Hi Ralph, thanks for the reply. We investigated sending a unit back to TI for Failure Analysis but in the first go around there was a degree of miscommunication with Mouser (where we got the dev kits) complicated by getting a warranty replacement and keeping a TI field application engineer in the loop. With a failure on our own board we are trying to pick this option up again, but have been told we need to see at least three failures in a production run of 200. I'm sure 3 failure in a smaller run would qualify, but it's not clear we could make that case in our current situation. In answer to your questions: 1/ Are there any potential sources for transient voltages on the system at all? With the dev kits we were connecting an external pretotype board connected to the dev kit using booster pack headers. We were powering the pretotype board with a bench supply. The prototype board was entirely USB powered from a computer USB 3 port. 2/ From the USB side, can you share schematics of that portion? On the dev kit there are a couple of transient suppression diodes and nothing else between the connector and the processor for the data lines. On our board there are a couple of transient suppression diodes and an EM suppression transformer on the data lines. The dev kit sources power from from the debug USB connector. A TPS62177DQC LDO is used to generate 3.3V_MAIN which powers most of the dev kit's 3V3. 3.3V_MAIN is the run through a 1 Ohm resistor with a shunt across it to enable current measurement for the processor. On our board the 3V3 is derived from VBUS using a LM3670MF. If you think actual schematics would be helpful let me know and I'll post them as images. 3/ Have you ever been able to recover operation of the device after failure? No, but we haven't tried swapping out processors. 4/ Have you tried any ABBA swaps with the failed device and a working one? No. We are investigating ways we could do that with our prototype board, but we don't have suitable rework tools in our office to let us swap the processors out. The regulator seems to be fine. In all cases disconnecting the load from the regulator restores it to normal 3V3 operation. On at least one dev kit I disconnected just the processor by lifting the 1 Ohm current sense resistor and confirmed the regulator was operating correctly under normal load.
↧