~ PFN- 21041 —: OPOLE A on' _I PULL-OFF PLUG - CHARGE+ A 0 —--- + D021 CHARGE- B 0 MONITOR- C -- - - -TO 9V. BATTERY ISK R7 I MONITOR+ D 0- 76 ACC. DRIVE CONTACT DETECTOR -9 VOLTS r 22nX -AAA ( RE( R75 IN752j (6.0 V.)D2o 20.f 1' C32 I I I -POLE B IO I I I c POLE CT OFF IN482A 47K R51 - 9 VOLTS 0 0 0 m ~ / BAL TEST FLY FUNCTION SWITCH iT BATTERY SUPPLY FSf (CLOSED IN FLIGHT) FLY SWITCH MODULATION SUPPLY Fig.. 29. Sphere-circuit schematic. 35

This change results in: 1. a 20-4sec pulse transmitted as a "stop" (i.e.,contact) pulse; 2. opening of the "AND" gate so that subsequent pulses can pass through; 3. resetting the circuit that provides for the simulated contact pulseo As a result of (2), the next clock pulse will be allowed to pass through the "AND" gate and trigger the RESET circuit. The RESET is a mono-stable multivibrator, with a period of 150 milliseconds, with the primary purpose of energizing the accelerometer drive circuit for that period of timeo At the end of this period-the reset flips back to its stable state, the accelerometer drive circuit is de-energized, and the bobbin is releasedo Upon returning to its normal state the RESET circuit triggers the DELAY circuit, another mono-stable multivibrator, into its quasi-stable state. The duration of the delay circuit period is about 2o5 milliseconds which delays resetting the controlling binary by this amount of time after the bobbin pickup fingers have been de-energized. As the delay circuit returns to its normal state it resets the controlling binary to its original state so that it may be triggered by the next contact pulse. The delay circuit is required because the fingers do not pull free of the bobbin the instant that the accelerometer coils are de-energized. In fact, it takes from 0.7 to o.5 milliseconds for this to take place, depending on the finger spring adjustmentso Occasionally on this break-away, minute bouncing between fingers and bobbin may occur. This bouncing is sufficient to cause small "contact pulses," which can appear to the intervalometer as true contact pulses if some way is not devised to ignore them. This is accomplished by the delay circuit which does not set the binary to accept a contact pulse until well after the fingers have pulled away from the bobbin. Upon triggering of the delay circuit,the step voltage so generated is amplified and in turn triggers the 10-microsecond modulator, and a "start pulse" is transmitted. This occurs at the same instant that the accelerometer coils are de-energized. To correct for true release (when fingers withdraw from bobbin), the time it takes for the fingers to withdraw (the 0.7 to 1.5 milliseconds noted above) from the bobbin is measured accurately and is subtracted from all subsequent transit-time data points. This lag time apparently remains constant for a given adjustment of accelerometer springso As described so far, if a contact pulse failed to appear or was inadequate for initiating circuit action in the intervalometer, the accelerometer would never again be recycledo To insure against this disastrous possibility, a circuit is utilized that provides automatic recycling in the form of an over-ride circuit. This, again, is a monostable multivibrator, with a time period of 2o5 seconds. The over-ride is triggered into its quasi-stable state by the binary as the binary is being reset. When a contact pulse flips the binary, the over 37

