I have a customer who is seeing ECC errors in the OTP region of the Delfino ‘TMS320F28375D_ZWT_337’ on an estimated 25-40% of hard power cycles.
On power up we are doing a read verify check of the various memories and the reads of the OTP consistently (25-40%) show the same ECC errors.
The location of the ECC errors are 0x701F8 (single bit correctable error) and 0x701FC (multi-bit un-correctible error).
If the control is re-started the ECC errors are cleared and do not return. We have not seen the problem on a software reset.
Normally, the only way a ECC error can ‘go away’ is if the location is written to.
Since this region is written to ONLY by TI, never by us, it appears to be either a glitch in your ECC detection (circuitry or software), or an actual problem in the OTP memory cells that some built in TI process is trying to correct.
Is this a known software glitch or some other type of problem TI has seen before?
Currently, since we can not write to the memory or work around it in any way, we only have the choice to ignore it.
Are there other errors we should ignore as well?
