Hello, we have a C6655-base design that has been on production for at least 5 years.
Because of component shortage, we had to qualify a new clock oscillator for the CORECLK.
With the new selected part, we have 10% of the production failing. It turns out that at at the very beginning of the CVDD ramp-up, the intial current reported by the IMON_A output of the UCD7442 is twice the amplitude we have on good board and at the end of CVDD rampup, a huge current surge occurs triggering protection, which cause the processor subsystem to shutdown.
What does it have to do with the oscillator ? Well:
- Taking a faulty board and replacing the oscillator with another brand or the original device clears the problem
- Taking the oscillator from the faulty board and mounting it on a golden board sample (built years ago) makes the problem appear on that board
- Having tjhe clock oscillator forced in run mode during the power-up clears the problem as welll
- We looked at usage note 10, and we a clean with respect to it
- We studied a post on what looks like a closely related problem (solved by after a faster CVDD rampup time) but this did not help
- We looked very closely on possible CORECLK glitches during the rampup process (oscillator in tristate), The following capture shows CVDD, IMON_A and the clock output from the oscillator (differential signal)
- We looked at possibly bad DC-couping capacitor but the common mode voltage before and after the caps are as expected.
Any hint on what could possibly wrong with the C6655 would be very much appreciated. I known the obvious short-term solution is "pick another oscillator".
Still, how can the processeur powerup be so sensitive to the oscillator ? If this is not the oscillator (as I first believe), then can it be explained that simply swapping oscillators between two boards cause the problem to follow the oscillators ?
We have plenty of capture to share in case it is needed.
Pascal