ride is returned to normal by the binary(53) abovej. However, if no contact pulse appears after 2~5 seconds have passed, the over-ride circuit automatically returns to its normal state, and upon so doing triggers the binary in exactly the same manner as would a contact pulse. From here the operation of the system is identical to that of a true contact-pulse-triggered binary action. With this over-ride circuit the system keeps on operating even if a contact pulse is missed, Thus once the system gets started it will keep recycling the accelerometer whether or not a contact pulse is received. The intervalometer system always starts correctly because the normal action of monostable multivibrators is to be momentarily in their quasi-stable state when power is first turned on. They remain in this state for 1/4 to 1/2 of their normal period, then flip back into their normal state. Of course, when the delay circuit functions, it sets the binary in the proper state to accept a contact pulse; and this in turn sets the over-ride in its proper state to carry out its functiono Meanwhile this turn-on action is taking place in the reset circuit which picks up the bobbin, centers it, and then releases it when the reset circuit returns to its normal state. From this point on circuit operation is as described aboveo Referring to the detailed circuit diagram, the input circuit consists of an emitter follower buffer stage, Tl, followed by a differentiating-amplifying configuration. The primary purpose of C is as a d-c block; however, along with the approximately 50K input impedance of the buffer stage, it helps to differentiate the contact pulses C2 along with R4 and the input impedance of T2 constitute the main differentiating circuit. T2 amplifies the differentiated waveform and passes it on to the base of T3s'The action of T2 is similar to that of a switch, in that it is normally cut off, but in heavy conduction during the contact pulseo The purpose of the input circuit is to take the d-c voltage containing the contact pulse, and change the contact pulse into a single "spike" waveform with which to trigger the binary. C3 is present to block any d-c current which might tend to flow into the base of T3, and R5 is a current limiting resistor to protect the base of T3 against high currents when T2 and T3 are both conducting. The binary is a standard Eccles-Jordan type bistable multivibrator Cornmutating capacitors were found unnecessary since switching speed without them was less than 2 microseconds, and speeds greater than this in this part of the system were not neededo Since most signals to and from the binary are of a pulse-type nature, capacitor coupling is usedo The exception is the coupling to the "AND" gate, where standard low-level d-c logic has been employed. Proper operation of the "AND" gate depends upon a greater voltage across D1 in series with T4 during conduction than appears across D3 in series with TT during conductiono The binary is triggered by a contact pulse when T4 is conducting and T3 is cutoff. The contact pulse appears as a negative spike at the collector of 38

T29 and thus is transmitted to the base of T3. This action turns T3 on, thus cutting T4 off. This in turn cuts D1 off. However, if T5 is still conducting, essentially all the 6 volts appear across Rise But when T5 becomes cut off, D2 also is cut off and current then flows through R21, thus turning T7 on and triggering the reset circuit into its quasi-stable state. Thus when either T4 or T5 is conducting, the reset circuit cannot be triggered into its quasi-stable state; but when both T4 and T5 are cut off, the reset is triggered and the accelerometer is thereby recycled, T5 is part of the clock circuit, which is an unsymmetrical astable multivibratoro The clock has a period of 350 milliseconds, T5 being cut off during the 10-millisecond clock pulses Due to the extremely long time constant of C4 and R15, and the resulting low current in R15, Ico leakage current in Tr resulted in very poor temperature stability of the clock circuit when germanium transistors were usede We had to resort to silicon transistors, which eliminated much of the temperature instability. Still better stability was obtained by choosing capacitors for C4 and Cs (primarily 04) with temperature coefficients which would best offset the transistor temperature dependence. The resulting circuit exhibited a total change of less than.5% period variation over a 40oF temperature range. Careful temperature compensation of T5 was unnecessary due to the relatively short time constants involved with this transistor. The reset circuit is a standard monostable multivibrator, with one slight modification. Normally a well-designed monostable multivibrator requires two separate power supplies: the normal collector supply voltage, and an opposite polarity supply for base bias through R22e If transistor T7, does not remain in the cut-off state, the circuit may become unstable and tend to oscillate. Because of weight and space-saving considerations, it was not practical to consider a separate power supply, and other methods of accomplishing the same end were investigated. The method finally settled upon was that of simulating an additional low-voltage power supply as illustrated by the Rla, D3 combination. D3 is a silicon diode, and in the configuration as used always has a forward conduction voltage drop across it of about 0.5 volt. With R22 returned to ground, T7 will always remain in the- cut-off state as long as TS is conducting and no trigger pulse is applied. This method exhibits no indication of instability and has been extremely reliable. The reset circuit is a fairly low impedance (high-power) circuito By making this a relatively high-powered logic circuit, only one intermediate stage of current amplification (T21) is needed to drive switching transistor T22 used to supply the 3 amps necessary to recycle the accelerometer0 The output point of the reset circuit is the T8 collector, which drives the base of T21 in the accelerometer drive circuit and also triggers the base of T11 in the delay circuits 39

