ESU Decoders With Coreless Motors

[Dwight in Toronto]

I’ve been working on equipping an N scale Dapol British tank engine (A1 Class Stroudly Terrier) with DCC & sound. 

It is diminutive, to say the least, and my approach is to put the speaker and decoder into an equally small sound car, tethered behind the locomotive.  The tether is detachable by way of a 6-pin plug & socket located in the bottom of the sound car.  The sound car is an old Graham-Farish open-top goods wagon that I’ve had since the late ‘70’s. 



At this point, the model is alive and functional, and I’ve had fun compiling the steam, whistle, bell and other sound effects from various sources. 

But I’ve run into a problem. 

There’s a supplier in the UK called Tramfabriek, who specialize in compact, coreless replacement motors (I believe that a few Railwire members may have retrofitted these to small switchers and such).  I ordered and installed a Tramfabriek motor & worm gear specifically intended for the Terrier.

Unfortunately, it turns out that ESU decoders do not work with coreless motors.  Even at speed step 1 or two, the loco takes off like a rocket, and stops as if it ran into a brick wall.  There is no slow speed operation whatsoever.  According to Sven van der Hart at Tramfabriek, this incompatibility is well known, and he forcefully steers customers to either Zimo or Doehler & Haass, both of which reportedly deliver stellar performance with his coreless motors. 

For my part, I standardized on ESU several years ago, and I’m far too familiar, comfortable and invested in the LokSound product line to try an alternative.  So, I suspect that I will be re-installing the original Dapol motor.  But before I do, I was curious to hear if others have run into this problem, or were aware of it, or know if there might be some sort of work-around? 

[greenwizard88]

Nice model!

I'm pretty sure that the HO scale Kato F40 uses dual coreless motors in the trucks, and it was offered with an ESU decoder at one point. So there's probably a way to make it work.

My google-fu brought up these two pages, which seem relevant. The first is this, with some CVs that you could try to adjust.
https://forum.mrhmag.com/post/esu-lokpilot-v4-0-and-coreless-motor-12210606

The other is the ESU decoder tester, which uses a coreless motor. They go out of their way to call it a high quality motor, the implications being that yours might not be  but it at least shows that there's potential.
https://www.esu.eu/en/products/decoder-tester/

[tehachapifan]

I have two N scale locos that have Bachmann coreless motors in them and they run like Swiss watches with ESU decoders installed (I probably had to do some CV adjustments). Not sure how different the Tramfabriek are from the Bachmann ones, but they look near identical (someone recently explained some differences though).

[Jim Starbuck]

ESU decoders handle coreless motors very nicely when configured to do so. I have a couple dozen engines and scratch built projects using coreless (many from Tramfabriek) and all using ESU decoders.
I believe what you are experiencing is caused by the BEMF settings on the decoder. The decoder also has presets for specific motor types.
Do you have access to a Lokprogrammer?

[Dwight in Toronto]

Encouraging feedback guys!

And Jim, yes, I spend a lot of time on my LokProgrammer.

Any and all further insight and suggestions are welcomed. 

FWIW, I’m presently fiddling with file SO143, but only the sound aspects - I have not done anything with motor control, and have not attempted auto-tune.  I simply tried running the loco using the file as downloaded.  The dismal operation was the worst I have seen compared to my other installs, so I concluded that Tramfabriek’s Sven van der Hart’s earlier comments had borne out - “don’t use ESU with coreless motors, they can’t deliver slow speeds and have a high starting speed, none of which can be controlled - you will see”. 

Interestingly, here are Tramfabriek’s website tips for using ESU decoders with their coreless motors (hope to try them soon):

So far, I haven’t been able to get ESU decoders run as smooth as all the other mentioned on this page, mainly at the lowest speed steps. I would advise to choose one of the other brands on this page.

