Question regarding Speed Matching and BEMF

[shark_jj]

I have been busy for over a month speed matching the entire fleet of locomotives on the Grand Trunk Southern.  It has gone quite well and overall I am satisfied.  My fleet is primarily Kato locomotives with Digitrax decoders.  I am using JMRI to do the programming and setting up Speed Tables for each decoder.  In the initial steps I change the factory setting of 6 in CV57 (Back EMF) to 0 and then I proceed to install the speed tables.  I read somewhere that this was the appropriate course of action.  As I moved forward I never changed the BEMF settings back to 6. 

Part of the reason for doing the speed matching was to restrict the top end speed of consists to a prototypical scale 50 mph since I have some operators who like to operate a little on the fast side.  With the BEMF off it means that a train proceeding upgrade may require the operator to give it a little more power, and a train going downhill may need power reduced so as not to run away.  To make a long story short I noticed that my operators were doing well adding power but weren't reducing it on the downhill grades and 50 scale mph downhill just didn't look right to my eye.  One of the ideas I came up with was to add the BEMF back in so that a train entering the grade at 30 mph should maintain its speed at 30 mph.  That's my understanding of one of the benefits of BEMF.

I reset CV57 in two Kato SD40's to 6.  I turned the throttle to Speed Step 1,  before the change these locos barely crawled at Speed Step 1, now they took off at an estimated 20 scale MPH.  I reset CV57 again, this time back to 0, turned the throttle to Speed Step 1 and they barely crawled.    Continuing the test, I reset CV57 to 6.  Coupled the consist to a 20 car train.  Turned the throttle to Speed Step 1.  The train moved off at 20 scale MPH and went around the entire layout, up a 1.8% grade, down a 1.8% grade and just maintained a steady speed. 

I've done a little searching around the Internet to try and understand how BEMF effects speed tables and whether I did the right thing by doing the speed matching with BEMF off.  Haven't found much that is helpful or informative.  My hope is that someone here who knows a lot more about electronics and decoders than yours truly can educate me a little bit and perhaps offer some guidance moving forward.

[peteski]

Friend of mine uses many Digitrax decoders and his experience with BEMF is identical to what you observed.  We tried to mess around with the BEMF CV settings but got nowhere.  He just keeps his disabled and like you has good slow speed control.

BTW the BEMF behavior has nothing to do with the 28-speed table. My friend only uses CV2, 5, and 6 for adjusting the max speed and as I mentioned, enabling BEMF takes away slow speed at step 1.

Maybe someone here has a solution, but if not, there is a forum dedicated to Digitrax ( https://groups.io/g/Digitrax-Users ).  Might be worth joining and asking there.

[Mike C]

  I use Soundtraxx decoders in all my locos . They have a CV that adjusts the starting voltage of the loco . Maybe Digitrax has a similar CV ?  If so try reducing the start voltage .       Mike

[peteski]

  I use Soundtraxx decoders in all my locos . They have a CV that adjusts the starting voltage of the loco . Maybe Digitrax has a similar CV ?  If so try reducing the start voltage .       Mike

If you're thinking about CV2, that is a NRMA standard defined CV that all the decoders have (at least all the decoders I have ever worked with).

[Maletrain]

I have also read that speed matching should be done with BEMF turned off.  But that doesn't make sense to me. 

Isn't BEMF designed to make the locos run at a constant speed, so that it would tend to smooth out differences in things like warm-up, etc.?  That can be really important for what the lowest reliable crawl speed step setting can be.

I always wonder if that "guidance" was confusion with turning off momentum before speed matching.  That makes sense to me.  And, once speed is matched, then the next step is to turn on momentum and match that, too.  Otherwise, one loco ends up dragging of pushing the other for a while when speeds are changed.

However, I have read that some decoders automatically turn off BEMF when consisted.  But, that would have to be NMRA consisting with CV19, or the decoder would have not way to tell that it was in a consist.

I did not find the place where I think I read about BEMF being turned off automatically by consisting, so if anybody has a reference to that, please post it.

[peteski]

However, I have read that some decoders automatically turn off BEMF when consisted.  But, that would have to be NMRA consisting with CV19, or the decoder would have not way to tell that it was in a consist.

I did not find the place where I think I read about BEMF being turned off automatically by consisting, so if anybody has a reference to that, please post it.

