After switching the FSS monitor channels to the Interface Controller box (with monitor read backs coming through the D25 connector) I found that the returned monitor signals were not working as expected.
The XT1221 card installed in the interface controller box only gave positive voltages that latched near zero. I think there is an issue with this particular unit. I've switch it out for a brand new card and have labeled in appropriately.
After switching I found that there were two further issues. The outputs were noisy and railed and latched to zero when driven above about 4.0 V. The first of these issues was a grounding one: the Acromag XT1221 inputs are differential and need to be grounded on one side for differential inputs. I fixed this and gave all the channels a common ground referenced to the common ground of the rack. After this the signal was a lot less noisy but railed and then dropped to zero when driven above 4.0 V. Turns out the final buffer stage of the signals feed back through the D25 cable only has 50 Ω resistance in series (with the return pin giving a further 50 Ω). It seems that the buffer is being overdrawn in current. I switched out R101, R105, R118, R130 50 Ω for 1 kΩ.
In addition to the above changes I discovered last night that I had gotten the polarity of the wiring wrong in the acromag inputs. The positive input was being sinked to ground, which is kind of bad and I don't really know how I was seeing any signal at all. I've fixed this and now see the full range of voltages with very little noise.
I'm not sure about the choice of 1 kΩ in series with the output (unity gain) buffer. The ADCs are high impedance at the intput (100 kΩ) so it should be fine. It does mean that there isn't a situation, say a short, where the output stage buffer will be drawn more than 15 mA.
After switching the laser fast monitor channel from the front panel FASTMON pin to the FASTM_P/FASTM_N pins I found that the PID controllers that use this channel to adjust the laser slow frequency became unstable. After turning down the gain the oscillations of laser temperature were very difficult to tune out.
It looks like the issue is with the fact that the FSS interface board (LIGO-D040423) low passes the monitor signal with a 0.8 Hz LP filter (200kΩ wt 1 µF cap). The narrowed bandwidth of this 'sensor' in the PID loop limits the bandwidth of the total OLG. We could turn down the P and up the I a little bit, but this seems less good than switching out some resistors.
I replaced four 100 kΩ resistors R99, R100, R116 and R117 with 510 Ω resistors. This brings the LP filter on the Fast monitor channels and mixer monitor channels up to 156 Hz: the new frequency is well above the ADC 10 Hz sampling rate but should still filter very high frequencies that are not of interest to the slow loop. I'm not sure what the ideal cut off point should be wrt digitizing signal and the loop, but this seems like a good first guess.
These modifications were made to North and South FSS interfaces boards (LIGO-D040423), serial 2010:005 and 2010:00? respectively. These changes are logged in the wiki page and with a label on the box back to this post.
Autolockers were not catching the resonances again. It turns out I had unplugged the excitation to the FSS acromag controller box that engages the binary channels. It had been plugged back in but often fails to activate the logic because it is still being powered from a 5 V plug pack.
Decided it was time to make the switch to 9V. I installed some voltage dividers to set the excitation out to the FSS interfaces to 4.9 V with a low voltage of 0.66 V maximum (unpowered). See PSL:2058 for the wiring, I used option D with R1 = 820 Ω and R2 = 680 Ω. The new wiring was tested to check the voltages were right, so I don't fry the FSS interface box again. I also made some changes to the soft binary channels in the acromag IOC: the front medm panel Test1 and Test2 switches are now flipped (NOT operation) so they make sense from the users point of view. Before the logic was inverted so turning TEST1 off actually activated this path in the circuit.
I also noted down the channels off the ADC card that picks off the monitors from the FSS D25 connector. This should reduce the number of front panel BNCs as it can all be routed inside the Acromag interface controller box. These have remapped from Aidan's acromag crate into the FSS interface controller box.