How do you speed match locos? : LUSENET : Wiring for DCC : One Thread

Hi- Can someone give a quick lesson how to speed match a pair locos? I have a Digitrax Empire Builder system and would like to match two N scale Atlas SD-35s with the factory installed Lenz decoders as well as two Kato PAs with Digitrax DN-145K decoders.

I've seen several references to speed matching in the digitrax docs but no really good description of how to do it.

Thanks, Jim

-- James Boyd (, April 19, 2001


See my answer to the same question on Tony's bulletin board, dated 16- APR-01. DonV

-- Don Vollrath (, April 20, 2001.

I could not find the message (referred to in the other response) on the TTX website, so I emailed them and am waiting for a reply. Could you re-post your info here?

I found out that Loy's Toys publishes a "DCC Encycolpedia", in which there is a writeup on this topic.

-- -John Montenigro (, August 17, 2002.

re – Matching Loco Speeds and Speed Tables

The goal is to adjust the DCC decoder so that many locos will run at the same speed while in a consist and receiving the same speed step command.

Remember...Model railroad loco mechanisms are nothing more than a permanent magnet DC motor geared down to rotate the wheels. Each manufacturer, and perhaps loco model or manufacturing run of the ‘same’ loco may have – 1) A particular motor with 2) A particular gear ratio and 3) A particular set of manufactured gears. The characteristics of these variables define how fast the loco will go with a particular voltage applied to the motor while pulling a particular load. Item 4), the brand & model of the DCC decoder, also becomes a variable for speed table generation as each modeler elects to use his favorite brand of decoder, and each DCC decoder manufacturer also elects to interpret the numbers his way.

1) From an engineering point of view, the motor voltage at the armature terminals will be Vdc = (K)* (rpm) +( Iamps)*(Rohms) (Equation 1) Where:

The CEMF constant (K) is relatively fixed for each individual motor design but will have minor variations from unit to unit. It will also vary slightly with motor temperature. This is basically the result of the strength of the permanent magnet and rated motor design rpm at 12 volts. One can also use the multiplier (K) to include the gearbox ratio and convert rpm to be scale mph or kph if desired.

Iamps = the current flowing through the motor in amperes. This is what creates the torque to pull the load and overcome gearbox friction. The torque per ampere ratio of the motor is part of the motor design selection but does not directly enter into the equation for voltage.

Rohms = the electrical resistance of the motor armature as measured at the motor terminals. It is relatively fixed by design of the manufacturer, and will increase slightly with motor temperature. The motor resistance will be significantly different from manufacturer to manufacturer. More powerful motors tend to have less resistance.

The (Iamps)*(Rohms) term, known as the IR drop, defines an additional amount of motor voltage necessary to maintain the same operating speed when load amperes change. Note that this is a linear characteristic in that the voltage change is directly proportional to the change in load amps. Decoders with CEMF compensation supposedly adjust motor voltage to match this term in order to hold rpm constant. Motors with more resistance need more voltage compensation to hold rpm constant with an equivalent change in torque.

Turning the motor equation around reveals that: rpm = (Vdc – I*R)/K (Eq. 1a) This defines the change in rpm at a fixed voltage as load varies, like when a loco starts to pull a load uphill. This is the speed- torque characteristic curve of the motor. With constant voltage applied, motor speed will sag proportional to torque. Motor resistance defines how much it will sag.

And/Or: Iamps = (Vdc – K*rpm)/R (Eq. 1b) This defines the additional current drawn by the motor at a constant applied voltage as speed is varied.

The loco manufacturer has hopefully selected the motor, gearing and loco weight such that any load from free-running without cars to maximum strength pulling with wheel slip is acceptable. This defines the range of ampere draw expected with the unit. Drawing more amperes may result in an overheated motor. There is little change in the rolling friction of model railroad cars once they overcome static friction. The motor equations then tell us that once moving with a constant friction load, locomotive speed will be increased in proportion to additional voltage.

