After some cable swapping this morning, I have determined which cable is bad. It's the Gur1 cable between the seismometer and the breakout box. This is a milspec -> 37pin d-sub cable. I'll pull out the cable and have a look at it after lunch.
So, I was wrong about which cable it was, probably in my rush to get some lunch. The actual culprit was the octopus cable that Bob made waaay back in the day (~2 years ago) to go from the "ADC out" of the breakout box (37pin Male Dsub) to 9 BNCs. As it turns out, the Gur1Z channel of that cable was broken on both ends!!!
On one end, we have the 37 pin Dsub. The cable used was so thick (way too thick for this application) that it made a super rigid connection between the wires and the connector, and any bending of the cable stressed this connection, despite the strain-relief of the backshell. The Gur1Z connection snapped off when I was gently wiggling the connections to check them out. Also, since the wires were all so thick, they didn't really fit into the hole in the backshell, so 2 or 3 of them were squished.....straight through the insulation so that several channels were shorted together / potentially shorted to ground. This may explain some of the nasty behavior that Rana and I had seen (although I might have forgotten to elog? My bad.) that even with the inputs of the breakout box all terminated, there was high coherence between the channels. Terminated inputs should give random noise, so this was fishy.
On the other end of the cable we have the 9 BNCs. I had finished redoing the 37pin end of the cable, and was 'beeping' it to check it out, when to my dismay I found that the Gur1Z channels (the inside and the outer shield of the BNC connector) were shorted together. I removed these 2 wires from the Dsub connector to confirm that the BNC was at fault. Koji looked at the BNC with me after I chopped it off of the cable. Bad news strikes again. To get the wires to fit in the inner pin of the BNC connector, the cable-maker had cut off several strands of the wire to make it skinnier. It appears that over the years these cut-off strands wiggled their way to touching the outer shield. This appears to be a danger for all of the BNCs on this cable: a little bit of torque (which one might expect during plugging and unplugging a BNC) and the 2 sides of the differential measurement will be shorted together.
I then decided to start afresh and make my own cable. I found some AWG26 8-twisted-pair cable laying around underneath the Yarm (since this was all I could find, I was just going to do the Gur1 and Gur2 channels, and leave out the Gur3's). The 37 Dsub side was easy, but I seem not able to connect such skinny wire to the BNC connectors in a robust way. Since this bad cable has so far cost me ~2.5 16-hour days of grief, I don't want my new version of it to also be bad. At this point, I await the advice of one wiser than I. I think BNC connectors are designed for something a little closer to ~20AWG, but I could be wrong. Also, they are obviously optimized for coax cable. So what I have now is never going to be great. Maybe tomorrow I can go to the Electronics Shop / Store and buy BNC connectors that are meant to be soldered-to. That would be awesome.
Since I currently have no functional cable to go from Breakout box to ADC, the Guralps are unplugged for tonight.
Conclusions for the day / evening: Frank, Alastair and Jenna are mostly absolved of blame, although the traveling to Bridge and opening and closing the box (which usually involves more plugging and unplugging of cables) probably didn't help this cable out too much. Also, Bob definitely owes me a Sugar Napoleon or something.
In other news, since the Gur2Z ADC channel is totally wacked, I have taken over (but not renamed) the Ranger channel for Gur2Z for now. Jan still has the Ranger hostage over in Bridge, so this is okay for now.