Both units (2001:001 and 2001:005, North and South) has issues with the HCT157 multiplexer chips (U3). U3 has been replaced on both interface main boards LIGO-D040423.
Three of the four channels outputs were working to switch the full excitation voltage on the north unit (2001:001). However, the ramp engage output (pin 9) was only reaching an on-state value of 4.86 V (compared to >5 V on all other channels). This was probably enough to transmit logic state to the field box but leaves open the possibility that there are undiagnosed issues with the unit: the best option was just to replace.
There was also a power issue on the north interface box (2001:001). I first replaced Q1, a high power PNP transistor D45H11, on the Low Noise Power Module (LIGO-D0901846-v4, rev C on version). This wasn't the issue. I then replaced U3 on the same board, a AD829 op amp, that was used a -10 V reference voltage to control a -15 V output (buffered by the transistor). This solved the dead negative voltage supply issue. It looks like this had been reworked before, there was some damage to the pin 1 and 8 pads on this sub circuit. However, these are not connected to anything. Its not clear what went wrong, its very possible that voltage polarity had been switched. Great care should be taken orienting the DC supply plugs the right way up with these boxes. It appears cheaper connectors have been used, either for the box or the plugs that go into it, that allow the sub-d connector to be forced in the wrong way up. At least the damage doesn't seem to cascade through the rest of the unit.
I tested all voltages and the logic on all the digital chips and the two units in the PSL lab appear to be back to normal operation. The interface units are back in the rack and plugged in. We are ready to turn lasers on and re aquire lock.
The vacuum can stabilization is also up and running. I've been tweaking parameters for the PID every time I walk past; mainly turning down the integration term that was too high. Either Craig or I should sit down with Kira tomrrow and debug her python logging code (it looks like it should work, just some python/EPICS quirk).
I took photographs of the repairs, but they are still on the SLR which I don't have at the time of writing.
At the end of assembling the FSS interface boxes I had this screw left over:
I will put it in a box with a question mark on it so that whoever find it in 5 years will know where it is from.