In case anyone is having odd issues with Maiana and MacArthur...
I had everything set up and working just great with the Maiana AIS and MacArthur hat. However the past few days I had intermittent issues with AIS. It was getting worse, and I finally tracked it down to the RJ45 jack on the MacArthur board. The jack had come partially loose from the PCB - the two large pads that physically secure the jack to the PCB, as well as 3 of the pads at the back. The traces/pads did not pull off the board - they are still intact. The solder seems to have just failed, which would seem to me to be a cold joint issue? There has been zero stress or impact on the jack or the cable plugged in to it, and the component itself seems fine - I should be able to simply reflow the pads and it should work.
I reported it to the OpenMarine team and will see what they say before attempting a reflow fix. Hopefully this is a one-off problem.
That's what I feared when we switched the RJ45 from a through-hole to an SMD part. The reason for the last minute change was that the metal shielding of the Raspberry Pi 5's USB port potentially touches the HAT's circuit board at this location.
As several points failed, I'm not sure whether bad solder joints can be blamed. I wonder if vibration could cause this failure mode.
If the pads are still intact, touching them up with a soldering iron and some fresh solder should fix it (for now).
(2024-03-19, 06:02 AM)Adrian Wrote: [ -> ]That's what I feared when we switched the RJ45 from a through-hole to an SMD part. The reason for the last minute change was that the metal shielding of the Raspberry Pi 5's USB port potentially touches the HAT's circuit board at this location.
As several points failed, I'm not sure whether bad solder joints can be blamed. I wonder if vibration could cause this failure mode.
If the pads are still intact, touching them up with a soldering iron and some fresh solder should fix it (for now).
Vibration shouldn't be the issue at least with mine - it has literally been mounted to the back of a touchscreen, sitting on my workbench, and basically hasn't moved from there since the day I received it. It won't be installed in the boat for several months. Especially with the large pads that secure the jack, the amount of force needed to make a good solder joint of that size fail would probably rip the pads right off the PCB. The contacts at the rear are quite small, and they haven't been bent or distorted hardly at all - they are completely intact.
I will see if I can get some close-up pics of the contact areas that failed tomorrow.
Oops, that does not sound good. Please send some pictures so we have more details on what to look out for in upcoming quality checks.
I just checked the RJ45 connector on the latest batch. It's going to be hard (impossible?) to resolder the mechanical pads just with a soldering iron. Heating the board from the bottom is probably your best chance.
I've made a note to extend the mechanical pads from under the part so that they can be more easily reworked.
(2024-03-19, 10:53 PM)Adrian Wrote: [ -> ]I just checked the RJ45 connector on the latest batch. It's going to be hard (impossible?) to resolder the mechanical pads just with a soldering iron. Heating the board from the bottom is probably your best chance.
I've made a note to extend the mechanical pads from under the part so that they can be more easily reworked.
Yup. I have both a heated plate as well as a hot air station - I think between the two I can make it work. Worth a shot anyway.
Below are a couple of pics, as close as I could get with my cell phone. Unfortunately the downlight for my microscope was not working, so I couldn't get better close-ups. On the rear, in case it isn't clear, the two most-inboard contacts are completely detached from the solder pad, while the next contact in seams to be partially detached but is still holding on a bit. The rest of the contacts outboard of this appear to still be attached. Both of the large pads underneath the jack have separated from the PCB. It does not appear that any pads or traces have been torn or lifted from the PCB.
A friend with more experience than me (he performed component level repair on audio gear as his job for many years) took a look at the pics and his first thought was a poor bond due to dirty/oily surfaces, along with possibly not enough heat in the joints.
For a contamination hypothesis, he pointed at the rear contacts, where they have come away from their pads. "Those contacts are tiny - any force sufficient to yank a component like this off of a well soldered board would almost certainly have ripped those contacts completely."
For the insufficient heat hypothesis, he pointed at the picture of the large contact pad that has two cylindrical columns rising from the PCB. These came from the solder squishing up through holes in the metal mounting part of the jack. He said that on a properly heated contact, the solder should have squished up through the hole and then formed a bit of a mushroom cap on top. But the top of those columns did not appear jagged as if the mushroom cap was torn off - they looked smooth - which led him to think that the solder cooled down right as it was being squished through the holes, and formed a column shape that did not spread out on top.
All he saw was the pics, but his thought was that it was a combo of not enough heat and some contaminated surfaces that likely caused a poor bond. It sounded reasonable enough to me
[
attachment=2737]
[
attachment=2738]
Based on the pictures, I agree with your friend that the connector probably wasn't soldered properly. The way the pins are bent while the others are still straight, I wonder if the part itself was chewed up and therefore not sitting flat during reflow.
I should have caught a loose connector during testing, as I plug a cable into it. I was definitely tugging on them while working through the latest batch today
(2024-03-20, 02:42 AM)Adrian Wrote: [ -> ]Based on the pictures, I agree with your friend that the connector probably wasn't soldered properly. The way the pins are bent while the others are still straight, I wonder if the part itself was chewed up and therefore not sitting flat during reflow.
I should have caught a loose connector during testing, as I plug a cable into it. I was definitely tugging on them while working through the latest batch today
By any chance do you have the part# for the orig jack that was going to be used? I see that the locking pin holes are still present on the PCB - I don't have any concerns with using some standoffs that are .5mm-1mm longer to keep well enough away from the Pi5 jack, and will use a couple layers of Kapton tape between the two.
The alternative RJ45 requires a different footprint, with through-hole pads for all the pins. The existing holes are just for the plastic posts of the connector.
If you haven't started rework yet, I'd like to exchange your board as is, so that we can review the defect with our contract manufacturer. Check your PM.
In case it is useful for the future... Here is a pic after I removed the jack. The two large pads definitely look pretty cold to me. the rear contacts are harder to tell, since I used a hot air gun to remove the jack. All of the pads were pristine after I removed the remaining solder - zero damage to the PCB
[
attachment=2739]