A speed limiter for a Universal Motor in an old Lionel Train

[NTL2009]

TL;DR Summary

I'm looking for feedback on this before I get too far. I want to set a max speed limit on this model train to avoid derailing.

I have a "262 Lionel Line" O gauge train, from the early 1930's. It has sentimental value, my Father-In-Law (RIP) got it as a gift when he was about age 10. I've set it up each Christmas, a short track around the tree. But it is difficult to control, the gearing is high (it goes fast), so reducing power to it to reduce speed results in frequent stalling at some points. It's a very fine line between too fast and stalling (like keeping the 0-100 knob from maybe 28-32, it's that 'tight'). I'd like to be able to set it and let it run, but I often have to modulate the throttle to keep it in limits. Typically, I need to advance the knob to ~ 50~60 to start from a stop. Especially now that my two grandsons were thrilled with it this year, I'd like to be able to let them run it next year.

FYI, here's a coupe videos of this model:


Some details: The controller I'm using is a fairly modern one, it uses an SCR/TRIAC to chop the 60 Hz 20V peak AC wave, so under normal operation it does see a 20V peak, which I think is good for providing torque under partial power. The train draws ~ 1.0~1.3 Amp running, and draws ~ 2 Amp to start. I tried a motor controller that puts out a 25KHz, 0~100% duty cycle wave, connected to a 15V 7A DC supply. No real difference in operation. I had thought about a much lower frequency chopped wave, but I doubt that would be much different than a 60 Hz chop (which I guess is really 120 Hz for a Universal motor). I also tried a constant current supply, and that seemed about the same as well.

My plan: I don't think I can do much from the supply side alone. I think I need feedback from the train speed. I thought about feedback from the train back to the controller, but that got complex. Now I'm thinking about a local electronic switch in the locomotive. Something that could easily be un-done to restore the train to original condition, if needed. My idea is to add an optical detector on the locomotive, a convenient place is underside where a 'push rod' comes through. I could obtain a pulse for each wheel revolution. One rev of the drive wheel takes the train ~ 4.25" (~ 11cm). I measured my slowest/fastest practical speeds for one lap to be from ~ 15 inches/sec to 25 inches/second. So that would give me ~ 3 ~ 6 intervals detected per second, which should be enough for good response. The 3 pole motor turns about 5:1 with the drive wheels.

I really just want to limit the top speed. I think I can use two 555 timers such that I keep power switched ON to the motor when the interval between pulses from the detector is > X mSec, and I'd cut power (internally) if the interval is < X mSec. I could have a small pot on the locomotive to fine tune the top speed. I think this should be fairly simple, though I haven't detailed out the exact configuration yet. And I can full wave rectify the input so I'm just working with DC.

The advantages I see is that should be pretty easy to undo if desired, and it doesn't require a special/modified power supply. I can use that just as I do now, and adjust it to control acceleration - it just wouldn't have much effect once the train has exceeded a set speed (the train would start cutting power internally at that point).

 

[anorlunda]

Do you enjoy digital projects? This sounds like a fun Arduino type project.

 

[NTL2009]

I could do it with an Arduino, I've coded up a few projects with those, but those are bigger than the one or two 555 timers that I think I can do it with (need to actually design and test that, but I'm pretty sure that's easily done). I'd still need the optical detector and motor switch circuit, so I don't think that is my best option.

 

[anorlunda]

I bet you could get wireless comm so that the speed controller is not on board the train.

There are Arduino forums where you could post that idea and get lots of specific suggestions.

 

[hutchphd]

Why does the train stall at certain parts of the track? Are they uphill? It may be more important to improve the conductivity of the track

 

[NTL2009]

I bet you could get wireless comm so that the speed controller is not on board the train.

There are Arduino forums where you could post that idea and get lots of specific suggestions.

Yes, that was my original plan. I figured I'd use that optical detector to send IR 'beeps", or ultrasonic "beeps" at each interval, and detect those with the Arduino. Then all the 'smarts' would be on the controller side.

But I kind of like the idea of having this as a "limit switch" on-board. I can have a disable switch to bypass it, and it works as normal. That way, it's not dependent on two separate systems working together.

Why does the train stall at certain parts of the track? Are they uphill? It may be more important to improve the conductivity of the track

It's short track with tight curves ( 8 - 9" straghts, 8 - 10" 45° curves), so it is tougher pulling cars through those curves. And as I mentioned, they gear these things to go really fast. Which I don't understand - it's a kid's toy, it should be rugged, geared to overcome friction, curves, etc. If I had the mechanical ability, I'd really like to gear this thing down from the 5:1 ratio to maybe 20:1. If I run the controller above ~ 40%, it starts running away and will throw itself off the track. I'm no motor expert, but Universal Motors have series field windings, and I think that series windings tend to 'run away' RPM wise? That doesn't help.

