Question re: Understanding Consisting

[shark_jj]

My locomotives, regardless of the number of engines in the lash up, rarely get broken apart during normal operations.  As a result, rather than consisting them, I just program all of the locomotives in the lash up with the number of the lead locomotive.  I have recently been fooling around with speed matching with some mixed results and I have begun to wonder if the fact that I have more than one locomotive with the same number, rather than consisting, is causing part of my problems.  I also realized that I don't really understand how consisting and the packaging of information to the decoder works.   For examples if Locomotive A (#1000) has a set speed table, and Locomotive B (also #1000) also has a set speed table which is different, is this information contained in the decoder therefore when the command station sends the signal for speed step 10 for example the two decoders will interpret it differently based on their individually programmed speed tables and react accordingly, or will the fact that I have programmed the decoders with the same loco number cause confusion. 

Programming all of the decoders in a loco lashup with the number of the lead loco seemed initially to be a simple approach to consisting but am I creating other problems that I am not aware of?  Advice from someone who understand DCC better than I would be appreciated.  I use Digitrax with a DCS100 Command Station, with a DB150 slaved to it, and JMRI for programming.

[CNR5529]

I have been using both types of consisting for a while, using the same address for multiple locos that are always together (A-B units for example) and using the command station consisting functionality for consists that get broken up, with satisfactory results. In both cases though, I had to make sure that the locomotives were speed-matched before running them together. Using the same address for multiple locos should not cause problems, as each decoder will still revert to its own speed table (or Vmin, Vmid and Vmax CVs) when given a speed step command.

The key is to program each locomotive individually (especially when multiple locomotives have the same address) so that you don't write one speed table on another locomotive accidentally. Just because they have the same address doesn't mean they would have the same speed table. Next, only use one locomotive as your "standard" to calibrate all others against, otherwise you will accumulate error in the matched speed performance. I set all my locos to default settings and disabled momentum, acceleration and/or coasting, found the one that was quickest at speed step 1, and that became the golden loco (you cant program it to go any slower). All others get matched first at speed step 1, then at the top of the desired speed range. Minor tweaking may be needed to match the speed curve, but I typically use a linear curve, so this is not a problem for me. Then go back and add momentum if desired (this will also require some matching). I use 3 different top speeds, requiring 3 golden locos: Mainline road power for passenger/freight which is basically = used on the Ntrak layout and has the highest top speed, Branchline road power with a slower more realistic top speed of 40 mph, and switchers which are capped out at 25 mph.

Hope this helps!

[shark_jj]

Thanks for the info Frederick.  I agree about speed matching locomotives separately and do that, I was more concerned about whether or not I was missing something using the simple consisting method I am using.  It doesn't sound like I am, so thanks for the confirmation.

[djconway]

Another thing to remember is to turn off back EMF if all the locos have the same address.

[shark_jj]

Thankyou djconway I wasn't aware of that.

[MK]

Speed table info and things like (Digitrax only?) V(zero), V(mid) and V(high) are all programmed into the respective decoder and thus loco.

If you have all locos set to the same address without tweaking each one's speed, then (most likely) you will have a lot of mismatching going on, each one fighting the other.  The same will happen if you consist them.  But once you set the speeds, whether you use consisting or program each loco the same address, then they will play nice with each other.

[CNR5529]

Another thing to remember is to turn off back EMF if all the locos have the same address.

I agree with this, but would add that it applies to all forms of consisting, not just when using the same address to consist locomotives. It is typically (but not always) a problem at higher speeds, where the decoder will try harder to maintain speed. More sophisticated decoders have all sorts of CVs to modify or automatically disable BEMF at given speed ranges. This has the benefit of enabling low speed bemf (where it is needed most) and preventing higher speed "in-fighting" between locos.

[jdcolombo]

The notion that you have to turn BEMF off in a consist is an urban myth.  I've never, ever done that on my ESU LokSound and LokPilot-equipped locomotives.  However, you DO need to have the locomotives speed-matched relatively closely to avoid the "oscillation" that sometimes happens when poorly-matched locomotives are run with BEMF enabled.  Most modern decoders do have a CV that you can program which will turn BEMF off after a certain throttle setting is reached, but I've never needed to use that.

