Two wires can operate signals and points at greater distances than rods can
Interlocking frames on the Continent looked very different from American and British practice. The short levers with wheels, and the operation of points and signals with two wires instead of rodding or single wires for signals, as well as the large compensation levers seen in groups by trackside near signal cabins, were distinctive. This system was developed in Germany, and was used to the virtual exclusion of others in Germany and Austria, and was widely used elsewhere, as in Spain. Two-wire transmission was investigated in both Britain and America, but was never significantly used in either place. The Belgian signal engineer L. P. A. Weissenbruch appreciated its ability to operate three-aspect semaphore signals, adopting it for that purpose in 1919. This paper explains the fundamentals of its operation, for those familiar with Anglo-American practice who might be curious.
The most important reason for the investigation of double-wire working in Britain was to overcome the Ministry of Transport's limitation of 350 yards for the working of points by rods. Because of this limit, a long loop had to have a signal box at each end, which was expensive. It was a great economy to abolish the extra box by using some kind of remote working of points. One way to do this was with electric point motors, at first low-voltage battery-operated, and later by hand generator, where mains power was not available (which was usually). The 350-yard limit did not apply to double-wire working, and it was much cheaper than an electric point motor. Points could be worked at 880 yards by this method. The Great Western Railway first used double-wire working at Park Junction (Newport) in 1929, and it was later used at Builth Wells (1932), Maindee Junction, Llantaram Junction, Pengam, Silverdale Junction, Builth Road, and finally at Johnston in 1935. The last was the only case in which several points were so worked at one location. There is a turnover frame in the Warehouse at the National Railway Museum in York, but its origin is not given, and the locking trays are empty.
Signals and switches are now usually operated by electrical motors; linear and rotary solenoids have also been used for operting signals. Electricity not only provides easy control, but also light and mechanical power in any desired amounts. Pneumatic and hydraulic forces, especially when electrically controlled, have been historically important. The earliest and most direct methods of transmitting the necessary forces have been mechanical, using wires and rods. Mechanical methods of operating signals and points will be discussed here, with the aim of explaining the two-wire transmission, very unfamiliar in America and Britain, but elsewhere widely used.
The need for operating signals at a distance arose with the first distant signals in the 1850's. The method adopted was to operate the signal, then usually a rotating vane, by pulling on a wire, as shown in the figure. A counterweight tended to keep the signal in its normal position perpendicular to the track. The wire pulled against this counterweight to turn the signal parallel to the track. In each position of the signal, the travel was limited by a stop. These two positions became known as "normal" and "reversed" respectively. Should the operating wire break, the counterweight automatically pulled the signal to its normal position. It was convenient to pull on the wire by means of a lever working in a quadrant. A latch fitting into notches in the quadrant held the lever in one position or the other. This latch was operated by a latch handle. In some cases, the lever had to be held reversed without a latch, so the signalman would not forget to return the signal to normal. This was a very satisfactory arrangment, used ever after to operate two-aspect signals. Although shown with the early vane signal, the method was as easily used with semaphore signals. The maximum distance at which signals can be worked by this method depends on conditions, but distances of up to 1500 m are possible.
A wire of cross-sectional area A and length L lengthens by a distance e under a tension T given by e = PL/AE, where E is Young's modulus, 30 x 106 psi or 21,000 kgf/mm2. The stroke of the lever must be greater than the stroke of the output by this amount. For L = 1000 m, P = 10 kgf and A = 4 mm2, e = 60 mm. More significant is the thermal expansion of the wire. The coefficient of expansion of steel is about 12 x 10-6 per degree C. The total change of temperature between a cold winter's day at -10°C and a hot summer's day at 30°C is 40°. The corresponding expansion in 1000 m will be (12 x 10-6)(106)(40) = 480 mm. This is overcome by means of a device that keeps a small tension on the wire, and grasps it each time the wire is pulled, so the displacement is independent of temperature. These "compensators" were never regarded as wholly satisfactory. As a practical matter, a short length of 1/4" chain is inserted when the wire goes around a pulley to increase the life of the wire. Pulleys are used to change the direction of the wire. The typical wire used was #9 steel, or about 4 mm diameter.
