Fullscreen HTML5Med Res (??MB)HTML4

D-200 RADIO TRANSMITTER 7KH6RYLHW6SXWQLNVDWHOOLWHODXQFKLQVWDUWHG WKHoVSDFHUDFHp,WFDUULHGWZR:UDGLRWUDQVPLWWHUV %HFDXVHRILWVKLVWRULFDOLPSRUWDQFH,GHFLGHGWR FUHDWHDQDXWKHQWLFUHSOLFDRIWKHWUDQVPLWWHU DVVHPEO\GHVFULEHGLQWKLVVHULHVRIDUWLFOHV A Vintage Radio Story, Part 1 By Dr Hugo Holden S putnik-1 was an awe-inspiring accomplishment in the field of space exploration in 1957 and a credit to the Soviet engineers who designed it. The Sputnik-1 satellite confirmed that not only could an object be deployed from a rocket into space in a basically stable orbit, but that it could also carry a functioning radio transmitter. The transmitted signal could be easily received by many shortwave radios on the Earth, as long as they were within view of the satellite. Since the paths of radio waves and light are generally reversible, it also indicated that satellites could be used as radio relay stations in space. The idea that a satellite could be placed in a geostationary orbit was postulated by Arthur C. Clarke in 1948. Yet few people took him seriously at that time because he was a science fiction writer. Sputnik-1, as well as inspiring the world, triggered the formation of NASA. The impact of Sputnik-1 on 98 Silicon Chip space science and popular culture was very significant, even making it onto stamps (see Photos 3 & 4). I first saw images of Sputnik-1 in the early 1960s as a boy. It stirred my imagination in electronics, general science and space travel. I didn’t imagine back then that one day in the future, I would have a go at reconstructing Sputnik-1’s radio transmitter and “Manipulator”. The D-200 radio transmitter The satellite was as simple as possible, carrying two independent radio transmitter modules inside one D-200 transmitter unit, transmitting at 20.005MHz and 40.002MHz. One module is seen in Photos 5 & 6; the other is on the reverse side of the unit. Batteries and a cooling fan assembly surrounded the D-200. Essentially, the battery assembly formed a large octagonal structure inside the spacecraft and the transmitter was in the hole in the middle (see Photo 2). The inside of the 0.58m diameter polished spherical body was Australia's electronics magazine pressurised to 1.3 atmospheres (1.3 bar/1300hPa) and filled with dry nitrogen. The carrier wave was derived from a separate crystal-controlled oscillator in each module. The antennas were close to ¼ wavelength dipoles, folded into a V shape with the Satellite body in between, although they were physically shorter than exact ¼ wavelengths of the operating frequencies. The angled arrangement of the antennas on the satellite body helped it fit into the nose cone of the launch rocket. The effectively bent dipole also had a more uniform signal distribution than a straight dipole antenna’s typical ‘figure-8’ pattern. The transmitter output power was 1W per module. However, the two transmitter modules were alternately switched on and off by an oscillating relay system called the Manipulator (манипулятор). These unusual relays are the two cylindrical objects seen near the top of the D-200 unit in the photos. There was no RF carrier modulation, just simple interrupted carrier wave (CW) transmission. Due to the two transmitters being alternately switched on and off by the Manipulator, no more than 1W of radio-frequency power was transmitted at any time. There were three 2P19B miniature pentode valves in each transmitter module; one for the oscillator and two in push-pull for the RF power output stage. Radio wave propagation The designers used two transmission frequencies and two transmitter siliconchip.com.au Photo 1: Sputnik-1, the first artificial satellite, fully assembled. Photo 2: what was inside Sputnik-1. You can clearly see the octagonal battery pack, which had the D-200 transmitter module in the middle. modules for redundancy but also to ensure that under the worst expected conditions in the ionosphere, on a winter afternoon at that time of year, one of the signals would make it through the F layers. The F1 and F2 layers are regions in the ionosphere bombarded by UV light from the sun, where the pressure is low and free electrons and ions can move for a long time before recombining to become neutral atoms. These ionised layers react with electromagnetic waves and can absorb some of their energy, reflect them or let them pass through, depending on the angle of incidence and the frequency. The layer ionisation depends on the season, time of day and the year. The 11-year