I second pam_add's suggestion. If those tests do not imply a problem with the platesetter, try testing remakes of the same plate starting at different points in your workflow. I would start at the end - tell your platesetter to make 2 copies of the same plate if that's possible. If it can plate twice without the RIP having to send it the image twice and the plates don't match you'll have it narrowed it down to the platesetter's data processing. If the RIP has to send it twice you'll have it narrowed down to the platesetter or the communication between the RIP and the platesetter. At the beginning of your workflow, you could recreate the job from scratch and see if you get a matching plate. If you get unmatched plates when remaking from a given workflow step and identical plates from the step immediately following, the error is probably being caused somewhere between those two steps.