Well, I finally got sent out to support a customer site using an iiWA. I wasn't there to actually work on the iiWA (I've still never had the basic training course), but since I was there, I got grabbed to look at an intermittent issue they were having with all their iiWAs.
For reasons I'd rather not waste time going into, the main power going into these iiWAs gets shut down fairly often, when someone has to open a panel on the line that happens to (among other things) act as the power distribution panel to the Sunrise controller. Often, these power-downs are very short, short enough that the controller gets main power back long before it's completed its controlled shutdown.
Most of the time, this works. But a significant percentage of these power cycles result in the iiWA showing E-Stop and Operator Safety faults, even when the relays wired into the iiWA's safety inputs are confirmed to be closed. So far, the only reliable way to clear these errors has been to switch the iiWA off again (at its own power switch on the back of the Sunrise cabinet), wait for the SmartPad to go dark, then power back up. This clears the safety errors, even when we ensure that nothing else on the line was touched (no reset buttons, no safety interlocks, nothing).
I was only able to do very limited testing, but can confirm that even if the iiWA was fine before the first powerdown, it could still reboot into this faulted condition, even if all the safety relays were closed before power was brought back up.
So, my first thought is that cycling power down, then up again, too quickly on an iiWA might sometimes create something analagous to a single-channel fault, where (on the old relay boards) the only way to clear it was to remove power entirely from all the relays and then apply power again.
Or, is it possible for the iiWA internal safeties to get into a single-channel fault (or something similar) under these conditions?
I'm just spitballing here, but I'm hoping someone with more hands-on iiWA time than myself may have seen something like this before. To me, this 'feels' a lot like single-channel faults on the old FE20x relay boards, or the ESC board on KRC2s, but that's just a gut feeling.