A few years ago, I ran some 16 GA leads to the 'far end' of the track (which isn't very far!), to avoid IR loss on the steel/nickel track. That didn't seem to help much. This year, I soldered jumpers across the joints on each of the four sections I have mounted to boards. That did help, it is more consistent in speed across the lap, but it's still 'tweaky' to get the speed just right.

ETA: Not only that gearing, but in the box from my Father-In-Law was a rheostat controller. I can only imagine that putting resistance in series with a supply would make this issue much worse. These motors are going to draw more current under load, which would then drop the voltage, limiting the current further. I think the 1950's controller I had for the 1950's Lionel trains had a sliding contact that tapped progressively from the windings of the secondary, so that would have been a low impedance source, and I think would have worked far better than a resistor.

 

[hutchphd]

I think the 1950's controller I had for the 1950's Lionel trains had a sliding contact that tapped progressively from the windings of the secondary, so that would have been a low impedance source, and I think would have worked far better than a resistor

As a kid my American Flyer set had sliding pickup and my neighbor's O GAUGE Lionel set had a beautiful double slider toroidal wound transformer. The rheostat is not ideal but all the old singer sewing machines used them quite nicely and I think they were series wound motors.
Both train sets certainly went fast enough to fly off the track but you had to be nearly full throttle... Maybe 90% throttle...so I think the PWM set-up may somehow not be ideal. The American flyer put out 15 V rms max.
Yes the stators on the AC-DC motors are series wound with the commutators but I am unaware of runaway effect (but I'm no expert here).

I tried a motor controller that puts out a 25KHz, 0~100% duty cycle wave, connected to a 15V 7A DC supply.

Did you look at the voltage at the motor as you tried to vary the motor speed? I don't understand why this doesn't work unless the controller is not working as you surmise.

 

[NTL2009]

NTL2009 said:
I tried a motor controller that puts out a 25KHz, 0~100% duty cycle wave, connected to a 15V 7A DC supply.

Did you look at the voltage at the motor as you tried to vary the motor speed? I don't understand why this doesn't work unless the controller is not working as you surmise.

Oh it worked, the duty cycle varied just as expected on a 'scope. It's just that it didn't work noticeably different than straight DC (constant V or constant I), or the chopped AC from the TECH II O-gauge rated train controller I had.

 

[Tom.G]

Brainstorming here. How's this for a control approach?

• Startup: Current control, maybe Constant current up to some (predetermined) speed (as sensed by average voltage)
• Running: constant power; value set by the operator; with a hard voltage upper limit to avoid derailment

Cheers,
Tom

p.s. Hmm... sounds like a bench power supply with current limit and voltage limit

 

[Nik_2213]

Hi !

Been a while, but I built a 12 Volt PWM controller for some Hornby OO/HO. It was much, much better than the supplied rheostat, especially at low speeds, but still exasperatingly non-linear.

( Given magazine's circuit used germanium transistors, efficiencies were low and heat-sink remarkably large... )

Also, IIRC, the pulse frequency had to be tailored to the motor type. There was a broad 'sweet spot' based on the motor's induction, pole-count and maximum RPM. Too low, 'cogged'. Too high, eddy-current losses...

And, yes, a hulking great diode to catch back-EMF spikes, without which precaution, the circuit promptly died...

IIRC, there was a later up-grade using a 'make before break' multi-position wafer switch with resistance ladder determined by iteration to give more intuitive response. Think of it as a 'hardware look-up table'...

IIRC, analogy was made with eg trams or subway step-controllers which had 'experimentally determined' values of resistances or turns between the taps to give a 'more friendly' response.

FWIW, if you used a two-wafer switch, half the second's poles (minus one) could be wired to energise DPDT reversing relay...
==
Sorry, tad terse due two sub-editing cats. I think they want second-breakfast...

 

[Rive]

I could do it with an Arduino, I've coded up a few projects with those, but those are bigger than the one or two 555 timers...

I would try to look up similar (frequency-to-PWM) applications of some PIC microcontrollers. With some of those, it'll be smaller than an 555

 

[Dullard]

In some situations, 'brush noise' may be used as motor speed feedback.

