Making trains run smooth with capacitors (electronic-flywheel)

[kvp]

The small capacitors on the motors are for noise filtering, so the motor noise won't disturb analog radio and tv broadcasts.

 

If you follow the description above, then for keeping a motor running for a whole second, you would need for a typical 250 mA motor 250000 micro farads. A typical 2 cm long, 1 cm thick 16V capacitor (usable for 12V analog systems) has about 250 micro farads capacity. This means you'll have to put 1000 capacitors in your train, so by completly filling the inside, it would be about 2 meters long. This would give your locomotive 1 second of power, assuming every capactior is connected to the locomotive. For half a second, you only need a 1 meter long chain of capacitors, which is doable but not really useful and looks very ugly.

 

Supercapacitors are better, you can even get 250000 micro farads large ones, but they are bulky, so can't really be used in N scale trains. And you still need about 2 or 3 5V ones in series to supply your train with 10-12 volts, so it's still a challenge even in H0.

 

A typical small flywheel can store more energy, but it would still only work for about a second or so. The real problem is not storing enough energy, but the high friction loss nature of the drive system. If the drive train is suddenly disconnected from the wheels and all the wheels are left to roll on their own, the stopping distance can go up to several decimeters, the same distance we see when pushing the unpowered cars by hand. It's usable with any locomotive that could roll freely when pushed by hand. Most N scale ones don't.

 

A good solution could be to use small lithium polymer batteries, that can supply 3.7V (you need at least two of these) with a capacity of 100 mAh or 100 mA for an hour, that means around 20 minutes of running at 250 mA. They are not capacitors, so they have to be charged and discharged in a controlled manner, so you would need a charge/discharge circuit, usually with a digital controller. This circuit can be automated (to just stop the train after some time) or remote controlled through radio or infrared. The latter is used for small rail-less remote controllable toys, even in N scale. Running time is usually around 2 to 5 minutes with comparable charging time.

 

As you go up in size, the possibilities are incrasing, so when you get to LEGO size, you have the option to use off the shelf LiPo batteries, that can power your train up to two hours. This battery is slightly larger than a H0 scale locomotive, but for 0 scale (LEGO scale) that is perfect since it fills the inside of a typical locomotive quite nicely. I have several 1:42 scale LEGO locomotives that run from rechargable batteries and are controlled through infrared.

 

In short, simply by adding capacitors into an N scale train, you can't get much running time after power loss, but by adding a digital control circuit and a LiPo battery, you can make them run quite nicely even without track power. Some commercial trains, like maerklin's H0 toy trains for children do exactly this. They are infrared controlled and run from batteries. Imho the same can now be done in N scale too. (the same small batteries that are used for the Tomytec moving bus system can be installed into Tomytec train collection trains very easily, just remember that they are not rechargable and all controllers must be disconnected from the layout when running a train from battery)

[HantuBlauLOL]

I see.. Btw why Kato locos seems to be very smooth stop even with just 2 flywheels? My Tomix trains fitted with 2 flywheels Doesn't seems to stop that smooth..

 

Using RC seems to be a good idea. 2 channel receiver is small enough to be fitted in N.

 

Small non chargeable batteries isn't effective imho.. Anyway a Li-ion rechargeable batteries are small enough, but still see able in multiple units..

[cteno4]

There is a train rc standard that I think is in the works (I think nmranet/openlcb had one as well) and there was one company with an HO system for sale and of course the larger gauges. Batteries are still not quite to n scale needs for any extended use if you want it all in one car. But using the tomix power couplers multi unit consists could allow for current Li batteries to be distributed in a train and since they are usually made in flat wafers could be kept out of the windows. But adding all this quickly adds up in the price of each train, but at three decodes needed for a bound consist that puts you at $50 right there...

 

One chap had posted a couple of years ago on his use of banks of tantalum capacitors in his locos to give and electronic flywheel for short interruptions and also to smooth the power to his in engine cameras. Sorry I could not dig it up in the forum archives, I'll try again later, he had a lot of nice pictures of his work. He packed them into every spare little space in the loco. But this was only good for split second hiccups.

