Atlas GP38 with Intermittent DCC Troubles

[MK]

Similar to what Nick is asking, what DCC command station are you using?

I ask as I *just* had my 20 year old Zephyr (0G) die on me.  One spot got hot to the touch and then I smelled burning electronics.  Opened it up and it looked like a component in the power regulating side of things.  A guess here, as I think Peteski would know what burned out if I post a picture.

Anyway, the initial symptom was the loco slowing down and then stopped.  Pushed it a little thinking dirty tracks and it went an inch and then died again.  Then headlight flickering and no motion and finally completely dead.

I was fussing with the loco and track without realizing the display on my Zephyr was dead and then "What's that burning smell???"

[peteski]

Similar to what Nick is asking, what DCC command station are you using?

I ask as I *just* had my 20 year old Zephyr (0G) die on me.  One spot got hot to the touch and then I smelled burning electronics.  Opened it up and it looked like a component in the power regulating side of things.  A guess here, as I think Peteski would know what burned out if I post a picture.

I might have a clue as to what burned up. 

I believe Dwight uses NCE Power Cab (same DCC system I use).  The wall wart supplies little over 13V DC so I suspect the DCC track signal will be around 12V. I never actually measured it.

From past troubleshooting experience with Dwight I'm also assuming that all the other locos he owns operate correctly (even if placed on the track along with the misbehaving loco).  But I guess that is a valid question.  Assuming things can lead to trouble.

[Dwight in Toronto]

Hey guys - I use an NCE PowerCab.

And yes, dozens of other ESU sound-equipped locos all run fine, even as the troublesome Atlas is acting up.  As mentioned earlier, I usually have this particular loco in consist with two others, where it is merrily contributing to pulling the load.  At least until such a time where it craps out and suddenly becomes a major load in its own right, with the remaining two locos now struggling as pushers against the dead weight. 

I can measure the DCC track voltage, but I’m pretty confident that that is not an issue.

[jagged ben]

Dwight, pardon if I missed it, but have you tried factory resetting and reprogramming the decoder?

One thing I've wondered about is ESUs 'overload protection'.  I recently installed a Lokpilot Micro in an Atheran FP45 and the first time I went to speed match I had issues.  Thing had been running in circles at top speed for a few minutes while I adjusted CV 6 when it started stuttering every couple feet, and before I had the foresight to stop it the motor cut out entirely, and wouldn't run again.  I was worried my motor isolation had failed, but upon close inspection that seemed not to be the case.  Finally did a decoder reset and the thing came back to life and seemed to run fine again. 

The FP45 is a heavy locomotive with a fast motor so my theory is I tripped the overload protection (by running at top speed for too long).  ESU claims 'overload protection' on all its decoder outputs but gives no indication of how it works.  Presumably it works by ceasing output on overload, or pausing it.  In my case it seemed like it progressively ramped up the 'protection', to the point a reset was required.   I don't know how to confirm that's what happened to me, but thought I'd share.  Just searching for possibilities here.

[Dwight in Toronto]

jagged ben - thanks for sharing your experiences. 

I humbly submit that I feel that I’ve achieved somewhat above-average familiarity and competence with ESU’s decoders and the LokProgrammer.  However, I have to admit that doing a ‘factory reset’ is something that I have not tried. 

With other decoders, standard reset practice was to set CV8=8, but I’ve seen plenty of forum dialogue that suggests that such is NOT the case with ESU.  Popular advice and opinion is that ESU’s are uniquely different in this regard, and that there is NO factory default, as such.  I’m not entirely clear on the various in’s-&-out’s; perhaps others here on TRW can provide an insightful explanation.

I could certainly try re-writing the entire sound project, but I think I’ll investigate a few other things and maybe resort to that when all else fails.

[peteski]

As I understand CV8=8 does a "sound project reset".  By that I mean that when you create a project and save it, the LokProgrammer asks you if you want to change the default settings.  If you agree then loading the sound project will also save another copy of that "virgin" project  in the decoder's Flash memory.  So then when someone mucks around with all sorts of CV settings, and wants to restore them ti the original values  in that sound project, they have to  write 8 to CV8.  That copies the "virgin" copy to the working copy of the project residing on the decoder.