 

[Tom.G]

In some situations, 'brush noise' may be used as motor speed feedback.

Much easier to just measure the back-EMF when the drive is disconnected, as between pulses if using a PWM drive. You have to wait for the switching transient to die down first.

 

[NTL2009]

Much easier to just measure the back-EMF when the drive is disconnected, as between pulses if using a PWM drive. You have to wait for the switching transient to die down first.

I'm generally familiar with this technique, but haven't found specifics on how to make or buy something that does this. The references I saw actually has some sort of sensor on the motor, I'm not sure how to use the back EMF. This is also a very noisy environment - the two pick-ups on the center rail, and wheels on the outer rails create a lot of intermittent contacts.

I took the first step on the logic with two 555 timers on the bench (w/o the power drive components or motor connected). Maybe I'm just getting old, but I scratched my head over coming up with something that would just cut power when the sensor sent pulses less than (for example) 300 mSec apart, and would pass full power if the pulses were grater than that time. And it has to default to the ON condition. A micro-controller is starting to look more attractive!

Tonight I came up with something that might work, hard to say w/o trying it. I wired a single 555 timer as a re-triggerable one-shot. So the output stays high (which I use to turn power OFF) if the pulse interval is less than one-shot timing (train is moving fast). But it also turns OFF for that amount of time after each pulse. The end result would be 0% Duty Cycle when train is running just fast enough to keep the 555 re-triggered, and if the speed drops to 2/3rd that rate, the DC would be 33% (the 555 times out during that 1/3rd extra time between pulses), and DC is 50% at 1/2 speed.

I'd need to check again, but I think it runs at a good speed at ~ 30% Duty cycle with a pulse power supply. So that might be a good sensitivity to keep it running at that speed that is ~ 2/3rd of my one-shot time - faster will start reducing the duty cycle. I guess I can't really tell if this is too far under damped, over damped, or Goldilocks.

 

[Nik_2213]

Best of luck !

I've tried using back-EMF sensing to regulate speed of big windscreen-wiper motors re-purposed to drive a wheeled robot's base, wasted several months. An optical feedback disk per motor would have been more productive. Happens that by using the motors' existing speed taps, I got enough control as-is...

 

[skeptic2]

Many years ago I worked for an employer who made radio controls for locomotives among other things and wanted an O gauge demo unit for shows. The main requirement was that it not lurch at start-up.

I ran a signal generator, set for square wave output, through an amplifier to power the train to find the frequency that would run the train at the lowest voltage. That turned out to be about 70 Hz. I created a 70 Hz sawtooth oscillator and created pulse width modulated pulses by feeding the sawtooth wave and a varying threshold voltage into a comparator. The output voltage was considerably higher than the minimum needed. The higher voltage and short pulse width at start up overcomes the breakaway friction and allows the locomotive to start slowly and the O gauge locomotive did indeed start very slowly. The maximum speed should be able to be set by limiting the voltage of the threshold.

Note: The speed of the locomotive is controlled by the voltage. The torque is controlled by the current available to the motor. To avoid stalling, make sure the controller can supply more than enough current needed for the worst parts of the track.

 

[Nik_2213]

"... That turned out to be about 70 Hz. ..."

Yay ! Sweet spot found !

 

[NTL2009]

Thanks to all for the input/feedback. I'm putting this project on hold, at least for now. A couple reasons:

A) After soldering all the joints on the track sections that I could, it ran a little smoother. I'm going to also solder jumpers about 1' long across the 4 sections where I break it down for storage, to help a bit more. Looks like that will be good enough (but boring!)

B) I took the engine apart so I could determine where I could break into the wiring, to make this local control, and one thing I suspected was true. The field coil frame is directly connected to the chassis, so I can't break into connect a FW-Bridge. I suppose I could put the power circuit in the bridge, but then the motor "moves" from high-side to low-side for each polarity shift, complicating the drive circuitry, and this is just getting too far beyond what I want to deal with. Or commit this to only working with a DC supply connected in a one polarity.

FYI, the 90 year old insulation just crumbled with every wire movement, I'm really surprised it hadn't shorted out somewhere already. Took some effort to rewire it, one wire snakes through the frame to a hidden connection to the pick-up rollers that takes major surgery to get to (many press fit parts). Fortunately, some helpful people at the ogrforum.ogaugerr.com (O-Gauge Forum) explained how I could break away the old insulation, and snake a piece of heat shrink over the wire, without dissembling the pick-up plate. Whew!

