Stefan B, Keiko K

The non-functional FSS Binary Output (BO), see klog 5053, turned out to be a bad Contec DO-32L-PE (see http://www2.contec.co.jp/prod_data/do32lpe/c01.pdf ).
PIN 1 (ON 0/1, i.e. the negative reference for channels 0 to 15) was not properly seated inside the PCI chassis. And once we had fixed that, bit #4 was stuck grounded.

We replaced the card, which fixed the problem.

The one remaining item we found is the grounding scheme. The shield ground for all Binary Output cables (and probably also for the analog I/O cables) is tied to the chassis ground on both sides, generating a pretty big ground loop. For now we inserted the Binary IO Checker Board (which breaks the shield) in the FSS BO cable. Similarly we inserted a short shield-free cable in the PMC BO cable. This is however a temporary solution, and we need to decide what our grounding scheme between remote racks is.


Some background:
1) The disconnected PIN resulted in channels 0-15 being dead when just connected to an independently powered Binary IO Checker Board (JGW-D1807767-v1).
2) Channels 0-4 of the TTFSSI POWER BOARD (JGW-T1604761-v3) have a pull-up resistor to 5V. When connected to the checker board, this results in the first 5 LEDs always being ON (current flows from 5V to 15V, turning on the LED). The channels are properly switching despite this.
3) Peculiarly, on the broken Contec DO-32L-PE, this 5V also served as fake negative reference for channels 6-15 when the TTFSSI POWER BOARD was connected. Despite bullet 1), channels 6-15 this appeared to work when the TTFSSI POWER BOARD was connected, confusing us for a while.