Since I can't seem to cut & paste to my own post, I'll just re-type the short version of what I wrote to BLI tech support.
As far as I can tell, the issue is two fold.
1. I haven't had to clean the layout track, or loco wheels in years. They "look" clean, but apparently there's enough oxidation at this point to interfere with consistent good conductivity.
2. With these particular BLI locos, all 3 Burlington E8, the road number board lights come on when the decoder is being written to, and then go off. If I don't wait for the lights to go off, AND do not save the changes before removing the loco from the programming track, AND do not change the Digitrax PR3 back to Interface Mode - before I remove the loco from the programming track, then sometimes, more often than not, somehow causes the loco address to change to 127, maybe because the PR3 is a little slow at completing its task with some decoders. Just my unqualified opinion.
I've read that the newer Digitrax PR4 is at least 3x faster, AND you don't have to manually change the Programming Mode back and forth between Program and Interface (mainline). It looks like Digitrax may have become aware of this issue, hence the PR4.
I replaced my PR3 with a PR4 when it first came out, and I found it did work a lot better.
99 might actually be a jmri read error.
Most of the time it returns 255 when it can't read a register.
But cv1 is kind of special. Some command stations support 1-127 for short address. Some only support 1-99.
The dcs52 is agonizingly slow on the program track before it completes it's setup and reads. It's waiting for the decoders to power up I think.
What hardware are you using to read the values? And what mode?
Don - Specializing in layout DC->DCC conversions
Modeling C&O transition era and steel industries There's Nothing Like Big Steam!
I think I've solved the issue.
1. Even though my track and wheels looked clean, they were not. I haven't had to clean the track for maybe 6 or 7 years now. I gave everything a good cleaning by hand with "odorless" (LOL) mineral spirits.
2 The DigiTrax PR3 seems to be part of the problem. As I became more proficient I guess I must have begun doing the programming proceedures faster & faster. That, along with some locos having more sophisticated decoders, taking longer to read, write and save the data, I think I may have been pulling the locos off the programming track too soon. DigiTrax says the new PR4 is much faster.