I didn't think there is a true reset  to reset the decoder to how it came from the factory as a blank decoder.  Well , even the  blank factory decoders actually have the factory (test?) sound project in them.  But that project is wiped out when a "real" sound project is uploaded to  the decoder.

It is a bit confusing (and different from other decoders. I hope I got it right.

[jagged ben]

Admittedly my experience was with a Lokpilot not a LokSound, so that arguably simplifies things.

If anybody can dig up anything on ESU 'overload protection' and whether it has a reset procedure I'd be curious.

[Dwight in Toronto]

Continuing on from the most recent ‘trouble’ event.  As before, I disconnected the motor leads, “boosted” the motor with a shot of good ‘ol DC, reconnected the wires, and the loco again ran fine on DCC.  I put it on LokProgrammer and reinstated BEMF. 

Knowing full well that this, of course, is in no way a fix, I just tried something else.

The problematic chassis is from an Atlas GP38.  I also have an Atlas GP40 - it’s virtually the identical chassis, which I bought new, and installed the same ESU 58751 decoder just 3 months ago, and which has never experienced any problems.  So, I just swapped decoders between the two.

The suspect, older chassis with the newer decoder immediately ran problem-free.

The newer chassis with the suspect, older decoder immediately emitted a high frequency oscillating sound as soon as the wheels touched the DCC trackage.  I captured this in the attached short, hastily-shot video clip:

/>

This is an entirely new phenomenon, which never once occurred in all the time that the older decoder was in the older chassis.  I don’t have the faintest clue what this could be, but it doesn’t sound good, and it probably doesn’t bode well!  It was hard to tell where the sound was coming from - I’m tempted to say the speaker, but I’m not sure.  The throttle was at speed step 0, and I had not yet entered the loco address.

So, beats me … why this would show up just because the decoder was moved to a different chassis is puzzling.  Before putting the loco on the tracks, I ensured that there was no continuity between the frame halves, and that the motor leads were also isolated.

There was a question about NCE PowerCab track voltage … fwiw, it was just over 16V AC.

[dem34]

Continuing on from the most recent ‘trouble’ event.  As before, I disconnected the motor leads, “boosted” the motor with a shot of good ‘ol DC, reconnected the wires, and the loco again ran fine on DCC.  I put it on LokProgrammer and reinstated BEMF. 

Knowing full well that this, of course, is in no way a fix, I just tried something else.

The problematic chassis is from an Atlas GP38.  I also have an Atlas GP40 - it’s virtually the identical chassis, which I bought new, and installed the same ESU 58751 decoder just 3 months ago, and which has never experienced any problems.  So, I just swapped decoders between the two.

The suspect, older chassis with the newer decoder immediately ran problem-free.

The newer chassis with the suspect, older decoder immediately emitted a high frequency oscillating sound as soon as the wheels touched the DCC trackage.  I captured this in the attached short, hastily-shot video clip:

/>

This is an entirely new phenomenon, which never once occurred in all the time that the older decoder was in the older chassis.  I don’t have the faintest clue what this could be, but it doesn’t sound good, and it probably doesn’t bode well!  It was hard to tell where the sound was coming from - I’m tempted to say the speaker, but I’m not sure.  The throttle was at speed step 0, and I had not yet entered the loco address.

So, beats me … why this would show up just because the decoder was moved to a different chassis is puzzling.  Before putting the loco on the tracks, I ensured that there was no continuity between the frame halves, and that the motor leads were also isolated.

There was a question about NCE PowerCab track voltage … fwiw, it was just over 16V AC.

Any worn traces on the board? Over the last N scale gathering I had a similar issue. And with me and @EL3632 looking carefully over the model I noticed the board had a wear mark where part of the frame must have been rubbing against the board in shipping and over time caused a small short.

[peteski]

If you disconnect the speaker you will determine if the strange sound emanates from it or somewhere else (the motor is probably the only other components which can emit audible noise). 