ESU settings for Tramfabriek coreless motors:

CV  02 =   2 (Start voltage)

CV  51 =   6 (‘K Slow” Cutoff)

CV  52 =   0 (BEMF Param. ‘K Slow’)

CV  53 = 120 (Control Reference voltage)

CV  54 =  10 (Load control Parameter ‘K’)

CV  55 =   6 (Load control Parameter ‘I’)

CV  56 = 255 (BEMF Influence at VMin)

CV 116 =  25 (Slow speed BEMF Sampling period)

CV 117 =  50 (Full speed BEMF Sampling period)

CV 118 =   5 (Slow speed BEMF Measurement gap length VMin)

CV 119 =   5 (Full speed BEMF Measurement gap length Vmax)

 (With the kind cooperation of ESU)

Updated 7 SEP 2022

[Jim Starbuck]

Dwight,

It’s good to see those parameters listed as specific CV values. I haven’t programmed anything using just CV in so long I forgot how.
In the motor settings page in Lokprogrammer toward the bottom is a text line for settings using selected motor types. At the bottom of that list is coreless motors with small diameter. Writing that will get you in the ballpark then go from there and compare that to the values you posted above.

Jim

[peteski]

Dwight,
One thing I don't see you mention is whether you tried the ESU's famous auto-tune procedure to adjust the motor parameters. That would have been the first  thing I would try.  That should be the first thing done with any ESU install.

I'm not sure if you have the ESU decoder tester, but it actually has a coreless motor on it, and the decoder controls its slow speed just fine.  You can see it spinning slowly on speed step 1.

If the auto-tune did not improve the motor control then as others have  mentioned, there are manual CV settings you can try.

But having said that, my friend's N scale Kato FEF3 has a Kato's coreless motor and we still have not made it run smoothly (like ESU decoder should be).  Slow speed is ok, but at normal speed it surges slightly.  We just haven't spent a lot of time on tuning it.  There are many variants of coreless motors where the number of windings and their inductance varies, so there is no single set of CV setting that will work with all the coreless motors.

[Dwight in Toronto]

Pete - thanks for chiming in.  I mentioned above that I have not tried auto-tune.  The reason is that I found posts on the ESU io.group where auto-tune attempts on coreless motors either didn’t even make the loco move, or, made the performance even worse.  I don’t have ESU’s tester (I have a home-made one).

I hope to fire up LokProgrammer later today, and will report back.

[peteski]

Pete - thanks for chiming in.  I mentioned above that I have not tried auto-tune.  The reason is that I found posts on the ESU io.group where auto-tune attempts on coreless motors either didn’t even make the loco move, or, made the performance even worse.  I don’t have ESU’s tester (I have a home-made one).

I hope to fire up LokProgrammer later today, and will report back.

I missed that. - sorry.  Since you have the original sound project saved (I assume), even if auto-tune messes things up, you can restore all the original CV settings back by writing decoder data back to the decoder.

[tehachapifan]

I have a number of N scale locos with my own ESU installs and, while I love the decoders, I have been less than impressed with the auto-tune feature.

[peteski]

I have a number of N scale locos with my own ESU installs and, while I love the decoders, I have been less than impressed with the auto-tune feature.

As I see it, doesn't hurt to try it first, then revert to the saved CV settings, or start modifying the setting auto-tune configured, or manually set them to the settings shown in the decoder manual.

[Dwight in Toronto]

Dwight,

In the motor settings page in Lokprogrammer toward the bottom is a text line for settings using selected motor types. At the bottom of that list is coreless motors with small diameter. Writing that will get you in the ballpark then go from there and compare that to the values you posted above.

Jim

I did exactly what Jim suggested, and the improvement was not only immediate, but several orders of magnitude better than the deplorable performance that accompanied the “as downloaded” project file. 

