I’m considering adding a stay-alive capacitor circuit to my Märklin loco 39206, BR 01.5 - and maybe more of my locos, eventually. Got a small size C-track layout. Sometimes my locos struggle to go over turnouts without stopping, especially at low speeds. As this is a well known problem with Märklin, I hope there’s someone out there who have tried to do something similar. The factory installed decoder for this locomotive (sitting in the tender) does not come with a connector for stay alive capacitor, unlike the mSD3 /mLD3 decoders for retrofit (for example the 60975). I’m not very keen on replacing my decoder(s) just to get this connectivity. So I’ve been trying to figure out how the decoder is working and where to connect in with my add-on circuit. It seems that the decoder has its own rectifier from AC digital to DC, and a smoothing capacitor of 470µF. Not to be mixed with the rectifier on main board below (main board rectifier is not used in this loco?). This gives the +Ub, aka DC-bus, used internally on the decoder and for motor switching transistors. Also being fed back to main board via the 21pin MTC connector. I suspect +Ub is controlled by a transistor or something. The positive side of the smoothing cap is not connected directly to MTC pin 16 (+Ub), so there must be some kind of switch between. My guess; in case of “overcurrent”, measured by a small low ohmic resistor for current sense just after rectifier and capacitor, this transistor will cut power out to avoid stuff burning up (short circuit or high loads). Also off by default (normally open) when decoder is not powered. Negative side of smoothing cap is connected directly to MTC pin 20 (gnd).
So ideally my stay alive circuit should be connected in parallel with the smoothing cap, and feed in power whenever track connection is bad. Then the displayed locomotive current consumption in CS3 will be correct, regardless if the power comes from track or stay alive capacitor. Also, if there is any overcurrent, the “safety switch” will prevent capacitor from feeding power into a short.
BUT:
It is a pain to get my solder tip onto the positive side of the smoothing cap, really tight space. Alternatively; I’ve found a good spot to solder on another capacitor, easy access, but this one is connected to MTC pin 16, which is +Ub after the current measuring resistor and safety switch. Guess this will also work, as long as the “+Ub switch” stays on (no overcurrent or other major issue..) But feels a bit more risky to solder in here.
And hoping there’s no forwarding diode that will prevent stay alive cap to feed back power in opposite direction on +Ub line.
To avoid any potential problems with loco programming, I’ll put in a switch to turn on/off stay alive circuit. Meaning, if the microcontroller need to lose (track-) power for a split second in order to reboot (?), the stay alive circuit should be turned off, not to keep the microcontroller alive at this point.
Am I missing something here? Have anyone tried something similar, or have relevant info on how the decoder is working? Does anyone have a schematic for Märklin decoders, even if not exactly the same model? It’s a pain to identify components /reverse engineer these small decoder boards, any info would be much appreciated.
Yes, I am aware of the downsides of having a (large) stay alive capacitor, wrt dead sections, emergency stops and so on. Hope my cap is not too big… Then again, in normal operation my locos will mainly stop based on position feedback and digital signals, not so much relays and dead sections.

Stay alive circuit:
Planning to use the smallest supercapacitor module I could find, with integrated cell balancing, plenty high voltage for Märklin, and (hopefully) a sensible amount of capacitance. Testing will tell. It will be a pain to squeeze into the loco tender, maybe I’ll have to move it into a box car or something. Not critical for initial testing, though.
https://www.digikey.no/n...SM154Q030W075PB/17047446
47Ω charging resistor in series with supercapacitor, to limit inrush current and slow charge. Parallel Schottky diode to feed power to +Ub when track connection is poor. High current capable and low forward voltage drop.

Marklin 21MTC (NEM 660) 20260306.pdf (466kb) downloaded 8 time(s).

Hello Gunnar,

I'm glad to see someone else getting into this area of adding stay-alives to Marklin locos equipped with Marklin decoders.

Here's a bit of background. I have only Marklin locomotives and have converted my older DCM (6090, 60902) locos with LokSound decoders and ESU PowerPacks. I love the LokSound decoders and the ESU programming capabilities.

However, the newer Marklin MFX and MFX+ locos have nice running characteristics and decent sound but don't have stay-alives. I really don't want to spend money on replacing these good decoders. Hence, I started down the rabbit hole of adding my own stay-alives to MFX decoders. (BTW, I'm glad that Marklin has finally made the move to include stay-alives on their newer locos.)

Getting back to your questions. You are correct in your statement -- pins 16 (U+) and 20 (digital GND) are used to power stay-alives. In your circuit, I'm not sure why you'd need a big 3W resistor and that capacitor is sure expensive.

Picture 1 is a circuit diagram of the setup I use. However, for this to work, some space is required. This works for most steam locomotives with tenders. I'm pretty sure it would work for your 39206. I have this setup in my BR01.10 (39103), BR44 (39889), BR50 (37897), E151 (39581 -- C-Sine), V200 (39821 -- C-Sine), and a few others. If space is a problem, then I revert to a simpler setup (see picture 2).