When it became necessary to work points, facing-point locks, and similar devices remotely from a central location, a new method had to be devised, since it was unsatisfactory to work against a counterweight in these applications. Not only a tension, but a compression as well had to be transmitted. The wire was replaced by a rod, initially the easily available 1" gas pipe with its screwed fittings. A straight lever could change the stress from tension to compression, and bent levers could change the direction of the rod. The fact that both tension and compression were available made temperature compensation easy. It was only necessary to reverse the stress at one or more points, and to ensure that the total distance in compression was equal to the total distance in tension. The operation of a three-aspect upper quadrant signal is illustrated. It would be possible, but not convenient, to operate such a signal by two wires working against counterweights, but this was never done.
Gas pipe had the disadvantage that moisture inside the pipe would encourage rusting, and this was not easy to monitor. The pipe was eventually replaced by a channel section that did not have this problem. Friction is much greater with a rod, so the limit of operation is about 350 yd or 320 m, as is legally specified in Britain. In both America and Britain, two-aspect signals were operated by wire, and turnouts by rod. With the adoption of 3-aspect signals in America around 1900, wire operation could no longer be used, and rod operation was substituted for signals close to the control point. Electric motor signals were becoming available at that time, and they were used for distants. Wire operation soon disappeared in America, but was retained in Britain for manual signalling.
Around 1870 a method of operating turnouts by means of wires was developed in Prussia. The principle of this two-wire transmission is illustrated at the right. A wire passes around two pulleys, one at the operating lever, and the other at the output, that rotate together. It is shown operating a two-aspect upper quadrant semaphore, but of course this could just as well be a three-aspect signal, or a turnout, or anything else. The upper-quadrant semaphore arm provides its own restoring force, but it is not necessary here except in case of derangement. An essential element is the tensioning apparatus or Wire compensator (German, Spannwerk) that is seen beside the tracks when this method is used. How these devices function will be described in detail below. The tensioning apparatus compensates for thermal expansion. Solid wire is generally used because of its resistance to corrosion; great care must be taken not to nick it. Wire rope or chains is used to go around pulleys.
There are four essential parts to the system: the levers, the compensators, and the signal and point operating mechanisms. Two wires are used for each function controlled by a lever in the signal cabin. The wires are constantly under considerable tension that is normally equal in the two wires. That is, they are never slack, and this facilitates operation. It is not true that one wire was slack, as some references assert. When a lever is moved, the tension in one wire is made greater than in the other, and the force is transmitted to a wheel at the function, which rotates to carry out the function, after which the tensions in the two wires are again equal. The wire compensator is the mechanism that maintains the tension by means of a heavy counterweight for each wire of the pair. One compensator design is shown in the Figure. The pivots at points marked a, b and c are fixed in the frame of the compensator. When the temperature changes, the two counterweight levers move together to keep the wires taut. In hot weather, the counterweights sink; in cold weather they rise. If the tension in one wire is made greater than in the other, by moving a lever, one counterweight tends to move relative to the other. This relative motion is stopped by the jaws or blocking device, which is rotated slightly by any relative motion so that their teeth engage the serrated rod, which holds the counterweights fixed as long as the tension is different.
A contingency that must be guarded against is the breaking of one of the wires, as we have just mentioned. Since the wires are normally under tension, they store considerable elastic energy, which will be released if a wire breaks. The wires may conveniently be called the traction wire and the return wire, which move the pulley in opposite directions. Suppose the traction wire breaks. Then, the return wire tends to rotate the pulley anticlockwise, either returning it to "normal" or moving it to a stop when the follower reaches the end of the slot. The outcome is, then, not dangerous. On the other hand, if the return wire breaks, the traction wire tends to move the pulley in a direction that clears a signal or moves points to select a diverging route, both of which are unsafe actions. In German signals, the operating wheel groove has a restriction that will stop rotation at a safe position. In Austrian signals, the operating wires are automatically disconnected when a wire breaks. In still another method, a spur lever is pivoted to the pulley. The return wire is pinned to the end of this lever, and the traction wire is held in a hook. If the return wire breaks, the traction wire rotates the spur lever instead of rotating the pulley and comes off before any damage can be done. Every output device for two-wire transmission must have a similar feature. We shall discuss certain arrangements below.
