On model railways, as on the prototype, it is useful to know whether any given section of track is vacant or occupied by rail vehicles. The ideal is a system which provides an electrical indication of the presence of vehicles in the section. This can be used as an input for mimic diagrams, automatic signalling and other security systems and the automatic operation of level crossing (grade crossing) gates or barriers.
On prototype railways track circuits are often used to provide such an electrical indication of section occupancy. In its simplest form (illustrated diagrammatically above) a track circuit consists of a low-voltage battery connected to the rails at one end of the section (which is delimited by rail breaks) and a relay whose coil is connected between the rails at the other end. If the section is vacant the battery voltage energises the relay coil, but if there are any rail vehicles in the section their wheel-and-axle assemblies cause a short-circuit which de-energises the relay coil. The relay contacts are normally connected to security systems in the signalbox (tower).
On first consideration it might seem that anything analogous to a track circuit is impossible on a two-rail model railway for two reasons. Firstly, the track already carries a voltage, namely the power supply for the trains. Secondly, because the track carries the power for the trains all rail vehicles apart from locomotives of necessity have insulated wheel-and-axle assembles, which would not activate a track circuit. So, this is where lateral thinking comes in: we let the power for the trains be the track circuit voltage; as for the insulated wheels, we settle for detection of locomotives only which for most purposes is sufficient. If there is a particular need to detect some other vehicle, it is usually a simple task to make it sufficiently conductive to show up in a track circuit.
As on a prototype railway, the section to be track-circuited is delimited by rail breaks or (preferably) insulated fishplates, which may be in one rail or both. Because of this, each track-circuited section needs its own supply from the controller(s) (throttles). The feed from the controller to the section passes through the track circuit unit which works by monitoring the current flowing. If it detects current flowing, this is taken as indicating that there is a train in the section. If it detects no current flowing, the section is assumed to be vacant. To allow monitoring to continue when the controller is set to stop or is switched off or disconnected, allowance is made for a trickle of current (called the auxiliary current) from the track circuit's own power supply also to pass through the section. A resistor limits this to typically about 1 mA, which is too low to cause any unwanted movement of trains, but nevertheless sufficient to activate the track circuit.
Controller current and auxiliary current pass through a current detector which is the heart of the track circuit unit. The design of this is an interesting challenge, for it must pass and detect current ranging from 1 mA (a typical auxiliary current) to 250 mA (the typical current drawn by an HO/00 locomotive at cruising speed) in either direction and without causing a voltage drop high enough to impair the performance of the trains. The usual way to detect a current is to pass it through a resistor and use the voltage raised across this to bias a transistor into conduction. Since it takes about 0.7 V to bias a transistor into conduction, a 68 R resistor would be needed to raise this from an auxiliary current of 10 mA (high for an auxiliary current). But then at 250 mA the resistor would cause an intolerable drop of 17 V! So a conventional resistor is out of the question - we need a non-linear resistor, one that automatically adjusts its resistance to limit the voltage across it to 0.7 V. And that is precisely how a silicon junction behaves. Since a junction conducts in only one direction, a reverse-parallel pair is needed to cater for the controller current which may be in either direction. Some track circuits use a pair of diodes in reverse-parallel (the so-called 'twin-T' arrangement), but TEKTOR uses the reverse-parallel base-emitter junctions of a pair of transistors whose collectors are bonded. Inevitably these transistors must be high-current types (for HO/00 and smaller scales choose types rated for a collector current of no less than 2 A). Current in either direction through the base-emitter junctions stimulates collector current (now of uniform polarity) from one or other of the transistors. So we have a working bidirectional current detector with a maximum voltage drop of 0.7 V; in practice the change in speed of a train moving between a track-circuited section and a non-track circuited one is barely noticeable.