sunspot cycle affects them too, because it affects UV levels. The designers’ calculations were based on the satellite being above the horizon, 700km above the Earth’s surface and 3000km away. The designers concluded that it would require 1W for the signal to pass through the F1 & F2 layers from the satellite to the observer (radio receiver). They did mention in the design document that with a super-sensitive professional receiver, 10mW might be adequate. But the average member of the public would not have such equipment. The designers were clearly intent that average citizens, especially in the USA, should be able to tune into the satellite’s transmissions. The selection of 20.005MHz by the designers was a stroke of genius because it was 5kHz away from America’s time-frequency channel WWV on 20.000MHz. This would naturally beat siliconchip.com.au with Sputnik-1’s carrier wave transmission, creating a 5kHz audio beep that could be heard on a garden-variety shortwave radio without a BFO (beat frequency oscillator) if it was tuned to the 20MHz region. Many American citizens could grab a shortwave radio and tune close to WWV to hear Sputnik-1, if the satellite was in ‘radio view’. Battery power Sputnik-1 carried three specially-­ made silver-zinc batteries inside the octagonal housing. One battery Photos 3 & 4: North Korean and Soviet stamps featuring Sputnik. It was a big deal at the time! Photos 5 & 6: the D-200 transmitter unit that flew on Sputnik-1, shown from two different angles. You can see the two large relay cans on which the Manipulator is based at the top. The transmitter circuitry is in the section below. Australia's electronics magazine November 2023  99 powered the ventilation fan, while the other two formed the low-voltage battery for the 2P19B valve filaments. It also had a high-voltage battery to power the plates, screens and suppressor grids of the 2P19B valves. A 21V tap on the high-voltage battery powered the Manipulator circuit. The batteries were designed to power the craft for at least 14 days. However, after its launch on October 4th, 1957, Sputnik-1 transmitted continuously for three weeks; the transmissions stopped on October 26th. The satellite did not fall to Earth until January 4th, 1958. Sputnik-1 had a fairly elliptical orbit; the satellite’s apogee was 947km with a perigee of 228km. What ended Sputnik’s transmissions? The 7.5V filament battery for the valves was rated at 140Ah, while the total filament consumption was about 180-200mA for the two transmitter modules combined. The filament battery should have lasted about 700 hours or 29 days at that rate, but the current drops with voltage, so it could probably have lasted more than 30 days. However, the calculation to full discharge might not be helpful because the oscillators in the units would have stopped at about ⅔ of full discharge, after around 20 days. As the valve filament temperature drops, so does its transconductance and at some point, that would stop the oscillators. The tapped HT battery supplying the Manipulator with +21V had a negligible current draw, less than 1mA at 21V. On testing the single transmitter with its output loaded to give 1W of RF power, the average 130V supply current, operating at its usual 50% duty cycle (under Manipulator control), was in the region of 24mA. The total average screen current for the three valves was in the order of 7mA. That makes the transmitters’ on-power consumption from the HT battery 3.75W (7mA × 90V + 24mA × 130V). In the transmitter’s off condition, the 130V current (due to the oscillator anode) measured 7mA and the 90V current (for the screen grid of the oscillator) measured 3mA. The power then was 1.18W (3mA × 90V + 7mA × 130V). What about the Doppler Effect? Could the Doppler Effect have affected the historical audio recordings when the satellite was low on the horizon and moving away from or toward the observer? If the transmission frequency is ft, the observed frequency, fo, at the receiver is ft x c ÷ (c + v) for the transmitter moving away from the receiver and ft x c ÷ (c – v) when the transmitter is moving toward the receiver. The speed v of Sputnik-1 was approximately 8000m/s and c (the speed of light) is close to 3 × 108m/s. Ignoring curvature of the path, when the satellite is travelling away from the receiver, the observed carrier wave will appear to drop in frequency by 0.0027%, or when travelling toward the receiver, increase by 0.0027%. Applying that to the 20.005MHz carrier frequency, it would appear