As Jim surmised, the loco was now definitely in the ballpark.  Thankfully, the uncontrollable runaway velocity at speed step 1 was gone, replaced by more typical, controllable bottom-end acceleration and deceleration.  Not perfect, mind you … speed steps 1 through 4 are overly herky-jerky, jiggling along in a manner unlike anything I’ve ever seen before (kinda like those goofy, felt-bottomed football players that you placed on a vibrating playing field back in the 70’s!).  Luckily, things smooth out nicely from step 5 and up. 

I then tried Tramfabriek’s numbers.  It seemed to me that steps 1 through 4 may have been even slightly slower, but the high frequency jerky, ‘stop-motion’ effect remained.  I then tried mid-point values between the two baselines, and if there was any improvement, it was too subtle for me to discern. 

I’m not savvy enough to know which motor control CV’s do what, so I’m at a loss on how to smooth out and fine-tune those crawling speeds.  But holy smokes, I sure feel a lot better today than I did at this point yesterday!

Any chance that those who have achieved satisfactory performance with their LS5-controlled coreless motors could list their motor control CV values, as per the Tramfabriek format above?   Yeah, I know it’s a lot to ask.  But it would sure be interesting, revealing, and helpful. 

[Jim Starbuck]

Dwight,
Here are the values I have loaded for my Model 40 switcher with a 7x16 Tramfabriek motor and a Lokpilot v5 DCC Micro.
What decoder are you trying to set up as I may have the same one in another engine. The LokSound and LokPilot act slightly differently.
Most of that low speed jittery can be worked out with CV51&52. Usually lower values help. I move one at a time by one number value then test and save it each time I get a bit better performance.
I also usually pull the top speed way down around 100-125 then split the mid setting between the low and high.

CV02=5
CV51=4
CV52=2
CV53=120
CV54=15
CV55=15
CV56=230
CV116=75
CV117=150
CV118=15
CV119=20

[Jim Starbuck]

He’s another one with a LokSound Nano and Tramfabriek 7x16.

02=2
51=20
52=5
53=110
54=25
55=30
56=255 (sometimes helps to lower that~225)
116=50
117=150
118=15
119=20

As many of these as I’ve done, I still couldn’t tell you what each specific parameter actually does. It’s something I need to learn though. My method is mostly trial and error. I won’t lie, I’ve spent hours fiddling with settings to get the best performance while other times it seems to just fall into place. It sometimes helps too when it’s getting frustrating to set it aside and come back to the next day. Ask me how I learned this.
When I get one acting pretty good but not quite there, I’ll save that configuration as a different file name like version 1 so I can always at least go back to where it was doing well at one point then keep making changes on the original and saving subsequent improvements as I go.
One thing that is consistent is that no two are exactly the same but using the motor type presets is usually a good place to start.
I’ve never tried to use auto tune.

Jim

[Dwight in Toronto]

Jim - thank you so much for the templates … great stuff!

Neither delivered perfect performance, but as you have found, a lengthy session of LokProgrammer trial & error paid off.  Your switcher numbers were a pretty good starting point, and I progressively swapped in cv values here & there that other folks had suggested, and then tested single digit increments ad infinitum.   

Eventually I zeroed in on a suite of numbers that I think may have culminated in the epitome of what’s possible for slow speed running with my Dapol Terrier:

cv’s 2, 5 & 6 = 10, 58 & 125
cv 51 = 0
cv 52 = 0
cv 53 = 110
cv 54 = 20
cv 55 = 15
cv 56 = 255
cv 116 = 30
cv 117 = 50
cv 118 = 40
cv 119 = 40

For cv’s 118 & 119 (“Slow speed length of measurement gap” and “Full speed length of measurement gap”, respectively), I tried values ranging from 15, 20, 25, 30, and 40, none of which seemed to make any discernible difference.

After several hours of fine tuning, I managed to eliminate pretty much all of the high frequency jittery dithering that was detracting from speed steps 1 through 4.  Those crawling speeds are now “good enough” for me.

I still need to fine tune the chuff rates, after which I’ll try to post a short video of the end result.

Quite the interesting exercise, and a challenge to one’s perseverance.