Digitrax decoder woes?

[NYC54]

I have 2 Atlas newer model GP-40 loco's with the Digitrax DN166I3 decoders. They both ran fine until recently. I am having trouble with one.
This loco runs perfect if by itself. Nice slow speed both forward and backwards. It's when I try to make a consist that I'm having problems.

I took it apart around Thanksgiving and cleaned it up and greased it up. It's ran fine until the last week or so.
Last week I made a consist using this loco and 2 of the newer Atlas GP-9 with the ESU LokPilot 59925 decoders.
The consists ran great running forward. When I reversed direction is when my troubles started. The loco would sit there and
"Short out". I'm running an NCE SB5 system using 2 wired ProCabs and my Power Cab as the 3rd throttle when needed.
I have an NCE EB1 as my circuit breaker. When I say "Short Out" I mean the loco sits there and the lights blink and other running loco's stop.
The EB-1 led is blinking as it is trying to reset it self. It's a continuous loop until I remove the loco.

So I clear the consists and take the loco with my Power Cab over to my programming track and I get nothing. The lights will function but loco won't move
I go to programming mode and see that the short loco address is some weird number and not 003. It still shows 3523 as the long number which is correct.
So I reset the decoder by doing CV8 with 8. I go thru re-addressing the loco and putting all my CV number back in. It runs great on the programming track.

I try to consist it again with the 2 ESU equipped GP-9's and it does the same thing. Runs great forward but as soon as I change direction it "Shorts out" again.
Again the Short address is some weird number. I take the decoder out and reseat it and reset it again. It runs fine on the programming track.

This time I thought I would consist it with 2 other loco's with NCE decoders. Again it runs fine until I change direction and again it "Shorts out".
Now this time the Short address stayed as 003 and the long address is correct as well. I clear the consists and it run fine.

So short of me taking the decoder from the other GP-40 and trying that in the loco I'm at a loss as to what to do.
Listed below are the CV's that I change to get this to speed match with my other loco's.
CV2 =7... CV57 = 102... CV5 =65... CV66 =125... CV6 =35 and CV95 =125. All other CV are left at the default settings.
Anybody have suggestion I would appreciate it.

Thanks, Tony

[Maletrain]

Any chance that the loco that runs fine in forward direction but then "shorts out" when reversed in a consist is actually somehow getting pushed by the other locos in a way that makes some sort of contact in the misbehaving loco?  What happens if you do not have the loco in a consist, but put it in reverse and hold it to a stop by letting its coupler push against an immovable object?

Another thought: are the couplers you are using metallic?  Could there be a short from one loco to the other when they are pushed together hard enough?

The blinking EB1 does seem to prove that there is a short circuit being created.  And, since it doesn't happen when the loco is in reverse but is not in a consist, it does not sound like it could be a decoder problem.

[NYC54]

Maletrain... I had this loco in a consist with the shell off so no couplers. The other 2 were coupled together with about 10 freight cars. I'm running the stock Atlas couplers.
As I said in a consist with the shell off and it runs fine forward but shorts out in reverse. Clear it from the consist and it runs fine both directions. I've tried 2 different consists
with 2 different locos each consist with 2 different decoder brands all with the same results.

Edit:...
I just came in from consisting this loco with just one other loco and still had the same problem.
I'm done for the night as my layout is in my heated garage. I think tomorrow I will consist it again but change the direction of the loco and see if that makes any difference.
I had it setup as the lead loco. I will try it tomorrow as the 3rd loco facing the opposite direction and even try it as the middle loco and see what happens.

[peteski]

This makes no sense to me.  Consisting DCC loco does not affect it mechanically or electrically (as to cause shorts). All the same components of the model are used in solo and in consisted operation. The only difference is in the internal decoder (software)  configuration.

Re-reading your "short out" description it does not sound like an actual short, just an overload.

Weird or not, if you have 2 identical models with the same decoder and only one has the problem then to me it seems  like that decoder is a suspect.  Since those decoders are easy to remove and install I would swap the decoders between the locos and see if the problem follows the decoder or stays with the loco. That way you will know what needs to be fixed/replaced.

[tehachapifan]

It's been a loooong time since I messed around with Digitrax decoders, but I seem to have some recall with at least one brand of decoder that blinking lights could indicate a speed step mismatch. As in, some locos are set to something like 14 steps and others at maybe 128 steps. As I recall, this could cause behavior similar to what you describe when in a consist. Are the lights blinking in a rhythmic pattern?

[peteski]

If speed step mismatch is the problem (reading out CV29 and verifying it woudl be helpful), then here is a thorought explanation of the effects:
https://sites.google.com/site/markgurries/dcc-welcome-page/advanced-topics/decoder-motor-drive/mixing-up-speed-step-modes

[NYC54]

I put this loco on my programing track and CV29 =038.
I put it's twin sister (the one that work perfectly in a consist) on the track and CV29 =038 as well. I use 28 speed steps on my throttles.

I'm waiting on my train room to heat up some before I go out there. I will swap decoders and see if the problem persists.
I will report back later this afternoon or early evening.

To answer Maletrain question: (What happens if you do not have the loco in a consist, but put it in reverse and hold it to a stop by letting its coupler push against an immovable object?)
On the programming track it runs fine butted up against the stop. It sits there grinding away at all speeds like it should. This is the chassis only no shell on it.

Thanks for all the suggestions so far. I may just buy another decoder just to have a spare on hand as these 2 loco's are the only ones that use that model.

Tony

[Mike C]

 I can think of 1 more test . Consist the loco with one of the others but do not couple them together . Now try running them , does it still do it ?