The setup in picture 1 provides plenty of stay-alive power to help a loco bridge most gaps. The capacitors that I use (Digikey #4688-CXP-3R0105R-TWX-ND) are pretty small, so I can fit 5 of them in a tender (see picture 3). 15 volts with a Zener diode for protection works well. I have not had any issues with this setup.

The setup in picture 2 is for locos that don't have space for the super caps. I try to fit as big a capacitor as I can in the space available. By "big", I don't mean size but capacitance. I tend to try and fit at least two 1000uf capacitors and if not, then only one (Digikey #1189-1583-1-ND). Manipulating placement of these capacitors is the hard part (see picture 4).

If there's no room for any of the above, then I'll try to use at least a few smaller capacitors. However, going below 500uf is not worth it -- just not enough "value for effort". Even anything below 1000uf is marginal.

I hope this helps. I'm still learning, so please feel free to ask questions or provide suggestions.

Cheers,
Andry


Picture 1


Picture 2


Picture 3


Picture 4

Hi Gunnar, Andry,

Indeed, I have been using MTC pins 16 and 20 to connect my capacitors.

I have applied the picture 2 diagram shown by Andry, but also supercap circuits by Fischer-Modell, Lüssi or Train-O-Matic (SPP nano).

In older locos with 6090/60902 decoders, I derive U+ and GND from the 4 large diodes that form the bridge rectifier (also for providing a flicker-free function return (orange) for the lights, if they were still connected to brown.



I haven't worked on many locos with such a smoothing capacitor, but in the one I found (36198 Vectron), it was directly wired to pins 16 and 20 of the MTC, so a good spot to connect my supercap circuit in parallel.

However, I have learned that the buffer circuits don't work in some of my locos (some with 6090 or 1st generation mLD), and I can't figure out why.
The buffer circuit connections (taken as described from pins 16/20 or directly from the bridge rectifier) correctly provide +17V, but in those locos, neither a DIY circuit (picture 2) nor any of the commercially available boards seem to work.
Anyone with similar experience?

Hi Andry, thanks for excellent info. I find this very useful, maybe I should have spent some time playing with electrolytic capacitors instead of going straight for the supercap module 😊 Anyway, it’s already ordered, will get in a few days I guess. If it’s way too much capacitance, I will consider an electrolytic cap instead (or several smaller in parallel). But there’s a huge difference in capacitance on your two setups, for sure there must be some difference when running over turnouts or sections with poor connection, at low speed? I thought 1000-2000 µF would only help to avoid short dips and blinking lights, not for running the motor for more than a split second. A supercap setup will be quite different, probably running the motor for a few seconds. Am I right?
In picture 2 you have 5 caps in series, each 1F, giving a total capacitance of 0.2F. This is roughly 90 times more than the 2200µF electrolytic cap… Just saying. My module is 0.15F, almost the same as yours. I get a little higher charging voltage though, rectified +Ub is around 18.4V I think, and I won’t need the 15V Zener diode. Have you considered adding two more cells to take advantage of the full +Ub voltage? If space allows.
I think the 1N5818 Schottky diode could be slightly better, as it has a lower voltage drop and super fast switching. But not a big deal, the 1N4001 is also solid, and differences could be negligible.
Have you considered adding high ohmic balancing resistors to your supercap cells? Like a voltage divider circuit. These supercap cells are never 100% identical, and the total voltage does not necessarily distribute evenly across cells. Worst case is that a cell gets more voltage than the others, and gets strained /destroyed over time. Same thing with electric vehicles and charging of Li-ion batteries, they also need cell balancing like this. But there’s probably some wiggle room from nominal voltage, data sheet should say something about that. The supercap module I found has an integrated balancing circuit, so it seems very robust.
And yes, I agree my 3W charge resistor is probably oversized! It will see a 7W peak for a very short time when track is powered up and capacitor is charged from empty. But Watts go down fast, as cap is charged and voltage diff over resistor drops. So I could probably decrease to 1-2W without problems, just playing it very safe at first as I have not done this before and don’t want to fry anything. The resistor should not get very hot unless I have super dirty tracks with dead spots all over the place, so the stay alive circuit is working overtime!

Hi Leo
I think you are perfectly right. The inductive coil is used to separate digital commands (read&write) from the power line, and the capacitor is used to smooth the DC used for power. This is a typical setup when combining power and digital communication on same line (communication is "superimposed" on power line).
But the coils used on stay alive units I think is for a voltage step-up /booster, as these modules typically have a low voltage from supercapacitors that need to be transformed up to correct voltage level for decoder&loco.
The booster probably have switching transistors that drives the primary side of coil, would be my guess.