Tracking bugs and failed connections in relay logic is not really bad if one understands what is supposed to be happening. Adding neon light indicators across coils can be very helpful. And if you ever get into doing that same logic with a PLC it will spoil you for going back. Watching coils and contacts on a screen is certainly a more fun way to do diagnostics.

 

Just wanted to post an update to (a) thank everyone for the advice and (b) let you know I finally figured out what was causing the voltage drop. It turned out to be a embarrassing case of Assumititis Noobius.

The culprit was not the relay board or the power supply. It was an assumption about the MIC2981 chips I was using to drive the relays.

In order to understand what was going on, I built a little power supply board with some 5v regulators on it, so I could provide regulated 5v to each part of the system (the relay board, the MCP23017 IO expanders, and the MIC2981 drivers that provide the signals to the relays) and isolate the problem.

Probing around, I found all the voltages were OK except the outputs of the 2981s. When not actually connected to a relay (so not under load), they showed the 5v I expected. But when driving the relay they only showed about 3-3.5v.

At which point, a little googling revealed my error. What I didn't understand (because I'm still learning to read datasheets) was that, under load, the vOUT on the 2981 signal pins is up to 2 volts less than the vIN. I naively thought that the vOUT would match the vIN.

And since 3-3.5v is right on the edge of the voltage needed to operate the relays, unreliability (and confusion) ensued.

Since I was already using my power supply to feed ~7v to the 5v regulators, I tweaked the setup so the 2981's were getting that directly; the outputs of the 2981's under load went to 5v, and the board started working perfectly.

And thus the student was enlightened. Or in this case, enLED'd.

 

Just wanted to post an update to (a) thank everyone for the advice and (b) let you know I finally figured out what was causing the voltage drop. It turned out to be a embarrassing case of Assumititis Noobius.

The culprit was not the relay board or the power supply. It was an assumption about the MIC2981 chips I was using to drive the relays.

In order to understand what was going on, I built a little power supply board with some 5v regulators on it, so I could provide regulated 5v to each part of the system (the relay board, the MCP23017 IO expanders, and the MIC2981 drivers that provide the signals to the relays) and isolate the problem.

Probing around, I found all the voltages were OK except the outputs of the 2981s. When not actually connected to a relay (so not under load), they showed the 5v I expected. But when driving the relay they only showed about 3-3.5v.

At which point, a little googling revealed my error. What I didn't understand (because I'm still learning to read datasheets) was that, under load, the vOUT on the 2981 signal pins is up to 2 volts less than the vIN. I naively thought that the vOUT would match the vIN.

And since 3-3.5v is right on the edge of the voltage needed to operate the relays, unreliability (and confusion) ensued.

Since I was already using my power supply to feed ~7v to the 5v regulators, I tweaked the setup so the 2981's were getting that directly; the outputs of the 2981's under load went to 5v, and the board started working perfectly.

And thus the student was enlightened. Or in this case, enLED'd.

Voltage drops happen in every conductor that has current flowing in it. Always. In many instances the drop is small and does not matter, while in other instances it matters a great deal. My one employer learned that a rather expensive hard way when a very expensive product failed to meet the specification because of a lot of small voltage drops. I may some day relate the whole very sad tale, but the point is that at 100 amps, 0.01 ohm leads to a whole 1 volt drop.