The over-ride circuit is very similar to the reset circuit. A point of interest about the over-ride circuit is the decoupling network, consisting of R27 and C7, associated with the collector of T9 and the base of To10 The override circuit was purposely made a high-impedance circuit since the drive requirements on it were small. Consequently low-level pulses can trigger this circuit. In the course of operation of all the circuits in the system, rather high-level pulses drop the terminal voltage of the battery source enough to produce small pulses on the 6-volt regulated line. Occasionally these pulses become large enough to trigger the base of T1o through R28 and R29o The decoupling network of R27 and C7 eliminated this problem. No other circuit was faced with this problem as no other circuit attained nearly as high an impedance level as did TIo of the over-ride circuit. R26 and D4 serve the same purpose in the over-ride circuit as RIa and D3 do in the reset circuito The binary is the drive and load circuit for the over-rideo The delay circuit is also a standard monostable multivibrator, with D5 and R33 serving the same purpose in this circuit as D3 and R1s do in the reset circuito The biggest difference between this circuit, the reset and the override circuits is that NPN transistors are used here, whereas PNP transistors are used in the other two circuitso PNP transistors were arbitrarily selected for most of the circuitry. However, NPN transistors were employed wherever their use would simplify triggering, driving, or any other phase of the logic systemo The delay circuit is such a case. The triggering pulses available, and the output pulses needed, suggested that NPN circuitry would be simpler and more efficient in this caseo The delay circuit is triggered into its quasi-stable state by the collector of T8 of the reset circuit through R25 and CIo as the quasi-stable state time of the reset circuit terminates and Ts switches from the "off" state to the "on" state. The step voltage generated as the delay circuit switches into its quasi-stable state triggers the 10 microsecond pulse modulator and a start pulse is transmitted. After 2o5 milliseconds the delay circuit switches back to its normal state, and upon so doing provides a negative pulse to the base of T4 through Rs9 and C29 turning T4 back on (and T3 off) so that the next contact pulse can again turn T3 on. This latter action by the delay circuit also aids in turning Tio of the over-ride circuit off (thus initiating the quasi-stable state in the over-ride circuit)' but the main drive for this action comes at the base of T9 from the binary through C17 as the binary is being reset by the delay circuit. 5.3 MODULATOR The section of the system block diagram entitled modulator processes the stop, start, and clock pulses so that they may be transmitted as 20-, 10-, and 5-ksec pulses, respectively. The stop-pulse step voltage is amplified and impressed upon a 20-jtsec pulse forming network (PFN), and the start-pulse step 4o

voltage is amplified and impressed upon a 10-Vsec PFNo The clock pulses are used to-trigger a 5-.sec blocking oscillatoro The resulting 20- 9 10-, and 5-p sec pulses are then used to drive high-power switches so that high-power 20-, 10-, and 5-=isec pulses are placed across the primary of a power pulse transformero The secondary of this transformer provides the B+ voltage for the transmittere Referring to the circuit diagram, the start-pulse step voltage is amplified by T13 and its associated circuitry; the stop-pulse step voltage, by T14o The principal components of the two pulse-forming networks are~ (1) T15 and FFNI for the l0-osec start pulse, and (2) T16 and PFN2 for the 20-psec stop pulseo Operation of the 10-~tsec unit is as follows: Due to capacitor C14 and resistor R4sh Ti5 is normally cut-offo PFN1i being a lumped-constant open-circuit 5-Resec delay line consisting of series-parallel combinations of inductance and capacitance, charges up to the full 9 volts supply voltage, so that no current flows in R52 and D6o The charging path of PFNi is through R50 and R52s The amplified start-pulse step voltage is applied to the base of T,59 thereby switching T15 into the saturated conduction stateo PFNi then discharges through T15, D6, and the primary of PT, for 10pLseco The reflected PTi load, along with the saturation resistance of T15 and R52, constitutes the proper load for PFNio Thus no reflections occur and one-half of the original PFNI voltage (4o5 volts less the transistor drop of.about 0ol volt) appears across the primary of PT for 10-iseco Upon complete discharge of PFNi the voltage across PT1 terminates abruptly, D6 becomes nonconducting, and the saturating current for T15 is supplied by the supply voltage (-9 volts) through R50o Finally, when the drive current on the base of T15 terminates, T15 again cuts off, and PFNi charges up to -9 volts again through R50 and R52s and the circuit is ready for another start pulseo Recharging'time of the PFN is about l0-4seco Operation of the 20-Vsec unit is identical except that it is triggered by the amplified stop pulse, step voltage and has a 20- sec outputo The diodes D6 and D7 prevent pulse feedback to the opposite PFNo The pulse transformer PT1 transforms the 404-volt, 1O-psec (or 20-pLsec pulse down to l-lo5 voltso In this manner maximum current drive to the bases of T19 and T20 is achieved along with approximately properly matching the 100-ohm characteristic impedance of the delay lineso Ds prevents excessive back voltage on the base to emitter junctions of TIe and T20 due to induced voltages in the secondary of PTIo The 10- and 20- sec signals applied to the bases of T'l and T20 are sufficient to drive these normally cut-off transistors into conduction saturationo This forces the complete 85 volts (less the voltages drops in R55 and R56 and the drops in T19 and T2o) across the primary of PT3, the output power pulse transformero This voltage is transformed by PT3 into the 360 volts needed for the transmitter B+o D10 serves the same sort of function as D8 in that it protects T19 and T20 against wrong polarity voltages from collector to emittero The purpose of R55 and R56 is equalization of currents through T13 and T20o Total current through the primary of PT during conduction is 008 ampo 41

