Comparing decoder motor control features, namely Back EMF, Torque Compensation, Dither etc.

Main North HOME


Stated simply, if you want your loco to move off from stationary at speed step 1 – every time, irrespective of the train load, use a Back EMF decoder from Lenz, ESU and now (Sep 07) from TCS. There can be a significant price difference, but with the release of the TCS T1 and MC2 at A$35, the difference is acceptable. 


The relationship between PWM frequency CV 9, noise and performance (oomph).


DCC decoders use Pulse Width Modulation – PWM techniques to control the motor, lights etc in the loco. PWM is cheap to manufacture, does not create much heat and gives good overall control. This PWM voltage from the decoder may cause a buzz/hum from the motor depending upon the frequency of this PWM voltage. This buzz/hum from the motor/mechanism varies, depending on the mechanism used.


The frequency of early DCC decoders could be adjusted between about 200 – 2,000 Hz (cycles per second) and many operators complained about the “noise” from these decoders. The noise from some locos is so loud that it had its own sound system, not too prototypical but it was cheap. The frequency could be adjusted that did make a little bit of difference, but the noise was still very evident in the noisy mechanisms.


For successful quieter operation, manufacturers increased the PWM frequency to about 15,000 Hz hence the term “quiet” decoders. The noise is still there, but higher than our audible hearing range. Most, if not all manufacturers now use this higher PWM frequency. Getting rid of the noise problem created a new problem. These “quiet” decoders caused the motor to loose its “oomph” (torque). Now was this a bigger problem than the original “noise” one? This also required adjusting CV 2 – Volts Start, to get the loco to move off at speed step 1 or close to it.


An example of this is on my early NCE Version 3.2/3.3 software “quiet” decoders, I could adjust the PWM frequency using CV 9. Adjusting CV 9 to the higher values with 255 the lowest PWM frequency but this turned the “quiet” feature of these decoders OFF. This generally gave the decoder more “oomph” This lower PWM frequency gave a very “noisy” motor but the loco moved off at speed step 1. This defeated the purpose of introducing these decoders, but due to the loss of “oomph”, modellers played around with CV 9 to give the best compromise between noise and oomph. Adjusting CV 9 to 0, gives the highest frequency of 15,625 Hz, the loco motor is “quiet” but I had to adjust CV 2 to 40 (using 128 speed steps) on my NCE Procab to provide the loco to move off at or near speed step 1. 


Manufacturers introduced an extra enhancement to the PWM voltage to the motor to overcome this loss of torque – “oomph”, by adding an extra “magic” pulse of different frequency and amplitude. NCE calls it Torque Compensation and uses CV 116 - kick rate (frequency) and CV 117 - kick strength (amplitude). TCS call it Dither, and use CV 56 - frequency and CV 57 – voltage. Digitrax call it Torque Compensation etc.


This now gave a quiet decoder that gave super slow speed and great performance (oomph) WITHOUT the hum/buzz from the motor/mechanism. Manufactures coined the term “Silent Running” decoders. The NCE and TCS decoders that come with the Torque Compensation and Dither do not support CV 9 (PWM frequency) and “programming’ this CV does nothing. The decoder just seems to ignore this. For NCE all decoders now are shipped with version 3.5 of the software.


Low Speed Performance.


In the past I have used, NCE’s D13SRs and N12SRs  and TCS’s T1s and M1s as my motor decoders. Depending on space availability and location against cost, would  determine which one. Each of these decoders offers torque compensation or the similar dither for this enhancement to the slow running speed characteristics of these decoders. I adjust the torque compensation or dither volts and frequency CVs. The final product is better than a decoder without these features. My initial decoders were NCE and I adjusted the relevant CVs to achieve the best result.


NCE’s Torque Compensation and TCS’s Dither are a "fixed" amount of "enhancement" signal that does work. Using my electrical/automotive experience, this type of correction is what is called "open loop" (no feedback) This “enhancement” will not change according to what is actually happening, like when the load changes,


Note: If you don’t adjust torque compensation or dither CVs for volts and frequency, or operate these decoders with the minimum values, the loco may need a higher throttle setting (higher speed step setting) to get the loco to move off from stationary. This may be interpreted as a slower decoder. As the dither or torque compensation introduces more “oomph” to overcome the initial inertia of the loco and possible mechanical resistance because of the higher PWM frequency, CV 2 may have to be adjusted to get the loco to move off at the lowest speed step setting. Also setting dither and/or torque compensation CVs, will affect CV 2 – the Volts Start setting. 


Back EMF is a “variable” amount of “enhancement” and value is determined by what is actually happening NOW - real time. This form of adjustment is called "closed loop" and this is feedback. It does this by measuring the voltage generated by the motor, when the power is removed, and that is why it is called Back EMF. Incidentally all motors generate back EMF. Since the current in a later "quiet" decoders is turned on and off at about 15,000 - 25,000 times a second - PWM, this happens very quickly and the PIC inside the decoder changes the "adjustment" to the base power signal to the motor, many times a second. This is a “real time” adjustment, happening all the time.


