Stefano Tests 2 New Load Compensated Decoders!

Updated January 2002

Dear Model Railroaders,

BEMF technology for DCC decoders is improving the overall quality of available decoders. In this short review I shown the results of the measurements of slowest speed for two new BEMF decoders: Lenz LE130XF(LE1025C) and Zimo MX61R v2000. The test follows the guidelines I have specified below in the previous report.

Both these decoders have names similar to their old ancestors: LE130 and MR61R. However, they are much different from them, especially in the BEMF control. The main difference of these decoders is the resolution of the analog/digital (A/D) converter for the voltage feedback. Both manufacturer use Microchip CPUs, but Zimo uses 8 bits of A/D resolution, while Lenz uses 12 bits for improved control.

Lenz uses a very smart approach to get 12 bit BEMF resolution. My personal understanding is that, instead of sampling the analog value of the BEMF current with an AD converter, Lenz measures the time delay of zero crossing voltage with the internal timer of the CPU. In this way Lenz can achieve high bit resolutions and even improve using faster PICS. This is very smart !

For the test, the various CVs have been set to improve slow speed and reduce noise as much as possible. PWM frequency has been decreased from the default values to increase torque of the motor at low angular speed. Results are amazing if compared with top technology of two years ago.

Here are the values:

LENZ LE130XF (LE1025C)

FLAT:	Vmin=0.274+-0.004 Km/h {0.171+-0.003 mph}
PUSH LOAD:	Vmin=0.259+-0.004 Km/h {0.161+-0.003 mph}
GRADE:	Vmin=0.242+-0.004 Km/h {0.151+-0.003 mph}
GRADE PUSH LOAD:	Vmin=0.224+-0.004 Km/h {0.140+-0.003 mph}

Discussion: The Lenz decoder has a minimum speed that is a little higher than the Zimo. The loco is extremely smooth and silent, much more silent than the old LE130. There are no signs of over-compensation, and the loco slows down a little with cars and grades. The CPU has FLASH program memory and the software can be updated by Lenz

ZIMO MX61R v2000

FLAT:	Vmin=0.156+-0.004 Km/h {0.097+-0.003 mph}
PUSH LOAD:	Vmin=0.159+-0.004 Km/h {0.099+-0.003 mph}
GRADE:	Vmin=0.145+-0.004 Km/h {0.090+-0.003 mph}
GRADE PUSH LOAD:	Vmin=0.139+-0.004 Km/h {0.086+-0.003 mph}

Discussion: The Zimo decoder has the slowest speed I have ever seen in a DCC decoder. Great work has been done on the software and on the circuitry, respect the old MX61. Unfortunately the loco is a little noisy: increasing the PWM frequency would decrease the noise, affecting the minimum speed, unfortunately. There are no signs of over-compensation, and the loco slows down a little with cars and grades. The CPU is OTP (one time programming) and the software cannot be updated.

COMPARISON OF THE NEW DECODERS WITH THE ONES I TESTED PREVIOUSLY
Yellow bars are the new decoders, while magenta bars are the old ones:



Stefano Curtarolo

Disclaimer: I am a N/H0 scale DCC model railroader with no connection with Lenz and/or Zimo.

Stefano tests 3 load compensated decoders.
by Stefano Curtarolo, Cambridge, MA, April 20th, 2000

The following is the test of minimum speed for a H0 locomotor with three different BEMF decoder: Zimo MX61, Lenz LE130, Digitrax DH142P.

Date/place: 20th April, Cambridge MA
System: Digitrax Chief in H0 mode voltage
Engine: Life-Like SD7 Proto 2000, NMRA Plug
Distance: 129+-0.25 mm of flat track
Load: car with 150 grams of lead
Grade: 34.5+-0.5 mm height over a length 1816+-0.5 mm, percentage 1.9+-0.03%
Speed: all speeds are expressed in scale kilometers/hour, the scale miles/hour are between brackets {}.
Methodology: 10 time measures with photo cell switches. Average between 2 decoder tests.

ZIMO MX61R/N - Cv 2,3,4,5,9,56,57,58 = 01H,00H,00H,03H,00H,FFH,00H,FFH

Discussion: The Zimo decoder is well designed. The compensation gives a speed that is almost constant with the load/grade. The minimum speed is always in the range 0.75 - 0.77 Km/h. With the highest possible PWM frequency (16KHz, Cv9=00H), the loco is almost silent. The minimum speed is the highest between the three decoders but it is constant.

LENZ LE130 - Cv 2,3,4,5,9,29=00H,00H,00H,001H,0FH,06H

Discussion: The Lenz decoder gives a slight over compensation. The minimum speed increases with grade (+33%), load (+56%), and grade/load (+76%). With the maximum possible PWM frequency, the movement is smooth and the motor does produce a very low noise. The minimum speed is very low but it changes with grade/load.

DIGITRAX DH142P - Cv 2,3,4,5,55,56,57=00H,00H,00H,00H,90H,50H,05H

Discussion: The Digitrax BEMF is hard to setup. I have tried to get a good response at high and low speed (as in a real layout use), and I found some good average parameters Cv 55,56,57=90H,50H,05H. Anyway, the engine cannot keep its minimum speed as it reaches the load/grade. I set the throttle to 07 for the flat speed, while I increase to 09,10,11 for the other cases, since the engine gets stuck as it reaches the grade/load. This is very similar to reality.

You can see in square parenthesis the minimum flat speeds given by the different throttles. The engine does not run very smoothly and it's quite noisy (compared to the other decoders). The overall effect (average BEMF + extra throttle) is quite under compensated. Better results can be achieved adjusting the BEMF parameters for the low speed response, but this would introduce over compensation (and irregular speed) in the higher regime.

The minimum speed is the lowest between the decoders in the flat case. The other cases cannot be compared since the throttle is not kept at a constant value.

Stefano Curtarolo
TMRC governator - www.mit.edu/~tmrc

Disclaimer: I am a N/H0 scale DCC model railroader with no connection with any DCC manufacturers or dealers.

We welcome comments or suggestions from readers; please write or call.