Saturation resistance of the 2N1715 is about 10 ohmso Absolute maximum rated current for 2N1715 is 1.0 ampo However, it was found that 2-2N1715 used in the parallel configuration as shown in the circuit diagram required less base drive current for the same saturation characteristics-thus the use of the parallel configuration. The use of 2 transistors in place of 1 also affords some degree of redundancy. The 5-Lsec pulse is generated when the leading edge of the 10-millisecond clock pulse triggers the base of T17 through R61 and CIs. The configuration of T17 is that of a blocking oscillator with the feedback winding in the emittero The output of the blocking oscillator is applied between base and emitter of Ti8, and is a 5-usec pulse of enough power to switch Tie from cut-off into conduction saturationo This, of course, applies a 5- sec B+ pulse on the transmitter in exactly the same manner that T19 and T20 do the 10- and 20-pisec pulseso Width control of the transmitted clock pulse is possible by adjusting the value of resistance R59o As R59 is made smaller, the pulse width becomes larger and vice versa. D1, attenuates reverse polarity voltages from the base to emitter of T17, whereas Reo is needed to insure proper cut-off (and thus stability) of T17. D9 also is a back-swing protecting diode. R58 was found necessary to provide adequate damping and eliminate ringingo The delay lines for the 10- and 20-psec pulses and a blocking oscillator for the 5-Lsec clock pulses were chosen as followso It was desired to have extremely stable width pulses for start and stop time markerso Constant width clock pulses were much less necessary, since clock pulses serve only as convenient ground-station telemetry tracking signals and accelerometer recycle time reference markerso Pulse forming by delay lines involves no active elements in pulse-width determination and thus reliable constant width pulses can be much more easily attained0 Hence the 10- and 20->sec pulses were formed in this mannero On the other hand, pulse forming with blocking oscillators is simpler to design, and much less expensiveo Since the characteristics of blocking oscillator pulses were entirely satisfactory for the clock-pulse application, these pulses were formed in this manner. 5.4 ACCELEROMETER DRIVE The accelerometer drive circuit is the circuit used to energize the acs celerometer coils used in recentering the bobbin. It consists of T21 for drivesignal amplification and T22 as the high current switch. The 150-millisecond reset pulse is delivered to the base of T21 through Re62 In flight the switch FS1 is closed so that the emitter of T21 is tied directly to the base of T22T21 is normally cut-off, so that no current flows into the base of T22* The purpose of R65 is to keep the emitter of T21 from floating when FS, is open. The purpose of R64 is to keep the base of T22 from floating when FS5 is openo With FS, closed the R64-R65 parallel combination prevents IEo of T21 from flowing in the base of T22 and becoming amplified0 R63 is simply a current-limit 42

