We tested PicoZed well below +20C. The C-grade version was tested down to -25C and the I-grade down to -45C. I would like to try and duplicate this issue here on my bench. Is this a problem that you can duplicate with an Avnet Carrier and the pre-built image that we have posted for PetaLinux? If so, I will try the same thing.PetaLinux 2015.4 Compressed BSP, z7015
I have 5 SOM and we have seen this issue on 2 of the 5. I can try reverting back to the pre-built image and will post the results.
We've been experimenting with the cold start eMMC issues mentioned above. I can confirm that we have seen this issue after testing enough boards. I estimate that this affects 25% of all PicoZed SOMs.
We have also found a work-around to eliminate the failure. The solution is to switch the Zynq SDIO clock from the typical high-speed mode to standard-speed mode. While we do not think we have the Hold problem identified in this Design Advisory based on measurements that we have taken, the suggestion to switch to Standard Speed Mode does work.
We have posted updated copies of the PetaLinux 2016.2 BSPs that puts U-boot and PetaLinux both into SDIO Standard Speed Mode for the eMMC, which uses the Zynq PS SD1 controller on PicoZed. Please check out these updated BSPs, posted on the PicoZed Carrier pages:
We will also release an Errata that states that PicoZed does not support High Speed Mode clocking for the eMMC.
I am using the AES-Z7PZ-7Z015-SOM-G Rev C04. When the board is cold started (power up after being powered off for a while) it will boot into Pentalinux but will not load the OS from the eMMC. After a couple of power cycles it will eventually boot al the way up.
I put the SOM and Carrier in a temperature chamber and I could repeat the failure when the set point was less than 20C.
Below is a log of what I am seeing when it fails:
U-Boot 2015.07 (Sep 01 2016 - 14:31:20 -0500)