Made a digi conversion of my very first Märklin loco, the classic beginner's loco... model 3000.
Only problem with the engine conversion was the lower screw, it was _both_ too long (locked up a gear wheel on the other side) and had a too large head (made contact with rear left wheel). Nothing unsolvable.
I also put LED:s in it, both front and rear, looks really nice.

When I first started running it, it didn't run smooth at all, seems like the default values are "as few settings as possible" or something. :P
So I made some adjustments in the basic loco settings menu (don't remember its name) where you can change Vmin, Vmax, Acc, Dec.

Loco suddenly ran as it should, with smooth acceleration and deceleration.
No surprises this far...

But after I had run it for 15-30 minutes, it suddenly stopped abruptly???
No "STOP" error in the MS2 and LEDs on the loco kept shining.
Then after a few seconds it started to move again, by itself, smoothly, like nothing had happened.

And then after a minute or two it did the same thing again.
And again.
Etc...

Emailed Digitrax support, who said it didn't sound like a decoder problem???
I argued.
In the next email they asked me to reset the loco settings in the controller... Since I only have two digital locos this far (And the other one runs fine on the same track btw!) I decided to reset both the MS2 and the Digitrax decoder.

All smoothness gone again of course... BUT it just kept running without a single stop. Had it go round and round on my little test oval for at least half an hour without hiccups.
"Ah, nice, the reset worked."
So I adjusted the Acc and Dec values again.
And then the problems began again. :(

I have since then tested back and forth, and it seems it's the Acc setting that f things up.
Then I found out that it should be possible to change some CV value directly to make the same settings, so I tried to do the advanced thingie but now the MS2 started to whine about a "STOP error" even though it doesn't seem to be any.

Digitrax doesn't seem to want to help me anymore, at least I got no answer on my last email...

So I turn to you instead. Does anyone here have any similar experiences and perhaps a solution?

Or is the decoder simply faulty, in spite of what Digitrax claims?
Or incompatible with the MS2 in spite of the fact that both are supposed to support the DCC protocol?

Frustrating! Partially because I'm still a beginner with this digital MR stuff, it feels like "There should be a solution but I know too little to figure it out, gah!"

One of the replies from Digitrax also suggested that I used the wrong values. I assumed that since they talk the same protocoll, at least the basic settings should be fool proof, but that was of course just an assumption from my side.
And that's why I decided to investigate the CV setting (CV3 if I remember correctly) but got stuck on the error message from the MS2. I had previously changed another CV value successfully, to change the default direction of travel (instead of re-soldering the wires to the motor... lazy!). But after the reset of everything, I didn't touch any CV values directly, only used the graphical interface with a scale from - to +.

My first thought actually was that the decoder was overheating, so I have tested with as much air as possible around the decoder, but it didn't make any difference. And directly after one of the malfunctions, I touched it and it wasn't abnormally warm. I think. There's a red shrink plastic cover around the electronics but I doubt it can insulate _that_ much. And I don't want to remove it since that will most likely woid any warranties...

For the engine conversion, I used the Märklin 60943 kit (with 5-pole rotor).

Sorry, can't change those. As I said, the MS2 started to whine about a "stop error" when I tried to change the CV's.
Weird thing is, I was able to change CV29 earlier (to reverse normal direction of travel) so something seems to have changed during my experiments. :/

It seems like the MS2 can write values blindly, since I can still change "the basic settings" as I call them, but when I try to access the CV's directly I get the stop error, and I assume it's because the MS2 has problems reading the current values off the decoder for some reason?