THE SIGNAL BOX
LOCKING FRAME TESTING
Notes for the 21st Century S & T Engineer regarding testing 19th Century interlocking technology.
by Peter Woodbridge
Compiled with assistance of notes of tuition gratefully received from earlier generations of signalling testers and a little practical experience.
Brief Introduction to Mechanical Locking.
There are two basic ways of applying mechanical locking to a lever:
Whilst there was a variety of interlocking methods invented, the vast majority of locking are forms of tappet locking, though in each case there is a variety of methods of actually transferring this locking to the lever.
A typical arrangement has a steel blade called a tappet connected directly to the lever, generally just below the signalbox floor. Sometimes there are several tappets connected to a lever at various heights but there are limits to where they can be connected- too far from the pivot and the travel is large giving problems with conflicting ports, too close to the pivot and the locking has to be immensely strong because of the mechanical advantage of the lever action aiding an enthusiastic signalman.
Each tappet fits in a race which is a vertical slot milled across a locking tray. This is a cast iron plate which has horizontal dividing strips (called ribs) separating it into many channels which hold the locking dogs or nibs. These are the actual locking pieces which engage in the tappets to perform the interlocking. Locking dogs were usually cast iron though, due to the difficulty in obtaining it, mild steel is normally substituted nowadays. They come in many different shapes and sizes; each type of frame has a range of different types of dog (the simplest Stevens frame has 7 patterns, Brighton locking has 17 and a Westinghouse frame has nearly 100).
Direct lever locked frames often have several layers of locking trays each with its own set of tappets in order to contain all the locking required. Certain designs of lever locked frames, and all designs of catch handle locked frames, require the tappet to be connected by means of steel linkages and escapement mechanisms. Sometimes these lead to multiple locking trays with either a succession of tappet blades (e.g. GWR 5 bar) or multiple tappets driven from the same mechanism from the lever (e.g. GWR 3HT).
Pros/Cons of indirect connection are:
Generally all levers in a frame will usually stand Normal (i.e. to the back of the frame). Points are generally set for straight running and signals are (for all usual frames) at danger when the levers are normal.
Most mechanical locking is done via the points levers (level crossings, miscellaneous releases etc. are generally regarded as points for this purpose):
Lever leads tell the signalman which levers have to be in the Reverse position before the relevant lever is free to be pulled. These generally do not state that the lever is locked whilst certain other levers are reversed, the signalman is expected to keep a tidy frame and replace (i.e. normalise) levers when they are not specifically required.
There are of course exceptions to this rule; in particular FPLs (facing point locks) are sometimes designed to "stand locked" and sometimes designed to "stand unlocked". On the GWR the most usual situation was for the FPL to lock the points in position when reversed and hence it is quite usual to see the blue levers reversed and all else normal in a frame when no train is signalled. This is the more consistent approach as far as interlocking is concerned, allowing signals reading over the points to be released by the FPL, though it does result in there always being the odd lever reverse which can be a nuisance in a narrow signalbox if there are too many people about.
Nomenclature used for the interlocking that one lever imposes on another is:
Outline of Operation
When a lever is pulled its associated tappet moves along the race in the locking tray. There are notches or ports in the side of the tappet and locking is achieved by dogs that are constrained to slide at right angles to them in the channels.
Dogs are made so to be a good sliding fit in the channel. There must be no sloppiness allowing them to twist within the channel between the ribs of the locking tray; however they must not be so tight as to hinder their free movement along the channel.
Dogs act as wedges which fit into the ports of the tappets and can then prevent the tappet from moving if that dog is firmly held in position by the locking requiring to be imposed on that lever. Dogs are shaped with a nose on one side which has a carefully shaped ramp or bevel on it. It is this bevel that allows motion of the tappet to impart a sideways motion to the locking thus allowing the dog to move fully of the way of the tappet if the lever is free to be moved. Bevels must be cut to the right dimensions or else the locking wont travel to its proper extent.
Whether a dog locks a tappet or allows it to move depends on the locking imparted to it by bars that connects it to other dogs. These bars are also housed in the locking channels, most types of tray accommodate four bars- two above the tappet and two below them. On Stevens frames only three bars (one bottom, two top) are used and whilst the earlier Great Western frames (3HT, 3VT) also only had three bars, the later variety (5VT) had five (two bottom, three top).
Note that dead locking (sometimes called absolute locking) results in automatic converses. If locking is provided to cause lever 5 to lock 12, then the same pieces of metal also impose 12 locks 5. The two dogs are physically linked together and cannot move relative to each other- at any one time one of them must be in the ports of its tappet (preventing the associated lever from moving) so that the other can be clear of its tappet. Whilst this reciprocal locking is inherent in mechanical lever frames it has to be explicitly provided if locking is performed electrically (whether this be via electric locks on lever frames, relay interlocking driven from a panel or within computer software from a VDU based Signalling Control Centre.
It is worth noting that anything that has to be explicitly provided can be inexplicably forgotten, so always check for it. A well intentioned designer, trying to avoid having to implement new mechanical locking because of installation skill shortage, might decide to provide electric locks instead, but then only design half of the locking required.
Other categories of locking are:
Bothways locking is achieved just as for normal dead locking but, as the name implies, there is a port cut in the tappet for both positions of one of the levers. Hence, provided that lever is fully in either position, the second lever will be free. Once that lever is pulled however, the position of the first cannot be changed.
