did you ever encounter some RP2040 GPIOs just not... working? i.e. you tell it to output a logic high level, but the actual pin is close to GND? we have some pocket keyboards that are like that. there is no short on the board. so i'm wondering how this can happen. damage during soldering, handling?
@Laberpferd@timonsku with my fluke, ohmmeter, i measure a resistance to GND on the broken pins, varying from like 10k to 10M, and on the working pins there is no measurable resistance
@mntmn@Laberpferd@timonsku hm, but if that ESD diode became self-sacrificial, that would either be a severely underdimensioned protection, or a serious ESD surge that I wouldn't expect pcbway to have on series prod.
Q: is there a reasonable way you could estimate the worst-case negative pulse amplitude caused by opening a keyswitch under load? How much current do you put through your keys while scanning the key matrix? What's the (order of) inductivity of the longest trace?
@funkylab@Laberpferd@timonsku so far i think these keyboards are DOA, it doesn't happen because of continued use. assembled, flashed, some columns not working--dead pins.
the circuit is like: column GPIO outputs logical 1 (3.3V). on closed switch, this connects through a diode to a row GPIO input. on the broken boards, one or two of the column GPIO outputs cannot produce a logical 1.
@funkylab@Laberpferd@timonsku i just desoldered the first chip from a board. the problem is not on the board itself, there are no shorts. so it is the chip
@mntmn@funkylab@Laberpferd@timonsku I have damaged MOSFETs before with a poorly grounded soldering iron which had ~90V to earth at several hundred microamps right on its tip.
@jaseg@mntmn@funkylab@Laberpferd hm yea, if they have floating irons that could def. be an issue. The workers likely wear grounding straps so they would realize the potential if they touch a ground point on the PCB during soldering.
@timonsku@jaseg@mntmn@Laberpferd but that's what I mean with "unlikely that PCBway has that problem"; these aren't a bunch of folks sitting at home with a 10€ soldering iron.
@funkylab@jaseg@mntmn@Laberpferd They might very well be, PCBWay has no assembly lines, they outsource this to various PCBA places. Which is why they can be quite inconsistent when it comes to assembly quality.
@timonsku@jaseg@mntmn@Laberpferd oh wow, wouldn't have expected that; keyboard assembly takes forever (and gets into quality/consistency problems) if you need to solder every pin of every keyswitch manually, which is why I'd have assumed there's enough concentration to either have one wave solder bath machine yourself, or one go-to subcontractor that does…
@funkylab@jaseg@mntmn@Laberpferd wave soldering only becomes cheaper at really high qty where the setup time and dialling in becomes worth it. Wave soldering is really tricky to get consistent and not every design is suited for it.
Not something you do if your joint count is in the thousands.
@mntmn@Laberpferd You can use the diode mode of the multimeter too to get the voltage drop of the diode and compare it to a working pin, will likely be very low or nothing at all.
@mntmn@timonsku@Laberpferd If it's a blown ESD diode, the chip's current consumption should also be significantly higher. You might even see a difference when looking with a thermal camera.
Add comment