as 20.0046667MHz or 20.00553347MHz. The beep’s tone is generated at the receiver as a beat note of two frequencies, so it could therefore change in pitch from around 5.53kHz as the satellite breached the horizon to 5kHz (overhead) to 4.66kHz with the satellite going down on the far horizon, due to the Doppler effect. It would probably be less of a shift in practice due to the curved path. The beep rate (not beep pitch) of 2.5Hz would not change as the satellite went from horizon to horizon, as it would only shift over a range of 2.500066675Hz to 2.49993335Hz. The listener would never notice that. Period changes due to battery discharge were much more significant over time. Some of the historical audio recordings of Sputnik-1’s signal have more of a spooky ‘phasing in and out’ effect typical of multi-path shortwave radio reception. It was thought that the Doppler effects and the two different transmission frequencies might also help provide more information on the ionosphere. In some of the historical recordings of Sputnik-1, people are turning the BFO knobs on their radios during the recording, altering the beep pitch. That confused people about the transmitted signal’s nature and misrepresented what happened. To make matters worse, on tape loops, the pulses appeared on some to change spacing abruptly, but that is due to poorly spliced loops. 100 Silicon Chip Australia's electronics magazine With two transmitters alternately switched on & off, the total power would therefore be 4.93W (1.18W + 3.75W). I assumed for simplicity that this power came entirely from the 130V battery terminal, meaning the current drawn from the HT battery for Sputnik-1 would be close to 38mA. The HT battery was rated at 30Ah. Therefore, it should have taken about 789 hours or about 33 days to completely discharge or perhaps a day less, accounting for the tiny current consumption by the Manipulator. That is not dissimilar to the calculated life to complete discharge of the filament battery, at around 30 days. The probable running time for the circuitry, before the voltages were too low, is about ⅔ of that, accounting for the 21-day practical life. Since the filament power was 1.5W (7.5V × 0.1A × 2), one could say that Sputnik-1 used 6.5W to produce its 1W RF output. Sputnik-1’s operational duration of three weeks well exceeded its design life of 14 days, which is very impressive. It took a 50kg battery pack to do it. The Manipulator Since the release of Sputnik’s D-200 transmitter design document over a decade ago, electronics historians have mainly focused on the transmitters and largely ignored the Manipulator circuit. I’ve only read brief remarks on it, such as “relays switched the transmitters on and off”. It appears that nobody has investigated the Manipulator or exactly reproduced it and documented its features before. That’s partly because there was a paucity of information in the design document on the theory and function of the Manipulator. The Manipulator alternately switched off the screen supply voltages to the transmitter module’s two 2P19B output valves, thereby killing the transmitter output when the screen voltage abruptly fell to zero. Its circuit comprised two commonly available (at the time) Soviet-made twin-coil super sensitive magnetically latching change-over relays, the PnC4 model PC4. Sputnik-1 did not transmit information on satellite conditions, such as telemetry information. However, it had three simple switches (called “error switches” in this document) that could change the Manipulator’s siliconchip.com.au duty cycle and frequency if certain extremes of pressure & temperatures in the spacecraft were exceeded. A separate internal thermal switch operated the ventilation fan system, switching it on if the temperature exceeded 30°C and off if it dropped below 23°C. In Sputnik-1’s flight, none of the error switches deployed, so the signal from the two transmitters remained with close to a 50% duty cycle for each. However, the switching frequency dropped as the battery powering the Manipulator discharged over time. Relays as oscillators Magnetically latching relays had to be used for efficiency in this satellite application. The principle of using a relay as an oscillator, with a capacitor in the relay coil circuit and some resistors, appears simple enough; you will find many relay oscillator circuits on the internet. It is not so simple to produce a perfect 50% duty cycle from them. The reason is that the charge and discharge cycles of the capacitor are