The duty of the wire compensator is to maintain a specified equal tension in the two wires independently of thermal expansion, and in case of wire breakage to ensure that the unbroken wire is pulled far enough to ensure that security arrangements work properly. The most widely used type, illustrated at the right, is the lever wire tensioner, with an adjustable weight "a" on one end of a lever usually inclined at about 45°, and a freely rotating pulley "b" carrying the signal wire near the other. The signal wire is guided by two fixed pulleys "c" to and from the movable pulley. There is one lever for each wire, which normally stand side by side. When the operating lever is moved, the tensions in the two wires become different. If there were no other arrangement, one lever would move upward and the other downward to equalize the tensions, and the wire pull would not be transmitted. To avoid this, a blocking device "d" carried at the end of the levers moves along a serrated rod. When the tensions in the two wires differ, the device rotates slightly to one side or the other and teeth contact the serrated rod "e", preventing the levers and the movable pulleys from moving up or down, so that the force is transmitted from the operating lever to the operated device. When the movement stops, the levers are again released.
If one wire breaks, the vanishing of the tension causes its movable pulley to tend to rise. However, the blocking device prevents this. As the other lever begins to take up the wire released by the breakage, its pulley will rise, releasing the blocking device, and the two levers now move together as the weights descend and the unbroken wire is taken in. This ensures the proper operation of the security arrangements.
It is clear that the two counterweights are always close together. The reason they are separate is only to detect a difference in tension of the two wires, which only requires a slight motion to operate the blocking device. Wire tensioners are located on the lower floor of the signal boxes or, quite often, in the open.
The levers, one of which is shown in the Figure, are 2' 6" long, and move through 180° from the normal position N to the reversed position R. This is the reason for the term 'turnover' to describe them. The resulting motion of the transmission wire is often 500 mm. The fulcrum is at a height of 3' above the floor. The total travel is 7' 9", much larger than the usual 4' 6" pull on a British or American lever. The first and last parts of the pull are easy, so the heaviest part of the pull takes place when the lever is in a favourable position. A reversed lever is well out of the way, so there is no difficulty with pulling a lever between two reversed levers. The locking is operated by the catch handle. When the catch is raised, the locking moves 1", and the pivot comes into line with the fulcrum of the lever. There is no motion of the locking as the lever is pulled, but when the catch is released again at the end of the stroke, the locking moves a further 1" to complete the locking, since it is now 180° from its original position. We see that the latch locking with initial and final motion is carried out in a very simple and effective way. These levers may often be seen in the open on a station platform.
The output pulley or operating wheel is shown at the left. These are usually enclosed in metal boxes at the bases of signal masts and near points. They may be vertical or horizontal. Different diameters can be used to give different angles of rotation. There are several ways to arrange the output levers, with straight or bent levers, to derive the required motion. This is not the only way to connect the two-wire transmission, but it is the most satisfactory way for general purposes. In this case, the pulley is designed to rotate clockwise through 180°. As it does so, the follower is guided by the curved slot and approaches the axis of rotation. The diameter of the pulley is typically 500 mm. There is only a small motion during the first part of the rotation, then the majority of the movement takes place, and finally there is little motion again at the end of the rotation. This allows the operating lever to move enough to operate the interlocking mechanism at the beginning to lock conflicting levers, and at the end to release compatible levers. Any sort of interlocking can be applied, but this will not be discussed here.
A typical signal mechanism consists of a cam, or operating, wheel with a groove that guides a small roller at the end of a lever. When the wheel rotates, the groove guides the pin towards the periphery of the wheel, which moves the signal arm off by a crank and up-and-down rod. The first and last parts of the rotation do not produce much motion. The cam slot is prolonged beyond the normal stroke and returns again to near the axis. We shall see that this is provided to ensure safety in case of a wire breakage. Two wire pulleys are generally used, around which the wire makes several turns. This ensures that if a wire breaks, the other wire can reliably rotate the wheel to the stop where the signal displays its most restrictive aspect. If the wire were simply passed over the wheel, it would fall off when either wire broke. A point mechanism (not shown) also has a wheel, but it contains a pin instead of a slot, which works in a slotted lever to move the switch-and-lock mechanism for the points. There are several popular mechanisms for moving and locking points. In Germany, a turnout must be capable of being run through from the heel direction without damage when not correctly set. This is often realized by locking only the point that is against a stock rail. Pressure on the other point transfers the lock to this point. Note that the cam wheel can rotate by different amounts, depending on the design. Most common are 180°, or 90° in opposite directions.