Well, that is somewhat accurate.  Factory default is to disable BEMF when advanced consist is active, but you can change it.
Here is an excerpt from https://www.digitrax.com/tsd/KB926/cv55-cv56-cv57-scaleable-speed-stabilization-back-/

CV57 controls the amount of droop or speed loss as load is increased, by limiting the amount of change or compensation that the decoder is allowed to implement.  CV57 controls the droop separately for both regular addresses and consist addresses.  The first hex digit controls the amount of droop in effect when speed is controlled using the standard decoder 2 or 4 digit address.  The second hex digit of the CV value controls the droop in effect when the decoder is part of an advanced consist for speed and direction control. Values for each digit can range from 000 to 015 decimal or x00-x0F hex.

So CV57 is split into two 4-byte "nibbles".  Half of CV57 is for when the model is operated as a standalone unit, and the other half for when in advanced consist.  Since the factory default is 06, the advanced consist half of the CV is zero.

Also see https://sites.google.com/site/markgurries/dcc-welcome-page/advanced-topics/consisting-information/bemf-consisting/digitrax-bemf-advance-consisting

[shark_jj]

Thanks everyone this was helpful.  Thanks Peteski for the link to Mark Gurries page and the excerpt from the Digitirax website.  The insights from both of those and your friends similar experience leads me to believe I am better served by leaving CV57 set to 0, i.e. BEMF off.  Your excerpt also clarified for me what I had been seeing, a drop off in top end speed under load.  By setting a max speed of 50 scale MPH, once the consist was under load I wasn't achieving that, there was a significant drop off.  Just an estimate but I would say top speed was in the 30 to 35 MPH range, except for the downhill where as previously stated it would pick up speed and reach the 50 scale MPH.  It appears that if I am going to leave BEMF off I may have to increase the top range of the speeds I am setting to 60 - 65 scale mph.    It appears that the answer to my initial problem is better training of my operators.   

[kiwi_bnsf]

Good question @shark_jj !

I will offer a bit of a counterpoint. I have spent a lot of time speed matching Kato, Atlas, Intermountain, and Fox Valley locos equipped with Digitrax decoders over the years — they were my standard until ESU V5 arrived on the scene.

I see quite a lot of misinformation/voodoo on various forums regarding speed matching. I can say with 100% certainty that it is possible to speed match and consist Kato locos with Digitrax decoders with BEMF enabled. It is also possible (and in fact easier) to get excellent slow speed performance with BEMF enabled. Katos can definitely achieve one-tie-per-minute creep on speed step 1 with BEMF enabled.

I often read that "locos with decoder BEMF enabled will fight each other in a consist". This simply isn't true if you do a half decent job of speed matching. I regularly run 5-6 loco consists with mixed brands and decoders with long trains on steep grades with zero issues. I override Digitrax decoders to always have BEMF enabled when consisted.

If you want to consist Digitrax-equipped locos with other brands (like ESU which is now the factory standard for most Atlas, Scale Trains etc), then BEMF will really be necessary to get consistent speed matching.

Without BEMF enabled, you are pretty much at the mercy of the loco's intrinsic motor qualities, and there are limits to what you can achieve with a speed table alone. That's fine if you have an all-Kato roster, but if you a diverse roster like me, then BEMF is your friend.

In summary: speed match without BEMF if you are going to run without BEMF; Speed match with BEMF enabled if you are going to run with it enabled.


Here are some example CV values from my reference Kato SD40-2 Mid Production equipped with a Digitrax DN163K1C decoder:

Kick start [CV65]  =  2
Motor Drive Frequency  [CV9]  =  0
Static compensation for speed stabilization    [CV55]  =  128
Dynamic compensation for speed stabilization    [CV55 correction: CV56]  =  48
Solo operation droop compensation for speed stabilization    [CV56 correction: CV57 bits 0-3]  =  7
Advanced consist droop compensation for speed stabilization    [CV57 correction: CV57 bits 4-7]  =  7
Speed Compensation Control  [CV61 bit 4] = 0  (F5 has no effect on speed compensation)

I use a near-linear speed table that starts with CV76 = 9, and ends with CV94 = 76


I have a slightly different template for my older mechanism Katos (SD45/SD40-2 Early/SD40 etc):

Kick start [CV65]  =  1
Motor Drive Frequency  [CV9]  =  0
Static compensation for speed stabilization    [CV55]  =  96
Dynamic compensation for speed stabilization    [CV55 correction: CV56]  =  48
Solo operation droop compensation for speed stabilization    [CV56 correction: CV57 bits 0-3]  =  6
Advanced consist droop compensation for speed stabilization    [CV57 correction: CV57 bits 4-7]   =  6
Speed Compensation Control  [CV61 bit 4] = 0  (F5 has no effect on speed compensation)

Speed table that starts with CV76 = 1, and ends with CV94 = 82


Anyway, hopefully the above allows you to quickly test with BEMF enabled to see if you like it.

Hit me up if you want some defaults for BEMF with older pre-ESU Atlas locos.