ing resistor to keep from over-driving T21o The 150-msec current pulse from T21 is amplified again in T22, and flows through. the accelerometer coils (which are in parallel with D ) causing the accelerometer fingers to function. The purpose of D12 is to eliminate any inductive pulses which could harm T22o The only critical portion of this circuit is the parallel combination of RB4 and R65. This value must be large enough so that most of the T21 emitter saturation current flows into the base of T22, but must be small enough so that the IEo in T21 cannot raise the base voltage of T22 high enough to cause T22 to begin conductiono The value of this resistance is, then, dependent upon the magnitude of IEo of T21 and the emitter-to-base cut-off voltage of T220 The design values shown have been adequate for all combinations of 2N456A transistors used for T22 and 2N1184 transistors used for T21o 5 5 PULL- OFF. PLUG This assembly provides for operating the sphere system with an external power source, and for monitoring the sphere power supply voltage, prior to flight~ This is done so that the batteries in the sphere do not have to supply power while the system is operating during the rocket firing count-down. It also allows a continuous pre-flight check on the sphere battery voltageo This plug is pulled out a few seconds before firing and power is then supplied by the internal batterieso An extra benefit of this scheme is that it actually allows observation of proper circuit functioning during the time that the pulloff plug is engaged-thereby giving an additional, last-minute check on the proper functioning of the system. The plug is a pressurized 4-terminal plug, two terminals for providing the load-floating current, and two for monitoring the battery voltageo These are shown on the circuit diagramo The components D21 and R76 are needed for protection against accidental pull-off plug pin shorts during or after the plug is pulled from the sphere prior to rocket takeoff. 5.6 POWER SUPPLIES All power supplies used in the system are derived from the bank of 6 HR-1 silvercells in series (BI through B6)0 Each of these cells individually maintains a nominal voltage of lo5 volts across its terminals-making 9 volts for the series banko This 9 volts is used directly for the -9 volt supplies indicated throughout the entire circuit diagram0 The regulated 6-volt supply is obtained across D20 and filter capacitor C32. The diode D20 is a 600-volt zener diode. The dynamic impedance of Do2 (IN752) is about 7 ohms; thus regulation is not perfect because the series dropping resistor R75 is only 22 ohms However, it has proven very adequate for this applicationo The 85-volt modulator supply voltage is obtained from the 9-volt silvercell supply by means of a d-c to d-c convertero This circuitry is shown o:,.

the circuit diagram with the label "Modulator Supplyo" The converter circuit, employing transistors T24 and T25, is of conventional designo The circuitry following the power supply filter consists of dropping resistor R74 a smoothing capacitor C31 and an 85-volt zener diode D1is The principal purpose of the zener diode is not for regulating purposes, but rather to insure that the nominal 85-volt supply voltage never substantially exceeds 85 volts and thus prevents possible damage to transistors T18, Ts19 and T20 in the modulating unito The converter transformer TR3 was designed for operation with an input voltage of 28 voltso Since it is used here with an input voltage of 9 volts, its design output voltage is also reduced by 9/28 to approximately 85 voltso 5. 7 SWITCHING There are two mechanical switches in the.complete sphere assemblyo One is the simple "open or closed" switch (FSI) used to keep the accelerometer inactive until the sphere is ejected from the nose cone, and herein referred to as the "fly switch." The other mechanical switch is a clarostat three-pole, four-position function switcho The fly switch is a special pressurized subminiature microswitch, used in the "open when depressed-closed when released" configuration, It is incorporated in the accelerometer drive circuit as shown between the emitter of T21 and the base of T22, and actuated by a spring-loaded plunger that, when released, is flush with the smooth outside surface of the sphereo To depress the switch mechanism, the plunger is pushed inwards about an eighth of an incho As the sphere is being installed in the rocket nose cone, a 1/8-in. steel ball is inserted into the hollow in the sphere caused by depressing the plunger, and bears against an assembly in the nose cone, thereby keeping the fly switch plunger depressed and the switch openo Upon ejection of the sphere from the nose cone the assembly retaining the steel ball is no longer present, and the spring forces the plunger back outwards, thus closing the fly switcho The purpose of the fly switch is, of course, to keep the accelerometer inoperative during the high accelerations of rocket take-off, second-stage ignition, and high-velocity drag forces so as- to avoid damage to the pick-up finger bearingso Normal operation of the entire system commences upon ejection of the sphere from the nose cone and the closure of the fly switch FSo Until closures of FSi the logic circuitry is recycling every 2.5 seconds due to the action of the over-ride circuit The function switch is used for turning power to the circuitry on and off, and to set up certain testing conditions. The functions performed by the function switch, as shown on the circuit diagram, are: Position 1 Off Position 2 Balance Position 3 Test Position 4 Fly 44