 

Cheers

 

Jeff

[cteno4]

Maddening! I can't find those posts from the chap using big tantalum cap banks in his locos! I think it was mainly euro models he was doing and I'm almost certain it was here he posted about them and his loco mounted cameras. Anyone remeber or can find these posts?

 

I've often wanted to see if there is a way to mine really good threads for the main info to store/post for easier location, frustrating when good stuff gets buried in the forum.

 

Jeff

[cteno4]

Found it it was vadim's work. Here is a post of his cap bank for his camera train to smooth out the power.

 

http://www.jnsforum.com/community/topic/6745-idea-e-guided-loco-decoupler-for-standart-knuckle-couplingsshunting-with-cam/?p=75612

 

Jeff

[kvp]

That is 12 x 100 uF capacitors, around 1200 uF. It's good for a low power consumption camera, but can't smooth out much for a motor. (for 1 second of unpowered motor operation, you have to use 200 times as much) But smoothing caps are used for decoders, mostly to keep the decoder loosing its state while the flywheel moves the locomotive over a pickup problem. On the other hand, those small cameras are interesting.

[ToniBabelony]

To revive this topic. I have finally started to experiment with capacitors/condensers as flywheels. I have managed to cram two 0.1F 5.5V supercapacitors into an N Narrow body, combined with a Kato B-train motor:

 

  

 

The capacitors are in parallel, providing a capacitance of 0.2F, which is quite a lot for a little motor like this. However, I'm a total electronic noob when it comes to calculations, but it seems the capacitors need a few seconds to charge (maybe about five or six), to see some significant discharging. The distance ranges from about 1 cm up to about 15 cm, according to the charge and the voltage. 

 

I haven't added a diode yet, but will do so for proper running, so reversing can be done safely (without charging the super capacitors) as well.

 

This is a good solution for travelling over points and dirty sections with some wagons attached, but not really for realistic running properties. I'm thinking about throwing a bunch of them in a 2-car Tomytec train in the unpowered car for some fun later. If that is a success, I'll make it a standard for my models, since I have bought 200 of these supercapacitors...

[cteno4]

Toni,

 

Sounds good. About what you would expect for the super caps and the pocket motor. Sounds like it's plenty to get the small wheelbase 2 axle mechs to ru across dirty patches and thru points better!

 

200! You must have gotten a great deal! I usually see those at a buck or more apiece!

 

You could run lines from the other cars to the engine and use the tiny 1x2mm magnets to connect the wires.

 

Cheers

 

Jeff

[ToniBabelony]

Hi Jeff,

 

The unit needs to run a few rounds to get it to charge up plenty, which is a downside of a high capacitance, but since this isn't intended to do switching or short runs, this isn't an issue for me. It's fun to hear the motor rev up a little on a 'cold start'.

 

I managed to found a bag for extremely cheap from some chap on Yahoo! Auctions, so if anyone needs a few, I think I can spare some! Electronic components are ridiculously cheap here, since they have a massive market and are made locally (either around the corner (sometimes almost literally) or in a neighbouring country).

 

The idea with the magnets is a great one! I'll certainly try to have a system of wires and all-wheel pickup on all cars with capacitors where possible. This however defeats the purpose of the capacitors as safeguards for points and dirty spots though, but more capacitance and more electrical contacts is better in every way. Make every train an EMU!

[cteno4]

Yes a super cap will need charging! I'm tempted to try a couple to just see how it runs!

 

That's great they are so inexpensive in Japan, so little of that sort gets out much on ebay or online sources outside Japan.

 

Caps plus more contacts, good! Maybe you can load in enough caps to get your full flywheel action.

 

Jeff

[kvp]

Very large caps have a large inrush current and can overload or trip the controller if there is no current limit on them or between the controller and the tracks. Modern controllers do have this and this might be the reason for the slow chargeup. The caps are not CL compatible though and if you add the protection diode, you won't be able to reverse at all. The latter is the reason supercaps are best used with full bridge lighting circuit or as buffers behind a DCC decoder. (both have and need current limiters and diode bridges)

[ToniBabelony]

Caps plus more contacts, good! Maybe you can load in enough caps to get your full flywheel action.

