Two failed processors, looking for tips to chase down the cause

[480VAC]

We are using an nrf51822 processor, running off of 2.7 volts from a switching power supply, powered with 12.4volts. The processor has simple gpio connections, and a battery monitoring circuit, consisting of 348k + 100k resistor divider from the 12.4V battery.

The failure has happened on two of four boards tested to date, though the boards tend to work for a while, then stop.   When the boards stop working, the processor is a short across the vdd power supply.  I'm not sure what could be causing this, and because we used the wafer-scale components, I can't swap out the processor for a new one.

Any advice is appreciated.

Thanks

[AmbystomaLabs]

I would look at your buck converter as the source of the problem.  A shorted nRF can be easily caused by over voltage.  Even small transients over 3.6V while your buck converter stabilizes could cause the nRF to fail.

You should start by monitoring Vdd_nRF on an o-scope during various power up scenarios.  Make sure the sweep is sufficiently fast to spot transients.  If you can't get your scope to do this, you can always set the sweep trigger to 3.6v and see if it triggers.  Then you would know you have an over voltage situation.

Barring this, I would then look at ESD issues.  Often people don't provide ESD protection in their labs and that combined with nylon carpeting and suddenly you have failed parts.

[480VAC]

Thanks for the tips - I resurrected a board, removed the processor and checked the power supply.  The voltages look good, with no overshoot in any startup or shutdown condition I could create.  

Since I don't have a processor in-place, are there conditions that running code could induce?  Maybe transmitting on the radio?  If so, how do I encourage the processor to create that load ste-condition.

If not that, then I guess I'll have to chalk this up to ESD, which I hate to do, because I'm usually wrong when I think that.

[AmbystomaLabs]

I agree ESD is often pointed to as the cause erroneously.  However, I have seen a lot of poor ESD situations in labs where real failures occur.  So, you should at least review your ESD practices.

You should also double check your buck converter.  Make sure the inductor is appropriately sized with the correct saturation current spec. Also, double check you have sufficient input and output capacitors of the correct type for the converter. Instability in DCDC converters is common when the design isn't perfect, especially grounding, and will result in high ripple and over voltage transients.

On the nRF, there isn't much you can do to fry it, but these are some thoughts:

1. The RF output is not unconditionally stable into all loads.  So, if you have a very poor output match you could get the RF output to oscillate and fail in a manner you describe.

2. While the GPIO can source/sink over 5mA in high drive mode the device is not designed to do this for all gpio simultaneously.  Nordic have never nailed down a max current number on the devzone, but have suggested it is somewhere around 50mA.  This is why it is always best practice to sink large loads rather than sourcing them.  The ability of the device to sink is greater than its sourcing capability.

3. You should review the onboard DCDC design of the nRF to make sure you followed their guidelines. 

4. You should review the grounding of the nRF especially the Epad.  Mistakes in grounding can cause instability in the DCDC and other circuits.