2) & 3) The gear ratio (including wheel diameter) defines the motor rpm at any particular loco speed. It is selected by the manufacturer and may even vary from unit to unit. We don’t really care what the actual motor rpm is but our expectation is that the manufacturer has selected components that will result in a reasonable scale mph speed at close to rated motor voltage. What is critical is that the gearbox friction doesn’t waste torque or cause disruptions in operation due to mechanical gear-teeth hang-ups or backlash. Clean, well meshing gear teeth, free of burrs and stiff grease are necessary for smooth performance. The worm gear is particularly prone to high friction and gear binding due to mechanical misalignment. It must remain stationary with no axial play to avoid bucking and chattering during operation. Torque wasted in the gear box simply steals power away from our locos and foils our attempts to control speed.

4) Most new DCC decoders support the Start voltage at CV2 and the Maximum Speed setting at CV5. Many (but not all) support a Mid Speed setting at CV6. Most units, old and new, have an uploadable 28 step speed table, CV67-94, for customizing a particular speed curve. Most of today’s decoders will vary the motor voltage with 256 steps of resolution from full OFF to full ON using PWM from rectified track voltage. For operation with 14 or 28 step systems, the decoder control unit will calculate and select one of the 256 total possible steps by mathematical interpolation. In either case, the DCC manufacturer has interpreted the NMRA standard in their own way to determining exactly what motor voltage is produced with what table selected speed step.

So...How does this all come together to determine the best method and procedure of matching loco speeds? We may want to run Atlas, L-L, Stewart, & Athearn locos all in the same consist. So we must agree that making all locos in a consist the same make or type is not desirable. And each loco may indeed require a different type or brand of decoder, just to fit inside. The motor equations and differences between motor brands tell us that different locos will have different voltage-speed-torque curves. So even if we adjust each unit to run at the same exact speed without pulling any cars, they will still tend to run at different speeds when subjected to pulling the same string of cars, or going up and down hills. When coupled together in a consist they obviously must run at the same speed, or wheels will slip. As train car drag increases, the consist locos with low resistance motors will quickly draw more current as rpm falls off and assume most of the load. (Eq. 1b) Units with high resistance motors will draw a smaller additional amount of motor current with the same sag in speed. The differences in their voltage-speed-torque motor characteristics will show up as variances in torque, or pulling power. So the best we can hope for is that each unit in a consist will pick up a share of the load, based on their design strength.

When a consist goes downhill, a different problem arises. By nature, a DC motor can become a generator if load torque pushes the motor speed faster than equation 1a predicts. (Part of the dynamic brake principal on the prototype) However, the worm gear on our models does not transmit torque power from wheels back to the motor very well. The usual result is partial binding and oscillation of the worm gear as it bangs back and forth within the mechanical backlash space, absorbing energy. This causes noise and pulsations in thrust and speed. From the outside, noise and visible banging and bucking of couplers occurs. This is a mechanical problem that cannot be totally avoided by electrical adjustment.

At this point the choices for speed matching are: A. Match speeds at no load. B. Match speeds at mid load. C. Match speeds at full (almost wheel slip) load.

Most discussion is about modelers trying to do A, but option B seems to be a more intelligent choice than A and is certainly easier than C. Reason – The real objective is to use consists to pull loads, rather than run them without a train. Actual consist speed will be established by the stronger units, i.e. – the motors with least resistance. So individual consist units should be adjusted to run at the same speed while pulling typical loads. One should establish a ‘standard load’ of cars and pull them while adjusting decoder speed tables. A 5–10 car train should do. Use the same string of cars for adjustment of each loco. Because there is so much variance between locos and so many variables that affect the actual load placed on the motor, there is not much point of attempting to speed match them at every speed step. Anything beyond getting them all to start at step 1 and run at the same maximum speed with a typical load is wasted effort. All the intermediate speed set-points may be uniformly calculated with acceptable results.