 

Yes, but more caps also means a longer charging time, plus it only works when winding down. I'll go ahead anyway, because... I can! xD

 

The caps are not CL compatible though and if you add the protection diode, you won't be able to reverse at all.

 

The diode will not matter in regards to running direction when you think beyond a very simplistic circuit. The diode will only be there to protect the capacitors. If you had read this thread, you probably have come across this post from the previous page for the inspiration of the circuit I want to recreate. This circuit will also redirect an overflow to the motor, rather than back to the controller. Reversing will not be affected in this circuit.

[kvp]

In the mentioned circuit the leds are only on in one direction only. If you add a diode to the capacitor only, it will be unable to discharge. If you allow current back towards the track, then it will allow reverse charging. A bidirectional circuit can only be made with polarity free caps or an additional direction follower circuit. (DCC decoders in analog conversion mode use the latter)

 

PS: its possible to make a polarity free cap from two elcos if you connect them at their ground terminals and connect power leads to the supply sides, but this decreases the value of a single one. (but makes them ac compatible)

[inobu]

Well, here goes......

 

The capacitance is not really addressing the smoothness. It is the number of poles or armature that creates the smoothness especially at slow speeds. The magnetic field is what turns the armature core/rod. Increasing the number of armature is what eliminates the jerking motion. That motion is caused by the armature passing the magnet field.

 

For example two ( - ) armature verses (+) 4

 

A capacitor can act as an UPS (uninterrupted power source) when the wheels lose contact with the track but the functionality of the DC motor will remain the same based on its engineered design. Along those same lines the capacitor mitigates the CL function as it merely interrupts power long enough to make light visible to human sight but short enough not to drive the magnetic field in the DC motors core. The capacitor will help by filling in the so called power gaps.

 

The capacitance value should be calculated to fill in the duration of the suspected power loss. (Be it track bump or lighting function)

 

 

Inobu

Edited December 14, 2015 by inobu

[ToniBabelony]

Thank you Inobu for your insight. Capacitors don't bring in much smoother running of the motor, but smoother running of the train over unpowered/dirty sections indeed. Especially with short wheelbase vehicles with limited axle numbers this is an issue, so hence the capacitors.

 

The best solutions for slow running capabilities are big gear transmissions, a mechanical flywheel and a good motor. Capacitors are fun to experiment with and I'd love to find cheap/affordable and small 5-polar motors or better (IIRC the motors in Kato's B-train power units are 3-polar), but if all these things aren't easy to find/available, resorting to these solutions (and experimenting with them) is an option (IMHO).

 

Another upside of the capacitors, next to the extra power, is that they add weight to small vehicles. Double functionality! :)

[ToniBabelony]

A very quick video with the effects of supercapacitors only visible when charging them a bit (see the end of the video). Having a train run for a longer period will also achieve this, but it will take about ten rounds or so at ±4V to get a noticeable result. Will do that later.

 

[inobu]

Ok, I see. I was scratching my head on the smoothness but I get your point.

 

Inobu

[kvp]

Something is not right with your power source and capacitor combination. The caps should charge immediately, effectively taking all the current from the motor until they are charged. This gives a slow start, slow stop pattern. Unless the caps are current limited, which would make them useless during discharge.

 

(charge and discharge times should be the same, discharge time matching the length of travel after power source disconnect)

[ToniBabelony]

This is just deducted from my own observations and my own logic, which might not comply with theory and/or other people's workings of their brain, but I came to this conclusion:

 

Charging the capacitors takes time, because it has a huge capacity (0.2F, or 200,000µF) and is provided with a low voltage. Next to that, a motor is running which has a high resistance and thus taking up a big percentage of the current flow. This makes charging the capacitors go very slow and thus over a short amount of time, not much is stored to be actually functional when discharging. The charging also doesn't affect the performance of the motor much when running.

