Power up initialisation failure/PMIC reset

[Karl Karpfen]

Hi,

I know there is a problem where the Ethernet (PHY?) in some seldom cases isn't initialised properly on power-up ending up in a state where it can't be used and therefore Ethernet is disabled. Now I have a BBG-cape where this happens much more often than I know this from a plain BBB. Two questions on this:

1. The BBG is powered externally by feeding 5V into VDD. Could a not-so-steep power-on rising edge of VDD make this problem happen more often?

2. When I'm in this state and press the reset-button on BBG, the problem is solved when the board is back. Is there a possibility to invoke something similar out of software by accessing the PMIC somehow? If yes - what would have to be done?

Thanks and keep the good work going! :-)

Karl

[Karl Karpfen]

I played around a bit with this and it is quite strange:

When I reset the board by pressing the reset-button on BBG, on next boot-up the Ethernet PHY is working smoothly.

When I do the same programmatically by invoking a cold reset on SoC (by writing a 0x02 into reset register 0x44E00F00, the PHY stays in same unusable condition. According to TRM of AM335x such a reset should pull output NRESETIN_OUT to LOW which should be the same like pressing the reset-button.

Any idea why one is working and the other not although they should be the same?

[Karl Karpfen]

After playing around a bit more with this I found it is a problem of BeagleBone Green. With a BeagleBone Black my hardware shows this PHY-problem for about once every 50 boot-ups, with the BBG it happens once on four power-ups.

So: has anybody an idea what is different between both that could have an influence on this?