not always equal due to varying source resistances. This can be matched by diverting the discharge via an additional contact to a load. However, matching these exactly on each half-cycle is still a challenge. There are also electromechanical properties of the particular relay and the delay to magnetically latch and unlatch to consider. If you apply a voltage to the coil of a relay, you will notice a delay before anything happens. Part of this delay is the current rise time due to the inductance of the relay coil, while the magnetic field is being established. Another aspect is the time it takes to accelerate the mass of the armature (the moving mechanical arm) and for it to arrive at its new mechanical position. Typically, in a relay, the armature carries the relay contacts. Depending on the relay design and physical size, this combined electromechanical delay process could take from 1ms to 300ms or more. This raises the interesting question: how did the designers of the Sputnik1 Manipulator get the relay oscillator to produce a near-perfect square wave pattern? siliconchip.com.au Photo 7: an exploded view of a Sputnik-1 replica. Source: https://w.wiki/6tVc Part of the answer is that they used a symmetrical electrical circuit incorporating latching relays in a master/ slave configuration. Latching relays contain a permanent magnet that holds the armature (and its contact) in position once latched. This also makes them very energy efficient. Only pulses of current are required to change the state of the relay, or a drive waveform with a higher leading edge that can decay later. The wasteful direct holding current needed to hold a conventional relay (with an armature return spring) in one state is not required. The usual way to reset the latching relay is by either applying an opposite polarity pulse to the same coil that set its position, or applying a separate pulse to another coil on the relay bobbin with an opposite phase to the first. In addition, for a balanced square wave oscillator using magnetically-­ latching change-over relays, a perfect magnetic balance is needed in that both ‘halves’ of the relay must have a near-identical coil current sensitivity to initiate a state change. This balance is heavily affected by the mechanical adjustment of the relay’s magnetic pole pieces. The Manipulator’s designers used a system where each half of the full operating cycle relates to charging an 8μF capacitor. This matches electrically to the symmetrical (mirror) circuit. It then only requires that coil pole Australia's electronics magazine pieces on each side of the relay are in an exact position so that the magnetic forces balance. They could alter the oscillation duty cycle away from a balanced 50:50 condition by modifying the resistor values on each side of the charging circuit feeding the master relay coil. This allowed them to transmit the possible “error” or fault conditions. The Manipulator system using two twin-coil magnetic latching relays is astonishingly energy efficient. They quoted a power consumption of under 20mW in the design document. The relays in a master/slave configuration are somewhat analogous to a master/slave flip-flop. The DC resistance of the coils in the slave relay, close to 6kW, provides the charging resistance for the timing capacitors for the master, which saves on parts too. When the timing capacitors are sufficiently charged, the voltage across their terminals becomes high enough, in conjunction with a series resistor with the master relay coils, to cause the master relay to change state. In the design document, they argued against having a valve-based Manipulator because it would consume more power. They also argued against a gas-discharge valve relaxation oscillator because the lamp required is more sensitive to acceleration and vibrations. The system had to survive accelerations of up to 20g. November 2023  101 The final design had six possible patterns or duty cycles and frequencies for switching the two transmitters. However, as noted, none occurred during the 21-day flight to the point of flat batteries. Oscillator period The design document (siliconchip. au/Shop/6/224) refers to a Manipulator period of 0.4 seconds. However, it was unclear if that was the whole period of a Manipulator cycle or the period that one of the transmitters was turned on. If the latter were the case, though, Sputnik-1’s received signal, heard as beeps at the receiver, would have only been 75 per minute. Examination of the amateur radio audio