Fail on first access to i2c (SOLVED)

[oxo]

Not hot or cold.

I am screwed now, because it's been working since friday morning, and I can't get it to fail again.

[PCM programmer]

That actually is helpful. It likely means that your lab conditions are
different than field conditions in some way. Try to imagine what
the differences are, and try to duplicate them in your lab. Visit the
field location and look around.

1. Temperature
2. Humidity
3. Static electricity (carpet, humidity, people, etc).
4. Power source stability and cleanness.
5. Shielding of the circuit board from nearby EMI pulses.
6. Human intervention
7. Animal or insect intervention

[temtronic]

If this is just ONE board that now works, check the PCB for connections and solder joints. I had some PCBs that failed in the field,on very cold days.Turned out one of the hand sldered joints was never done. The trim pot wiper was just in contact and NOT soldered to the PCB.Once it got into the Real World and got real cold, PCB contracted and wiper was no longer in he circuit. This of course had to happen in the middle of the night, middle of Winter in a cast iron foundry.
Since you've isoloated it to just 2 chips( the PIC and the gas gauge),use a high powered magnifier and a dental pick to confirm they are soldered properly.
As it's now working, as Mr. T say 'something' has changed which is good, it's a matter of finding out what now.

Be systematic, make a list, check off items as you go.

Jay

[newguy]

Since no one else has mentioned it, this could also be caused by a cold solder joint(s).

In my first real job the quality guy gave us the standard spiel regarding common failure modes and cold solder was number one closely followed by ESD damage.

[oxo]

Here's an update..

Problem is solved.

It turned out to be caused by the charger not bringing the flat battery up above the minimum supply voltage for the gas gauge chip.

If the code writes to the gas gauge chip before the minimum voltage is achieved, the chip locks up, and the only way to get it back is to remove the battery power.

So now I have a 50 second delay after applying the power before initialising the gas gauge.

Thanks to all for helpful suggestions

[Ttelmah]

Well done.
PCM's 4th suggestion.

Mark the thread as 'solved' (edit the title).

It's more common than you might think. The PIC generally wakes at quite low supply voltages, and a lot of chips need significant time to actually get started properly.