Thus by definition alone, Back EMF will be better than Dither, Torque Compensation etc as it is "real time" closed loop feedback, enabling continuous

adjustment according to all the variables and in the case of our locos – variable load.


When adjusting the decoder’s dither or torque compensation etc CVs, you will get varying results dependant on “actual” load at the time. In model railroad terms, when operating a train with a full load of wagons, shunting a few wagons and operating “light engine”, the loco “start off” results will be different. See also my NR loco experience about super slow running complaint.


For me, the Back EMF decoder, is all about getting the supreme low speed control with constant take offs at speed step 1 - every time and is not about "cruise control" going up and down hills, but the two go "hand in hand".


So that is the theory and these are the differences in Dither, Torque Compensation and Back EMF. I find the results from Back EMF decoder will be determined by the type and condition of the drive line used. This type of control is so good, that it can hide some mechanical problem like a binding mechanism etc, but so what. When the problem gets bad enough, the decoder will not be able to hide it any longer by its “magic”.


Adjusting the decoder for better speed control except Back EMF decoders.


I use CVs 2, 5 and 6 for my loco preferred performance and speed matching, as this is close enough for me when double heading. Sound decoders from QSI and Soundtraxx do not support CV 5 and CV 6. I have to make my own arrangements here, depending on the decoder. I may even use a Decoder Pro constructed speed curve (CVs 67 thru 94).


CV 2: Volts Start - Adjust CV 2 for the lowest speed that the loco can maintain at speed step 1.

CV 5: Volts Max – Adjust CV 5 for the maximum speed of the loco.

CV 6: Volts Med – Adjusting this CV determines how the loco responds through the middle speed. Adjusting this to less than half the maximum speed will give a “speed curve” with smaller speed step increments in the lower half of the “speed curve” while the upper half will have greater speed step increments. This gives the sense of better low speed control.


Decoders with Torque Compensation and Dither: Adjust these CVs to give the best slow speed control that will get the loco to move off regularly at the slowest speed step (hopefully near speed step 1).  


Price comparison between decoders and choice in Australian dollars.


For me in Australia, decoders cost are: NCE’ D13SRs about $23, N14SRs about $40. TCS’s T1s about $30 and M1s about $43. Lenz Silver about $55, and Lenz Gold, about $60. TCS has, from Sep 07 introduced Back EMF to its range of decoders and at the moment the T1 and MC2 have Back EMF that is standard. Click here to see all about TCS Back EMF. Programming CV 61 with 3 allows you to turn off the Back EMF by operating function button 6. This is a great feature, if consisting locos.


Depending the available space, if I could not fit a D13SR or a TCS M1, I would have to fit the smaller N scale decoders.


Obviously the electronics involved with Back EMF and the “European” price, makes these Lenz Back EMF decoders more expensive. This above comparisons is not totally fair, as price has not been considered. I have ONLY made comparisons of the features.


Not everybody will have the same priorities as me in regards the way a loco starts off. Some are happy at just using the decoder at the default value, some are happy to adjust just CV 2, some want to use Decoder Pro for a User defined speed table, etc. Everybody will have to make a choice on which decoder he/she fits to a loco, but after fitting decoders to my somewhat small fleet, when compared to others, and to the installations that I do for others, I have found that when I include the cost of the loco, an extra say $20 to $30 for fitting the Lenz Back EMF decoder that will give the best possible performance and always start at speed step 1, worth it. This $20 to $30 is “bugger all” in the big picture, when for me locos start off at $245 for diesels and $400 for steam locos. Hopefully the future QSI locos that I am getting from Eureka Models, will have the back EMF feature turned on, as for me, this make a great deal of difference.


I will be retrofitting my “better” most commonly run locos that have a problem with starting with these Lenz Back EMF decoders, but certainly don’t intend doing all my fleet. I will check for binding, tightness etc in the mechanisms first on these problem locos. I am intending to fit sound to all my locos. So each loco does become a little expensive.


I have three Austrains 4-6-0 36s that don’t have any tender pick ups fitted from the factory (I have fitted extra pick ups to two of them) and the one that I have fitted the new Soundraxx Tsunami to, always stops same set of Peco Insulfrog points (turnouts). These locos basically are the same as a 0-6-0 tank engines or similar locos that have only pick ups on 6 wheels. I have 3 diesel K & M mechanisms that only pick up on 6 wheels also. I am going to try a Lenz USP. Report soon in these web pages. I have added a “Stay Alive”  capacitor to some of my Sountraxx DSD-100LC sound decoders that causes the sound to reset at the slightest hint of dirt, but does not affect the loco performance. I have also fitted the Stay Alive to one NCE D13SR in one of the K & M mechanism diesels.


This is just my own experience. I am not suggesting it is the ONLY way to do things. I am explaining the options. Everybody will have their reason and way of doing things and it will depend on many factors.


Note: When consisting locos, the effect of Back EMF should be decreased as the loco speed increases, if this possible by adjusting CVs OR deactivate the Back EMF , otherwise there will be a “jerking” operation of the consist. I will add detail here later.