The key to understanding consisting is to first understand that all locomotive operating parameters are stored locally in each individual locomotive's decoder.  Thus the settings for CV2 (start voltage), CV5 (high voltage limit) and CV6 (mid voltage), or the settings for a 28-step speed table, are all "local."  Ditto for things like what function key turns on the headlight, rear light, and so forth.  That means that the first step before consisting two locomotives is to set them up so that they match as closely as possible in terms of speed.  The way I do this is pretty simple.  First, I take the two locomotives and determine which one runs SLOWEST at maximum throttle.  You want to match the faster loco to the slower one, not the other way around (because you can't speed up the slower one, but you can slow down the faster one).  Next, I set up the "speed curve" for the slower locomotive - in my case, I don't want any diesel to run faster than 60 smph on my layout, and I use a linear speed curve.  So I put the slower loco on the track, and adjust CV5 (max voltage) until the engine runs at 60 smph at maximum throttle.  Then I set CV6 at exactly half of CV5 - if CV5 is 120, then CV6 is 60.  Then I set CV2 so that the engine just barely creeps at speed step 1; with modern diesels, this usually is a setting of 1-3.

Then I program the faster loco with the CV values I've set for the slower loco.  I put both of them on the track, with the faster one in front, separated by a couple of inches.  I then consist them (NOT USING THE SAME ADDRESS!).  I turn the throttle to maximum, and then use Ops Mode programming (programming on the main) to adjust the faster loco's CV5 until it matches the slower loco.  Then I cut the throttle to half, and do the same for CV6.  Finally, I adjust CV2 so that the faster loco creeps at the same speed as the slower one at speed step 1.

Now you're done speed matching.  If you use momentum, then you will also need to match the locos in how they respond to momentum.  Again, I use the slower loco as the "model" and set the momentum the way I want it for that loco (CV3 and CV4).  Then I match the faster loco to the slower one (I do this by putting both on the track, separated by about a foot, turn the throttle to max, then quickly turn it to zero after the locos have hit max speed, and observe the stopping behavior; I use ops mode programming to adjust CV4 in the faster loco until they match, and then use the same value for CV3).


At this point, you can safely consist your locos and you shouldn't have any BEMF oscillation problems.  You can use any of the three different methods of consisting.  The first is to set the locos to the same address to use what I call "hard consisting."  I use hard consisting on all my consists, since I never break them up on the layout.  Hard consisting is relatively simple, as long as you don't mind the consisted locos responding to all the Function commands you send - for example, if you press F0, BOTH locomotives will turn on their lights.  If you have sound decoders, then BOTH locomotives will blow their horns with F2, turn on the bells with F1, and so forth.  You can change this by customizing the way one of the locos responds to function keys (remember that how the decoder responds to function keys is stored locally), but that gets you into some potentially complicated programming.  You also have to remember that if you have one of the locomotives in your consist running in reverse, you'll need to enable "reverse running" in CV29 so that it will run in the same direction as your lead engine (your lead engine is running forward; the "reverse" engine will be running in reverse).

You can also use "Universal" consisting (Digitrax's term) which is actually "command station assisted consisting" in that the command station memorizes which locomotives are assigned to a consist, and sends the correct speed packets to each loco in the consist.  Note that in this kind of consisting, only speed packets and direction packets are sent to each loco; function commands are NOT sent to each consisted loco.  If you press F0 to turn on the headlight, only the headlight of the lead locomotive will come on.  If the locos are sound-equipped, pressing F2 will sound the horn ONLY ON THE LEAD LOCOMOTIVE and so forth.  Universal consisting works fine as long as (1) your command station doesn't lose its memory (if it does, all your consists will be gone) and (2) you don't want to take a consist to a different layout and run it as a consist on that layout.  Since Universal Consisting depends on the command station, if you take your consist to a different layout, that layout's command station won't know the locomotives are consisted, and will treat them as individual locos. 

The final consisting method is "Advanced Consisting".  Advanced Consisting uses CV19 to tell a decoder that it is part of a consist.  If CV19=0, then the decoder knows it is NOT part of a consist.  If CV19 is >0, then the decoder knows it is part of a consist, and will only respond to DCC command packets sent to the consist address.  Unfortunately, CV19 cannot accommodate a 4-digit address.  So if you use Advanced Consisting, the "consist address" will have to be some number between 1-127.   Some DCC systems, like NCE, then use the command station to "alias" the CV19 number to the actual engine number; this works fine, but if you then take your consist to, say, a Digitrax-equipped layout, it won't respond to the engine number - you'll have to select the CV19 number as the loco address to get the engines to respond.  The advantage of Advanced Consisting is that the consist is not dependent on the command station; if you have three locos set to consist address 19, you can take them from one layout to another, and they will always respond as a consist to the "19" address.  In addition, most modern decoders also have two additional CV's (CV21 and 22) that allow you to tell the decoder how to respond to function keys when it is in a consist.  For example, you can program the locomotives to all turn on their lighting when F0 is pressed (or not); or have all sound their horns when F2 is pressed (or not) as you wish.

