DIE revision identification for the TMS320C6747

Guy Caron

Other Parts Discussed in Thread: TMS320C6747, OMAPL138

Is there a 'foolproof' identification register on the TMS320C6747 that identify the DIE revision that is user SW accessible?

I know that the boot rom revision is somewhat changing with the DIE revision (D800K001, K003 and K005), but this appear to be 'iffy'.

Also, the SYSCFG DEVID and REVID don't seem to provide any indication of the DIE version.

There are 4 DIEIDRn registers, but they are undocumented and 'for internal use'.

Please advise.

Also, this info should be added to the WIki page related to the 'premature IO aging' of the chip, as I am pretty sure peoples will want to know if they are running on a bad die and possibly ask the customer for an exchange.

over 12 years ago

0 Sunil Kamath over 12 years ago

TI__Expert 7470 points

Guy,

The Boot ROM revision is indicative of the DIE revision also. There is a debug gel file available on section 8.1 of the wiki link given below that reads the Boot ROM and prints out the silicon revision.

http://processors.wiki.ti.com/index.php/Debug_and_Trace_Capabilities_on_OMAPL1X/AM1x/DA8x_devices#Debugging_GEL_file_for_boot.2Fapplication_crash_related_issues

For OMAPL137 and C6747/5/3,

ROM ID = 1 for Rev1.0

ROM ID = 3 for Rev2.0

ROM ID = 5 for Rev2.1

The even numbered ROM IDs are for the OMAPL138 and C6748/6/4.

Regards,

Sunil Kamath

0 Guy Caron over 12 years ago in reply to Sunil Kamath

Prodigy 105 points

I know how to read the Boot ROM version, All I need is a confirmation that this will still be a reliable indication of the DIE revision in the future. My concern is with identifying the premature aging parts (2.1 and earlier) relative to the new ones (supposed 3.0?) coming 2Q2013.

0 Guy Caron over 12 years ago in reply to Guy Caron

Prodigy 105 points

So, is the bootloader revision the official TI recommended way to identigy the die for the C6747?

0 Mukul Bhatnagar over 12 years ago in reply to Guy Caron

TI__Guru* 83815 points

Guy

Going from PG2.x c6747 to PG3.0 there is no ROM change, so the ROM ID will remain the same. Ideally PG2.1 vs PG3.0 should be distinguishable by the JTAG ID (DEVIDR0) field , that allows distinction between every major silicon revision ie. PG1.x to PG2.x to PG3.x , however we confirmed that this is not being done for C6747 PG3.x silicon and the only chang is in the IO buffers. So unfortunately there might not be a way to distinguish this in software and you would need to rely on ordered parts and case marking etc

Regards

Mukul

0 Guy Caron1 over 12 years ago in reply to Mukul Bhatnagar

Prodigy 20 points

Wow, this is not good. Will the GPIO REVID change in some way?

0 Mukul Bhatnagar over 12 years ago in reply to Guy Caron1

TI__Guru* 83815 points

Guy

We are following up on one more aspect to see if we can make any change on the efuse configuration to allow software distinguishability. I will not have the answer on this till the end of the month. Will update this thread, if that provides an avenue of what you want to do.

No other peripheral revision id register (including GPIO), will be changed.

Regards

Mukul

0 Guy Caron1 over 12 years ago in reply to Mukul Bhatnagar

Prodigy 20 points

I will (anxiously) wait for the update...

0 Guy Caron over 12 years ago in reply to Guy Caron1

Prodigy 105 points

So, will there be a way to identify the new 3.0 silicon compared to the 2.1 revision?

0 Guy Caron over 12 years ago in reply to Guy Caron

Prodigy 105 points

No info yet? Should I worry that the parts will be available later than the expected 2Q2013?

Processors

Processors forum

DIE revision identification for the TMS320C6747