what is needed to construct a current keeper for 2 rail O scale motors?

If you go to the OGR Search page and type "keepalive" into the search string, you will get many hits for posts on 'O' scale sized keep alives that have been posted across various DCC and DCS control forums over the years.

For example, this thread has an example circuit.

Keepalives, O-Sized | O Gauge Railroading On Line Forum (ogaugerr.com)

Scott

Hello Scott,

Thanks for your reply but that is exactly where research led me. The applications were for other than what I was looking for. For example the booster and decoders I am not familiar with all of that. An HO current keeper is five 3 volt capacitors in series (15v) with an diode (Unknown) a resistor (unknown)and a zener diode (unknown). I don't know their values but I believe the same process could be used to keep a decoder with an O scale larger motor going for 3-10 seconds. That is what I was hoping some of the electronic guys may have already cooked up or have a schematic with values so I can get some Moser parts and start building. Soundtraxx told me I could put two of their current keepers in parallel and that helped but only about a two second run. That is an expensive way to go with 50+ engines. The four axle truck engines just don't operate with the current keeper due to the larger motor current draw. I am surprised that no one has fielded a current keeper for 2 rail O scale and that includes my 2 rail MTH engines which also suffer dirty wheel and dead frog stoppages. Thanks again for your reply!

First, ditch those five 3-volt caps and get one 25 or even better, a 50 volt one that has the largest mf rating that will fit in your application. I prefer a cap that is rated at least twice the maximum circuit voltage.

Dave,

Thank you for your reply! I would do that but I need to know what size resistor, diode and zener diode to use with it. Are you able to advise on those or some of you electronic guys? I am a mechanical Engr (ie., I need to see it) the magic of the electronics world has been used by me successfully for years as a recipe only! Other than knowing what a diode does and a resistor, I have no clue what a zener diode does or how to size those three items to go with the 50v capacitor. I do however, know how to hook them up. Many thanks as I close in on this project.

Frank

I understand that the new LokSound Version 5 decoders come with a keep-alive feature.  Their advertising says that "a PowerPack installed directly on the decoder helps to bridge dirty rail sections."  While this falls short of the original posters design/build question for these devices, it may be useful to some here.  Like the original poster, I would also like to see one of our resident geeks provide a parts list and diagram here for do-it-yourself construction on existing DCC decoders for us 2-Rail operators.  Preferably oriented towards those of us who have trouble changing the batteries in a flashlight, but still manage to install DCC decoders in our locomotives.

I don't think you are grasping how these keep alives are designed and what the trade-offs are.  First, these keep alives are made with super capacitors.  Super capacitors have very high storage capacity, in 1F, 2.5F, and 5 Farad values (not microfarads like a typical electrolytic capacitor is rated for but rather full farad values).  However, these super capacitors used in the keep alives do not have high voltage ratings.  Many have a voltage rating of only around 5.x volts.   Now.. you can put these in series to raise the overall voltage; however, the overall capacitance value goes down.  Capacitors in series are calculated as 1/C = 1/C + 1/C + 1/C ....   You put them in parallel to add capacitance but then the rated voltage stays the same.

Here is a sample using three 2.5 Farad 5-volt capacitors in series in order to get a 15-volt super capacitor:

Note:  The physical size of each of these capacitors are 3/4" x 3/4" x 3/8" (and now stack 3 of these together for 15-volt value - that is huge to cram into a loco shell or tender body).

1/C = 1/2.5F + 1/2.5F + 1/2.5F

1/C = 0.4 + 0.4 + 0.4

1/C = 1.2

C = 0.8333 Farads @15 volts

Therefore, you can see why there are none of these types of stacked super capacitor circuits for O scale out there.  You only get 0.8333 Farads of capacitance which won't even give you 0.2 seconds of power even though you use three huge super capacitors.  When they are bundled together, they are huge, and you don't even get the equivalent of what is available in a stock ESU Loksound decoder.  I think they have around a 1 Farad onboard super capacitor in ESU Loksound V5 L.

Therefore, in order to get longer run times with higher capacitance values and maintain a manageable package size, then circuits with voltage boosters like the ones provided in the link of my previous post above (repeated again below) are used.  This link has designs and parts lists for both ESU Loksound 3-wire models and 2-wire, Soundtraxx and MTH PS3 versions but uses voltage booster circuits to boost and reduce voltages at the right places.   According to the test data in the link, you can get up to 6 seconds of runtime with the bigger 5 Farad capacitors.