Position 1 simply disconnects all power from the circuits. Position 2 energizes only the accelerometer pick up coils by placing them directly across the 9-volt supply. In this position it is possible to balance the sphere with the accelerometer bobbin centered. Position 3 puts the entire system in normal functioning configuration with the single exception that the fly switch is bypassed with a short circuit. This allows test cycling of the accelerometer with the sphere in the nose cone. Position 4 engages the entire system in normal functioning configuration and ready for flight. The operation of all circuits under this configuration is as described in detail above. It will be noticed that one of the poles, A or B, is apparently redundant in that the pole arms of these two both go to the positive side of the 9-volt supply. The purpose of pole B is to energize the converter circuitry whereas pole A switches power to all other parts of the system. It was found that, if this separation was not employed, interfering pulses were present on the supply lines. These pulses were found to come from the (relatively) highcurrent, high-speed switching action of transistors T24 and T25 in the converter. The best solution to this was found to be the use of separate supply lines (both positive and negative) for the converter and for the other circuitry. This makes necessary the use of both A and B poles as shown. Pole B was given the additional task of connecting the accelerometer coils across the 9-volt supply. Pole C connects the fly switch properly in all positions. It will also be noticed that the "ground" side of the circuitry (the positive side of the battery) has been on-off-switched rather than the "hot" side (negative side of the 9-volt supply). This was done to eliminate the need for a fourth pole when using the function switch in position 2, and creates no system problems. 5.8 COMPONENTS (See Table IV) 5.8.1. Transistors.-Most of the circuitry is logic switching circuitry using d-c level logic with pulse triggering. For this application transistors were needed that had the following characteristics: (1) high switch speeds (2) high current capabilities (5) large hFE for both large and small collector currents (4) 8 to 10 volts rating for reverse base to emitter bias (min) (5) economical (6) either NPN or PNP (7) either silicon or germanium Both germanium and silicon transistors (and NPN and PNP) were available to meet the first three specifications. However, to meet specifications (4) and (5), we had to resort to germanium. After an extensive search and testing of various transistor types, PNP type 2N414 was chosen as the main transistor 45

TABLE IV* COMPONENT FUNCTION Transistor and Diode T, T2 T3(T4) Ts(T ) T7(Ts) T((T 1o) T Il(T,) T l3(T 14) Application Emitter follower Amplifier-switch Bistable Multivibrator Astable Multivibrator Monostable Multivibrator Monostable Multivibrator Monostable Multivibrator Amplifier switch Switch Blocking Oscillator Power Switch Amplifier switch (power switch) Oscillator Converter Oscillator "AND" Logic "Power Supplies" "OR" Logic Reverse Voltage-Protection Zener diode Rectifier Bridge Rectifier Configuration Quiescent Biasing. Emitter = -5 volts Cut-off; saturated when contact pulse occurs Cut-off (saturated); saturates (cuts off) when triggered by contact pulse Saturated (cut-off) during most of 350 ms cycle. Cut off (saturated) during 10 ms clock pulse Cut-off (saturated); saturated (cutoff) during 150 ms recycling interval Saturated (cut off)-(quasi stable state-2.5 sec) Cut off (saturated)-Csaturates (cuts off) for 2.5 ms at bobbin release] Cut off; saturates for start (stop) pulse Cut off; saturated during start (stop) pulse Cut off; saturated during 5 usec clock pulse Cut off; saturated during 5 Usec (start and stop) clock pulse(s) Cut off-saturated during 150 ms recycling Class A biased Alternately saturated and cut off Normally conducting Always conducting at least 6 ma Normally nonconducting Normally nonconducting 7.0 (6.0) volt regulated supply Low forward voltage drop High voltage-high current Tls(Ts9,T2o) T2l(T22) T23 T24(T25) Di(D2) D3,D4,Ds Ds,D7 D9 Dlo Dll D11 D1 D13(D0o) D14 D15 Da2 D17 D18 Di9 Zener Diode Keeps supply from exceeding 85 volts DAi Reverse Current Switch Protects against accidental shorting of pull-off terminals *The "application" column the type of circuit in which the element is used. The "configuration" column comments on operation of the element during the interval of time that the bobbin is in the process of freely falling toward the cavity wall unless specified.