C) I have some other more pressing and fun projects, and my wife would really like the train out of the living room!

I might still try a pulsed source, like the 70 Hz suggested above (or whatever seems best for this train), but the actual speed feedback is what I was interest in, from a "geeky-fun" view. It's just getting to be too much effort for the reward. It's very tedious trying to wire a circuit small enough to fit in the cab. But I'm crying "Uncle" (for now).

Thanks again for the insights!

 

[Nik_2213]

I had a serious fight with those big screen-wiper motors to break their 'frame' earthing, attach my own return lead. At least they were modern-- IMHO, doing such for 7-decades old wiring borders upon 'heroic'.
Well done !

 

[essenmein]

Interesting, didn't know older model trains used universal motors, makes sense I guess if they pre date readily available permanent magnet materials.

The problem you have is that these are series motors, they behave distinctly differently to a permanent magnet DC motor.

For a permanent magnet DC motor voltage basically determines the speed, almost independent of load torque (assuming you can supply the current). This is because the magnetic field strength is fixed by the permanent magnets, therefore you get a constant X rpm/volt. Then as you increase load torque, the current increases, and the speed drops a little due to winding resistance introducing an internal voltage drop.

A universal motor does not have a constant field strength, the field strength is dependent on the load torque, this is because the field current and rotor current are the same since the windings are in series.

So at the two extreme ends of speed here is what happens (assume fixed voltage applied).

0rpm: Rotor generated back emf is zero volts (not moving), so now maximum current flows through the stator and field winding creating the largest available torque for the machine.

Maximum speed: Current is low which means field current is low, this reduces generated BEMF as speed is increasing resulting in very high no load speeds.

So as speed increases, the field current, and thereby field strength, is going down, this reduces back emf generated per speed, resulting in a non linear response to load torque and allows almost exponentially more machine speed before you run out of voltage.

Basically universal motors are self field weakening as their speed increases, which allows them to have enormous no load speeds, easily >20krpm.

This is what it looks like (fixed voltage):

So, unlike a regular PM DC motor, a fixed voltage (ie a PWM of fixed duty cycle) does not set speed the same way. As you've found, at a given fixed voltage if the load changes speed changes dramatically.

The only way to control it really is with some sort of speed feed back to change the applied voltage as load changes.

 

[NTL2009]

Interesting, didn't know older model trains used universal motors, makes sense I guess if they pre date readily available permanent magnet materials. ...

Your entire post was very informative, thanks. It confirms what I thought I knew about these motors.

In my research of Lionel trains from this era, I learned that another reason for the universal motors was that these were designed at a time when some significant % of homes in the US still did not have electricity. So the same train could be run from a battery and rheostat, and later used with a transformer from house current. That may just be a secondary advantage, driven by your observation of the lack of widely available, suitable permanent magnets.

My locomotive has a forward/reverse switch, which swaps the polarity of the brushes. A clever design was the Lionel "E-Unit". It was a cam rotated by a solenoid and pawl/ratchet mechanism, to advance one tooth every time the power was dropped and re-applied. Each increment of the cam would cycle through contacts that would cut power to the brushes, then connect the brushes in one polarity on the next power-cycle, cut again on the next, and then connect in reverse polarity on the next cycle. It apparently had enough mass and sensitivity to hold in with the short power interruptions you get moving along the track, though I'm pretty sure problems were fairly common when crossing a switch (turnout), as those had to have some dead spots in the center rail.

 

[essenmein]

Random thought, depending on how accessible everything is and how risk adverse you are, you could try to convert it to an external or shunt excited DC machine.

If you have two supplies you can test it, if you can separate the field winding, trying powering it from one DC supply (be gentle!), and then use another DC supply to power the brushes/rotor. This way field strength is no longer dependent on load current and should run more like a PM DC machine.

If it works you can wire it up so you half wave rectify and put a smoothing capacitor on your PWM signal to supply the field, ie your field maybe via a current limiting resistor, would see essentially constant voltage, and the raw PWM signal would be applied to the rotor, which would now see the average DC from the PWM.

This way peak PWM voltage sets the field current and PWM duty cycle sets the rotor voltage, and presumable no more self field weakening and crazy speed changes!

This may or may not be worth the effort...

