Blowing up Chips

A test fixture uses an FPGA to stimulate a board under test and analyzed the results for testing.  Once in a while, maybe after testing 300 units, the FPGA gets fried.  The symptom is a near short from Vcc (3.3V) to ground, often with a similar short on an I/O pin.  

I can't find anything that would cause this.  The entire circuit is 3.3V other than a pair of RS-422 chips which has 5V outputs, run through 10k resistors to FPGA inputs.  Often the failures are on other I/O pins.  

Any idea what I could be missing?  I will try doing a better job of isolating the 5V outputs, but these fixtures ran for years, testing 1,000s of units without problem.  

-- 

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

On Tue, 19 Jul 2022 21:49:14 -0700 (PDT), Ricky
<gnuarm.deletethisbit@gmail.com> wrote:

>A test fixture uses an FPGA to stimulate a board under test and analyzed the results for testing.  Once in a while, maybe after testing 300 units, the FPGA gets fried.  The symptom is a near short from Vcc (3.3V) to ground, often with a similar short on an I/O pin.  
>
>I can't find anything that would cause this.  The entire circuit is 3.3V other than a pair of RS-422 chips which has 5V outputs, run through 10k resistors to FPGA inputs.  Often the failures are on other I/O pins.  
>
>Any idea what I could be missing?  I will try doing a better job of isolating the 5V outputs, but these fixtures ran for years, testing 1,000s of units without problem.  

Look at the last 'board under test'.

Ruggedize the fixture to survive it.

RL

On Friday, July 22, 2022 at 11:15:29 AM UTC-4, legg wrote:
> On Tue, 19 Jul 2022 21:49:14 -0700 (PDT), Ricky 
> <gnuarm.del...@gmail.com> wrote: 
> 
> >A test fixture uses an FPGA to stimulate a board under test and analyzed the results for testing. Once in a while, maybe after testing 300 units, the FPGA gets fried. The symptom is a near short from Vcc (3.3V) to ground, often with a similar short on an I/O pin. 
> > 
> >I can't find anything that would cause this. The entire circuit is 3.3V other than a pair of RS-422 chips which has 5V outputs, run through 10k resistors to FPGA inputs. Often the failures are on other I/O pins. 
> > 
> >Any idea what I could be missing? I will try doing a better job of isolating the 5V outputs, but these fixtures ran for years, testing 1,000s of units without problem.
> Look at the last 'board under test'. 
> 
> Ruggedize the fixture to survive it. 

Not sure what you are trying to say.  Are you suggesting the unit under test is causing damage to the test fixture?  I had not seen any indication of this, but I suppose it's a possibility.  

This is the last run for these units.  We may be asked to respin the design using parts that are still in production.  I'd be interested in improving these units, but I'm looking forward to the next design as much as anything. 

-- 

Rick C.

+ Get 1,000 miles of free Supercharging
+ Tesla referral code - https://ts.la/richard11209

On Fri, 22 Jul 2022 10:44:33 -0700 (PDT), Ricky
<gnuarm.deletethisbit@gmail.com> wrote:

>On Friday, July 22, 2022 at 11:15:29 AM UTC-4, legg wrote:
>> On Tue, 19 Jul 2022 21:49:14 -0700 (PDT), Ricky 
>> <gnuarm.del...@gmail.com> wrote: 
>> 
>> >A test fixture uses an FPGA to stimulate a board under test and analyzed the results for testing. Once in a while, maybe after testing 300 units, the FPGA gets fried. The symptom is a near short from Vcc (3.3V) to ground, often with a similar short on an I/O pin. 
>> > 
>> >I can't find anything that would cause this. The entire circuit is 3.3V other than a pair of RS-422 chips which has 5V outputs, run through 10k resistors to FPGA inputs. Often the failures are on other I/O pins. 
>> > 
>> >Any idea what I could be missing? I will try doing a better job of isolating the 5V outputs, but these fixtures ran for years, testing 1,000s of units without problem.
>> Look at the last 'board under test'. 
>> 
>> Ruggedize the fixture to survive it. 
>
>Not sure what you are trying to say.  Are you suggesting the unit under test is causing damage to the test fixture?  I had not seen any indication of this, but I suppose it's a possibility.  
>
>This is the last run for these units.  We may be asked to respin the design using parts that are still in production.  I'd be interested in improving these units, but I'm looking forward to the next design as much as anything. 

That's why you inspect first, before testing, and 
replace life-limited connectors before they misbehave..

It's obviously more of an issue testing sub-systems that 
consume or store energy, require low-impedance power, or 
inhabit environments with poor static control.

Backwards or incorrect parts, regulators and switches 
that don't, power sequencing that isn't. Simple shorts.

Test jig should be bullet-proof.

RL