Actions
Forum Thread Details
  • Replies 10 replies
  • Subscribers 355 subscribers
  • Views 1351 views
  • Users 0 members are here
Related

UltraZed 3EG VCCO_PSIO failure

kgervel1

I'm using an UltraZed 3EG SOM on a custom baseboard modeled on the IOCC which has been extensively tested with multiple SOMs.  We had successfully programmed the QSPI and eMMC with our software and powered on and off several times.  Then upon one power up, only the red D4 LED (for POR) was lit, and I also thought I saw some smoke from end of the SOM where D3 and D4 are.  On my baseboard, I still have a PG light on from the SOM.  On further investigation with my IRPS dongle, Output A on U19 (one of the SOM power supplies) is disabled because of an overcurrent condition (see attached).  Output A provides the VCCO_PSIO net at 1.8V, which powers the VCCO_PSIO core in the Zynq, as well as QSPI chips and a few other things. All the other IRPS outputs look okay.  When this happened, I remembered that several months ago, we had brand new SOM that had never worked, and the POR light was always on.  I then pulled it out today and checked with the IRPS dongle, and sure enough, the exact same thing.  So potentially seeing the same issue on two SOMs.  I tested this baseboard with a previously working SOM and everything is working okay, so I think the issue is on the SOM, not the baseboard.

 

 

image

 

image

  • Hi,

     

    Could you tell me what the PSIO rail is powering on the carrier? The rail is designed for 500mA output. It could be possible that the device failed (especially with the smoke) before OCP activated damaging the device.

     

    Could you give me a rough estimate of the current draw from the carrier?

    Thanks
    Chris

  • in reply to ctammann

    Chris, the PSIO rail that failed is the one on the SOM, not the carrier.  It is the IRPS located at address 0x13 in the screenshot in my first post, and it doesn't power anything on the carrier board. It corresponds to U19 on the Ultrazed-EG SOM. From the SOM schematic, it looks like it powers the QSPI and the Ethernet PHY on the SOM, along with a few other minor things.  I believe it also provides the 1.8V for VCCO_PSAUX,VCCO_PSIO_500, and VCCO_ PSIO_503 for the UltraScale+.

     

    My carrier board is fully functional when I put a different SOM on it, so I think the issue is with this specific SOM.  All of my carrier hardware is powered from the IRPS power supply labeled 0x12 in the screenshot.

  • in reply to kgervel1

    image

     

    Here is the PowIR screenshot with a functional SOM attached to the carrier board.

  • in reply to kgervel1

    I have a power check tool for FPGAs that may work in your case if you want to identify possible intermittent failures.  It is called PowerLoad and it configures the FPGA to operate as a programmable electronic load for its own power supplies.  Here is a link to more info:  http://piscanner.com (slow loading website please be patient).

    Regards,

    Cosmin

  • in reply to kgervel1

    Apologies, I got the rails mixed up and forgot this one did not get passed to the carrier. Nothing is jumping out at me so please let me look into this a little further. Out of curiosity, have you created an XPE for your design? I'd like to see what the expected current draw for the Zynq would be for your design.

  • in reply to ctammann

    I do have an XPE that was just an initial shot at it during the hardware design phase. I'm not sure how to upload it here so you can see it.  But here's a shot of the front page with the summary, anyways.image

  • in reply to kgervel1

    Hey Katey - I haven't been able to find any history for this issue and haven't seen it myself yet. Can you tell from visual inspection where the "smoke" came from? I've been looking through the schematic, but again nothing in my notes or that I've come across from other customer inquiries has given me any leads.

    Thanks

    Chris

  • in reply to ctammann

    It's not obvious from looking at the board what might have been the source of the smoke. At first I wondered if it came from my carrier board, but the board is fully functional with other SOM's, and that also doesn't really make sense for the SOM power rail that is down.  In any case, neither the carrier board nor the offending SOM have any obvious marks that would indicate where the smoke came from.  I will say that this SOM did not have a heat sink/fan installed on it yet.  Could it be possible for the Zynq itself to have smoked? I would think I would see other power rails having a problem, though, in that case.

     

    Something else I wondered about--we have a different SOM where the QSPI has gotten into a bad state and can't be written to.  The QSPI is powered by that same power rail that has failed on the first SOM--so I wondered if that's coincidence or if it could be a related issue.

  • in reply to kgervel1

    Hey Katey,

    So sorry for the delay, I must have missed the response notification for your last post. The Zynq without the heatsink could very well be what failed. From what you've described that definitely sounds like the most likely suspect. Given the rail that failed I agree it sounds SOM related and not the carrier. I haven't seen a QSPI failure trigger a power failure but it is certainly possible. You could try removing that device and see if the rail powers up. However, the most likely explanation is a failure of the Zynq itself. The Zynq runs pretty hot so if it was operating without the heatsink and fan the AUX and CORE rails would the the most likely failure points.