recordings on the internet, early in the flight of Sputnik-1, indicated the beep rate to be around 144-150 per minute. This confirms that 0.4 seconds was for an entire Manipulator timing cycle and that each transmitter had an on-time close to 0.2 seconds early in the fight, with fresh batteries. The Manipulator’s oscillation frequency slows as the power supply voltage is lowered. The oscillator runs at half speed once the voltage drops from 21V to about 13V. Most of the recordings indicating that each transmitter was on alternately for 0.2 seconds were in the early phase of the flight of Sputnik-1, and the slower recordings, where it appeared to be closer to 0.3 seconds, were in the later stages as the battery voltage was dropping. The oscillator stops when the applied voltage is below 9-10V with the PnC5 relays. The design document mentions that the factory guarantees four million relay operations. In the nominal mode, the number of operations for 14 days should add up to about three million. There are 1,209,600 seconds in 14 days; three million divided by that number gives 2.48Hz, close to the 2.5Hz corresponding to an entire oscillator cycle. In summary, there is overwhelming evidence that the Sputnik-1, at least in the few days after launch, with fresh batteries, transmitted alternating bursts of unmodulated carrier waves at 20.005MHz and 40.002MHz that were very close to 0.2 seconds long each. However, some internet sources quote 0.3 seconds, likely corresponding to later in the flight. When the transmissions were received on a radio with a BFO, they became “beeps”. The pitch was typically determined by the BFO knob position on the amateur radio, while the ‘beep rate’ was close to 2.5Hz or 150 beeps per minute. Error switches The error switch configuration is shown in Fig.1. Normally-closed switch E1 would open below 0°C while normally-open switch E2 would close above 50°C. Normally-open switch E3 would close if the pressure inside the craft dropped below 250mmHg (1/3 bar, 333hPa). That would indicate Sputnik-1 had sprung a leak, possibly perforated by a small meteor. I had to deduce how these switches were connected to the Manipulator to agree with the duty cycle patterns in the design document. Those patterns were recorded on what appeared to be 35mm rolling film with a time marker signal on it. Using recordings on the internet of Sputnik’s transmitter taken a few days into its flight with fresh batteries, I determined that the time marker signal is 100Hz. Fig.2, taken from the design document, shows how the error condition switches affect the Manipulator timing. When side A is active, the 40MHz transmitter is on; when side B is active, the 20MHz transmitter is on. To have created these film recordings, the designers would have used a dual trace CRT, with the output of the central relay contact on the slave relay deflecting the beam vertically. Unlike an oscilloscope, there would have been no horizontal beam deflection. They likely used a positive and negative voltage supply connected to the two slave relay contacts. The film would have been rolling past the CRT’s face to expose it. The added calibration signal ensured that the film speed was not a factor in the measurement. It is more easily seen in close-up Fig.3. Most likely, the calibration pulses were derived from a full-wave rectified line power source since the line power frequency in Russia is 50Hz. Alternatively, they may have been created by a divided crystal source. Fig.1: the ‘Manipulator’ oscillator circuit based on two relays, a ‘master’ and a ‘slave’. It oscillates at close to 2.5Hz with a duty cycle very close to 50% unless one of the fault switches (E1-E3) changes state from its default. 102 Silicon Chip Australia's electronics magazine siliconchip.com.au Notice the short ‘dead time’ pulses, centred vertically, when neither slave relay contact is closed. When none of the error condition switches were active (as they turned out not to be in the actual flight), the duty cycle of the Manipulator was close to a square wave, alternately switching on each of the transmitters at close to 0.2 seconds on time and 0.2 seconds off time for each transmitter. PnC5 latching relays Photo 8 shows some PnC5 relays, which have the same form factor as the PnC4. Photo 9 is of one of the relays out of its canister, showing the structure, perhaps visible more plainly in the drawing, Fig.4. Each coil has two windings. It is possible to apply pulses of the same polarity to the different windings to