Hi,
we just hit the exact same problem as the author of the related discussion (See here). Although with a slightly different device.
Some TRF7964A on some of our PCBs occasionally transmit an erroneous value for the last byte of a block in an ISO15693 Read Single Block command. The original thread does not seem to be resolved unfortunately.
I ticked all the boxes that the TI engineer brought up at that time, but to no avail. I ended up explicitly writing all TRF configuration registers (with POR value if applicable) as part of our initialization routine. Even those that only the TRF7970 models expose. That did not change anything unfortunately.
I can confirm that setting weird values (e.g. 0x01) for the RX Wait Time Register improves the situation to some degree, but it still happens from time to time. I realize this is reckless behaviour, just thought I'd mention it.
Curious if 5 years later this rings some bell with you NFC people at TI or if you have any other suggestions as to what we might try.
Cheers
Björn