Looking at your writeup, seems that problem following the decoder, and it started making the funny noise as soon as you installed it in the previously reliably working chassis. Correct?

Is the scenario repeatable? I mean when you put the original decoders with the original chassis, does the noise stop? Then if you swap the decoders  again to the noise-generating setup, does the noise start up again.  Seems to me that something is shorting to something (sorry to be vague), or the decoder is malfunctioning.

I question that there is 16V DCC signal at the track.  Read the specs of the wall-wart you are using.  The track  voltage can be only as high as that voltage minus about 0.5V or more.  But your general-purpose multimeter might be providing false reading as DCC signal is not a typical sine-wave AC voltage.

[Dwight in Toronto]

dem34 - I carefully examined the board with the highest magnification lense on my visors, and I see nothing untoward.

Peter - since it was easier for me to de-solder the gray motor lead rather than the speaker, I first tried that.  And the sound did not reappear when the loco was placed on the tracks.  While on the tracks, I tried touching the loose gray lead to its motor terminal - nothing happened.  So I resoldered the gray lead, put it back on the tracks, and the sound reappeared.  I again desoldered it - the sound was absent, and touching the gray wire to the motor post again did nothing.  So I resoldered it once more, placed the loco on the tracks, and there was silence. 

I entered the loco address on the NCE … horn, lights, bell and prime mover fired up, and normal running was restored.

At this moment, both locos are running simultaneously on individual test ovals - the older board in the newer chassis, and the newer board in the older chassis.  I’ll just leave them like this, and will let ‘em run until such time as something acts up again.  Strange behaviour indeed.

My gut feel is that the older decoder has some sort of intermittent problem, and should probably just be replaced.  But I’ll continue to poop around with these a little longer and report back.

The power supply I’m using is the one that came with the NCE PowerCab - the specs on the wallwart label say output is 13.8V DC and 1.35A.  With probes on the tracks, my multimeter indicates 16.83V on the AC scale.  That’s all I know!

[peteski]

Man, you sure attract some odd decoder problems Dwight!   

Your Power Cab wall wart is identical to mine. Like I mentioned, the track voltage cannot be higher than what the wall wart supplies.  But it is also well known that multi-meters are designed to accurately show AC voltage of a sine-wave, not the funky DCC waveform. So their reading will be inaccurate.  Going by the wall wart rating, the track voltage is likely around 12V.

[Dwight in Toronto]

Lol - “odd decoder problems” … ain’t it the truth!

I simply cannot fathom why the lightest, shortest tap of a soldering tip, to a wire and component nowhere near the decoder itself, can possibly influence the behaviour of said decoder.  But that is what has transpired in this case, time after time.  So weird!

[Dwight in Toronto]

Both locos ran on their respective test loops, without incident, for a good 90 minutes straight yesterday (May 15).  This morning, I turned them around so they could accumulate some time, running forwards, in the opposite direction (just because!).

The loco with the suspect decoder immediately emitted that high pitched buzzing sound as the NCE was trying to boot up.  I say “trying”, because it twice shut down and rebooted as per short circuit detection.  As it was rebooting the 3rd time, I flicked the motor and all was good. 

Both locos are again doing test oval loops.  I’m going to order a new 58751 board.

[peteski]

Both locos ran on their respective test loops, without incident, for a good 90 minutes straight yesterday (May 15).  This morning, I turned them around so they could accumulate some time, running forwards, in the opposite direction (just because!).

The loco with the suspect decoder immediately emitted that high pitched buzzing sound as the NCE was trying to boot up.  I say “trying”, because it twice shut down and rebooted as per short circuit detection.  As it was rebooting the 3rd time, I flicked the motor and all was good. 

Both locos are again doing test oval loops.  I’m going to order a new 58751 board.

Sounds like that will be the solution. Any chance you could sent the decoder to ESU for repair/exchange?  It would be less expensive than a new decoder.  The problem could be proving to ESU that the decoder misbehaves intermittently, although you do have a video showing the problem (and this entire troubleshooting thread here you could point them to).