Contactor drive PCB - MCU cuts out when closing contactor occasionally

Question

I have a PCB that I am currently supplying 12V which has control of multiple contactors (12V coils with built in economizer so a set voltage should be all that is needed). The board closes all 5 contactors at start up after checking all is ok (300ms apart to allow for max pickup current time so that the higher driving loads do not overlap).

The issue I am having is that occassionaly on startup the board closes the first contactor and immediately opens (The sound is instant I just haven't checked on a scope as I don't have the product on hand just yet!). This opening is because the MCU resets. What is weird is that it does this once and maybe a second time but then it works absolutely fine then goes through the routine and closes all 5 contactors and keeps them closed with the rest of the PCB working perfectly from then on.

I was mainly just seeing if I am doing something or thinking of something obviously wrong? Mainly as I would normally expect an issue on opening contactors, not closing.

Here is a top level overview of what the PCB does, I can share schematics if more useful?

The only issue I can see is that the diode is on the PCB and the low side of the contactor is connected to chassis GND to could the path be going though that instead of the diode? Although, as mentioned I would expect an issue with opening and not closing.

Maybe a ground loop issue as the whole PCB is on a shared GND plane that is solid on the 2 inner layers and stitched through to top and bottom as well, do I need to seperte this more?

Thanks in advance!

EDIT:

I managed to squeeze a probe in today to measure a few points. These are all captured when the contactor cuts out the MCU, luckily (Or unluckily) it was fairly repeatable.

This is the 5V supply to the MCU with my probe reference on chassis GND:

The GND of the PCB with the scope reference on chassis GND:

Here it is at a lower division to show more:

So it looks like the closing of the contactor does great some sort of interference that is seen by everything in the circuit. I am just not sure how to fix this in the board revision.

For this unit I don't have tiem for new boards so I was thinking of using ferrites around the contactor cable to see if any of there spikes are filtered out.

Answer 1

Thank you for sharing board design screenshots. Those are very helpful.
However, I'm not that impressed by the segmented board design, this puts components in sub-optimal placements, such as TVS diodes for contactor 3 and 1 are nowhere near the connector and basically useless.
There are also a few very long traces on the bottom. But on a four layer board this may be acceptable.

From what I can tell, there also isn't a lot a "big" capacitance to go around. Most ECU's (looks like one) need to support several milliseconds power glitching. And you've already determined that you need 300ms delay between relay activations.

Unfortunately you will need to get an unsealed board and start probing.
Even if we could tell what the problem is through our crystall ball, you can't fix it while potted.
(don't ask how I know)

• Simplify the firmware, make sure it isn't a firmware bug.
• Check what the mcu reports as reset source.
• Check to see if you have brownout, the lack of electrolytics makes me suspicous.
• Measure power supply ripple during relay activiations.
• Check other switching regulators do not crash.
• Validate your clocking does not fail. Try running the chip on internal clocks and test again.

And if you get really desperate:

• Investigate if the fields of the relays do not cause disturbance. You can do this by switching a contactor with another relay and psu and have the wires go over/under the unit.
• Another cheap check is the piezo-electric lighter test. ('click' it nearby)

This is all assuming electrical, can the shock of relay activation cause it? Hammer the table with the unit on it?

You can also take your board to a emi consultant and have their instruments try to replicate the issue. Some are very helpful and have seen many issues.

---

We've had an issue like this, a customer kept returning board that would boot-up in self-test failure. It's odd, we could never replicate this problem with those boards.
We've spent weeks trying to turn it on and off seeing if you could get it into this failure state. Changed firmware, added delays, made selftests slower/faster... Eventually we visited the client, he had more than a hundred meter of cable from the 24V source to the panel with our board and a plc. This caused powerup to be very non-linear in the first few moments.
We've added a 24V isolator to the panel and the problem has been resolved ever since.

Answer 2

You are powering the unit with 12Volts. Does the DVM show a voltage drop when the contactor closes? If so Add a capacitor to the input power 2200mfd/ 35 volts is good to test with. If this solves the problem, that indicates you have a power suply that has bad caps and is failing.