set/reset the relay. Alternatively, you can apply pulses with opposite polarities to the same winding to achieve a similar effect. I could not acquire an exact PnC4 relay as used in Sputnik-1; however, the PnC5 relays I did manage to buy are almost identical. I discovered that the main difference is that the two pole pieces, P1 and P2, are adjusted slightly differently. I think there would also have been a difference in how the armature was suspended. The PnC4 would probably have used a friction-­ free pivot. When the pole pieces P1 and P2 are open enough, the PnC5 does not latch, and the armature returns to a neutral position. The armature is suspended on a thin metal strip and acts like a taut band suspension. However, closing up the pole pieces just a little on their adjustments allows the armature to latch in either position. Then, the PnC5 relay behaves like the PnC4 and becomes a latching relay. After I made this initial discovery and adjustment, it became clear that the overall sensitivity of the relay also depended on the combined average position of the pole pieces. If one considers using a capacitor as a timing element, ignoring the 75kW resistor in the capacitor charging process (as it is large compared to the resistance of the slave relay coils at about 6kW each), we can test some assumptions. In most RC timing circuits, a capacitor is seldom charged beyond one to two time constants to reach some siliconchip.com.au NORMAL Frequency = 2.5Hz; 100Hz reference pulse Side A 0.2s 75kΩ ERROR 1 t < 0°C; approximately 2Hz Side A 0.31s 91kΩ Side B 0.2s 75kΩ Side B 0.2s 75kΩ ERROR 2 t > 50°C; approximately 8Hz ERROR 3 P < 250mmHg; approximately 8Hz Side A 0.09s 75kΩ Side A 0.025s 14.5kΩ Side B 0.025s 14.5kΩ Side B 0.09s 75kΩ ERROR 1 & 3 t < 0°C, P < 250mmHg; approximately 6.5Hz ERROR 2 & 3 t < 50°C, P < 250mmHg; approximately 15Hz Side A 0.033s 14.5kΩ Side A 0.125s 91kΩ Side B 0.025s 14.5kΩ Side B 0.033s 14.5kΩ Fig.2: the various possible Manipulator oscillator waveforms, recorded by the original designers on 35mm film. Fig.3: a close-up of one of the Manipulator waveforms; note how the dead time is visible as dots where neither relay contact is closed. Photo 8 (above): four Soviet PnC5 dual-coil SPDT latching relays. There are 16 pins on the base as some other relays from the same series have multiple sets of coil windings. Photo 9 (right): the PnC5 relay mechanism out of its can. Australia's electronics magazine November 2023  103 Fig.4: the general configuration of the Soviet PnC4/PnC5 dual-coil latching relays used in the Manipulator. Their large coils make them very sensitive. threshold to initiate a state change. The reason is that the voltage profile across its terminals starts to flatten out after that and timing errors become more significant. One RC time constant charges the capacitor to 63% of the supply voltage, two time constants to about 86.5%, three to 95%, four to 98% and by five time constants, the capacitor is 99% charged; its terminal voltage changes little after at that point. I found that, once properly adjusted into a latching version with correct magnetic balance, the PnC5 relays worked in the Sputnik circuit but required a 36kW resistor, rather than 75kW, to achieve the correct 2.5Hz frequency with 8μF capacitors. This indicates that I achieved a relay sensitivity a little lower than I could have with the correct PnC4 relays. The sensitivity increases opening the pole pieces, but if one goes too far, the relay won’t latch reliably and it reverts to a non-latching condition. This is the effect of the taut band suspension in the PnC5 design; a small amount of extra energy is required to overcome that. Given the master-slave arrangement, for test & measurements only, I deleted the slave relay and replaced its coils with two 6.2kW resistors. That had little, if any, effect on the behaviour of the master (oscillator) relay. I was interested in the coil current required for the relays to change state. I made a voltage recording with a fully isolated scope across the 8.2μF capacitor in the initial test setup – see Fig.5. I later changed to using the original Soviet pairs of 4μF 160V PIO (paper in oil) types for the transmitter replica. Considering coil 1 (pins 1 & 2 of RLY1), the master relay, capacitor C1 charges when the relay contact feeding C1 is closed. Eventually, the master relay deploys when the threshold is reached and the relay changes state, magnetically latching to the opposite condition and initiating the charging process