DM355 system boot failure in 35 degree C or 10 degree C

Miguel, thanks for your information. In lab I verified new UBL are effective to solve the boot up temp-related instability issue. Lately I received three returned failed DM355 products(new UBL) from customers. After check, this failure was caused by boot failure and happened every boot. These product were tested ok in strict factory test before shipment. This really frustrated me. Here are some debug information, help it was helpful.

1# board (never boot up ok)

UART information

sometimes it shows this without stop

UBL: booting to U-Boot
UBL: detected valid U-Boot magic number

Sometimes it shows:

UBL: Failed to read app descriptor
    UBL: NANDBoot() failed

The boot LED keeps flashing. This board can't connect with JTAG.

2# board (never bootup ok)

Sometimes it shows:

UBL: Failed to read app descriptorek
    UBL: NANDBoot() failed

Sometimes it shows:

Starting kernel … 
     Uncompressing Linux................................................
     ................. done, booting thek

This board can't connect with JTAG

3#  board (sometimes boot ok)

Error information when fails:

UBL: Failed to read app descriptor
    UBL: NANDBoot()

AWEN LEE,

I am not quite sure I understand if you are asking me more questions on the 3 additional products or if you are providing just extra information before you had success case?

                "Miguel, thanks for your information. In lab I verified new UBL are effective to solve the boot up temp-related instability issue. "

miguel->This tells me the new UBL resolved the boot up temp-related issue, is this correct understanding?

"Lately I received three returned failed DM355 products(new UBL) from customers. After check, this failure was caused by boot failure and happened every boot. These product were tested ok in strict factory test before shipment. This really frustrated me. Here are some debug information, help it was helpful."

miguel-> Are you saying that this 3 additional products fail with new UBL?

                Are you using the same board(EVM or custom board?)

                Are you using the exact same NAND memory chip in each dm355 fail case?

regards,

miguel

miguel->This tells me the new UBL resolved the boot up temp-related issue, is this correct understanding?

Yes, new UBL is effective to most of the temp-related boot issue, but not to all.

miguel-> Are you saying that this 3 additional products fail with new UBL?

Yes, the 3 products with new UBL are returned from customers

                Are you using the same board(EVM or custom board?)

               We designed the maiboard by ourselves.

                Are you using the exact same NAND memory chip in each dm355 fail case?

               The NAND memory chip is the same. Also we replaced the flash chip,but it didn't work. From the error information, it has nothing to do with flash chip.

Awen Lee,

Thanks for the clarifications.  Here are some additional questions/suggestions..

1) Has the custom main board followed all the DDR routing guidelines as described here (http://focus.ti.com/lit/an/spraar3d/spraar3d.pdf), if not this could cause problems.  (at the worst case you can provide the layout of your design for TI to review and you can try to provide them via your local TI sales rep)

2) What silicon revision of dm355 are you using? you can find out by looking at the product symbol as described in the device errata doc (http://www.ti.com/litv/pdf/sprz264e )  Some earlier Rev A revisions of the silicon had RBL issues related to ECC, this is explained in the errata.

3) are you using DDR2 or mobile DDR in your board? if mobile DDR some people have found solutions to properly initializing mobile DDR as mentioned in the following forum post. ( Re: DM355 serial / SD card flash utility supporting mDDR)  If you are not using mobile DDR then ignore this point.

4) Before attempting NAND Boot mode, have you tried simply connecting with JTAG with a different boot mode(UART or MMC/SD, to avoid Linux boot via NAND) and see if you can run some of the spectrum digital evm testcases such as the DDR testcase?  This will help you see if the DDR setup provided with the EVM gel file is allowing you to run the DDR testcase sucessfully.  You can also try to modify that DDR testcase to write/read to the entire DDR memory space and see if there is any instability there.

5) at the worst case scenario (if nothing else works) you could return the devices to TI for further factory testing, you can work that via your local sales rep.

regards,

miguel

Miguel,

Thanks for your quick suggestions. We try to verify them one by one.

One thing I need to mention is the similiar failure happen in PCBA production line, some of them can be fixed by reheating the DM355 with flux.  These three products mentioned above were definitely ok in factory test(at least 30 times power on test in different scenario), but always failed now every time. I think some physical feature changed of the board or chip, but it is hard to understand it  is related to soldering quality.

Anyway I will let you know if there is any progress.

Hi,

Taking into account that this problem is solved by reseting the unit. Could it be possible to reset unit from UBL code? 

Regards,

Paco