I have replaced a lot of MTC21 decoders and have not had any problems getting them to work. Perhaps I have been lucky. As Tom states, there is more than one standard for the 21 pin connector. Edit: I do not know if Marklin uses more than one standard.

I did have trouble identifying which output was under which decoder function. Slider changeover relays (2 functions btw), power for sinus drives, interior lighting, all needed to be configured with trial and error until I bought the ESU decoder tester and could quickly identify output-function pairings. Marklin tries to put outputs under the same decoder function. Sinus drive power is usually F4, but this is not steadfast and true.

Thanks Tom. That particular loco (BR53) as converted has smoke on AUX1 and the Condenser Fans on AUX2, nothing on AUX3 or AUX4.

An simple question...
Do Märklin have buffering capacitor too in the lighting sets?

They don't have a capacitor in my G1 coaches......

I´m not really sure,if Railcom with feedback do have disturb too,when there is buffering capacitor in the lighting sets.

You're splitting hairs there a bit Tom! Even if it isn't written down formally, it has always been accepted convention that you will get faster and more reliable registration if only 1 mfx locomotive is on the track when registering.



However, I knew I had seen it written down somewhere and indeed I have. See page 17 in the Ecos 2 manual (even if it is ESU saying it and not Märklin).

"11.2.1. Entering M4 locomotives
..............

We recommend placing new locomotives on the layout one at
a time. If there are several new locomotives on the layout then
the registration process may be unnecessarily delayed."

I think Tom is saying that you must have only one new loko on the track. That is very different from only one loko on the track during registration. If it is the latter, then you must remove all lokos except the new one from the track. Very impractical and what affect would that have on re-registration? Would you then loose sync with prior registrations and end up with the other lokos re-registering when placed on the track? All this strange and abnormal operation is exactly why MFX is such a mess. Do this, but do it exactly like this, in this prescribed order or you will end up with your lokos having to re-register. Why must we have so many conditions on a system sold as plug and play?

I often have a rolling road or some pieces of track plugged into the programming track output on the Ecos / CS2. That way it is easy to unplug the layout, leaving only the programming track connected.



You're taking me out of context there - the splitting hairs comment referred to you questioning whether there were written instructions or not, to have only 1 loco on track when registering a mfx loco. Nothing to do with what you had to do at your club to get your locos registered.

Last night I received a new loco, the French Crocodile, BB 12000 series, Trix 22336, and this comes with an MFX decoder. I must say that there is no need to do the programming with this decoder. It auto-programs itself and its ready to run once it is placed even on the main track. I used to like the procedure of opening up the loco and installing the decoder and then putting the loco on the programming track to program it.

I don't know if I can say that I dislike mfx, I feel ambivalent.

Interior lights with big capacitors are very likely to kill the mfx feedback signal. The feedback is achieved by varying the current consumption of the decoder. This variation is then converted to a signal voltage in a signal transformer in the controller (or in the mfx capable boosters). A big capacitor in the circuit will even out the current variations so that they get too weak to be picked up by the controller.

This is the same electrical principle as in the Digitrax system.

The Railcom works differently. Here the feedback is a voltage driven by stored energy in the decoder during short periods of "dead" track voltage. This can of course also be disturbed by other things on the track, but since the principle is fundamentally different from mfx feedback it can be completely different things that disturb it.

By the way, the mfx protocol itself can very well be operated without feedback, by using manual registration. The fact that this option is not offered in the controllers seems to be only a marketing decision.

With a digital oscilloscope you can observe the signal behaviour and also see different disturbances. Measuring on mfx feedback is a bit tricky since it is small current variations on top of a large current, which is also varying, but slower. An oscilloscope is however not really sufficient if you want to analyze the data patterns on the signals going back and forth. Then you need some kind of protocol spy tool, with capability to take longer recordings than a normal digital oscilloscope can. I built such a spy tool for my investigations but it only supports analysis of the forward signalling so far.

To really get a feeling for what is going on we are talking about several minutes. Of course you can see some interesting things with much shorter traces too, but you will only get fragments then.