of C2 via contact 2. Fig.5 shows that this occurs when the voltage (marked in white) across the capacitor’s terminals has climbed from 9.5V to 18.5V. Therefore, 9V is required to cause the PnC5 Master relay to change state, in conjunction with the 36kW resistor and the 6kW coil resistance. That corresponds to a coil current of 214μA (9V ÷ 42kW). It’s close to but not quite as sensitive as the original PnC4 relay, which would have toggled at a mere 111μA. The capacitor discharges at a slower rate because, in the interval when contact C1 is open, the capacitor is discharging into the relay coil via the 36kW resistor. The yellow markings in Fig.5 show that the inverted exponential charging 21V (SUPPLY VOLTAGE) 18.5V 9.5V 0V 0.1 second/cm 0 RC 2RC 3RC 4RC Fig.5: a scope grab showing how the voltage (marked in white) across the relay coil varies during oscillation. The yellow annotations show roughly how the RC time constants correspond to the waveform. 104 Silicon Chip Australia's electronics magazine Fig.6: as the magnetic fields of both coils interact, we can sum them like this to see how the magnetic field strength varies over time. siliconchip.com.au curve seen is close to that of a four time constant RC curve. The charge time approximately matches an 8μF capacitor charging via 6.2kW (the slave relay coil) from a 21V source. Superficially, this does not seem ideal for setting a timing threshold, where one or two time constants would have a steeper approach. This is just considering the magnetic effects of the current in one of the master relay coils, but what about the other coil? As the applied voltage and therefore the current via one coil is climbing, the voltage on the other coil is falling, and the currents have opposing magnetic effects due to the polarity relationship of the two coils. If we chop up the scope recordings and invert the wave on coil side B, then add it to the wave from coil side A, we get a better idea of how the master relay approaches a state change. The approach to the threshold is much steeper, more like a two time constant inverted exponential curve, as seen in Fig.6. I have never seen any other large latching relay types that can change state with coil currents in the order of 100-200μA. Even the most sensitive relays I have seen before require at least 500-1000μA coil current, most much more. After finally finding the PnC4 data sheet for the part number PC4.520.350 used in Sputnik-1, it confirmed that the relay coils are 6.5kW ±1.3kW and that the relay operates in the range of 87-174μA, consistent with the Photo 10 (left): I made this relay test/adjustment jig using two bases that match the PnC5 relay pins. Photo 11 (right): the underside of the relay test/ adjustment jig showing the components and wiring that form the oscillator with the two relays. conclusions that I had made about it, switching at around 111μA. I suspect that the makers of these relays supplied specially tested and adjusted versions of the PnC4 relay to the Soviet Space Agency. I found out for myself that the pole-piece adjustments for the master relay are critical, especially for a perfectly symmetrical switching waveform. Once they are adjusted, though, the relay behaviour seems very predictable. Custom adjustment circuit To assist in setting up the PnC5 relays and adjusting their pole pieces, I built a custom circuit to monitor the duty cycle, shown in Fig.7. It also required a test jig with sockets to hold the relays – see Photos 10 & 11. Part of the setup involved using dummy 6.2kW resistors to take the place of the slave coils. The voltage developed across those is used to activate a comparator, with a 1V slice level, giving a stable 5V peak-to-peak output. A custom circuit using an op amp, shown in Fig.8, helped me make the required adjustments. The actual unit is shown in Photo 12. The output of the OP295 op amp swings rail-to-rail. The signal is heavily time integrated. The exact duty cycle was affected a little by the Fig.7: this shows how the major components are wired to the relay bases, for both the test jig and the actual Manipulator recreation. siliconchip.com.au Australia's electronics magazine November 2023  105 Fig.8: this test circuit aids in balancing the relays so that they give a 50% duty cycle in the Manipulator. operating frequency, so I made the relay pole piece adjustment at the operating frequency, close to 2.5Hz. One might expect that with an exact 50% duty cycle, the output from the integrator should be 2.5V with this