I am using the uMod form factor as I am working with prelease silicon and samples are in tight supply. Need to have the chips on a convenient carrier to move between projects until I get the next rev.
During the past week of bring up, I have had to remove the board several times from the ATP carrier for debug/rework. My observation is that once you mate/unmate the a board in the TE 2199230-4 M.2 socket about 3 or 4 times, it will need replaced.
Typically I observed an increased impedance through the connector. It was initially apparent as this chip has a built in DC-DC that would not start up due to an additional 30ohms I observed on a power connection. Today, it was one of the USB lines (roughly 100 ohms).
This is an 0.8mm stackup built using the template reference in the sparkfun library.
Swapping the connector fixed the problem. I gone through 3 connectors so far.
Have others had this same observation? Perhaps I am not meeting the 25 degree insertion requirement?
Cool - I tossed this to our product development team and they had the following as responses/feedback…the main takeaway is probably that the footprint of the MM slotting/pins/etc need to match ours exactly, chamfering the edges (just sand em down a bit) should be a good first step that can readily be implemented/tested:
‘What kind of power is being putting through it? I’ve never run into those kinds of issues, let alone within 3-4 times.’
‘I know we have some in Ops that are used for multiple batches before needing to be replaced, like 1000+ insertions. Also the testing we did early with MM development, we stopped testing after 700+ insertions without a failure.’
‘Not chamfered edge will do it too. Could be putting too much pressure on the inside of the connector. If he’s using our footprint it should work fine, but based on the photo I’d presume that isn’t the case’
Thanks for the response. I did you the SparkFun template / Eagle library for the card. Any goobers in the outline are just the result of the PCB fab tolerances. I did check several of the raw PCBs against a teensy uMod. It seems that some of the routing tolerances on around the key are sloppy. Not all the the boards, the the one I use for a build was worse than others.
I think the edge chamfering is a good clue and makes sense. I had to do it manually as the board house I used wouldnt do it on this sized board.
Thanks for the data on your testing. Good to know that it went through through a lot of cycles. The power here is pretty small (on the scale of mA for an MCU)
I have new PCBs on the way for more builds and will look closely at what comes back from the fab house.
Rev B & C had a much more aggresive chamfer. Everything seems ok now. Did a lot of swapping of a few prototype boards in and out of the ATP things behaved well.