Step 1 – Decide on a target step-speed profile. Determine the desired top speed and a way to measure it. A marked off measured distance on a loop of track and stopwatch works well. Calculate the time it should take to traverse that distance at the desired top speed. Personal preference will determine if you want a linear or a logarithmic increase in speed response from cab throttle steps. [Remember, cab throttle steps are different from the PWM voltage step values actually presented to the motor by the decoder. This is what speed tables are all about.] One useful profile is a two slope linear relationship about a mid speed set-point. Selecting the mid range speed (step 14) to be 20-30% of overall speed range seems to yield adequate adjustment range for slow speed yard work without noticeable speed step jumps above half throttle. Once the minimum speed and maximum speed values are determined in Steps 3 and 4, one may quickly calculate the rest of the 28 step speed table. See the accompanying spreadsheet for an aid to calculate 2-stage linear or parabolic rise intermediate steps.

Step 2 – Clean and adjust the loco gears. Take them apart. Examine each gear under a magnifying glass. Remove any burrs. Reassemble and select which gears mesh with what others for best performance. Adjust the worm gear bearings and shims to eliminate axial play and achieve good meshing to the bull gear. Re-lube with quality gear oil. Clean, remove burrs, and re-lube universal joints, motor and wheel axle bearings. This will help any loco perform better by reducing friction load on the motor. It will usually minimize or eliminate worm gear backlash banging.

Step 3 – Determine the minimum speed set-point value. Set the loco by itself on a test track. Select the 28 step operating mode. Select the alternate speed table. [Or use the standard table to program Vmin,mid,max. You must be test running from the appropriate table that you intend to use.] Select speed #1. Use Programming-On- The-Main to adjust the value of CV67 [or CV2] to obtain the slowest creep speed possible. Record that value as Vmin. Note: This is the time to play with the PWM frequency at CV9 and or the Kick Start voltage at CV65 to achieve reliable starting without jack rabbit jumping. On some decoders the value in CV9 affects the translation from CV values in the speed table to actual motor voltage, so determine the desired PWM frequency first. The writer has discovered that reducing the PWM frequency to the lowest possible value sometimes produces the best results. Reason – It continually kicks the gears to overcome friction, like the pulse power of yesteryear. On many models, the acoustic noise produced is not objectionable. Newer models may like Silent Running better.

Step 4 – Determine the maximum speed set-point value. Add the ‘standard’ test load of 5-10 cars. Set the value of CV67 [or CV2] to be 200, which should be about ¾ speed. [If you will be using the Vmin, mid, max method, you can temporarily set Vmin (CV2) to be at the maximum speed test value at speed step 1.] Select speed step 1. (Careful – It is set for maximum speed!) Be sure to let the loco fully accelerate before timing over your measured distance. Clock the time to travel between marks. If the time is too short, use Programming-On-The-Main to decrease the value of CV67 [or CV2]. If the time is too long, increase the value of CV67 [or CV2]. When the travel time is within your acceptable tolerance, record the value of CV67 [or CV2] as Vmax

Step 5 – Load your observed values of Vmin and Vmax into the spreadsheet to calculate Vmid and the other 25 values. If you are going to use the NCE Vmin,mid,max method, the values shown for Speed Steps 1, 14, and 28 are the calculated values for CV2, CV6, and CV5. For a complete 28 step custom speed table, read the calculated values from the proper column and program them into CVs 67-94 of your loco.

Repeat Steps 2-5 for every loco that you intend to consist together.

Enjoy - DonV

-- Don Vollrath (, August 19, 2002.

If you have a computer and some form of Loconet connection (MS100 or LocoBuffer), this may help with the speed table, and matching speeds to another loco. John Kabat came up with a program called racetrack.

Click on the link to racetrack in the menu. Or just click on

The basics, you start with a 22" radius circle, place the "master" loco (the one you want to match TO) and the new loco directly apart. Start the program and tweak the speed tables to match.

You can then speed match different loco's together no matter the brand!

-- Joel Jameson (, February 18, 2003.

Moderation questions? read the FAQ