TPS62111 Power Good (PG) pin failure

I've never heard of an issue like this.  You should get waveforms of Vin, Vout, EN, and PG when it doesn't work.

Sometimes manufacturing issues cause strange things like this--pins/parts not soldered down well, etc.

You might also try an external pull-up, in case the internal one is for some reason not present or disabled.

There is an external 1 Mohm pullup in the circuit.  I've checked the Vin, EN (tied together), Vo and PG on a good and bad unit.  All are the same except the PG pin doesn't go high on the failed units.  Vin comes from a Low Loss Controller and p-channel mosfet at 6 volts and I don't see overshoot.  The rise times are 5 to 10 usec from the cabling and filter capacitance we have on the battery.  I also measure the resistance from PG to the rail and it's ~70K Ohm on a good unit and ~1-2K on 2 failed units.  I tried to paste the schematic in here.

Your schematic didn't seem to come through.  Inserting images can be tricky on these but uploading a pdf usually works.  

Could you provide a waveform of Vin, Vout, SW, and PG on power up for both a good and bad unit?  This might show some clues.

Unfortunately my scope doesn't save the settings or the grids but the scales are the same on both (200 usec per div for a total of 2 msec shown.).  The scales are 2 volts per div for the input to the regulator and the other two being Vo and PG are 1V per div.  The 1st signal to leave ground is Vin, the 2nd is Vout and third (on scope_8) is the PG pin.  Scope_8 is the good one, scope_9 the bad.  

4812.scope_8.pdf

3531.scope_9.pdf

here is the schematic

4807.214243_D page 4 and 5.pdf

I'm sorry but your schematic seems to show the PG output wired to the EN of another IC.  It doesn't look like it's connected to the SVS.  Could you confirm your connections?

Page 2 shows the TPS62111 PG pin tied up to the 3.3V rail via a 1 Mohm resistor.  On the same page it is also used as an enable for a separate RF power rail of 1.8 volts. Additionally on page 1, it's tied to the Master Reset (MR/) pin of a reset controller (U46) which controls the reset (MCLR) line that allows the CPU to run (not shown).  The EN pin of the RF Power Rail regulator (U21) has no internal pullup and provides <<1 uamp load on PG.  The Reset Controller (U46) has an internal 70-90K pullup on the EN pin that PG is tied to.

The MR pin has a different net name than the PG pin.  Thus, my question.

Earlier, I should also have noted that we released a new family of 17V input devices last year.  Being newer, they are smaller and have a newer control topology.  TPS62160 family or TPS62140 family should work for you.  If you need even less current, then TPS6212x devices would fit.

I would recommend using those devices anyways, even if you didn't have this issue.

If you're talking about the /5.4C following the PG, that is the page number that the tag goes to - it's not the real net name.

Ah, ok.  Very confusing way to indicate the page numbers.

I can't see how you could be damaging the PG pin while you're operating.  So, the next things to look at would be how you solder it on the board, ESD precautions, mechanical stress of the device (during assembly or from the PCB during operation), etc.

One other thing I have is a picture of some transients when inserting the battery.  It appears we are making-breaking-making contact when inserting the battery causing some +/- transients due to wire inductance and circuit capacitance.  Could this be the root cause or would we see some other failure mechanism?  Note the Drain on D2 is the same as Vbatt just hooked up at a different location and VCC is the Vin on the regulator.

Yes, these are the kinds of things to look at.  To see if this damages the IC, it is best to use a low inductance probe technique right on the IC pins.  This allows you to see the actual signal that the IC sees.  It is possible that such a waveform, which violates the abs max ratings, would damage the device.

Some customers use TVS or ESD type diodes on the input to handle such things.