Cheers

[peteski]

Tim,
According to Digitrax technical info in https://www.digitrax.com/tsd/KB926/cv55-cv56-cv57-scaleable-speed-stabilization-back-/

CV56 controls the dynamic compensation or how much the decoder considers the historical difference between the current speed and the target speed when determining the next speed command to send to the motor. This setting is like a damper or shock absorber on the spring that helps to restore the spring to its new position.  Higher values cause more rapid adaptation to the target speed and lower values cause slower adaptation to the target speed.  CV56 will have no effect on decoder operation until you program CV57 as described below.  Excessively high CV values programmed to CV56 will tend to let the locomotive to "hunt" around a new desired speed when a change of speed is commanded.  We recommend that you use the lowest CV value in CV56 that gives the desired performance.

CV56 is not for "solo" operation - it is for BEMF all modes (solo and advanced consist).  Then the CV57 comes into play as indicated below:

CV57 controls the amount of droop or speed loss as load is increased, by limiting the amount of change or compensation that the decoder is allowed to implement.  CV57 controls the droop separately for both regular addresses and consist addresses.  The first hex digit controls the amount of droop in effect when speed is controlled using the standard decoder 2 or 4 digit address.  The second hex digit of the CV value controls the droop in effect when the decoder is part of an advanced consist for speed and direction control. Values for each digit can range from 000 to 015 decimal or x00-x0F hex.

If either digit is 0, speed stabilization is OFF.  A value of 015 or x0F hex is speed stabilization FULL ON.  If the droop CV value is too high, you may see locos jump from one speed to the next if they encounter an obstacle or problem with track work. If the value is too low, there will be very little speed stabilization effect at all. A higher digit makes the droop or speed fall-off less.  A typical value for most locomotives not part of an advanced consist is CV57=005 (same in decimal & hex), but the actual value that is best for a locomotive and train size must be determined by observation and experimentation.  Note that this value of 005 for this example means that no speed compensation is used when this decoder is in an Advanced Consist.


In both of your examples CV57 is either 6 or 7, which means it is only utilized in solo operation and disabled (the upper 4 bits of CV57 are zero).
You don't  realize that your models have BEMF off in advanced consist.

[kiwi_bnsf]

Hi Peteski,

The labels for those CVs come from JMRI Decoder Pro (not me).

I realise that I made a typo when copying from Decoder Pro (which in my defense makes the raw CVs quite hard to correlate)…

Kato SD40-2 Mid:

Kick start [CV65]  =  2
Motor Drive Frequency  [CV9]  =  0
Static compensation for speed stabilization    [CV55]  =  128
Dynamic compensation for speed stabilization    [CV56]  =  48
Solo operation droop compensation for speed stabilization    [CV57 bits 0-3]  =  7
Advanced consist droop compensation for speed stabilization    [CV57 bits 4-7]  =  7
Speed Compensation Control  [CV61] = 0  (F5 has no effect on speed compensation)

I use a near-linear speed table that starts with CV76 = 9, and ends with CV94 = 76

Kato SD45/SD40-2 Early/SD40 etc:

Kick start [CV65]  =  1
Motor Drive Frequency  [CV9]  =  0
Static compensation for speed stabilization    [CV55]  =  96
Dynamic compensation for speed stabilization    [CV56]  =  48
Solo operation droop compensation for speed stabilization    [CV57 bits 0-3]  =  6
Advanced consist droop compensation for speed stabilization    [CV57 bits 4-7]  =  6
Speed Compensation Control  [CV61] = 0  (F5 has no effect on speed compensation)

Speed table that starts with CV76 = 1, and ends with CV94 = 82

Apologies for any confusion!

[kiwi_bnsf]

You don't  realize that your models have BEMF off in advanced consist.

If I had BEMF disabled in consist I would definitely notice — locos would be dragged along resulting in prototypical wheel flats.

[shark_jj]

Thanks Tim and Peteski for sharing your insights, I am sure I am not the only person on the forum who is finding this information educational and helpful, though it does raise several questions in my mind particularly with regards to CV55, CV56, and CV57. 

First simple question.  I am using JMRI.  If you look at the screen shot below is Forward Trim and Reverse Trim as shown on this screen, CV55 and CV56?  As a second part of this question, can I actually set a value in these fields if I have set CV57 to 0.   This screenshot also shows the Speed Table for a Kato SD45.  You will note that it reaches top speed at Speed Step 14.  This is an effort to have the Percentage of Throttle number on the Throttle screen reflect the scale speed of the loco. i.e. if the Throttle screen reads 50, then the loco is actually doing 50 scale mph.  If the throttle reads 25, then the loco is doing 25 scale mph. 



