Model train controller

[Terry]

No, I can't see any logical reason for it.

OK, when C1 is fully charged so that no current flows in the diode (but see below) a small current can still flow via R3 to charge C1 to a slightly higher voltage, but it isn't necessary. C1, being an electrolytic, will have a slight leakage current but it doesn't make any difference if this is supplied via the diode or R3. The effect on the operation of the circuit will be the same in either case (negligible!).

The only possible reason for R3 is that, if R2 is reduced to a lower value, the diode will be reversed biased so R3 could allow C1 to discharge to the new value set by R2, but it would be a long time constant!

[Crackle]

Thanks Terry

Mike

[Crackle]

I have hit a slight problem with this design. Now I have got the inertia control working, I wanted to incorporate the circuit into one of the three LM317K controller circuits in the home made triple controller which was made back in the 90's, by my brother in law.
The idea being the inertia control would be switched in to one of the controller circuits if desired.
I found that with no load the 317K worked well with the circuit, as it should because I believe the TO3 version of the 317 is identical electrically to the small plastic one.
But when a 300mA load was applied the voltage failed to rise beyond about 2.6v.
After treble checking everything, and being baffled, I removed the protection diode D1 and found the circuit worked fine.
I then had the thought, smoothing capacitors were never fitted to this controller when it was made. Possibly because they are not shown on the application circuit, or also possibly because he wanted to not smooth the rectified 15v DC save the voltage rose too high.
Anyway fitting a 2200uF capacitor to the rectified DC has cured the problem of the inertia controller not working and has allowed the D1 protection diode to be refitted. But it has also resulted in excessive heat in the 317K regulators which were not fitted with any heat sink at the time.

The only solution is a complete dismantling of the printed circuit, (which is a bit like a prototype breadboard) and redesigning the layout to incorporate heat sinks.
However with the extra smoothed input voltage the regulation is now better under load than it was before in its original design with just the 15v input into the regulators.

Mike

[Crackle]

I have finally finished this project of refurbishing this model train controller.
There has been a lot of fiddling around making a mounting for the 3 push buttons that I wanted to fit.
I have completely redesigned and rebuilt the circuit board, changed the component values and added extra smoothing.
I also changed the small 1 amp bridge rectifiers, which failed before, with 6 amp types. As they were the same price I got 1000v types so I have a couple of useful spares now.
With the higher input to output voltage difference, as a result of smoothing the supply, the regulators were getting excessively hot relatively quickly. So I have made a heat sink out of an old aluminium echo sounder mounting bracket, and drilled a few holes to allow a little ventilation.

The whole affair works well, just getting warm after a half hour at about 1 amp total load. (300mA each channel)
There is also now better regulation of the voltage under differing loads.
I have discovered the lower value resistors for R1 give better regulation.
The output voltage from the new smoothed 21v supply was rather high on full speed, over 16v, so I have added a 4.7k resistor in parallel with the resistance track on all the 5k potentiometers. That has bought the max voltage down to 12.9v with R1 at 270 ohms.

Two of the channels work in the same principle as before, variable voltage of 1.25v to 12.9v, and with a switch to set the channel to have a max voltage of around 6.7v for half speed.

The third channel has the same normal voltage control but this time the high/low switch connects in the push button controlled "inertia" circuit .
When the inertia circuit is switched in, the train accelerates from stationary to full speed (12.5v) in 40 seconds. With the first button pressed the train slows to just over half speed (8v) over 25 seconds, the second button takes the train to a crawl (5v) over 15 seconds, and the final button allows the train to come to a stop (2.8v) over 5 seconds.

So I am happy with that at the moment, I may tweak the resistor values at a later date when it has had a bit more use.

Here are some photos of the controller.

This one is of the inside before my alterations, the red tape on the transformer is there just in case I accidentally leave the mains switched on whilst working on the insides.
   

This one shows the new final circuit layout.
       

and here are a couple showing the finished controller.
       

This was the plan for the component layout, viewed from component side, the copper tracks are below (the scan of the RS stripboard was inverted)
   

Mike

[decowoodie]

I tried making a 317-based controller when I was at college as I filled in my spare time with model train layouts. Maplin magazine published a better controller using low-frequency pwm techniques which I built, just a couple of 8-pin op-amps and a transistor, and this gave superior control especially at low speed, so much so, I ditched the 317 controllers. The pwm was audible at low speeds as a buzz, rising to a whistle, and the lower frequencies seemed to overcome the stiction in the motors.