Students on our team reported that our Systemcore unit screen intermittently stopped functioning. In response to this, I did the following:
- Remove Systemcore case (3DP top cover, bottom metal half, M.2 door)
- Disconnect MHF connector from CM5
- Remove single screw holding screen daughter board to main PCBA. At this point, I noticed that the daughterboard felt oddly... loose. Not rigidly mounted to the PCBA. Attempting to lift the daughterboard vertically off the main board again did not work (lifting gently of course).
- Looking under the board, I immediately noticed that J17 had lifted off the main board, though was still mechanically connected through the traces that route out from it. I got tweezers and held J17 while lifting the daughter board, at which point it easily de-mated.
As you can see, basically every pin that is not routed out of J17 has completely detached from the PCB. Other pins with traces are still connected through those traces:
Some thoughts (as a professional EE):
- J17 has no mechanical features for strength like many other connectors do. This makes sense as it is supposed to be a high density board-to-board interconnect (and usually in pairs, with more than one support mount, like the CM5) for high-speed signals. Here though, it has few pins routed through it, and (though I don't know for sure) probably not high-speed (multi gbps) connections. First suggestion would be simply switching to a different connector with mechanical boss, tabs, or through hole solder posts, such as Hirose DF12N, which has solder tabs specifically to prevent pad lifting, instead of what looks to be their DF40 series. Seems to be plenty of space for such a swap.
- There is no ground pour on the top layer, or even a localized pour around the connector, and many pins on J17 simply connect to their pads below. The copper in a PCB is generally adhered to the fiberglass base with an epoxy, and with such little surface area, those pins do not provide much mechanical strength in keeping J17 attached to the board. Without changing connectors, I'd expect big improvement from simply adding grounded fill around the connector and attaching the unused pins to ground, and there appears to be space to do so.
- Belt-and-suspenders would be doing both of these.
- Also could do localized epoxy around the connector, but care needs to be taken that the epoxy doesn't migrate up the pins and cause issues with the contacts, so I wouldn't really recommend that.
When exactly J17 started to come loose from the PCBA I am not sure. Given the evidence of intermittent connection before disassembly, I assume that some sort of impact, or maybe even damage during factory assembly, was the original cause of the issue. That being said, my disassembly likely further separated pads
Students on our team reported that our Systemcore unit screen intermittently stopped functioning. In response to this, I did the following:
As you can see, basically every pin that is not routed out of J17 has completely detached from the PCB. Other pins with traces are still connected through those traces:
Some thoughts (as a professional EE):
When exactly J17 started to come loose from the PCBA I am not sure. Given the evidence of intermittent connection before disassembly, I assume that some sort of impact, or maybe even damage during factory assembly, was the original cause of the issue. That being said, my disassembly likely further separated pads