 

[essenmein]

Pictures:
What you have (series excited or universal):

External excitation (Vf and Vr are separate supplies):

Shunt excitation (field winging and rotor share the supply):

 

[hutchphd]

The use of a universal-type motor was also dictated by the lack of a cheap robust method to rectify AC power in the 1930's. I believe obtaining several amps of DC at 16 V typically involved a rectifying tube with filament.
The postWW2 American Flyer (S gauge two rails) trains all relied on the "escapement" mechanism to provide reverse for their (series wound) locomotives. As I recall there was a switch on top of the locomotive or tender to enable this feature. Had a lot of fun with those trains.

 

[Tom.G]

I believe obtaining several amps of DC at 16 V typically involved a rectifying tube with filament.

Interesting; unfortunately practical gets in the way. More likely a Selenium rectifier, at least after they were invented around 1933. A rectifying tube rated at several amps would be a LARGE Mercury Vapor rectifier!

According to the site listed below, forward ignition voltage is around 50V and then drops to 14V during conduction. That works out to minimum 36VDC for the train. Ohh..., then there is the Heater power, 2.5V at 5A for a 1.5A rectifier. Might as well use batteries.
https://www.diyaudio.com/forums/tub...mercury-vapor-rectifier-tube-post4640811.html

Cheers,
Tom

 

[NTL2009]

Random thought, depending on how accessible everything is and how risk adverse you are, you could try to convert it to an external or shunt excited DC machine.

If you have two supplies you can test it, if you can separate the field winding, trying powering it from one DC supply (be gentle!), and then use another DC supply to power the brushes/rotor. This way field strength is no longer dependent on load current and should run more like a PM DC machine.

If it works you can wire it up so you half wave rectify and put a smoothing capacitor on your PWM signal to supply the field, ie your field maybe via a current limiting resistor, would see essentially constant voltage, and the raw PWM signal would be applied to the rotor, which would now see the average DC from the PWM.

This way peak PWM voltage sets the field current and PWM duty cycle sets the rotor voltage, and presumable no more self field weakening and crazy speed changes!

This may or may not be worth the effort...

Very interesting. The PWM supply I tried switches at ~ 25 kHz, so the smoothing cap would not be large. I experimented with a 15V supply in front of that ( a cheap, universal laptop PS replacement), so ~14.3 V peak at the smoothing cap after the diode, plus a bit of line loss. Even the AC Train transformer I have that is a couple decades old, uses an SCR or TRIAC, so there is still a peak voltage available - but at 60 Hz, the cap would be larger, but not crazy - a very rough estimate of a 5V delta V over 20 mSec with a 1.5A draw would be 6,000uF. I see those are about 1.4" x .7" or there-bouts. A simple constant current transistor circuit would help smooth that further, if a resistor didn't cut it.

I'll try to find time to make some measurements on how much voltage the field drops, versus the armature. Then I could design from that.

Too bad I don't have an O-Gauge dynamometer :) Now I'm picturing my meter strapped to the train while I try to read it as it goes around, or strapping my phone to that to record a video! Hmmm, maybe I do need a Bluetooth meter? Or another Arduino project?

 

[arydberg]

Very interesting. The PWM supply I tried switches at ~ 25 kHz, so the smoothing cap would not be large. I experimented with a 15V supply in front of that ( a cheap, universal laptop PS replacement), so ~14.3 V peak at the smoothing cap after the diode, plus a bit of line loss. Even the AC Train transformer I have that is a couple decades old, uses an SCR or TRIAC, so there is still a peak voltage available - but at 60 Hz, the cap would be larger, but not crazy - a very rough estimate of a 5V delta V over 20 mSec with a 1.5A draw would be 6,000uF. I see those are about 1.4" x .7" or there-bouts. A simple constant current transistor circuit would help smooth that further, if a resistor didn't cut it.

I'll try to find time to make some measurements on how much voltage the field drops, versus the armature. Then I could design from that.

Too bad I don't have an O-Gauge dynamometer :) Now I'm picturing my meter strapped to the train while I try to read it as it goes around, or strapping my phone to that to record a video! Hmmm, maybe I do need a Bluetooth meter? Or another Arduino project?

The original trains used a 0 to 20 volt transformer with a sliding contact. I would get a 20. volt transformer and drive it off a Variac that puts out zero to 115 volts.

 

[NTL2009]

The original trains used a 0 to 20 volt transformer with a sliding contact. I would get a 20. volt transformer and drive it off a Variac that puts out zero to 115 volts.