The more complex part to this question is how actually do these two CV's impact your speed settings.  I have tried to look this up on Digitrax and by looking at You Tube Videos on Speed Matching and no one has explained it.   I was of the understanding that I used these if for example, I had two loco's consisted (I use simple consisting, both loco's with same decoder number) and they ran smoothly in one direction but were not speed matched in the other direction.  I could then use the Forward or Reverse Trim to adjust the direction that was out of Match to bring them closer.  Haven't used it a lot, so I have little experience to draw on, but that was my understanding. 

Next question Tim is with respect to CV65 Kick Start.  Assuming that I have a loco which starts to creep at Speed Step 1, what does Kick Start, CV65, contribute to the running of the loco.  My understanding of Kick Start, and I am showing my lack of knowledge of electronics here, so forgive me if I get terminology wrong, is that it give a jolt of current to the motor to get it spinning and moving.  If I am already crawling at Speed Step 1, why is that necessary?  I would say that a number of my Kato SD40's have Step 1 in the Speed Table set at 40 or 50 to get them to crawl.  I notice that all of your Step 1 settings are in single digits, does the use of CV65 allow you to reduce the number needed to get the crawl at Speed Step 1. 

You are showing CV57 as Advanced Droop Compensation.  Is this another term for BEMF?  Am I correct in my assumption that setting CV57 to 0 in fact turns BEMF off?

Lastly Tim, you mentioned setting for older Atlas locomotives.   I was going to send you a PM and my email address, which I can still do if you wish, but thought it would be better if you could just post those settings so that others who might be interested also get a chance to see them and possibly use them as a template.  I have had issues with speed matching with the older Lenz decoders that came with many of the early 2000 Atlas.  The combination of their slow speed motors and Lenz decoders, I can barely get them to run at around 25 to 30 scale MPH.  I use a bunch on some coal drags, otherwise its Kato all the way.  I have done factory resets and fiddled around with them a bunch with little progress.  I would be interested to see your settings.

Anyway, Tim and Peteski, thanks to both of you for sharing your knowledge.

[peteski]

If I had BEMF disabled in consist I would definitely notice — locos would be dragged along resulting in prototypical wheel flats.

LOL. It would take very long for the N scale locos to develop wheel flats (especially since they would not be totally stalled - the wheels would still be rotating), but the consist would run like crap!   

Ok then, how do you explain the description of the CV57 functionality in the Digitrax documentation?  You do understand what the documentation states, correct?

When CV57 is 6 (decimal) the hex value is 0x06  or 0110 0000 in binary.  As you can see the upper 4 bits are 0000.
To have value of of 6 for both solo and advanced consist CV57 has to be set to 0x66 or 0110 0110 in binary (that would be 102 in decimal).  I have no stake in this, but if the Digitrax manuals and KB articles are this wrong, someone should tell Digital about it.

I'm assuming here that you are using advanced consists (CV19 > 0), otherwise in other types of consist the decoder is still in solo mode.

[peteski]

John, as I mentioned, my experience with Digitrax BEMF (through messing around with it with a freind of mine) was not good and we turned it off (CV57=0).

The KB article I linked to earlier explains that CV55, 56, and 57 are all specifically dealing with BEMF.

The speed tables have nothing to do with BEMF, but since active BEMF affects the locos behavior (and speed), it will indirectly shift the speed speed range of any speed tables you have configured (whehther it is the 3-step por 28-sstep table).

[shark_jj]

Peteski, I read the KB article when you first mentioned it, and since you mentioned it again, I read it again thinking I must have missed something.  Truth be told, I don't really understand what Digitrax is trying to tell me in that article.  Not an uncommon situation since I don't find their manuals to be particularly user friendly.  They seem to be written on the belief that everyone has a high level of understanding of electronics and in particular as it applies to DCC.  What I do understand is that BEMF doesn't change the speed settings, but as I observed at the start of this article, it can clearly impact their implementation.  I also understand Tim's point that I can overcome what I observed by speed matching with BEMF on or conversely, I can do as you and I have already done and just leave BEMF shut off. 

Since Tim mentioned JMRI and CV's 55 and 56, I just wanted to determine if they were in fact the Trim functions shown on the screenshot I posted, since I don't really know for a fact what they do.  I certainly could be dead wrong and those Trim functions and CV's 55 and 56 could be totally separate issues.  I just don't know and was looking for some insight.  Similarly I have a general understanding of Kick Start, CV65, but don't really understand why or when I should be using it.  I do belong to the JMRI user group and could post my question there, but honestly, I've found by asking here at the Railwire DCC Forum that yourself and others have such a wide breadth of knowledge that I can usually get an answer.