After a couple of years working well in the field, last week we got multiple failing boards back with an AM3354.
Below a short description of the failure and how I found it :
All boards are from different production batches (and revisions) but do have the SAME failure.
We use the Vaux2 LDO output (3.3V @ 150mA max) from the TPS65910A to power all the VDDSHV6 power inputs on the AM3354. Lucky enough we have a bead on this line so I can measure the consumption :
On a good working board its 3mA at boot, 46mA max running and average of 39.7mA.
Now we got more than 10 failing boards back from the field (running from 1 to 2 years), where this Vaux2 was around 0.4 V. Step I already done :
-
Removed the bead.
-
Vaux2 output was retuning to 3.3V, so PMIC ok
-
Removed all small ceramic caps on the VDDSHV6 power line and checked them for shorts, all where ok.
-
Put an external 3.3V supply on the VDDSHV6 lines, in the first second it took 0.5A on two boards and almost 3A on a third.
-
After that the current dropped to normal and the AM3354 booted normally
-
Replaced the bead with a short (only because I’m no able to resolder the bead in 0402 size).
-
All three boards are running again from the PMIC Vaux2 ? Tested most I/O and all is ok
Is there any reason for this increased consumption on the VDDSHV6 lines, latch up ?