I ran into a couple little snags when attempting your solution. the first being the melabs programmer doesn't allow directly modifying the hex file, so I had to use notepad and discovered it is all stored in Little-Endian but I actually got to remember something from college after 10 years So I think I did it correctly other than my check sum was no longer good, but hopefully it is OK.

My second problem is my chip doesn't use config1L so I got a bunch of errors and will have to compare data sheets later to figure out what you were doing in your example. Unfortunately I need to be productive for the next few hours and will have to get into this later today. I'll let you know if I have any luck

Thanks for your help

David