PIC32 loosing stability

[Bartec]

My first time here so:

Hi everyone!

From the very beginning:
Got damaged the 3.3 voltage regulator on my board recently (yeach very bad beginning) - the multimeter shows ~6V now. So I switched to powering PIC32MX695F512H from programmer (using PicKit3).
It seems that for some time before shutting power off (well quite long time unfortunately) processor was powered with those 6V BUT:
    - programmer connects to the target and get it`s ID
    - processor starts running and executes a few lines of code (which is sending 1 character via UART - second character comes with errors)
   - then it seems to crash

Is it possible that device is damaged somehow that it can work only for some time (it seems like working about 0.2-0.5 ms)

What else may cause that strange failure? May it be capacitors (damaged after voltage reg. crash)?
Any other ideas?

BTW: I managed once or twice to run the processor and everything seemed to works fine after using hardware reset few times - but it doesn`t seem to work anymore.

Thanks in advance!

[RCMR]

I've had PICs go partially screwy after regulator failures in the past.

Just toss the chip and put a new one in.  That's the quickest and simplest way to verify that this is the problem.

[LaurenceW]

Yes, first thing to do is try a new PIC, they are cheap enough. If THAT has the same issues, sprinkle a few supply decoupling caps around the device, make sure that you have tied the reset line high and left no other input lines floating. Still problems? You need to look at your code. Go back to something really simple like a blinking LED (this will prove whether the PIC is basically alive or dead) and work up from there.

Make sure that you have either handled watchdog interrupts correctly, or more simply switched them OFF for now.

In my experience, I've yet to fry a PIC from 6V alone. 12V? Now, that'll do it

[FlashEF]

Surely enough, 5V version wouldn't get damaged from 6V, but we are talking about PIC32 that has a 3V3 voltage supply rail. Thanks for the advice, we will try the caps first, and then the controller itself. The code is most likely not the problem, cause it works on an identical test unit...

[amyk]

Very likely you damaged something. The absolute maximum rating from the datasheet is 4V.

The PIC32s are a bit more fragile than the lesser product lines at 5V, because they're made on a smaller process and are designed for lower voltage operation (also means thinner gate oxide). Even if the core survived, it's not much good if the flash hasn't.

Speaking of toughness, the oldest NMOS 8-bit CPUs had absolute maximum ratings of 8V, with normal operation at 5V.

A 3V6 zener across the power rail may help reduce the chances of frying another chip in the future.

[ptricks]

Over volt a chip and it may appear fine but still be screwed up internally. I had a pic chip that was exposed to 9V,  the chip would still program and work except that port C3 could not be enabled.

[SeanB]

I had a whole box full of TTL survive an overvoltage of 15V for a few hours. Overvoltage trip did not operate as expected, and it let the full unregulated rail go through. Still was running, just was running "funny". Only one chip died, a PM5406PM3 open collector hex inverter, funny enough a device with a 30V output voltage rating.......

[FlashEF]

Any practical tips on replacing TQFP-64 that has 0603 ceramic caps close to it? I'll be using a hot air station. What should be the starting temperature? Around 250C?

[nctnico]

I've set my hot-air station to 360 deg. C which works fine for that kind of jobs.

[ptricks]

If you are using hot air to remove a part near some stuff you don't want to be heated aluminum foil works well. You can build a little barrier over whatever you want to shield to deflect the hot air away.

[Bartec]

Thanks for your advice!
Don`t really like the situation but well... that`s how it works
Replacing uP and adding zener diode is about to be a priority now.
Can only hope it will work.. some day!

[FlashEF]

Thanks for all the advice, you all were right. Replacing it, together with it's caps worked. The trick with covering stuff around the part with aluminum foil works great as well. Surrounding parts got barely warm. Starting at 360C was perfect as well, overall a quick fix.