18F6720 programming problem

[TL]

I have programmed an 18F6720 on a board with a hex file (1) using CCS ICD-U40 but I got verification errors in the program memory from addresses 0001A05C to 0001BFA4. This hex file has programmed successfully on many populated production boards previously using the CCS ICD-U40. Therefore, it is a proven hex file.

The interesting thing is that when I used another test hex file (2) with all the unused flash area filled up with data eg. 0x0000, AND used a different working hex file (3) in turn, the above chip (i.e. same chip) got programmed successfully without any problems. This shows that the above board is not faulty, or is it?

I have repeated the above actions with the Microchip PM3 and Warp13A programmers and the same results occurred.

Note: Hex file 1 and hex file 3 are similar in size and programmed correctly on previous production boards.

Could anyone explain why I cannot program the same chip with the hex file in (1)?

[Neutone]

I have seen symptoms like that before.

I had a problem once with a program that had more than one constant located in the same memory location. The program compiled without an error but this was a few years ago with an earlier version of the compiler. The programmer appeared to write both constants to the same memory location and then failed to verify the first constant it had written.

It might be worth extracting the HEX that did not verify and compare it to the HEX you were trying to write to the chip. This might give you a better view of the problem.

[TL]

Hi Neutone, thanks for your comments. Yes I shall follow your suggestion soon.