Keepalives, O-Sized | O Gauge Railroading On Line Forum (ogaugerr.com)

The resistor (56 ohms 0.5 watt) used in the above circuit is to limit the charge rate of the capacitor.  If you have too low of a resistor value or no resistor at all, then the charge rate of the capacitor will appear as a short to the DCC booster and trip the DCC circuit breaker due to the high inrush current, especially when the capacitor is fully discharged.  If the resistance is too high, then it will take too long to charge the capacitor after a discharge occurs during a power interruption.  Therefore, a happy medium needs to be selected.

The zener diode acts like voltage regulator.  Think of zener diode as dam holding back water to regulate the water level in front of the dam at the absolute maximum (pretend in preparation for a drought).  However, like a dam, it has a finite limit.  When the water level exceeds the height of the dam, then the water flows over the dam, so theoretically the water level will never be higher than the height of dam (by any appreciable amount).  In the case of the zener diode, if the voltage exceeds the rated value of the zener, it flows backwards through diode to ground (e.g. or over the dam) and essentially wasted energy.  If the capacitor charge voltage is much higher than the zener rating, you waste a lot of power.  Think of a heavy rainstorm and lots of water flowing over the dam.  Therefore, it is important to rate your zener value safely below the super capacitor limit to protect the capacitor from an overvoltage condition but also not so low that you waste a lot of power when the capacitor is fully charged, and you are just dumping excess voltage (current - which could be used to run your trains) to ground in a steady state condition backwards through the zener.  In the dam example, you would always want a just little water going over the dam so you know that the water level is regulated at the very top of the dam at all times.

The regular diode in the Soundtraxx current keeper designs is used as a one-way valve to allow current to flow out of the capacitor to power the decoder in the event of a power interruption but not allow power to flow backwards into the capacitor when power is restored.  We want to limit the inrush current to not trip the DCC circuit breaker, so we want the capacitor charging to take place through the separate 56-ohm 0.5 watt resistor path as noted above and not through the path that the capacitor discharges.

Therefore, go back re-read the thread in the link above and see if becomes clearer why this type of circuit is used in O scale rather than just stacking a bunch of super capacitors in series and adding a resistor and some diodes that would only yield an inferior keep alive in the end.

Scott

Hi Scott,

Many thanks for that great description of the components of a current keeper and for explaining a super cap vs a regular cap. I printed it as a keeper! I went back to the referenced link for building a current keeper but I have a few questions: I am using QSI decoders that have 5v and 12v functions, So for those would I construct the 5v or 12 volt version? For Soundtraxx decoders of 2A or 4 amp that have a socket for a keep alive. Could I hook up the version where you don't have 5 V to the current keeper socket if I added a plug instead of ground and U+? For the Pololu boost converter there were several versions. Is the one I need the #799 and if so where do I set the potentiometer that ranges to 25 volts? I use the 5 Amp NCE power and remote that I think is 14 v but less to the track. Is that diagram showing two capacitors one being a regular 10 uF capacitor and the other a super capacitor? I saw in there that a motor draws .8 Amp. I have 2-3 inch pitman motors and I thought they draw as much as 2 Amps? Does that and stall current have anything to do with using these current keepers with boosters and  5 volt step down boards?

Thanks again,

Frank

There is a 3-wire and 2-wire version.  The 3-wire version is for DCC decoders that have a 5V tap on the decoder, such as ESU Loksound decoders (or possibly QSI decoders - I have not worked with these decoders personally so I don't know what they have for available voltage taps).  The 2-wire version is for DCC Decoders that do not have a 5 Volt tap on the decoder.