The off-the-shelf train controller I have chops the 60 cycle sine wave, like this:

https://i.stack.imgur.com/2G4gT.gif

from:
https://electronics.stackexchange.c...quest-for-explanations-of-how-things-inside-w

which, I think, is better than transformer control by itself, as is maintains that peak to provide torque. But after a thorough cleaning of the motor, track, and (most importantly, I think) soldering most of the connections between track sections with a copper jumper wire, it is running consistent enough that I really don't need a speed limiter, or other 'solution'. But I learned a lot in the process, and kind of want to do it anyway, just because. But other projects are in the wings.

Thanks.

edit/add: Just thinking - that old sliding contact on the old Lionel controllers is probably better than a variac and 20V transformer, as I would think that the sliding contact on the secondary provides a lower impedance as you run the voltage down (fewer windings, lower impedance?), improving 'sag' as the train draws current.

 

[OldLionelFan]

Summary:: I'm looking for feedback on this before I get too far. I want to set a max speed limit on this model train to avoid derailing.

I have a "262 Lionel Line" O gauge train, from the early 1930's. It has sentimental value, my Father-In-Law (RIP) got it as a gift when he was about age 10. I've set it up each Christmas, a short track around the tree. But it is difficult to control, the gearing is high (it goes fast), so reducing power to it to reduce speed results in frequent stalling at some points. It's a very fine line between too fast and stalling (like keeping the 0-100 knob from maybe 28-32, it's that 'tight'). I'd like to be able to set it and let it run, but I often have to modulate the throttle to keep it in limits. Typically, I need to advance the knob to ~ 50~60 to start from a stop. Especially now that my two grandsons were thrilled with it this year, I'd like to be able to let them run it next year.

FYI, here's a coupe videos of this model:


Some details: The controller I'm using is a fairly modern one, it uses an SCR/TRIAC to chop the 60 Hz 20V peak AC wave, so under normal operation it does see a 20V peak, which I think is good for providing torque under partial power. The train draws ~ 1.0~1.3 Amp running, and draws ~ 2 Amp to start. I tried a motor controller that puts out a 25KHz, 0~100% duty cycle wave, connected to a 15V 7A DC supply. No real difference in operation. I had thought about a much lower frequency chopped wave, but I doubt that would be much different than a 60 Hz chop (which I guess is really 120 Hz for a Universal motor). I also tried a constant current supply, and that seemed about the same as well.

My plan: I don't think I can do much from the supply side alone. I think I need feedback from the train speed. I thought about feedback from the train back to the controller, but that got complex. Now I'm thinking about a local electronic switch in the locomotive. Something that could easily be un-done to restore the train to original condition, if needed. My idea is to add an optical detector on the locomotive, a convenient place is underside where a 'push rod' comes through. I could obtain a pulse for each wheel revolution. One rev of the drive wheel takes the train ~ 4.25" (~ 11cm). I measured my slowest/fastest practical speeds for one lap to be from ~ 15 inches/sec to 25 inches/second. So that would give me ~ 3 ~ 6 intervals detected per second, which should be enough for good response. The 3 pole motor turns about 5:1 with the drive wheels.

I really just want to limit the top speed. I think I can use two 555 timers such that I keep power switched ON to the motor when the interval between pulses from the detector is > X mSec, and I'd cut power (internally) if the interval is < X mSec. I could have a small pot on the locomotive to fine tune the top speed. I think this should be fairly simple, though I haven't detailed out the exact configuration yet. And I can full wave rectify the input so I'm just working with DC.

The advantages I see is that should be pretty easy to undo if desired, and it doesn't require a special/modified power supply. I can use that just as I do now, and adjust it to control acceleration - it just wouldn't have much effect once the train has exceeded a set speed (the train would start cutting power internally at that point).

 

[OldLionelFan]

I have been wanted to do this for sometime, and have found this thread very informative (if only to highlight the issues involved). I just wanted to bring to your attention a 1985 MIT open courseware lecture on the topic by the late Prof James K. Roberge entitled (Lionel) "Model Train Speed Control". It was the last lecture in his Electronic Feedback Course, and may be viewed at:

I did not fully understand electronics presented in the lecture, but that maybe because I didn't listen to the first 19 lectures. ;-) However, the live demo of his Lionel train maintaining constant speed going up and down hill is certainly compelling and entertaining. I do wish I too could build a circuit to do that!

Cheers,

 

[Nik_2213]

Nice !
IMHO, you still need to find the 'sweet spot' for the PWM frequency on load...

 

[arydberg]

I suggest a variac which is a 0 to 130 volt variable AC transformer. If you connect it to a 12 or 15 VAC transformer you will have a close approximation of what Lionel used in their transformers. They come in different wattage ratings. A small one is not too expensive and will probably work fine.

Al