A few additional features are needed to turn this into a practical track circuit unit. Many controllers deliver a pulsed output; with these the 'train in section' indication from the current detector as described above is also pulsed. Even with controllers that deliver a steady d.c. output, oil and dirt on the track and wheels result in moving locomotives making only intermittent contact with the controller. So if you were to use the raw output from the current detector described above to operate, say, a red/green colour-light signal guarding the section, the signal would flicker wildly between red and green, probably both appearing to be illuminated at once! For these reasons the intermittent output from the current detector must be smoothed to a steady indication before being output to other devices. This smoothing usually takes the form of a third transistor charging a high-value capacitor, the charge being used to bias a fourth transistor from whose collector the output is taken. The values of the capacitor and bias resistors are deliberately chosen to sustain the 'train in section' indication for a few seconds. Consequently the track circuit gives a steady 'train in section' indication as long as the section is occupied and this persists for a few seconds after the loco leaves the section and current detection ceases. There is method in this seeming madness. Most trains consist of a locomotive (which is detected) hauling a number of vehicles (which are not). When the locomotive leaves the section, for a few seconds its train trails into it so that the section is still technically occupied. This delay in clearing is a crude method of providing protection for the undetected vehicles as the train leaves the section.
Another smoothing element in TEKTOR is the capacitor in parallel with the base/emitter junctions of the current-detector transistors; this is shown as C1 in the circuit diagrams below. It is essential for the following reason. A model railway layout acts as a very effective radio antenna. In the indoor environment there are many sources of electromagnetic radiation - mains cables, transformers, cordless and mobile phones to name a few - and also further radiation comes in from outside - radio transmissions of many kinds, electrical interference from vehicles, electric railways and so on. All of this impinges on the layout and induces electrical 'noise' in the track. This consists of an a.c. voltage in a mixture of frequencies. Its amplitude is too low to interfere with normal model railway operations. A track circuit, however, is effectively an amplifier connected to the track and this electrical noise is sufficient to cause a 'train in section' indication. Capacitor C1 acts as a filter, allowing this noise to harmlessly bypass the current detector, while d.c. - representing current from the controller or auxiliary current - cannot pass through the capacitor and correctly results in a 'train in section' indication.
For the purposes of this description the following convention has been assumed. When a model train is running forwards, the rail whose voltage is positive with respect to the other is regarded as the live rail and the other as the return rail or common return. This distinction is vitally important when installing track circuiting.
Because of the way in which they work, track circuits have traditionally been installed in the return feed, requiring the return rail to be divided. For convenience it is often simpler to divide your layout into sections in the live rail, keeping the return rail continuous. Clearly, on a layout of this sort track circuit units would need to be installed in the live-rail feed. TEKTOR is described as a universal track circuit because it can be installed in either the live or the return feed to the section, but there is a restriction: TEKTOR cannot be used in live-rail mode if the controller delivers a smoothed d.c. output - in these circumstances train detection fails in the forward direction at all but the lowest speeds. This is because the difference between the controller output voltage and the track circuit power supply reverse biases the base-emitter junction of T3, preventing the 'train in section' indication. With controllers that deliver a pulsed output, however, monitoring continues between the pulses and TEKTOR's smoothed output delivers an accurate indication in live-rail mode as well as return-rail mode. Incidentally TEKTOR is compatible with command control systems, provided of course that the layout is divided into sections having separate feeds. Do not allow any accessories to take their feed directly from the track in a track-circuited section, as these cause permanent 'train in section' indications.
The following diagrams show TEKTOR configured for live-rail and return-rail operation respectively:
TEKTOR requires a smoothed power supply, which may be shared with other TEKTORs and with other circuitry to which it needs to connect, such as signal drivers. It has been tested and found to work correctly with supplies ranging from 5 V to 12 V. It may work with supplies outside these limits, but this has not been tested. (If you try it, please let me know the results.)
The output of TEKTOR is the open collector of NPN transistor T4. This becomes conductive (giving a 'low' or logical '0') when a train is detected. It is compatible with TTL inputs and with the inputs of the signal driver circuits shown elsewhere on this website. TEKTOR must share its return line with the systems to which it is connected.
The pre-set potentiometer is a sensitivity control. I have tried various ways of removing this, but none was satisfactory, so I'm afraid it stays. Begin with the slider set roughly mid-way. If the setting is too high (that is, slider too near the collectors of T1/T2) you may get spurious detection (that is, a 'train in section' indication when there is no train present). If the setting is too low, train detection will fail altogether. There is, however, a broad range of settings over which train detection is consistently accurate. Once set, no further adjustment should be needed unless you change the supply voltage or move the unit to another location.
© Roger Amos 2016