The reason for the difference is because when you look at the schematics, the supercapacitor used here is only rated at 5.4V and protected by a 5.1V zener diode.  Therefore, with the 3-wire version you can use the 5V tap directly from decoder models that have a 5V tap to handle the charging function of the capacitor through the 56 Ohm charge-limiting resistor to the '+' supercapacitor lead and still stay below the supercapacitor voltage rating of 5.4V max.  The 5V tap on the decoder is only used for handling charging the 5.4V supercapacitor while the train is running.  The DCC decoder is not being powered through the 5V lead by the supercapacitor.  The 5V is only output from the decoder, not a decoder input.  When the decoder sees a power interruption, then the power will dump into the decoder through the Pololu #799 and into the U+ and GND connections on the decoder.  Once the power is inside the decoder, the internal voltage regulator inside the decoder will supply 5V to the internal 5V rail in the decoder for decoder functions that use 5V.  Also, the 56 ohm charge-limiting resistor will act as a barrier for backfeeding to the decoder 5V rail.  Current takes the least path of resistance and in this case  the least resistance will be through the Pololu #799 and on to U+ at the DCC decoder.

For Soundtraxx decoders that only have 2-wire current keeper plug, then you need to use the 2-wire design that has the extra Pololu #2843 5V step down regulator plus the Pololu #799 Adjustable Boost Regulator 4-25V.  This extra #2843 regulator is needed because the supercapacitor is only rated at 5.4V and cannot handle the direct U+ (12-14 volts) coming from the decoder during the capacitor charging phase.  It must be stepped down to 5V first before going to the 56 ohm charge-limiting resistor and then on to the 5.4V supercapacitor.   The Pololu #2843 handles stepping the voltage down to a safe 5 volt value before entering the supercapacitor as stored energy.

You set the potentiometer on the Pololu #799 equal to the decoder U+ voltage when the decoder is powered up on live  tracks.  You will need to measure this U+ voltage on the decoder with a volt meter to know where to set the Pololu #799 potentiometer.   The Pololu #799 voltage should be set to the same value you read for the U+ voltage using a volt meter, as well.  Be aware that the U+ value may be slightly different on different brands of DCC decoders so the potentiometer may require a slightly different setting for different decoder brands.  The U+ voltage will probably be somewhere around 12-14 volts depending on your track voltage.  Read the full web site for the Pololu #799 as it covers how to set the output voltage.  Also read the entire thread to these 2-wire and 3-wire current keeper circuit as there are comments in thread that cover setting the #799 trim potentiometer.

The 10 uF capacitor is just a small filter capacitor to smooth out the ripple in the voltage feeding the decoder from this current keeper circuit during a power interruption.  The Pololu #799 Adjustable Boost Regulator is a switching-style step-up regulator from the 5.4V supercapacitor rating to the ~12-14 Volt U+ decoder tap during a power interruption and these switching-style regulators commonly add some high-frequncy noise (ripple) in the output voltage during the step-up process.  This small 10uF capacitor just filters out that noise so the voltage is nice and smooth once it enters the decoder through the U+ connection.   The microcontroller in the decoder is basically a small computer so it is sensitive to voltage noise just like any big computer.

The heavier draw motors, such as the Pittmans, will just draw down the capacitor that much faster and give less runtime during a power interruption but it does not impact this circuit design.  These 2-wire and 3-wire circuits will still work just fine for the larger motors too.  However, for these larger motors, you might want to use the larger 5 Farad capacitor if you can fit them in your installation so you have the most stored energy during the power interruption event.   However, the 2.5 Farad capacitor version might be just fine for the types of power interrupts that you see on on your layout so you just have to do some experimenting to see what works best for you.  Also, the 5 Farad capacitor will take longer to fully charge and may not be as responsive on occasions where power interruptions occur in quick succession.  It will take some trial and error to see what capacitor values work best for your layout and locomotive types.

Scott

Scott,

Many thanks for taking the time to prepare your very clear, detailed, quality  information on the purpose and what the components of current keepers do -another keeper! I have to believe you are an electronic or electrical engineer and perhaps professor?

So I believe for small/ medium O scale motors  I would try the 2.5F super cap and 5F for the large motors.

What guage wire do you recommend for construction 18?

My layout has a couple of blocks were voltage drops occur and engines slow down. But I presume I need to find the highest voltage at the rails for building this current keeper?

I believe these voltage drops occur in blocks that have 8 long track yards connected where the positive rail of yard tracks were gapped with an on- off switch but the negative rails of the yard tracks were not. So, am I correct that all of the accumulation of negative rails causes a lot of resistance in the block and causes a voltage drop? If so I probably need to gap the negative yard tracks and use double pole on -off switches on each yard track so I have consistent running voltage. I would like your opinion on this. Thanks again Scott!

