Analog Cab-Style controller

[medusa]

Hi all,

since there are a lot more analog drivers than I thought first, here is some more detailed input about an purely analog cab-style controller. Cab style driving a model train is the greatest thing at all. I'll never again miss my controllers. Feels so much better playing with them. Yes it is like driving a real train. :D
Inspiration came from the Tomix cab-style controller. Of course all this can be done digital also (even if probably the smallest ARM processor still is overkill for that job), but I simply didn't want to. I earn my money by writing low-level drivers and porting C++ algorithms to embedded DSP devices. After a workday, I certainly will not do anything digital at home. No way.



On the other hand, in the past still being a student I had to do some practical exercises about analog computers (yes, I'm that old). So, why not doing cab-style control by kind of such? Lets check:

- Model train standard controller: control of speed. Speed increases with voltage.

- Real loco or EMU: 2 controllers. One for motor power, one for the brake. Power control controls the amps of the motor which (ideally) means acceleration. Brake valve controls deceleration (which is acceleration with a negative sign).

So... we remember from high school physics lessons, speed equals integral of acceleration over time. Going to analog computers, all we need is an integrator circuit which gets power control and minus break valve settings as input, and it will output a speed signal like a standard controller.
Link about integrator circuit: https://en.wikipedia.org/wiki/Integrator



That's the simple circuit I set up. Published it earlier in 1zu160 forum under GPL license so feel free to use it.
The circuit has four pots, two for the power/brake controls and two for calibration. Calibration is for setting of the reference "zero" potential (left side, center) and for MOSFET gate voltage scaling (right side). The latter also can be used to set a maxium speed (i.e., output voltage).
Input voltages to the integrator are on left side, top and bottom pot. Please note that integrators revert the sign of the input, so "power" input is negative and "brake" input is positive.
Output of the integrator circuit drives a MOSFET which in turn controls the electric feed of the model train tracks. In the schematic, this is plain DC (but can be modded, see below).
Some detail about the integration capacitor: one polarity is shorted by a gemanium diode since it must not be possible to drive backwards by opening the brake valve. ;)

As you can see on the photo above I built two of them up to now, in a more archetypical design, one for my japanese trains (power on right side, brake left) and european trains (regulator wheel, brake on right side). Both have a directional switch. I have left parts for a third one which will get linear pots for modern "push" control of high-speed trains. It also will get modern LCD meters.

Some images about R&D and building phase of the first controller follow... you see I started with a very experimental set up first like many of you here in the forum do with their circuits, too.



Of course, such a simple analog circuit hast some pros and cons in real model railway duty. First and most important pro is, driving results in wonderful smooth acceleration of the train. And, turning the power control to zero, the train continues to "roll".
A fantastic feature of analog computers is, they model analogies of mechanical systems (at least if properly tuned). So, if I let "roll" a model train "unpowered" it will slow down after a time like a real train. This is because the OPAMP has a non-infinte internal resistance, being the analogy of the non-friction-free bearings of the real train. So, slow discharge of the cap lowers the output voltage in analogy of the friction in the bearings slows down a rolling real train. This comes for free with an analog computer.
And about breaking: stopping a train in a predetermined place needs some practice. Believe me. :)

A little bit con on analog circuit always is, it has to be carefully tweaked to the type of motor it has to drive. The japanese style controller has reference voltage set so it outputs about 2.5-3.0V with a motor car on the tracks just before driving starts. That's about the DC range of voltage where Kato motors (which is what I have) start to turn. However, it is next to impossible to catch it exactly so I stay a tiny bit below. Which means, by opening of the power control the train does not start immediately but stays a short period standing with just the voltmeter slowly rising.

My european loco models are all 2nd hand older ones of Minitrix, Roco and Piko with somewhat cruder motors. They do not start driving well under pure DC power like in the schematic above even with higher reference setting. But it is quite easy to cut off the "+" rail from the electric supply of the analog computer circuit and feed some pulsed DC into the rails. This works quite well since the MOSFET doesn't care much what stuff it controls. At least if the polarity stays right.
Just be careful an auxillary second electric input is properly insulated from mains since both inputs have to share a common ground.
This worked so far quite good for a Roco BR 181 model I have, it starts quite smooth now. My old Mtx E 03 isn't impressed at all but on the other hand it never started smooth at any time. Didn't have time yet to test more of them.

So... questions and comments are always welcome. I'm far from being perfect in these things. ;)
 

~Diane.

 

 

EDIT: the usual typos

Edited December 19, 2016 by medusa

[kvp]

Imho the starting voltage problem could be solved by pulse width modulation. This also lets you mix in a constant lighting signal. Easiest is with a tiny PIC microcontroller just like in Tomix cab controllers.

 

Does the wheel variant work like the real trains too, with 5 fixed positions of a rotary switch? (step down continous, step down single, 0, step up single, step up continous)

 

The japanese variant also has rotary switches, power and electro pneumatic brake (with release and emergency brake on the two extremes). It seems doable in full analog with a resistor divider arrangement.

[medusa]

A number of detail could be handled by a microcontroller... but guess I stated my opinion about that quite clear (even if I'm in holiday from today on).

 

The power wheel certainly could be done by a rotary switch w/ fixed resistors. But of course the step up/ step down logic you mentioned would require additional circuitry.

I'm not sure which locos have them. The examples In saw (I admit it was quite old ones) had a simple linear step switch so no automatic step up or step down.

 

As of my model train controller, the power control input is done by a simple pot. That's easiest and cheapest especially if planning to build more than one controller. Rotary switches can become quite expensive, the more steps the more money is the rule. ;)

I do not miss the steps so much but of course that's personal preference. All I gave here was a simple example how it could be done.

[kvp]

Imho you showed us a nice analog circuit. The 8 pin pic microcontroller and fet transistor solution is another approach that while require less components but way more programming (an a/d read+pwm write loop). A hybrid solution would be to make a pwm drive out of an 555 timer chip and drive it with the flat analog voltage of the integrator and use the pwm signal for the main drive transistor.

 

For wheeled and stepped locos, swiss, french and hungarian electrics, some german diesels and their control cars come to my mind, but older non mu-able ones had the classic linear wheel driving a servo motor directly and even older ones switching the main controller directly through mechanical linkage.

[medusa]

The hybrid solution sounds like an idea. Have still space left in the controller cases so it may be a good future expansion. Would also remove the necessity for an external second supply in the case I have to use pulsed DC for the old motors.