Align up: The case of the out-of-sync synchro amps
Tales From The Cube: An Apollo-program engineers struggles to straighten out a system that suffers from repeated misalignment of antenna synchro amplifiers.
By Arnold N Simonsen, Electrical Engineer -- EDN, November 13, 2008
A previously published Tales from the Cube column (Reference 1) brought to memory a similar experience I had with a relay-release time and an attempt to reduce inductive spikes.
As a young engineer in the late ’60s, I was working on the Apollo instrumentation ships, which NASA used to track and communicate with the Apollo lunar-excursion module. The ships had C- and S-band radars, a satellite-com antenna, a telemetry-tracking antenna, a VHF/UHF command-and-control communication antenna, and a large data-processing center. Each of the antennas provided azimuth- and elevation-position-synchro data to a central network.
This network allowed selection of any antenna’s position data for making one antenna act as a slave to another antenna. Large synchro amplifiers drove 23 small synchro amps to amplify the synchro-position data. The 120V, 60-Hz synchro amps required an electromechanical-alignment procedure. The amps could align to approximately 0.1°.
After successful system tests, we noticed that the amplifiers were out of alignment by large amounts—in many cases, 20° or more. Realignments were tedious and time-consuming. We resorted to pinning the shafts to keep them aligned, thinking it was a mechanical problem. Over time, the pins became loose and even bent, however.
The three-wire outputs of the synchros went into a relay-switch network. The relays allowed selection of which antenna would act as a slave. The relay coils each had suppression diodes to reduce inductive spikes, the noise on control lines, and switch-contact arcing. This procedure was standard and accepted.
While monitoring the switching transients, I saw that the relays’ dropout time was a lot longer than their pull-in time. This excess time allowed the outputs of two or more synchro amps to connect across each other during the dropout time of the relay. During switching, the amplifiers also made loud gear noises. This pull-in- and dropout-time difference was new to me, but checking relay characteristics confirmed my observation. Some studying of the use of suppression diodes indicated that diodes would even lengthen the relays’ dropout time. I tried capacitors and RC networks to no avail.
|
Then, in a diode catalog, I found a Thyrector device from General Electric. It was basically two back-to-back diodes for reducing inductive spikes. I tried using the device to correct the time difference, and it greatly reduced the relays’ dropout time.
My boss was reluctant to make any change because the customer had accepted the system. We had no recording scopes back then, so I set up a fast strip-chart recorder to monitor the relay pull-in and dropout times during typical switching sequences and showed him the two connected amplifier outputs. I repeated the test with the Thyrectors in the circuit, which showed a great reduction of the overlap time and no gear noises during switching. Success!
I replaced the suppression diodes with Thyrectors, and subsequent tests during sea trials showed no synchro-amp misalignments.
Arnold N Simonsen is an electrical engineer in Tucson, AZ. Like Arnie, you can share your Tales from the Cube and receive $200. Contact edn.editor@reedbusiness.com.
| Reference |
|
-
Slow relay release times can engender other problems if the contacts are controlling an inductive load. One normally places a snubber (R+C) across the load or across the relay contacts to prevent arcing. This works only if the breakdown voltage of the separating contacts increases faster than the actual voltage across them. If the relay release is slowed by a long tail in the coil current, the contacts may separate too slowly to prevent arcing. The arc may in turn momentarily connect the high inductive kick voltage to other sensitive components. I've seen a magnetic chuck system that repeatedly blew bridge rectifiers and an induction motor start/brake brake system that repeatedly blew the DC braking drive transistor, both due to relay contact arcing.
Dick Neubert - 2010-13-1 17:44:00 PST -
This resonates, I mean, I''ve run into the same dropout problem. Why is it that ALL of those common appnotes that suggest a clamping diode fail to mention the length of the suppression tail? The current circulates around the inductor-diode loop. We should have learned this in basic circuits but there was no note that said "This will shoot you in the foot" when your product malfunctions. What good is a 5ms relay transition time when the current wants to flow for 5 times that? The key item not mentioned in the story is that the (obsolete) thyrector could be specified with a threshold/clamping voltage, the way we might specify a MOV or TVS limit today. ANY extra voltage in the circulating current loop will dissipate the energy. Just optimize the voltage choice to keep it well below the breakdown of the attached components.
George Catlin - 2008-28-11 08:28:00 PST -
I've had the chance to work on some relay timing circuits that our customer (a major player in military/aerospace devices) uses internal to their top line relays, and note they use a similar circuit in their standard products too. What they do is replace the classic single suppression diode with a suppression diode connected anode-to-anode to a 36V zener diode. Normally, when a relay is disconnected the instantaneous coil current "flies back" into the suppressor diode and has a linear decay of di/dt=V/L, where the V is just the diode drop. When connected with the zener the drop is the diode drop plus the zener voltage, making di/dt much greater. This way the relay coil field dies much quicker and makes the overall relay turn-off time that much less.
Of course the turn on device now needs to handle a larger surge voltage at turn off, but it is still a controlled voltage for the short time it is on.
Ernie Murphy - 2008-21-11 06:31:00 PST -
I ran into the same problem with a high-speed electromagnetic shutter driver I was updating. The circuit had the knee-jerk clamping diode across the coil. The current in an inductor goes as the time-integral of voltage. At turn-on, 24 V was being integrated, bringing the current up quickly. At turn-off the voltage was only a diode drop, so decay was 30 times slower. I replaced the diode with a series diode and zener, letting the back EMF rise to 24 volts for symmetrical on - off timings.
Ted Crum - 2008-19-11 09:40:00 PST
