DCC Stay Alive Capacitors
I had an interesting conversation with Jens Emmermann at the 2mm Expo in Oxford during June 2013. Jens had various locos with him, some sound fitted, and some with additional Stay Alive components.
A Stay Alive circuit is a means of keeping power to a DCC decoder and the model's motor when pickup is lost. Most models with a small number of wheels suffer from momentary loss of pickup, from the tiniest specs of dirt, slight variations in track, variations in pickup. There are numerous mechanical techniques which reduce this problem; split frame, electrical wiper springs ("Simpson Springs"), compensation, pickup on every axle. But the problem still persists, and is worst in small locos with fewer pickups. Drastic solutions include wiper skids underneath a loco which drag on the track. The Stay Alive circuit is another option for DCC users, offering energy for a fraction of a second's interruption to power. That fraction of a second is sufficient to remove the need to use "finger prod" and "layout thump" to get a small loco moving again.
When designing a Stay Alive unit for a small loco, there are two main issues. Finding physically small capacitors with sufficient energy storage and attaching wires to the relevant places on a DCC decoder. The information for both of these is available on the 1001-digital, a website run by Carsten Berger. There are small 100uF capacitors, 2.6 x 2.6 x 3.4mm, and diagrams of various small decoders showing where to attach wires.
This circuit assumes a DCC system with a track voltage less than 16v. Many commercial systems are less than this figure, being around 13.5 to 14v. But there are a few with higher voltages (eg. Bachmann EZ, Roco Multimaus), and those might overload the circuit.
The design of a Stay Alive circuit is simple, a capacitor (or several arranged in parallel) to store energy, a resistor (around 50 to 100 ohms) to limit current when charging the capacitor, a diode to bypass the resistor when energy is drawn from the capacitor. With the small ceramic capacitors, a 16v Zener diode is also added to prevent over-voltage being applied to the capacitors.
Schematic for Stay Alive circuit. More capacitors can be added in parallel with the three shown.
My test loco was my class 02 shunter. This is small, has 4 wheels, and as such does sometimes stall on turnouts or track which is not absolutely perfect. The decoder sits above the motor, in the top of the bonnet. The loco has a little bit of space in the cabside, below the windows, which had held a lump of lead. As the chassis block is a piece of 1/4 inch solid brass bar, with only minimal milled space for gears and axles, the weight is quite high for the size. The space below the window is 8 x 8 x 4mm.
A rectangle of nine 100uF ceramic capacitors was assembled on a piece of very thin PCB. Because of the available space, it was better to stand the capacitors on end. The capacitors are not polarity sensitive (many other types of capacitor are polarity sensitive). The capacitors were added one row at a time, using a normal soldering iron to melt the solder below the capacitors, then fit the next row. Then the upper strand of wire was added to help join the upper pole. The final length of the PCB was trimmed after all the capacitors were fitted.
I fitted the resistor, bypass diode and Zener diode to a separate PCB,
with wires to the capacitor bank.
Block of nine capacitors, totalling 900uF, attached to second circuit board with charging resistor and bypass diode on top side, voltage limiting Zener on the underside.
The hard bit was connecting to the decoder. This was worse because I'd used a CT DCX75, which is probably the hardest of the small CT decoders to attach stay-alive wires. The newer DCX76/77 models appear to have clearer solder pads. Photographs for identifying the solder pads are on the 1001-Digital website (link below). This photograph shows the connection points on the DCX75:
The insulating film on the rear of the decoder was cut away around the solder pads. The positive is fairly easy to attach a wire, its large and at the edge of the decoder. The negative is more difficult, being small and in the centre. I took an old soldering iron tip and filed an extra-fine point on it for this job, the wire was tinned, cut back to have the minimum ahead of the insulation, and then carefully fixed down with the iron. I needed a lot of magnification for this job. If I do it again, I'll consider replacing the decoder for one with larger solder pads !
Finally, all components were insulated with thin paper masking tape (Tamiya
brand), cut to size and wrapped around parts.
The end results are a superb, and really worth the effort. Previously this loco would stall from time to time, whereas it will now traverse complex pointwork at very low speeds without any hesitation. Stalling when coming to a stop has been eliminated. I estimate that the loco runs for around 0.2 seconds without power from the rails.
Class O2 fitted with stay-alive circuit. The charging resistor circuit is the yellow blob in the cab, this will move out of sight when final adjustments are finished, but there is a small weight to dig out from the loco body before it can sit fully into place.
Amendment - April 2017. Subsequent to this article, I've done further experiments with different capacitor types. The conclusion I reached is that a 220uF Tantalum capacitor offers better performance than a 100uF Ceramic capacitor, even though the calculated "capacitance per unit volume" (ie. What you can cram into a loco) is far higher for the Ceramic. The reason for this appears to be the performance of capacitors as they reach their rated voltages, the ceramic's performance tails off very badly, whereas the Tantalum has a nearly straight line. Tests in a loco showed a better run-time for the same unit volume of space for the Tantalums. Further details can be found on my blog - http://www.nigelcliffe.blogspot.co.uk
Carsten Berger's website with information on capacitors, solder pad locations
for CT and Zimo small decoders, and an online shop for the small electronic
components. Website is in German, but there are free online translation
tools for those who cannot read German.