 

Discharging the capacitors over the motor, which is basically a resistor with a gigantic resistance, goes relatively quick opposed to charging. The motor has multiple points where this resistance is generated: 1) the turning of the poles; 2) spinning of the gears; 3) rolling resistance of the complete unit. After coming to a halt, this results in a ±0.3V capacitor charge (granted, a charge isn't calculated in volts, but I'm a noob, so please bear with me) which is not useful for the motor to turn any more. The leftover charge however I call a 'warm start', as this is a charge that doesn't have to be done any more when starting up again within a certain amount of time (the capacitors discharge slowly over a few seconds after reaching that spare charge).

 

Here is a video of the unit running on a 'cold start' (no initial charge from the capacitor) for about one minute on 4V. Stopping @1:05 with braking.

 

 

The electronic diagram is as follows:

 

 

I'm not sure if making it like this will make any difference, but I doubt it. It's been over a decade and a half since I had my last class in electronics:

 

 

---

 

P.s. I'm considering switching the parallel capacitors to a series, doubling the voltage, but cutting the capacity roughly to 0.05F, since I have the feeling the capacitors won't get a full charge on short runs anyway. Nope. That wasn't a good idea. More capacity is better than a higher max. voltage in this regard.

Edited December 16, 2015 by Toni Babelony

[kvp]

Uncharged caps have near zero resistance, so a high resistance motor in parallel should not recive much current until the caps are charged. So caps first then the motor.

 

If the motor is a very low resistance variant, and could eat up the power from the caps almost as fast as the controller can supply it, which would mean the controller is not powerful enough. (and you are also trashing the motor btw) This would give a slow charge as only the surplus power is stored. The discharge time will be rather short compared to the charge time though.

 

Also there is no such thing as motor that disconnects itself at a certain voltage. It has to drain the caps to zero even when it couldn't move anymore. (just converts the remaining charge to heat)

 

Have you tried physically disconnecting the controller wires insead of turning it off? Also could you measure the motor coil resistance with a multimeter?

[ToniBabelony]

Yeah, in theory it might be like that, but in practice it does not. You've read my posts, seen my videos and then decided to ignore everything.

 

Look. You're addressing problems and not offering help, nor suggesting solutions or improvements, nor giving solutions. So, I'll just move along and try to solve it on my own.

 

The capacitors are probably damaged from either soldering or something else, so I'll do the whole setup again with fresh components and diodes (that came in today).

[inobu]

I see what you are aiming for but once again feasibility and practicality will add into the final equations. There are a few factors that you should take into consideration (the theory that kvp mentioned) that causes problem for the proposed solution. 

 

Your test case in the video is not the same conditions in which you are trying to address.

 

The actual condition is the applied voltage (as far as the motor is concerned) is going to Zero and quickly returning to the applied voltage level. This voltage drop is created by a gap in the track or anomaly where contact is lost momentarily between the wheels and track. Your proposed solution is for the capacitor to replace the loss of applied power by discharging it until till the applied power is returned.

 

Problem

1. Input voltage is a variable.
2. Problem requirement is to sustain existing applied voltage level (decay of charge in a capacitor is similar to the decay of a radioactive nuclide) it will always decline to zero causing a jolt
3. Two consecutive dropout would not allow sufficient time for recharge due to circuit configuration negating the proposed solution.

When you look at these issues the feasibility declines and the practicality in creating a resolution exceeds the cost of redesign.

 

The simple solution is to increase the quality of the track bed construction and rail joiners than to design a controllable/variable UPS system on the train itself.

 

Inobu   

Edited December 17, 2015 by inobu

[ToniBabelony]

I see. That makes sense. It's too bad capacitors don't work with variable voltage too well. The best would then be to invest in improving the ride quality of the unit by adding more track pickups and/or getting a better motor, rather than adding electronic components for that cause.

 

For now, the system works fine for continuous running at a few volts in one direction, but not for switching (that would be a bit too much in N Narrow anyway) or short runs. It wasn't my intention to have it work in that regard, so I'm happy with how it functions at the moment.

 

I'll be out for a few weeks and after that I'm considering getting a bigger variety of electronic components (resistors, (button) LEDs, etc.) to play with and understand the workings of (super)capacitors a bit better. I'm sure they have a function as either electronic flywheels for motors or front/interior lights.