element for the logic circuitry. A 2N1090 was chosen as the NPN counterpart when complementary circuitry was required. In several places other transistor types were used when special applications or characteristics were required. The 2N1381 used for T1 has a higher hFE in the collector current region over which it operates than does the 2N414 or 2N1090, and is thus better for the input stage application. The 2N1231 used for Te is another special-purpose transistor. It is a PNP silicon switching transistor with a reasonably high hFE. It exhibits greater temperature stability than is possible with germanium transistors. The 2N570 used for T23 is a high-frequency transistor designed specifically for HF oscillator applications. The characteristics of the 2N1184 used for T24 and T25 (and also for T21) are very similar to those of the 2N414 except that it has a considerably higher current and power rating. The 2N456A used for T22 is a high current switching transistor. The 3 amperes flowing through the accelerometer coils all pass through T22, thus necessitating such a transistor. The 2N1715, used for T18, T19, and T20, is perhaps the most special of all the transistors used. For this application a transistor was needed that had: (1) a high emitter to collector voltage rating (90 volts/min) (2) a high collector current rating (1.0 amp) (3) high switching speed, and low storage time (less than l.0Lsec) Most transistors failed to qualify on at least one of the above specifications. The 2N1715, with a voltage rating of 100 volts, a current rating of 1.0 amp, and switching speeds of less than l-psec, proved to be ideal for this application. Other transistors were discovered that would have been satisfactory, but all had lower effective values of hFE, and thus would not saturate as quickly as the 2N1715. Moreover, storage time in the 2N1715 was less than any other usable type. All the transistor types suitable for application as T18, T19, and T20 were NPN silicon transistors, including the 2N1715. During all experience with this design, selected units were not required and no transistor failures occurred using the configurations and components as shown in the circuit diagram. 5.8.2. Diodes. —The diodes in the system are of four types:: (1) Logic and "circuit" diodes; (2) Current and voltage surge protecting elements; (5) Zener (voltage regulating) diodes; (4) Rectifying diodes. D14 is a 1N100A, and was chosen for its low forward resistance characteristic. It is a germanium diode and thus maintains a much smaller forward,onducting voltage across it than would a typical silicon diode. All other diodes 47

in the system are silicon diodes. D15, D16, D17, and D18 were chosen specifically for their high reverse voltage and high forward current characteristics for the bridge rectifier. The zener diodes D13, D19, and D20 are 7.0 volts, 85 volts, and 6.0 volts, respectively, and are used for voltage regulating purposes. The rest of the diodes in the system are all standard silicon diodes. Those making up (2) above are all the same type (lN2071). This seems to be an extremely good diode, as it can withstand several hundred volts in the reverse direction, and momentary surge currents of several amperes in the forward direction. The rest of the diode applications in the system jcomprising (1) abovejinclude a variety of configurations. However, the characteristics needed in each case did not differ greatly, and it was decided to choose, from the multitude of available diodes, one type that would be satisfactory for use in any of the logic circuitry applications. The type chosen was the 1N482A, and it has proven very satisfactory. D3, D4, and D5 are used as miniature "power supplies" (as described previously) whereas D1, D2, D6, D7 are used for standard "AND" and "OR" logic. Dll is used as a nonlinear resistance device in the blocking oscillator circuit. As with the transistors, not a single diode failure has been experienced under normal operation of this design. 5.8.5. -Capacitors.-,The following Types of capacitors are employed: (1) tantalum electrolytic; (2) ceramic; (3) mica; (4) aluminum electrolytic. Only C30 and C31 are aluminum electrolytic. The higher voltage requirement, along with high capacitance, without need for extremely low leakage, indicated standard electrolytics for these two. Mica capacitors are usedconly for the low-capacitance, high-precision capacitances in the contact detector circuitry. Tantalum electrolytics and ceramic capacitors are used widely throughout the system. All the smaller-value capacitances (less than one microfarad) are ceramics. All the low-voltage capacitance applications employ the new low-voltage ceramics, as a space-saving feature. All capacitances higher than one microfarad (and marked with polarity signs), except C30 and C3s, are low-voltage Texas Instrument tantalum capacitors. In particular C4 and C5 are TI tantalums and exhibit the "positive temperature coefficient" for aiding in temperature compensation of the clock circuit. 5,8.4. Pulse Tratsformers.-rThe pulse transformer used for PT1 and PT2 are the Technitrol (series TE) pulse transformerse The Technitrol series AMB (special small-scale transistor application pulse transformers) were tested and found unsatisfactory due to a systematic internal mechanical failure, The types used in this system (G13WA and 03QA) are heavier-duty types than the AMB 48

