TPS650250: Any known input vulnerabilities when powering AM335X devices?

Hi Franco,

I have assigned this thread to the expert supporting this device. However, due to the US Labor day, response will unfortunately be delayed. You can expect a response on Tuesday US time.

Thank you for your patience.

Best regards,

Samuli Piispanen

hello, I have measured the signals as requested on a good board as well as a unit where the DCDC3 output has failed.  In the case if the failed unit, the output as well as the feedback pin are around 360mV. Please see attachedTPS650250 - DCDC3 issues.zip captures from oscilloscope.

In scope captures, Channel one is the first item in the filename, Channel 2 is the second phrase in the filename, e.g. "good-VDCDC3-DEFDCDC3.BMP" would have the VDCDC3 signal on Channel 1 and the DEFDCDC3 feedback signal on Channel 2.

Thanks.

Franco,

I will need to review the material you have provided before I respond. Please allow me a few days to review and formulate an opinion on the next steps we can take to debug.

Franco,

Unfortunately, I do not see anything obvious in the scope shots you provided that indicates an issue with the device.

The only thing that is confusing is that in the image named "good-VDCDC3-DEFDCDC3.bmp" it shows VDCDC3 = 1.5V, but the "Good" behavior would be that VDCDC3 = 0.6*(392k + 86.6k)/86.6k = 3.316V, so it is a little odd that the measurement on a Good board is incorrect.

I also find it curious that the input power supply is not a stable 5V on a bad board, as shown in the image titled "bad-VDCDC3-VINDCDC3.bmp", because if the PMIC is misbehaving I would still expect the input power supply to be stable and nearly flat.

To determine if the TPS650250 IC is related to the issue, I will have to ask you to do a routine A-B-A swap:

• Identify a Good board (A), and a Bad board (B).
  • For the sake of simplicity, a unit that starts on a Good board will be called a Good unit; conversely, a unit that starts on a Bad board will be called a Bad unit
• Remove the Good unit from board A 
• Remove the Bad unit from board B and solder the Good unit down on board B
• Solder the Bad unit on board A
• Re-test: does the failure still happen on board B (the issue follows the board) or does the failure now happen on board A (the issue follows the unit)?
• Finish the test by swapping back to the original position:
  • the Good unit on board A: does board A still work?
  • the Bad unit on board B: does board B still fail?

If you determine that the failure is following the IC, then I recommend you contact TI Customer Service & Support Center and begin the process to submit the failed unit for our quality team to review and perform failure analysis, if needed. Please follow the process correctly to make sure the returned unit is accepted by the quality team.

hello Brian,

I have completed swapping out regulators as suggested. The result is that the fault follows the board, and thus no failures associated with regulator.  I need to investigate what might cause the AM3352 go into a mode where it pulls the 3V3 supply down to ~0.4V.

Thank you for your guidance in this regard.