Dealing with a test fixture that's been frying SMT resistors, under certain circumstances.
I designed the board where the resistor fries, and a couple of others in that fixture, but the guy who did the overall design (and the software) is currently unavailable except by Ouija board, so I've been having to trace through such documentation as is available.
I'm just not finding the erroneous current path!
In normal operation, the resistor will be dissipating around 0.48W. It's a big hummer, rated 2W, so no problem, right?
It's a small board, and the resistor in question isn't coupled to large areas of copper fill. In the context of a suitably laid out board, it's good for 2W indefinitely. In the context of this board, it's limited by heating.
And, in normal operation, heating isn't an issue, as the resistor is only carrying current maybe 20% of the time, tops, in few-second bursts. But, in the context of DUT in place, test station powered up, but software not yet running... well, it could be carrying current for minutes.
And so I need either a time limit or - happy thought! - enough of a pulldown resistor that when the NI DIO pod that controls this thing powers up, and all its I/O pins become input with pullup, the driver for the power relay doesn't see that as its cue to turn on.
Afterthought: the relay drivers won't really like that approach. But! It appears the pod has discrete pullup resistors. If this is the case, I can just remove the pullup from the line that turns on the power relay, and all will be as it should have been in the first place.
Update: gngngngn...! The pod has, according to the documentation, active output drive capability. Software-selectable, it says. But! I don't have meaningful access to the source code; setting bits to "output" and "1" in Measurement and Automation Explorer's test panel gets a passive-floating state; I'm not finding any way to configure the outputs for active drive; and the actual operating software won't run here, for reasons not yet determined. So, I gotta put the pullup resistors back, and maybe install a Big Red Warning Light as an interim protective measure.
And furthermore: as long as I'm ragging on NI in the postscripts, riddle me this: what kind of idiot designs a USB-to-something pod for industrial control, and doesn't include mounting flanges? This dang thing has one-each mounting (or perhaps strain-relief?) hole, in a tab that's not even flush with the bottom of the housing, so attempting to use it for mounting will give unhappy results. When we built this test station, we ended up mounting the pod with sticky-back Velcro tape, which ain't very industrial.