Which consisting method is best for your situation depends on how you use your locomotives and your DCC system.  Since I don't break up consists, I use the "hard" consisting method (each loco set to the same address); if you do helper service or break up consists regularly as part of your operations, then Universal consisting is probably the better choice for Digitrax systems, while NCE generally defaults to Advanced consisting (but then uses the command station to alias the CV19 address to the locomotive address).  Advanced consisting is a hassle with a Digitrax system, because you have to remember the separate 2-digit consist address; you can't just dial up the lead loco's number and take off.   In addition, there is no easy way to break up an advanced consist with a Digitrax system except by using ops mode programming to reset CV19 to zero.

It's too bad the NMRA didn't set up Advanced Consisting CV's to use a 4-digit address; if they had done so, advanced consisting would likely rule the world.  They didn't, it doesn't, and as a result this is a "whatever works best for you" kind of thing.

John C.

[C855B]

Great rundown, John. Since you brought-up BEMF and consisting, I have a question where I think I know the answer. Maybe.

I've rebuilt a chassis to use under a DD35 from a donor DDA40X. I'm using the original two motors, replacing the crude Bachmann decoder with a TCS M1, the motors in parallel. So like the prototype it's basically a pair of consisted locos under the same roof. BEMF off, or on? I suspect "off", but could use confirmation. Or should I throw another M1 under the hood and control both ends separately? (And, by extension, how would one go about speed matching? Set each end on parallel tracks and see if it tries to swap ends? )

Thanks!

[jdcolombo]

Great rundown, John. Since you brought-up BEMF and consisting, I have a question where I think I know the answer. Maybe.

I've rebuilt a chassis to use under a DD35 from a donor DDA40X. I'm using the original two motors, replacing the crude Bachmann decoder with a TCS M1, the motors in parallel. So like the prototype it's basically a pair of consisted locos under the same roof. BEMF off, or on? I suspect "off", but could use confirmation. Or should I throw another M1 under the hood and control both ends separately? (And, by extension, how would one go about speed matching? Set each end on parallel tracks and see if it tries to swap ends? )

Thanks!

Well, I have no idea how BEMF would work with two motors on the same circuit.  I think two motors would mess up BEMF's measurement circuitry, so probably best to turn it off.  But you could always try it on first and see if there's a problem.  The advantage of BEMF is slow-speed motor control - you can get even old motors to run like swiss watches with proper BEMF adjustments.  On the other hand, it's not like you'll be using the DD35 for switching duties at an industrial park, so it probably isn't that critical for this particular engine to begin with.

The real problem is whether you are taxing the M1's output capacity by having it run two motors at once.  I think the M1 is rated at 1 amp continuous, 2 amps peak, so you are probably OK, but older motors typically pull more current than newer ones.  In theory, an N scale motor could draw a bit more than 1/2 amp at stall, which would cause the M1 to either shut down with thermal protection or blow itself up (the former is more likely), but I'd probably put in two decoders and "hard" consist them with the same address just to be safe.  If the motors are well-matched, then all you need to do is program CV's 2, 5 and 6 with identical values in each decoder, and you should be home free.  If they aren't well-matched, then you've got problems whether you are using one decoder or two, and at least with two you might be able to fix any matching problem with trial and error iteration.  The way to do this would be to get an engine cradle with rollers and watch the wheels as you advance the throttle, but you should be able to hear (and see) if one motor is "dragging" behind the other without a roller cradle.  If it is, you'll likely get some sort of squeal from one end of the engine as the wheels on that end get dragged along.  But I honestly doubt that there will be enough motor mismatch to create a problem, whether you have one decoder or two.

John C.

[peteski]

Excellent info John!
Just to reiterate, in speed-matching and consisting, speed curves (and any of the motor control functions, except for the direction) are unrelated and independent from any form of consisting.

The speed matching has to be done when the locos are not consisted (using whatever method you find handy).  That process should result in all the locos which are to be consisted to run at pretty well matched speed at most speed steps they are directed to run at by the command station.  Consisting simply ties multiple decoders to a single throttle (by using one of the 3 consisting methods).  The speed step command sent to the consist will be the same for all locos in the consist, and since they are speed-matched before being consisted, they will run smoothly.

I would also not get too hung up on perfect speed matching (especially if consisting multiple brand locos, or locos with different brand decoders). It is doubtful you'll get them to be perfectly matched (on all speed steps) and small variations are quite acceptable. The mechanisms themselves will change their running characteristics when they warm up after running for a while.

Before DCC we (the model railroaders in general) ran multiple unit lashups without being able to well speed match the locos. I have done that myself for decades on NTRAK layouts and my models are still here and working (some converted to DCC).

[C855B]

Well, I have no idea how BEMF would work with two motors on the same circuit.  I think two motors would mess up BEMF's measurement circuitry, so probably best to turn it off. ...

