I now have a printout of the JMRI to EasyDCC commands for two locos. One with a year old basic NCE decoder using long address 349 works. One with a Lenz LE103 using long address 1477 does not. I will try decoding them on paper to look for errors.
What I did was try to run forward, stop, run backward, stop, turn the light (CV0) on and off. Those should be easy to decode manually on paper.
Sorry to add confusion, but after reprogramming the LE103 in 1477, it now was sometimes able to switch the headlight on using CV0. It only seemed possible while the EasyDCC Command Station was still circulating the packet from the EasyDCC throttle that turned it off. This was right after unplugging the EasyDCC throttle.
I'd use JMRI to turn on the light, a couple seconds later it would turn off. I'd be able to turn the light back on, I think by pressing the light command twice on the JMRI throttle to get another "off to on" packet generated. After a second or two, it would turn off again. Once about two minutes had passed, and the other packet was cancelled, I couldn't control the light with JMRI. At no time could I make the loco move, forward or backward with JMRI.
I may have to build one of those Arduino based packet sniffers. Data would reduce the number of possible explanations. I appreciate all the analysis, but it's getting confusing. With data to compare to that analysis, some understanding might follow quickly.
If it were only the LE103s that didn't work, I might just replace them with the Zimos MX60 somethings I bought years ago. But, the one Zimo equipped loco I have tried has also not worked right. I could control functions, but not speed and direction. That may be something more easily solved.
I keep hoping I'm one "Ah hah!" moment from a solution. It's beginning to seem I'm just wasting more and more time. But, those decoders were Standard compliant when new, and they SHOULD work! I didn't imagine I might have to replace all my decoders with newer ones to use JMRI....
Don Weigt
Connecticut
