Actions
Forum Thread Details
  • State Not Answered
  • +1 person also asked this people also asked this
  • Replies 2 replies
  • Subscribers 331 subscribers
  • Views 211 views
  • Users 0 members are here
Related

Power Up/Down Sequencing for Ultrazed 7EV SOM

mad_dog

Hello...I am having an issue with a custom carrier board and the required power sequencing of the SOM. Testing the sequencing with no SOM or second board mounted shows correct sequencing of all supplies starting with MGTRAVCC (0.85V) and finishing with 3.3V used to power VCCO_HD_47 and VCCO_HD_48 for HD IO banks 47 and 48. Power sequences up and back down in the order specified. Sequencing is controlled by an "always on" circuit on the board that does not touch the SOM. It just monitors the unit power switch and sequences up/down the supplies. We have a second board in this system that comes on immediately after 12V power is applied to the system (including the SOM) and this board has an independent 3.3V regulator to power itself. Circuitry on this second board has various outputs and lines with 3.3V pullups that connect to the SOM IO on banks 47 and 48. The result is these lines, driven immediately by the second board, are pushing the 3.3V VCCO_HD supply on the main board up to about 1.6V well before the other SOM supplies have sequenced up. When finally the sequencer turns on the VCCO_HD supply it then snaps right up to 3.3V. A similar issue happens on power down where VCCO_HD should be the first one turned off but it sits around 1.6V until the 12V power for the second board is removed at the very end of system power down.

We have seen a number of SOMs seemingly damaged in our system and I am suspicious that this is the culprit.Typically, we see MGTRAVCC shorted to ground on the SOM and some of the other MGT supplies show fairly low resistance to ground compared to a new, undamaged SOM. Can someone render an opinion on whether the issue described is likely to damage the 7EV SOM? The power up/down sequence is about 300mS in duration.

Thanks for any help...will be much appreciated.

  • 0

    First off - if your carrier card is somehow blowing up SOMs, I would immediately stop using the Custom Carrier Card. The SOMs are too expensive to be damaging.

    If you don't have one, I would procure an UltraZed-EV Carrier Card and make sure your SOMs aren't arriving pre-damaged since the Carrier Card is a known good HW environment for a sanity check. All of the SOMs are verified functional from the CM with a low overall failure rate.

    So your second board is feeding the IO of the MPSoC early which in turn is back feeding the VCCO_HD (3.3V) rail on the SOM/Carrier Card. That is not good and could probably lead to device failure you are experiencing. Typically, out of sequence only results in a higher current draw during the bring up, but this is a different case all together where you are back feeding through pullups.

    Have you tried physically controlling the 3.3V enable on the second board to delay this to the SOM/Carrier Card? You probably should have a load switch on that second board to pass the 3.3V to the SOM/CC in a timely manner. Perhaps monitoring the control signal that turns 3.3V on from the Carrier Card. I am curious why the 3.3V pullups come from the second board and not the Carrier Card VCCO_HD?

    Also, have you verified that you are not having manufacturing problems with the JX connectors or perhaps the footprint of your MGT rails since those are the ones that seem problematic. Have your manufacturer X-Ray those boards and check for opens/shorts where you specifically see the problems.

    I'm sorry you are going through this, but I am pretty certain anyone who has designed hardware has back feed a device at one time or another.

  • 0 in reply to iksevas

    Thank you iksevas for your reply. Yes it is certainly a design flaw and will be fixed in the next revision of hardware. One odd thing is the current design does not damage every SOM. We have built around 50 of these systems without noticing this issue but it has suddenly cropped up on the latest build. I am about to modify the hardware with "cuts and jumpers" to retest and see if the problem is resolved. I was hoping I might find someone here that had seen this before, where back powering an IO bank during the power up/down sequence has actually damaged their SOM. Thanks again for your reply.