Frank

Hi Frank,

For wiring up the current keeper itself.  You can use smaller wire since the wire lengths are real short.  Therefore, 22 or 24 gauge wire is fine for making up the current keeper and connecting it to the decoder.  Wire is rated with a resistance-per-foot value, therefore, the shorter the distance, then the less resistance you encounter and the more current a smaller wire can carry without seeing any appreciable voltage drop.

As for the your layout voltage drop issue, with straight DC layouts, voltage drop is impacted by one thing, resistance.  Either the main bus wire feeding this section of track is too small for the current load, not enough feeder wires are present and you are expecting the rail joiners to carry the current between sections of flex track that don't have feeder wirers attached or a combination of both.  If you can test with straight DC (not DCC) and you still see a voltage drop, then you know you have some type of resistance issue, most likely main bus wire size is too small.  For long O scale bus wire runs (e.g. greater than 30 ft.), you should use at least 10 AWG wire.  Shorter DCC bus wire runs can get away with 12 AWG wire size.

However, with DCC layouts, you can encounter a second form of "resistance" that impacts voltage drop that is not seen to the same extent on straight DC applications.  This "resistance" is due to the high bi-polar frequency of the DCC signal in parallel wires spaced a few inches apart.  When you have long parallel runs (for distances greater than 30 ft.) of north-rail and south-rail bus wires spaced a few inches apart like we commonly implement in our DCC layouts since it makes a convenient method to attach track feeder wire drops to the bus wire, then we can encounter something called parallel wire inductance.  Parallel wire inductance is a type of AC resistance that will produce a voltage drop in these long parallel runs even if the wire size is adequate for the required current load.  However, there is a way to negate the effects of this inductance and that is to put a twist in these long parallel north-rail and south-rail bus wire runs a few times per foot (see video below) to cancel out the magnetic flux that gets created and causes this "resistance" in these long parallel runs.  Adding these twists every so often obviously makes it a lot less convenient to attach track feeder wires, which is why a lot of modelers do not do this, but the alternative is to only run short lengths of closely spaced parallel bus wire or use outlandishly large bus wires to overcome the effects of parallel wire inductance.  If you apply the wire twisting cancellation method, then you are only left with pure wire resistance as the only other possible cause for voltage drop, just like on a straight DC layout.

Here is a good You Tube video on the topic:

https://youtu.be/N_sm6ttJDjE

There are also links in the video description for deeper dives into this inductance topic, if you are interested.

Scott

Hi Scott,

Thanks for another great reply!. Attached photos of my parallel wires. The whole underneath of the layout is a wiring rats nest like these. Where I used a single push button to throw all switches for each yard track (chains of switches with lots of diodes and resistors) that became an eagles nest! My rail joints were mostly soldered. I used 14 guage wire for the common of the main and yard tracks.  The positive blocks of the main and and yard rails are gapped with on-off switches and16 guage wire. The blocks where resistance is high occurs where there are 6 to 8 yard tracks, with the leads and yard tracks about 30Ft long and jumper wired common across to each track! So my common rail wiring is excessively longer than 35' for # 10 wire and worst is I used #14 wire with parallel wires to try to keep it neat, which did not turn out neat at all! Strikes me that I need to at least break up the very long common. Perhaps I could gap the main common from the yards and run a separate common to the yard track which jumpers x 6 or 7 parallel tracks. Only one track is ever used at a time in the yards. Or perhaps add another common 14 guage wire to the block main (where I notice a voltage drop). aybe on the Maybe on the opposite end? As you can see from the pictures- too late to add a twist for an 82 year old- too much trouble! I'll check out the video. Thanks again!

Frank

Hi,

Hope you do not mind my reply on this old post. These questions may help someone else looking at this in the future.

This post referenced thor's keepalive post. That circuit looks easy enough, but the Pololu #799 converter is nearing end of life and is $15.  I see another converter called MT3608 that seems to do the same thing for ~$1, albeit larger.

Has anyone used the MT3608 for this purpose?

How does one know if a capacitor is needed on the converter output?

If it is needed, is 10uF always the right value?

If it is needed, does it matter what style? (Electrolytic, tantrum, etc.)

Is there another equally cheap converter that is better? Sometimes things like this are too good to be true, but the only downside I see is the size.

Thank you for the help.