A two-arm semaphore can be a three-aspect signal, displaying Stop, Proceed, and Proceed at Low Speed (typically 30 km/h). The Stop aspect is upper arm horizontal, lower arm aligned with the signal mast. The Proceed aspect is with the upper arm elevated at 45°, the lower arm again in line with the mast. In the Proceed at Low Speed aspect, both arms are elevated at 45°. The usual way to operate this signal with a bifilar transmission is to use a cam wheel with grooves on both sides, one side controlling the upper arm and the other controlling the lower arm. This cam wheel has the form shown in the diagram. The side operating the upper arm has a symmetrical groove, which will elevate the upper arm for a rotation of 90° in either direction. The side operating the lower arm has a groove that does not move the lower arm if the wheel rotates clockwise, but lowers it to 45° if the wheel rotates in the opposite sense. Therefore, Stop is displayed with the wheel in the position shown. If the wheel rotates 90° clockwise, Proceed is displayed, but if it turns in the opposite direction, Proceed at Low Speed is shown. Three aspects are obtained with one pair of wires, which are pulled in opposite directions to display the two Proceed aspects.
There is a constriction in the groove at point "x", which is normally not reached by the cam follower. However, should one or the other of the wires break, the tension in the other wire will pull the wheel around to this point, where the cam follower will be wedged. At this point, however, both arms will be in their normal positions, displaying Stop, as required in this case. The tension on the unbroken wire is ensured by the tensioning apparatus until it has moved a sufficient amount.
The point mechanism has pawls that are held clear of the wheel by equal tensions on the wires. Should the tension on one wire vanish, the pawls hold the wheel fixed, and the points do not move. In addition, if the signals are cleared, the mechanical detection of the points and bolt prevent the the bolt from being withdrawn and the points moved. There are similar pawls on the lever in the signal cabin to protect the signalman in case of a breakage if the catch is released, and to warn of the existence of a breakage. We see, therefore, that the two hazards of temperature change and wire breakage are both well provided for.
Austrian semaphore signal arms were operated directly by wires taken up the signal post to the arm. The single arm signal, as well as distant signals and shunting signals, showing only two aspects, used a very simple Sicherheitsvorrichtung (Security mechanism) that protected against wire breakage. It should be remembered that both wires were under considerable tension, maintained by weight-operated tensioning apparatus, so that the wires would always have a positive tension and not hang slack or lose tension around the lever wheel. If the wire returning the signal to its most restrictive aspect should break, then without any security mechanism the other wire would quickly move the signal to its least restrictive aspect, a dangerous condition.
The very simple security device for 2-aspect signals is shown at the right. It consists of two levers, each pivoted in the part to be moved (at points c and e) and held together at pin b. In equilibrium, the wires exert equal torques about the axis d. If the wire at f breaks, the upper lever immediately rotates and slips out of contact with the pin b. The lower lever, no longer supported, swings down with the help of weight w and now acts on the other side of the axis d, pulling the semaphore arm horizontal. In normal operation, this weight also helps to counterbalance the weight of the arm, making it easier to raise. Should the other wire break, there is no immediate danger, but the levers also fall apart and the final result is the same.
The two-arm signal displays three aspects: Halt (Stop), Frei (Proceed), and Frei mit 40 km/h (Proceed diverging). Unlike the two-aspect signal, the normal middle position of the wires corresponds to the Halt aspect, while movement in one direction gives Frei, and in the other direction, Frei mit 40 km/h. That is, one 180° lever can be used for all three aspects. The top arm, pivoted at a, will move to horizontal under its weight, while the bottom arm, pivoted at b, is weighted so that it will hang vertical. The top arm can be adjusted to be horizontal by adjusting the screws at g, while its upward mvement is limited by a stop on the rod pivoted at f and running through rotating guide g. The two arms are connected by rod cd that pulls the upper arm upward if the lower arm moves outward. A slot at c (whose limits are indicated) allows the upper arm to rise while the lower arm remains vertical. The signal is moved from its normal Halt aspect either by raising the top arm alone, or by moving the lower arm outward, which raises the upper arm simultaneously.
