In the mid 1990s, working for a small aerospace company in Ottawa, Canada, I led the design team for the Canadian Remote Power Control Module (CRPCM), part of the Mobile Base System (MBS) for the International Space Station. I was also in charge of the analog design for this remote-controlled, on-orbit replaceable, circuit breaker panel (or “glorified fuse box” as I refer to it).
To current-limit and control loads on the 120-V dc power rail, we used banks of parallel power metal–oxide–semiconductor field-effect transistors (FETs), which had to absorb the fault power (up to 3,500 W) in current-limit mode, until the switch was shut down. The FET control circuitry rode on the load voltage, with an isolated power supply and control/status signal interface; one for each of the breaker circuits in the CRPCM.
For FET control, I used a simple op-amp circuit to sense load current and adjust the gate voltage for ON, OFF, or current limit mode. For fast fault response, I had no capacitance on the op-amp inputs. The circuit worked fine on the bench, limiting the fault current within microseconds, and holding it within spec for the allotted time until the FETs were shut off, opening the breaker.
We went ahead and built the engineering (preflight) model for testing. All was well until the circuit was installed in its aluminum case. At that point, the sub-picofarad coupling from the grounded case to the op-amp positive input caused instability as the control circuit bounced between zero and 120 V, trying to limit the test fault current. To fix the problem, I needed to add a larger capacitance to the negative input pin. But the circuit board was finished, and where does one find a flight-qualified 1-pF capacitor on short notice?
Fortunately, the board had some open vias near the op-amp. I took a 2-in length of insulated jumper wire, folded it into a loop, twisted the two ends together, soldered them into the appropriate vias, and then snipped the end off the loop. Voilà, an approximately 1-pF, high-voltage capacitor! That did the trick, stabilizing the circuit nicely without sacrificing response speed. We quickly installed these makeshift capacitors on each circuit, closed up the box, and the CRPCM went on to pass all its tests.
The six flight models and two spares were then built the same way, flight qualified and shipped to the contractor for integration into the MBS. With Shuttle flight delays, the MBS was mothballed for a few years, and I left the company shortly thereafter. However, the MBS did finally launch in 2002, and I have a nice picture of two astronauts installing it on the Space Station, with the six CRPCMs clearly visible.
In 1999, before the launch, I was called by the contractor asking whether the CRPCM was Y2K compliant. When I stopped chuckling, I was able to reassure the worried caller that the control software for the boxes had no real-time clock and did not track the date; it was dumb enough not to be a Y2K concern. As far as I know, the CRPCMs are still functioning properly, keeping the Space Station and its occupants safe from fault conditions on the MBS electric grid.