Having established a nice oval of track and got a train running by the slightly simplistic method of applying a 9V battery to the power cables, it was time to think about speed control. My nice Bachmann controller has apparently travelled to the dimension of lost socks during the move to the new flat, so I had two choices: Fork out for another one (not an attractive proposition for a poor student, especially as I wanted a PWM controller for low-speed running in N gauge) or put my studies for a degree in electronic engineering to use and try to build one. A quick look at the bank account told me that buying a decent-quality controller was out of the question, so it was off to Maplins... Surprisingly, the circuit I came up with was fairly simple, looking something (but not quite entirely - one component value has since changed from 100uF to 10nF) like this:

On this diagram, the top potentiometer (variable resistor) is the speed control, by adjusting the length of the "high" pulses, while the bottom one allows the length of the "low" pulses to be varied to suit different locomotives. In the event, however, I used a single 20 kilohm variable resistor, with the middle connection of the variable resistor connected to the point between the two of them on the diagram (Maplins not having the ones I wanted, and a friend having a spare variable resistor lying around)

After a trip to Maplin, the circuit was built on a scrap of veroboard, and the result was this:

A quick attachment to some switches and a battery, and a loco was put on the track. After a few faltering starts, there was movement! A quick calculation showed that the Farish 04 shunter ran at a maximum scale speed of 26mph. Given that the maximum speed of the prototype is given as 25-27mph, I was fairly happy with the results of this little experiment. Low speed running is pretty good too, with the only objection to slow-speed running being the need to clean track and wheels on my test track... All in all, this controller, which was only meant to be an experiment to see if it could be done, may very well end up in use on my layout!

(P.S. I should point out that this controller is only likely to work for N gauge and smaller, with adjustments to the supply voltage for T scale etc - the 555 chip I used can only supply about 200mA of current - plenty for an N gauge shunter, but possibly not for a 00 scale Pacific...)