Serious control problem with Atlas Gold Series S2s - help needed

[mark dance]

Hi all...I am reluctant to post this but I am near the end of my rope here with a major ops session coming up in a couple of months.

I have invested very heavily in four Atlas Gold Series S2s.  2 of them run so poorly that the are unusable.  One was returned to Atlas 3 months ago and arrived back two days ago worse then when it left.  There was no indication from Atlas of what if any testing had been done nor whether any problem was found.

The problem demonstrated is that the speed levels on the locos oscillate up and down without any changes being made to the throttle setting.  This is after wheel cleaning and auto calibrate.  The returned unit will oscillate at full throttle setting to zero and back up again to full speed on a roughly 1 cycle per 2 second frequency.  I have had no reply (yet) from Atlas on what to do.

The other two of the four S2 units work well. 

I have also contacted ESU's American rep (Matt) but my experience with the Rapido GMD-1s, which are also ESU OEM equipped, doesn't leave me with confidence that he will be able to help.  He isn't paid to support their OEMs I don't think. 

Ideas?

md

[Scottl]

Is there a way to substitute in another decoder (even with the shell off) to see if you can constrain this to the decoder or mechanism?

[mark dance]

These locos are (for me) a nightmare to work on. This is largely why I sent one back to Atlas. That it was sent back worse than it left, and with no info on what was done, is very very poor in my opinion.

My guess is they tested it electromechanically and saw it was fine and sent it back. The highly periodic and repeatable ramp up to full speed and ramp down to stop with no change to the throttle setting makes me think this is a decoder issue.

Matt Herman of ESU has gotten back to me (nothing earth shattering) while still no reply from Atlas. Doesn't that tell you something...

There is an Atlas rep who monitors this forum isn't there?

Md

[davefoxx]

Mark,

I can't see how behavior like that could be anything but a decoder problem.  If you can't replace the decoder and attempt to fix it yourself, I'd send it in to Atlas again.  I know it took three months, but, since it's otherwise useless for you, what do you have to lose? 

DFF

[Van Horne]

I have three of these and they are all phenomenal performers so it isn't an endemic problem.  There must be some subtle flaw in the decoders that, I suspect, can only be corrected by replacement.  I'm surprised Atlas hasn't been more helpful.

David

[Van Horne]

BTW, my three are also in CP, so, solely as a charitable act to help out a fellow Vancouver modeller, I could bring them to the op session . . .

David

[jdcolombo]

Hi all...I am reluctant to post this but I am near the end of my rope here with a major ops session coming up in a couple of months.

I have invested very heavily in four Atlas Gold Series S2s.  2 of them run so poorly that the are unusable.  One was returned to Atlas 3 months ago and arrived back two days ago worse then when it left.  There was no indication from Atlas of what if any testing had been done nor whether any problem was found.

The problem demonstrated is that the speed levels on the locos oscillate up and down without any changes being made to the throttle setting.  This is after wheel cleaning and auto calibrate.  The returned unit will oscillate at full throttle setting to zero and back up again to full speed on a roughly 1 cycle per 2 second frequency.  I have had no reply (yet) from Atlas on what to do.

The other two of the four S2 units work well. 

I have also contacted ESU's American rep (Matt) but my experience with the Rapido GMD-1s, which are also ESU OEM equipped, doesn't leave me with confidence that he will be able to help.  He isn't paid to support their OEMs I don't think. 

Ideas?

md

Hi Mark.

Here's what I would do.  First, I would turn BEMF compensation off (set CV56 to zero) and see how the locos run.  If the bad behavior disappears, then you have a BEMF issue.  If the bad behavior remains, then there is either a mechanical issue - e.g., a binding wheel or something, or the decoder is bad.  But getting TWO decoders with similar bad behavior is unusual.  BEMF can make things worse if there is a mechanical issue, as it tries to compensate for drag or other problems in the drive train.  If the locos run lousy without BEMF, there's a problem that Atlas needs to fix.

If turning BEMF off seems to improve things, then instead of using auto-calibrate, try some manual adjustment of the BEMF parameters.

Here's what you have to work with:
CV51 and 52 affect low-speed performance only.
CV54 and 55 affect everything.

First, set CV53 (BEMF reference voltage) to your track voltage x10 (for example, if your track voltage is 13v, set it to 130; if you don't know, set it at 140).  Also, turn any momentum off (CV 3 and 4 to zero) and set CV2 to 1, CV5 to 255 and CV6 to zero to turn off any speed table adjustments.  Turn BEMF back on by setting CV56 to 255 (that's full BEMF compensation throughout the speed range).

Now set CV51 to 0, CV52 to 15, CV54 to 50 and CV55 to 100.  These are ESU's current factory defaults.

Crack the throttle open to speed step 1.  Using Ops Mode programming, set CV51=1 and see if the speed behavior improves.  Yes?  Then try 5, then 10, then 15, etc. until things get worse (the loco will noticeably oscillate at some point).  Write down the number where things got worse, and go back one step (e.g., if things started to get worse at 20, go back to 15).  If setting CV51=1 made things worse to begin with, set it back to zero (though I've NEVER experienced a situation where CV51=0 was better than CV51=1).

