We checked the grounding of WFS and LSC cables to the table.
Disconnected the cable from the DC interface and checked if the ground pin on the cable was connected to the ground pin of the DC interface, and they were not.
Then we went to the table, disconnected the WFS cable from the feedthrough panel of the table, connected a DB15 cable to the feedthrough and locally supplied +-18V from a power supply.
No oscillation was observed, but none of the channels in WFSA was responding to the light. For WFSB, segment 1 and 2 (pin4-12 and pin3-11) were good, positive output of the segment 3 (pin2) was dead but the negative output (pin 10) was fine, and the segment4 (pin1-9) was dead.
We removed the heads from the table and opened them to find that:
For WFSA, in 3 segments AD8599 (U1 in D1101614, that's the dual opamp for the differential output stage) was blown up, and the remaining one segment had a fried AD8597 (U2, that's the trans impedance amp before the differential stage).
For WFSB, only one segment had what seems to be a fried AD8597.
Pictures will be posted by Lisa or Giacomo.
We don't know how these things were broken.
Don't do this:
I repeatedly connected the positive output of the differential output stage to the core of the BNC and the negative to the sheild, and connected it to an oscilloscope. This means that the negative output is short-circuited to the mains ground. It was also connected to the DC ground of the demod board as the same scope was connected to the single ended monitor output of the demod board.
This doesn't explain why the opamps were burnt (the absolute maximum rating of the opamp, AD8599, for short circuit is specified as "indefinite"), but it's simply a bad practice so don't do it.