dead gpio pins on 4/100 stm32407 chips

[russdx]

Posted on January 07, 2013 at 11:09

i have designed a little pcb which uses a stm32407vet. i have had 100 boards assembled. and 96 of them work great but 4 of them the gpio on 1 pin (different pin each board) is not working) same firmware on each board.

its using the A bank and the gpio goes straight to a 2x7 IDC header on the pcb.

have i some how killed the gpio when testing the boards? and zapping it with static by accidently touching the header? how static sensitive is the gpio on these stm32 chips?

I cant think what else would effect 4 boards out of 100 batch and kill of random gpio like this?

Thanks :)

[Uwe Bonnes]

Posted on January 07, 2013 at 15:35

[russdx]

Posted on January 07, 2013 at 16:00

Using a scope i have checked the actual pin on the device and its dead :(

Thinking about this with amount of stm32 dev boards out there with exposed headers. surely they cant be static zapped easily or there would be a lot of dead boards out there.

Just cant see why theses gpio ports are dead.

[frankmeyer9]

Posted on January 07, 2013 at 16:33

On a board, even in absence of additional circuitry attached to a pin, you have at least the header and the wiring on the PCB, which form a parasitary RC network, and thus a voltage divider.

And second, the uC is often placed where you can hardly touch a pin directly.

A naked uC is easy to kill with a 5...10kV body charge.

Just take care to always have a proper grounding during work. Most MOS chips have such instructions in their datasheets.

[russdx]

Posted on January 07, 2013 at 16:50

the header is self is very close to the stm32 ic, the trace is probably about 3-5mm long, the header idc header is through hole. touching the top is quite hard due to the sunken pins but touching the pins poking through the pcb on the underside is quite easy.

i must of just blown those io by accident some how, just cant think how else they would get damaged.

Im thinking of adding some buffer resistors in series with the header to protect the pcb from getting damaged in future. can you suggest a suitable resistance? 

[frankmeyer9]

Posted on January 07, 2013 at 18:47

[jj2]

Posted on January 07, 2013 at 20:28

Some do not trust smt headers - use thru-hole instead.  So your board goes thru the reflow oven - MCU and all other smt devices are soldered - but ''secondary operation'' is required for the thru-hole components.

Might an improperly grounded, voltage-limited soldering iron be used w/your headers?  We have observed a, ''lessening'' of ESD awareness - most often during ''quick/dirty'' secondary operations.  ESD protection - functional ground-straps for all personnel - must be emplaced.

I do like the idea of small value, series R, with your GPIO.  Small, multi-pack resistors can protect several GPIO in one go...

[russdx]

Posted on January 07, 2013 at 22:47

I think they was blown by static probably by me, the assembly quality is very very good and im sure they assembly them with all the correct ESD protection in place.

But i decided to upload the firmware on my carpet floor DOH i did touch bathroom tap first to lose any static but i guess i built some more up and zapped these 4 boards by touching the header at some point holding the board, there is also 2 via's on the bottom that connect to the gpio i could of touched these as well.

Ill be a lot more careful for the next batch :)

bit worrying how sensitive to static they are though. I hope the people i sell them to don't end up blowing them up. I have included a number of static handing with care warnings though!!!

Thanks for all your help :)

[russdx]

Posted on January 08, 2013 at 00:13

panic over, on closer inspection the pins where shorted to gnd with a tiny bit of excess solder on the idc header. nothing to do with static (although i will be more careful with that as well in the future)

i should of checked the pins first with a meter not just scoped its output.

its all ok :)

thanks for all your help :)