series and have proven very satisfactory. The G13WA(PT1) is a 22-millihenry primary inductance, three winding transformer, with a 3 to 3 to 1 turns ratio. The first and last windings have been used in series, with the middle winding being used as the secondary. PT2 (the 03QA) is a 1-millihenry primary inductance transformer, also with a 3 to 3 to 1 turns ratio winding pattern. The low turns number winding is used in the emitter circuit of the hbloking oscillator. PT3 is a 2 winding type D5315U4A custom built pulse transformer made by the Fischer Company. It has a 1 to 5 primary to secondary turns ratio. The transmitter presents a secondary load of about 2200 ohms, reflected into the primary as about 80 ohms. TR3 is a commercially available specialized transformer for d-c to d-c converter applications. It is made by the Triad Co. (TY-88) and is designed to operate on an input of 28 volts. 5.8.5. Delay Lines.-The delay lines PFN1 and PRN2 were designed and built in this laboratory and are 10-section, lumped-constant, 5 and l0-Isec, respectively, open-circuited delay lines. The circuit diagram for each is shown in Fig. 30. For each line there are 10 identical C's in the configuration as shown. In the case of PFN1 (5-usec delay line, l0 —sec pulse-forming network) the capacitors are 0.005 microfarad, and the inductors are 50 microhenries each. For PFN2 (l0-tsec delay line, 20-usec pulse-forming network) the capacitors are 0.01 microfarad, and the inductors are 100 microhenries. 10th SECTION _ T TT... TJ Fig. 30. Ten-section delay line. The capacitors are Centralab type DM low-voltage ceramics and the individual inductances were wound on General Ceramics type CF-102Q.1 toroids. The 50-microhenry toroids required approximately 32 turns of No. 33 enameled wire and the 100-microhenry toroids required approximately 48 turns of No. 33 wire. The characteristic impedance of each line is 100 ohms. The values of L and C were determined by the following criterion. The approximate formulas for lumped constant delay lines are: 49

T = n L where T = the delay time Z0 = characteristic impedance n = number of sections L and C are as in Fig. 30 above From this: TZ. L = - n and c = T nZ~ Thus, given the values of T, Zo, and n, the values for L and C were easily determined. Of course, T was known from system specifications, Z was quickly established from optimum circuit considerations, and the value of n was established by experiment. All these, in turn, immediately established the values of C and L. 6. ACKNOWLEDGMENT This work was sponsored by Geophysics Research Directorate, Office of Aerospace Research, U. S. Air Force whose cooperation and support is gratefully acknowledged. 50

7. REFERENCES 1. Bartman, F, L., Chaney, L. W., Jones, L. M., and Liu, V. C., "Upper-Air Density and Temperature by the Falling Sphere Method," J. Appl. Phys., 27, 706-712 (1956). 2. Jones, L. M,, "Transit-Time Accelerometer," Rev. Sci. Instr.,, 27, 374-377 (1956), 3. Jones, L. M., and Bartman, F. L., A Simplified Falling-Sphere Method for Upper-Air Density, Univ. of Mich. Eng Res. Inst. Rept. 2215-10-T, Ann Arbor, 1956. 4. Peterson, J. W, Analytical Study of the Falling-Sphere Experiment for UpperAir Density Measurements, Univ. of Mich. Eng. Res. Inst. Rept. 2533-2-T, Ann Arbor, 1956. 5. Jones, L. M., Fischbach, F. F., and Peterson, J. Wo, "Seasonal and Latitude Variations in Upper Air Density," Natl. Acado Sci., IGY Rocket Rept. Series, 1, 47-57 (1958). 6. Peterson, J. Wo, Schaefer, E. J., and Schulte, Ho F., A Simplified FallingSphere Method for Upper-Air Density, Part II. Density and Temperature Results from Eight Flights, Univ. of Mich. Res. Inst. Rept. 2215-19-F, Ann Arbor, 1959. 7. Jones, L, Mo, Peterson, J. W., Schaefer, E. J., and Schulte, H. F., "UpperAir Densities and Temperatures from Eight IGY Rocket Flights by the FallingSphere Method," Natl. Acad. Sci,, IGY Rocket Rept. Series, 5 (1959). 8. Jones, L. M., Peterson, J. W., Schaefer, Eo J,, and Schulte, H. Fo, "UpperAir Density and Temperature: Some Variations and an Abrupt Warming in the Mesosphere," J. Geo Res., 64, 2331-2340 (1959). 9. Jones, L M., and Peterson, J. W., Upper-Air Densities and Temperatures Measured by the Falling Sphere Method, 1961 Review, Univ. of Mich., ORA Rpto 03558-5-T, Ann Arbor, 1961. 51

UNIVERSITY OF MICHIGAN 3 9 03II0311111111 11 1 11 3 9015 03525 1035