Rather than using diodes, I’m wondering if a bank of stout wire wound resistors would be worth doing.
Back in the old days, Lionel sold rheostats, both for use with batteries and for use with the old multivolt transformers that had fairly course taps. I also remember that years ago a member on here(who I think was Dutch?) built a nice resistor controller from a bank of resistors and a lever/wiper assembly to wire them in series. I’ve played some with old Lionel rheostats.
A variable resistor set-up doesn’t work quite like any of the variable transformers or other means of modulating voltage we’re familiar with. Since voltage delivered to the track is heavily current dependent, the control isn’t as responsive, but I’d describe it as being “not as responsive” in a good way. As an example, let’s say you park the rheostat on a setting that’s 5Ω with a stopped train, and you’re feeding it 18V. At stall, the motor may draw say 2A(if it’s an old open frame motor) and you’ll deliver 8V to the track. As the motor starts to spin, current draw may drop to maybe 1A, at which point it will be getting more like 13V. This gives some momentum effect with any kind of change.
Here are some inexpensive(like a few dollars each) 100W rated wirewound variable resistors on Aliexpress. I’d be inclined to put some sort of active cooling on them(the old Lionel ones just rested on a big slab of asbestos and would get pretty toasty after a few minutes) but it would be a cheap way to try proof of concept at least.
You’d probably want to experiment with the exact one to use, as something like 5Ω max would be great under a load of a few amps, but you’d need more like 200Ω to get a low starting voltage with a low load. Something like a 100Ω might be a good compromise.
Wire wound resistors, we are heading into the swamp now! I will leave it at that for now. One refinement, the stall current of a small motor Gilbert steam engine is a bit over 4A at 12V. The DC resistance of the field plus armature is 2.8 ohms. The large motor stall current is between 5 and 6 amps at 12V, about the same as a dual motor diesel. I am sure the Lionel motors are at least at these values. The back EMF adds about 5 ohms when the motor is up to speed with a moderate load.
I have repeatedly considered rheostats, but I do not like systems which vary current instead of voltage. They tend to complicate matters and limit compatibility.
What sort of “compatibility limit” are you referencing?
And just to be clear, a variable resistor will change the voltage(Ohm’s law and all of that), it’s just current dependent in a way that other methods, like resistors, variable tap transformers, and triacs are not.
To be honest, though, a lot of this feels like re-inventing the wheel…a lot of inexpensive trains have gone to fixed track voltage with onboard remote motor controllers because it’s cheaper to do it that way now than to build a true variable voltage transformer.
I’m referring to the difficulties of finding a rheostat that is entirely suitable for Postwar trains with power-hungry pairs of universal motors, for cheap, newer trains with can motors, a light bulb, and nothing else, and for everything in between there. Not saying that it can’t be done, just that’s it’s a bigger hassle