Signals and Turnouts

For the “Big Railway”, this topic would be entitled “Signals and Telecoms” (or, in an earlier era, “Signal and Telegraph”) and would include the control and detection of switches as well as signals, and the detection of trains by track-circuits or other means. This is all interlocked so that conflicting and potentially hazardous moves cannot be set up.


Having spent my working life as a railway civil engineer, I simply had to get all of the visible details right for each of the engineering disciplines. Thus the railway will be properly signalled with working semaphore signals. Colour light signals have not yet arrived in my world.

Placing signals will always be a compromise, even on the best of model layouts. It is not difficult to place home and junction signals reasonably well but distants are a conundrum. With a line speed of 75 to 90 mph, the distant signal would be located a mile before its home signal, which is 69½ feet (21 metres) at 4mm to the foot. With a line speed of 100 mph, this would go out to 1¼ miles, or 87 feet (26 metres). Most modellers need to settle for shorter distances but, so long as the longest train is well past the distant by the time it is approaching the home signal, a reasonable compromise will be achieved. Most model layouts I’ve seen seem to ignore distant signals completely!

I can use fully detailed Ratio plastic signals indoors but these will be too fragile for outdoors. There I plan to use the brass kits made by Model Signal Engineering, now part of Wizard Models. I also have a few old Hornby Dublo signals which may be adapted for use where suitable.

There is a range of choices for operating semaphore signals on a model railway ranging in cost from virtually zero to around £15 per arm. A prototype semaphore signal is moved by a long wire which is elastic and the arm does not exactly follow the lever in the signal box. For a long pull (a distant signal), a signalman will pull the lever part way towards him to feel the stretch in the wire before giving it a good heave the rest of the way. This technique best overcomes the friction at all the guide pulleys and the signal arm can overshoot its off position before settling back on the wire tension. It will also bounce off its bottom stop when returned on (danger) when the signalman throws the lever back.

If the objective of a model mechanism is to replicate this slow and irregular movement, then the snap action produced by a simple double-coil solenoid as is traditionally used for turnouts will not be acceptable. For those signals close to the operator’s position (assuming a fixed control desk), wire-in-tube or string operated by some sort of scale or non-scale lever is a reliable, effective and cheap option. However, it has to be said that some of the more elaborate model lever frames are anything but cheap! For those further away, some sort of electrical operation becomes necessary and, if a panel of switches is being set up for those, then working the layout with two methods might become confusing. It would be possible to set up a lever frame with some levers pulling string and others switching electrical switches, but it is less complicated to motorise them all. Of course, if they are all to be able to be controlled by the DCC bus, then motorisation is essential.

Whilst researching different methods of operation, I came across a couple of systems designed to reproduce the motion of a railway semaphore signal in model form, Tony Witts’s Bouncer and Stephen Freeman’s signal servo, neither of which are now available. These sorts of systems seem to be the ultimate, with rise and fall speeds and bounce all being adjustable, although some YouTube videos I have seen suggest that they might be a little ponderous. A similar servo driver kit is also available through MERG (Model Electronic Railway Group) (for members only) and this is my choice of operating system.


The S&T department is normally only concerned with the part of the turnouts that move – the switches. They will be operated by servo motors after I read adverse comments on their longevity when driven by solenoid motors. Although slightly more fiddly to set up, the costs are now similar and the reliability should be much better. The slow motion will, as for the semaphore signals, be more realistic. They will have live crossings switched by the servo mechanism.

With a DCC layout, any short circuit will instantly shut down the power supply and some simple precautions will help to avoid predictable problems. One such trouble spot is where a metal wheelback makes contact with an open switch blade when traversing a turnout. Many manufacturers bond the switch blades to the crossing so that an open blade will be at opposite polarity to its stock rail. My switch blades will be permanently bonded to their stock rails, not to the crossing, to avoid potential short circuits.


Route setting will be from a fairly standard mimic panel and DCC encoded, with each turnout and signal identified with its DCC address, so that I can also operate everything from the controller handset.

I drew the line at full interlocking since this is largely invisible, although some signals will be electrically interlocked with the junctions they protect where that makes their control easier. DCC makes it possible to set up routes as macros, effectively the same thing. Railway interlocking is a specialised discipline, simple in principle but highly complex in practice, and that is not where my interests lie. I found a strong voice of agreement with this viewpoint in the book Model Railway Signalling by the great Cyril J Freezer, who is sadly no longer with us. This is his introduction to the chapter on Interlocking and Detection:

On the full-sized railway, points and signals are interlocked to prevent the signalman from inadvertently setting up conflicting routes or in some other way creating havoc. As the worst effects of such errors are greatly minimized on a model, it is not necessary to provide interlocking. Nevertheless, we need to be aware of the principles of interlocking and detection if only to see what aspects can play a useful role on our own systems. Apart from that, interlocking is a fascinating subject in its own right.

Before I go any further, I must point out the inherent disadvantages of interlocking as applied to the model. The first is the time it involves, not merely in the manufacture and installation of the equipment, but, far more importantly, in the planning and design. It is, alas, all too easy to devise an interlocking system of such fearsome complexity that it effectively prevents one from operating the layout.

The second and, in my opinion, far more serious consideration is that the installation of interlocking effectively freezes the layout design. It is not so much that you cannot alter the track layout, but that you are involved in so much extra work, over and above the initial inconvenience of uprooting and re-aligning the tracks and anything in their immediate vicinity, that, after mature consideration, you opt for the easy way out and leave things as they are.

In this connection we should remember the case of the interlocking on the Romney Hythe & Dymchurch Railway which was, until Captain Howey’s death, very much his magnificent model railway. Fired by a love of full-sized practice and, no doubt, encouraged and abetted by his engineer, Henry Greenly, who doted on interlocking, full locking frames were fitted at all principal stations. All went well until, in 1940, the military took over.

We must be perfectly clear – this was not a case of a bunch of roughneck soldiers getting hold of a fine piece of engineering; the line was worked by the Railway Division of the Royal Engineers of which it was said, not entirely in jest, that while you could tell a Longmoor-trained man anywhere, you couldn’t tell him very much. They encountered the locking, struggled with it for a few days and then proceeded to adjust it with sledgehammers. I have known cases where model interlocking has suffered a similar fate for exactly the same reason.

A man after my own heart, indeed!


Civil EngineeringIndexTrain Control