The method of operating the arms is shown at the right. The wires rotate the security pulley, which rotates about the same axis a as the upper arm, but is not connected to this arm. The security pulley moves a right-angle lever tbr pivoted at the axis of the lower arm b by means of rod st. If the security pulley rotates anticlockwise, point r on the lever moves to the left, pulling down on the connection to the upper arm, and raising it. This stops when the lever contacts stud p fixed in the lower arm, locking the lower arm vertical (preventing the movment of the upper arm from affecting the lower arm). The aspect displayed is then Frei.
If the security pulley rotates to the right, then the lever tbr acts on the stud p and moves the lower arm outwards. The upper arm is then also raised both by the motion of the lower arm, and by being pulled down by the lever rq. Clearly, the upper arm cannot be raised without the lower arm also coming to a diagonal position, since it is moved only by this action. The faulty aspect of a horizontal top arm and an inclined lower arm is also ruled out. The aspect displayed is Frei mit 40 km/h. The normal stroke of the signal wires is 250 mm in either direction, while a 180° lever provides a 500 mm stroke.
It is now easy to understand how the security device for the two-arm Austrian semaphore works. In the figure, note that the wires 1 and 2 tend to rotate the security pulley in opposite directions, that they overlap, and are fastened at two studs k and m on levers pivoted at h and f. Where wires pass around pulleys, as in this case, they are generally replaced by chains to avoid repetitive bending of the wires. The chains pass over the free ends of the levers at e and g, holding them down. The weight w, on the other hand, acts to push these ends upwards, acting through a chain of members acbneg. Weight w actually performs three functions. Since fulcrum c is suspended from the tail of the upper arm, w acts to counterbalance the weight of this arm, as in the one-arm semaphore. Should one or the other wire break, one or the other of the levers ef or gh can move, shedding the other wire from its stud. This means that if one wire breaks, the other is also disconnected from the security pulley. Finally, if this should happen, point a sinks enough that weight w falls off the end of it. Now w hangs on a chain that passes around a small pulley (not shown) and pulls the upper arm horizontal. This will also pull the lower arm vertical, due to the connection between the arms. Therefore, if either wire breaks, the signal returns to its normal aspect of Halt.
Now let us look at how these signals were operated in practice. At a typical small station, the points were often operated by hand, and main-line points were interlocked by means of locks and keys, as in the French Système Bouré. The keys fit into a central lock (Zentralschloss). If the key guaranteeing that the signals were normal was in the lock, then a key that permitted points to be changed could be removed. It was used to unlock the points. When the points were reversed, this key was trapped, but another key was released showing that the points were in the reversed position. When this key was inserted in the central lock and turned, it released a key that in turn permitted the signals to be appropriately cleared. This involved a lot of keys and key handling, but was very inexpensive. In these circumstances, the signals were often operated by a lever frame on the station platform, where they were easily available to the official managing the traffic (the Fahrdienstleiter, Fdl for short). This mechanism was called a Perronstellwerk or platform frame. These robust devices survived in the open, with at most small roofs over them.
An excellent example of this is displayed on the Steyrtalbahn Museum Railway at Grünburg in Austria. There are two two-arm semaphores, a distant signal, and a shunting signal, all of the old Austrian patterns, controlled by a platform frame, and a central lock that releases the platform frame by means of keys. The platform frame illustrated is similar, but the actual frame has more space to the right of the right-hand lever, where the locks are placed. This model 12-SA frame, dating from 1897, is similar to many found on Austro-Hungarian railways and seen in museums. Great care has been taken to ensure an accurate representation of signalling apparatus on the Steyrtalbahn.
Each lever is attached to a grooved pulley that carries a chain attached to the operating wires. The three-aspect levers can be pushed in so they do not protrude inconveniently in the normal position. The latch is very different from that illustrated above. On the left of the lever wheel is attached a disc with latch notches. The latch is held in by a weight to the left of each lever. When this weight is lifted, by means of a convenient projection, the latch is moved downward, and the lever can be moved, the latch piece riding on the periphery of the disc. The disc for a three-aspect lever is shown at the right in the normal position. If the latch is lowered a certain distance, the lever can be moved upwards to display the Frei aspect. However, since the latch is not fully raised, the lever cannot be moved downward to display Frei mit 40 km/h. When the latch is fully lowered, the lever could be moved in either direction, but a stop (as shown) prevents the upward movement when it strikes a locking piece, so only Frei mit 40 km/h can be displayed.