Now try the same thing with CV52.  First go lower (10, 5, 1) then higher (20, 25, etc.). 

Now speed the engine up to about 20% of full throttle.  Do the same iteration process with CV54 and CV55 (I would say that the liklihood is that you will need to reduce the factory default values in CV54 & 55, not increase them, so I'd start the iteration process by moving down, not up). 

This is all going to take a while, but it's what I did to get the smoothest running possible out of my S-2's.  BTW, here is where I ended up:
CV51 = 10
CV52 = 35
CV54 = 30
CV55 = 30

But every motor/drive train combo is a bit different, so don't expect the values I came up with to be the ones you end up with (if this DOES fix your problem, however, I'd be interested to know what values you DID end up with).

John C.

[GaryHinshaw]

These locos are (for me) a nightmare to work on. This is largely why I sent one back to Atlas.
...
There is an Atlas rep who monitors this forum isn't there?

What a drag.  I assume, from your comment, that swapping decoders between a well-functioning and poorly-functioning unit is difficult?  Were all 4 units programmed the same way, to your knowledge?  If so, CV settings are probably not an issue.

My theory is that there is a cracked trace on the ESU board that opens when it gets warm (e.g. when current flows), then closes back up again when current stops [and repeats].

Paul Graf (Atlas Paul) does participate here.  You could send him a PM.

-gfh

[jdcolombo]

What a drag.  I assume, from your comment, that swapping decoders between a well-functioning and poorly-functioning unit is difficult?  Were all 4 units programmed the same way, to your knowledge?  If so, CV settings are probably not an issue.

My theory is that there is a cracked trace on the ESU board that opens when it gets warm (e.g. when current flows), then closes back up again when current stops [and repeats].

Paul Graf (Atlas Paul) does participate here.  You could send him a PM.

-gfh

Gary -

The decoders are hard-wired in the S-2.  They can be swapped, but only with some difficulty and desoldering of wires.  Swapping decoders would be the best first step in troubleshooting if it weren't so difficult.

If this were just one unit, I might suspect a bad decoder.  But apparently there are two problem children, and I think the chances of having TWO decoders with exactly the same problem is very small (but possible, if other possible problems, like a mechanical problem or BEMF adjustment are ruled out).  I suspect manual BEMF parameter adjustment is needed.  Sometimes Auto-Calibrate works wonders; sometimes it doesn't.  I honestly never use it; if I have an issue with a locomotive that I know does not have a mechanical problem, I adjust BEMF parameters manually.  It always works, IF there is not a mechanical problem.

John C.

[davefoxx]

Pardon my ignorance, but what is "Auto-Calibrate"?

DFF

[jdcolombo]

By the way, Mark -

Have you made sure that these locos are not selected on two different throttles?  I've sometimes had weird behavior like this when the locomotive I was trying to run was inadvertantly also selected on another throttle (which you can do pretty easily with Digitrax UT throttles).

If you have a Digitrax system and haven't done a memory purge lately (or ever), I might try that first.  I've had very weird things happen at the command station level that a memory purge fixed.

John C.

[jdcolombo]

Pardon my ignorance, but what is "Auto-Calibrate"?

DFF

ESU has a procedure in which the decoder will automatically calibrate the BEMF parameters.  You do this by setting CV54=0, then hitting the F1 button.  The loco takes off like a banshee and then slows down and stops.  As it is doing this, the decoder is taking measurements of how the motor responds and setting the BEMF parameters accordingly.  (It's just a computer, after all - they are supposed to be able to do these kinds of things ).

It's wonderful when it works, but my experience has been that it only works really well about half the time.

John C.

[C855B]

Have you made sure that these locos are not selected on two different throttles?  I've sometimes had weird behavior like this when the locomotive I was trying to run was inadvertantly also selected on another throttle (which you can do pretty easily with Digitrax UT throttles).

Yes. I run into this all the time running JMRI/WiThrottle and forgetting I had the same number programmed on a UT. First occurrence took about an hour to arrive at the  , now it only takes about two seconds.

[peteski]

By the way, Mark -

Have you made sure that these locos are not selected on two different throttles?  I've sometimes had weird behavior like this when the locomotive I was trying to run was inadvertantly also selected on another throttle (which you can do pretty easily with Digitrax UT throttles).

If you have a Digitrax system and haven't done a memory purge lately (or ever), I might try that first.  I've had very weird things happen at the command station level that a memory purge fixed.

John C.

THat is an excellent advice.  If that was the case, it would explain why Atlas didn't find any problems.  Just in case something is still missed during troubleshooting I would temporarily set the misbehaving loco's address to an address you would  have unlikely used in the past (probably something like 1313, 9999, or some other address you don't ever recall using) and see how the loco behaves under that address.

[mark dance]

great ideas everyone.  I have to work the balance of the day but will try them this weekend.

By the way John, the digitrax "memory purge" is setting the command station op switch #39 to "closed", correct? 

I believe this wipes out all consisting as well doesn't it?  But it doesn't effect things like stationary controller addresses or fast clock settings does it?  Just want to know what I am getting into...

md