SM320C6713B-EP: SMS320C6713B: Cannot get correct IDCODE from SMS320C6713B

Manabe,

Unfortunately, I cannot answer this conclusively.

However, I did find some internal information that seems to indicate that the DSP is incorrectly outputting the BYPASS register instead of the IDCODE.

The information that I found is quite old, and it did not come to a conclusion.

Does the value match what would be expected for the BYPASS register?

We do not have the capability to regenerate or modify the BSDL file.

I do not expect this is indicative of any issues with the device.

Regards,

Wade

Wade,

Thank you for answering.
I will attach the JTAG waveform when acquiring IDCODE,please let me know if you have noticed any additional information.

 Waveform of acquired IDCODE.pdf

Regards,

Manabe

Manabe,

I want to understand what you mean by "normal DSP" and "specific DSP"

Does this indicate that some of the DSP's respond with the IDCODE, and some do not? 

If so, what are the differences?   Different date codes?  What is percentage that work as expected vs not?

Regards,

Wade

Wade,

Thank you for answering.

Explaining exactly the difference between "normal DSP" and "specific DSP", some DSPs output IDCODEs different from BSDL files when issuing IDCODE instructions(0x04) to DSPs using the JTAG tool. Although the board pattern design is different, DSPs with the same part number are used. As a result of trying to obtain IDCODE for 13 DSPs, IDCODE was successfully obtained for 6 DSPs. Details are shown in the attached file.

 　7318.IDCODE acquisition result list.pdf

It has been confirmed that the waveform of the IDCODE instruction(0x04) issued to the DSP is always the same. Please tell me what you notice.

Regards,

Manabe

Manabe,

I am not certain what the issue is. 

However, I was reviewing the silicon errata, and noticed this statement.

Boundary Scan: IDCODE is Only Loaded Onto Instruction Register When TRST Becomes Inactive
The IDCODE instruction is loaded into the instruction register only when the Test Logic Reset state is entered by
transitioning TRST from low to high, but not when the Test Logic Reset state is entered by holding TMS high and
clocking TCK five times.

It is possible that some devices may get the IDCODE from a TMS reset, but not robustly.

Regards,

Wade

Manabe,

Have you evaluated how setup is getting the IDCODE?  Ie, does it assert TRST, then obtains the data?  Or possibly if other commands are executed between the TRST and IDCODE it may be corrupted.  Additionally if the software uses a TMS reset sequence, then per the note it may not work correctly.

Regards,

Wade

You are welcome.

If you have the ability to reset prior to getting the IDCODE, I think that would be good to try based on the data I had found.

Otherwise, I am not sure what else to try.

Regards,

Wade