The interlocking is provided by stout bars moving transversely. In this frame, there are two such bars, one for the direction of movement (arrival or departure) and the other for the route (track 1 or track 2). The locking bars are moved by levers at each end of the frame. The direction lever, on the right, has a central position in which no signal can be cleared, allowing a key to be removed from its lock on the frame and carried to the central lock to release it. Locking pieces riveted to these bars interfere with the lever latches as required, preventing them from moving, or permitting only certain motions. The latch notch for the reversed positions are shallower than for the normal position, so that when a lever is reversed, the projection of its latch prevents the locking bar from moving. this is a typical German interlocking machine, very different from British or American practice where levers are interlocked with each other.
The two rightmost levers are interlocked by a very simple device, called a Quersperre or cross-lock, that prevents the distant signal's being cleared when the home signal is at Stop. This is managed by the small handle between the levers that throws a bolt right or left that passes through holes in the latch discs. When both levers are normal, the bolt is in the distant signal lever's disc and prevents it from moving, but is removed from the home signal lever's disc, so the home signal can be reversed when the other interlocking permits. In this position, the handle can be moved to insert the bolt in the disc of the home signal's lever, preventing the signal from being returned to normal. This removes the bolt from the distant signal, which can then be cleared. The home signal cannot be returned to Stop unless the distant signal is returned to normal and the bolt moved to the right. This is very similar to the device employed when typical US or British levers are side by side, and a tab on the home signal lever is in front of the distant signal lever, preventing it from being reversed unless the home signal is first reversed, and then ensuring that the home signal cannot be replaced unless the distant signal is first replaced.
Two levers can also be used to pull a bifilar transmission in two directions. In the diagram, lever A pulls the bottom wire when it is reversed, and lever B pulles the upper wire. The wheels on the levers are loose on the axes and provided with ratchets so they will rotate freely in the reverse direction. This is a very common arrangement for operating and locking points. With both levers normal, the points are not locked and can be operated manually. When one lever or the other is reversed, the points are moved to the required position and locked. This method of using two 180° levers was by far the most common method of pulling the transmission in two directions. Often, two independent signals were controlled by the same pair of wires.
Another method used cranks that were rotated in one direction or the other through 360°. Just above the crank was an index in the form of an arrow that would point to the right or left to show the direction in which the crank had been rotated, since the crank itself would look the same in both cases. Usually this was coaxial with a route lever handle that was normal when vertical, but was moved to one direction or the other to select a route. If the crank was not locked, it was pulled outwards toward the signalman, in which position it was clear of the neighbouring cranks and could be rotated, and finally pressed in again. This was equivalent to operating the catch handle on a normal lever. The crank rotated a drum over which the operating wires were led to move them. Rows of such cranks were often seen on the platform, where they were operated by the Fahrdienstleiter or equivalent. They were used only to operate signals and similar devices, not points, for which they were considered unsuitable.
A common arrangement had the Siemens and Halske block instruments in a row of four in an upper housing, with the signal handles at each side of a box containing the interlocking apparatus, which was directly dependent on the block instruments above. The two signals were the starting signals for each direction at the block post concerned.
B. Dieu, Histoire de la Signalisation Ferroviaire en Belgique, Tome II (Bruxelles: P. F. T., 2002). pp. 222-225 have some good illustrations.
C. Hager, Eisenbahn-Sicherungsanlagen in Österreich (Wien: Verlag Popischl, 1994). p. 43 shows two pairs of tensioning apparatus for operating semaphore signals.
E. Mazac, Einführung in das Eisenbahn-Sicherungswesen, Österreichische Gesellschaft für Eisenbahngeschichte Steyrtal-Museumsbahn, 2006.
D. Wurmser, Signaux Mécaniques, Tome III (Grenoble: Presses & Editions Ferroviaires, 2008). pp. 79-81.
Freiherr v. Röll, Enzyklopädie des Eisenbahnwesen, 2nd ed. (Berlin and Wien: Urban & Schwarzenberg, 1912). Articles Blockeinrichtungen, Stellwerke, Spannwerk and Signalwesen.
Composed by J. B. Calvert
Created 28 June 2004
Last revised 26 December 2008