I was hoping you had additional thoughts there about BEMF vs. two motors, but your assessment is the same as mine. The question is whether the BEMF firmware will figure out on its own that it is seeing garbage. But since we basically know it won't work as designed, I will just turn it off at install and call it done. Maybe you recall the problems several years ago as BEMF created issues with coreless motors; similar situation - unexpected feedback.

I've never had problems with the unmodified Bachmanns as far as each end not matching, so in reality I'm not going to spend a lot of time worrying that.

These are modern motors, so current shouldn't be any trouble, even at stall. The chassis is light, maybe too light given the number of wheels. Wheel slip will vex us long before we even think about stalling.

[peteski]

As I understand the BEMF circuitry, DCC decoders powers motors with pulses of voltage. In between pulses they sample the voltage across the motor leads. If the motor armature is in motion (spinning), the motor's armature will act as a DC generator and produce a voltage across the motor leads.  That voltage is then probably averaged over several sampling cycles and uses for the BEMF feedback.  If 2 DC motors are connected in parallel then the BEMF voltage from those motors should be an average of both voltages.  I think that even with 2 identical motors powered by a single decoder, the BEMF feature will still work but probably not as well as with a single motor.

Also, aren't both motors in that Bachmann loco mechanically connected by a drive shaft?

[C855B]

No, it's not generated voltage. If you've ever closely examined a common DC motor on a 'scope, you'll see these awful high-frequency, high(er)-voltage spikes that come backwards from the motor. This is the RF nastiness the Bachmann "Euro circuit" tries to filter out, preventing the rails from turning into antennas. The spikes correspond to the magnetic fields collapsing every time the brushes make/break with each commutator segment. This phenomenon is a super-accurate way of measuring armature speed. With two motors in the circuit, you'll see spikes from both not in phase, and the firmware will attempt to make sense out of them, which it can't. Good firmware will see the crap and shut off BEMF, consumer-grade firmware, definitely maybe.

The "ends" of the B'manns are completely separate mechanically. In fact, to use common gearing for both, the motors run backwards from each other. They only covered-up the donut hole to hide a flywheel for the aft motor; the casting and resulting plugs look like the original tooling was planned to feature the pass-through walkway of the 1:1.

[peteski]

No, it's not generated voltage. If you've ever closely examined a common DC motor on a 'scope, you'll see these awful high-frequency, high(er)-voltage spikes that come backwards from the motor. This is the RF nastiness the Bachmann "Euro circuit" tries to filter out, preventing the rails from turning into antennas. The spikes correspond to the magnetic fields collapsing every time the brushes make/break with each commutator segment. This phenomenon is a super-accurate way of measuring armature speed. With two motors in the circuit, you'll see spikes from both not in phase, and the firmware will attempt to make sense out of them, which it can't. Good firmware will see the crap and shut off BEMF, consumer-grade firmware, definitely maybe.

The "ends" of the B'manns are completely separate mechanically. In fact, to use common gearing for both, the motors run backwards from each other. They only covered-up the donut hole to hide a flywheel for the aft motor; the casting and resulting plugs look like the original tooling was planned to feature the pass-through walkway of the 1:1.

I'm pretty sure the decoder averages out that "crap" to get an average voltage.  Remember they sample the voltage between the power pulses for probably only few microseconds.  We know that this method works, so they must have a way to get some good reading from the "crap".

 BEMF is not the same as the collapsing magnetic field of a winding when the power is is cut off. It is the back electromotive force (BEMF) generated by the motor acting as a generator.  Quoting https://en.wikipedia.org/wiki/Counter-electromotive_force
The term back electromotive force is also commonly used to refer to the voltage that occurs in electric motors where there is relative motion between the armature and the magnetic field produced by the motor's field coilsof the permanent magnets, thus also acting as a generator while running as a motor. This effect is not due to the motor's inductance but a separate phenomenon.

This voltage is in series with and opposes the original applied voltage and is called "back-electromotive force" (by Lenz's law). With a lower overall voltage across motor's internal resistance as the motor turns faster, the current flowing into the motor decreases. One practical application of this phenomenon is to indirectly measure motor speed and position, as the back-EMF is proportional to the rotational speed of the armature.

In motor control and robotics, the term "Back-EMF" often refers most specifically to actually using the voltage generated by a spinning motor to infer the speed of the motor's rotation for use in better controlling the motor in specific ways.
 For more details read the info in the above link.

I'm not an expert in this field but I know that the electrical characteristics of a permanent-magnet DC motor are much more complex than just a simple inductor (due to the BEMF).  For more technical details and electrical characteristics see https://en.wikipedia.org/wiki/Permanent-magnet_electric_motor

But the bottom line is that with 2 motors in parallel in a single locomotive on a single decoder there will still be BEMF generated.  As I said, that won't be the optimal way to operate them but I think they will perform satisfactory.   Best way to test that is to try it.