circuit. However, when in perfect balance, the actual value achieved is around 2.66V because of the small gap in the timing where no contacts are closed (about 4ms on each side of the pulse) and the circuit being triggered by a low across the 6kW resistor, with the stage of inversion by the first op amp. A quick calculation suggested the measured (time-integrated) voltage would be 2.6V (2.5V × 208ms ÷ 200ms). The exact value of around 2.66V is of no concern, though, provided the voltages match precisely when the select switch is changed between the A & B sides. In other words, both halves of the relay must have identical magnetic properties and timing. When the relay is not in perfect ‘magnetic balance’, one voltage is lower than 2.66V, and the other is higher. This circuit could be doubled up, and the time-integrated voltage across each of the 6kW resistors could be fed into another comparator. However, it would need a window over which a range of voltages would be an acceptable difference. In practice, it was better to watch the meter and toggle the select switch to check that each half of the relay matched up. The sound of the Manipulator running With the complete Manipulator system running, the sound the relays make is very similar to a ticking watch or clock. It is easy to imagine Sputnik-1 flying around the Earth in 1957 at 8km/s with the relays inside it clicking like a clock. There is something quite magical about this, rather than it being deathly silent in there. You can hear the sound at the following links: • siliconchip.au/link/abmm • siliconchip.au/Shop/6/224 I doubt if anyone else would have recreated this circuit since Sputnik-1 launched. The design documents only appeared in the last decade, and it requires the now very difficult-to-get vintage Russian PnC4 or PnC5 magnetic latching relays, in good order and proper adjustment, to work correctly. Unfortunately, because these relays contained valuable precious metals, COMPARATOR SLICE LEVEL = 1V 0V 2V/cm Photo 12: this simple circuit, built on protoboard, helps determine when the oscillator duty cycle is at 50%. 106 Silicon Chip Australia's electronics magazine siliconchip.com.au most of them in Russia and Ukraine have been recycled because the plants doing it have offered good money for them. Power consumption As noted earlier, the design document stated that the Manipulator power consumption was less than 20mW. I measured a mere 14mW with the PnC5 relays and expect it would have been a little lower with the PnC4s. When I saw the 20mW figure and the 75kW resistors in series with the relay coils, I could hardly believe it and thought it might have been a misprint. I had to wait for the PnC5 relays to arrive from Ukraine to verify that the circuit really did work at such an astonishingly low power. If the slave relay contacts are connected to positive and negative voltage sources, the waveform shown in Photo 13 can be made, similar to the recordings of the original Manipulator on 35mm film. Note the small steps where, for a moment, neither contact is closed. You can see a video of the analog scope trace at https://youtu.be/k15GSKK_ UY0 The reaction to Sputnik-1 After Sputnik-1 was launched, the Americans were interested in seeing what telemetry might have been encoded into the transmissions. There was none, just alternate bursts of carrier wave at the two transmission frequencies at the 2.5Hz rate set by the Manipulator. Since none of the error conditions occurred, the Manipulator’s duty cycle remained at 50% during the whole flight. That could have disappointed the CIA or made them anxious, in case they had missed something secret embedded in the transmissions. Part of the genius of Sputnik-1 was its simplicity, and there is no doubt that the CIA, at the time, tried to overthink it. Next month At this stage, I had a working replica of the Manipulator, so the next job was to recreate the transmitter module. I would also need to build a copy of the metal housing that carried the transmitter circuitry and develop a suitable power supply. All of that will be described in the second and final instalment next SC month. ◀ Fig.9: the output waveform of the first op amp in Fig.8 during calibration. Photo 13: by connecting a bipolar supply to the outer relay contacts and the middle contact to the scope input, you get this sort of waveform. The steps in the middle of the ‘square wave’ indicate the dead time when no contacts are closed. siliconchip.com.au Australia's electronics magazine November 2023  107