Lpc2000

>It seems as though the Read Device ID is very reliable. It works 
>hundreds of times in a row and a blank check and erase (although 
>always blank) don't cause a 'crash'. So it is likely that it is 
>during Flash programming itself.

I had this same problem.  It turned that I had insufficient bypassing on
the RS-232 converter chip.  It worked fine to read the ID, which let me
to believe the communications were reliable.  But with longer
transactions (i.e., programming FLASH), it failed because of bit errors.

Beyond this, I never could get JTAG2 to work reliably.  I wound up
bringing up my code on a Nohau eval board, getting the serial port
working enough to serve as a debug/console port, and then bringing it up
the rest of the way on the target board.  This experience was right up
there in terms of embedded development hell.  If you get JTAG2 to work,
please let me know how as I am steering clear of it going forward.

Regards,

James Dabbs, TGA