There are several means of achieving conditional locking by the two main ones are:
Diagrams depicting Locking
All the locking contained within a lever frame is drawn on a Dog Chart. This is a pictorial representation of the locking trays showing the ports in the tappets and all the lock dogs and the bars to which these are attached.
The interlocking that is achieved by a locking frame is summarised on a tabular drawing called a Locking Table. There are many varieties of these diagrams, although most railways standardised on their own particular format. These generally consist of the following columns (although the wording and presentation do vary, and some types have additional columns):
Lever positions are always regarded as meaning Normal unless otherwise specified (most presentations suffix the lever number with a R if Reverse is meant, although the Southern Railway convention was to show the lever number within a ring to denote this). The need to show this generally occurs in relation to conditional or bothways locking:
Testing of Lever Frames, general
Lever frames need to be tested after any alterations to the locking have been made (New Works) and also on a regular basis (Maintenance) to ensure that the frame is in good condition and will continue to operate safely.
In the latter case the Locking Table and Dog Chart can be assumed to be correct (although the tester should be aware of any change to the train service pattern or other external influence which might mean that assumptions made when the frame was originally commissioned are no longer valid). In New Works testing, the tester needs to validate the locking achieved by the frame against their knowledge of the Rules and Regulations of the railway, the mechanical signalling interlocking principles etc.
The aim of the testing is to:
Organising Before Testing
For New Works or Remedial work it may be necessary to disconnect some or all of the points and signals from the frame and arrange for full handsignalling (and possibly a restricted train service as a result).
In minor alterations where the integrity of the locking has been disturbed, the best solution is often to leave all external functions on the frame but arrange for the operating department to provide a "check signalman". He acts as a second person to ensure that the signalman does not make an error without the benefit of an effective interlocking. In these cases the distant signals must be disconnected.
Routine maintenance testing (when the locking is being checked for wear but has not been disturbed) may not require any formal possession. It is usually best achieved "between trains" having reached a clear understanding with the signalman about which levers may be operated at any time and which must be left alone.
For small frames one assistant to pull levers and record the testing with another under the box to lift electric locks etc. is generally ideal. More complex frames would require more lever pullers (say 1 for every 30 levers in the frame), one ticker (recorder of the testing) and (1 or 2) lock lifters under the box. This allows the tester purely to concentrate on specifying the locking required, negotiate with the signalman and arrange the test sequence.
The ticker must understand the particular design of locking
Table and the significance of their task in ensuring that
a 100% complete test is undertaken.
Walking mud into the room, leaving the door open, touching a lever without a duster and sitting in the signalmans chair are just a few of the ways for the job to go badly wrong before it even starts.
Conversely arriving with cream cakes for the signalman (and his mates on the other shifts) can pay dividends later when you need an extra minute to get a test finished before the signalman is able to pull off for a train; it also virtually guarantees an offer of a cup of tea at frequent intervals throughout the work
Locking Room Checks
Each individual lock MUST be tested separately; if a lever is supposed to be locked by two functions you cannot tell whether only one is effective or both. Hence you must always prove that a lever is free just prior to doing the one thing which you are going to test locks it. ALWAYS PROVE FREEDOM BEFORE PROVING THE LOCK, PROVE THE LOCK AND THEN PROVE FREEDOM AGAIN. You need to do this to be sure that you know what you have tested.
As explained earlier, reciprocal locking is provided automatically by mechanical frames but it must be tested in each direction since the drive and locking directions are reversed wear on the various surfaces can have different effects on the security of the locking. If a lock is good but its reciprocal is slack suspect first wear on the linkage between the tappet and the lever rather than the interlocking of the dogs and tappets themselves.
Suppose that you need to test the dead lock between levers 5 and 12 and, having started at the low end of the frame lever 5 is the first one we are considering which is referred to as the object lever and it has just been proved free at the end of the previous test (rule1).
Since you will need to pull 12 to check that it is free (rule1), if you actually pull it right over whilst trying 5 to prove it is locked the lock can be tested in this direction initially. Replacing 12 will then allow 5 to become free and as soon as it is moved from normal attempts to pull 12 again should be prevented by the interlocking. It can then be replaced to prove 5 free again. This saves a lot of lever operations which can be very tiring after a full day and also lead to a considerable economy of time, which is especially important if you are having to work in margins between trains as you might have to wait ten minutes to do another minutes work.
Note that there is also a converse of a bothways lock in which a signal lever locks a point in whichever of the two positions it lies at the time. The converse is that the signal lever must be locked unless the point lever is fully Normal or fully Reverse.
Rule 3- permutate releases
A lever may require several others to be Reverse before it can be pulled. It is important not only
to test that the object lever becomes free when the last
of the lead levers is pulled but that each and every one of them is required
reverse. In order to discover any conflicting
ports careful methodical testing is needed to exercise
A common example is a distant lever (say 1) released by all (say 2,3,4) stop signals.
pull 2, try 1,
pull3, try 1,
pull 4, pull 1,
try backlock on 2, 3 and 4.
is an INCOMPLETE TEST and therefore UNSATISFACTORY as the only thing that has really been proved is the locking between 1 and 4.
You need to understand the design of the variety of the frame to determine what you must do. Generally you can complete the test by:
replace 2, try 1,
replace 3, pull 2, try 1,
replace 4, pull 3, try 1,
replace 2, replace 3.
This sequence is possible when there is sequential locking between the levers. Even this has not tested all possible permutations between the levers and for some frames all must be tested. This can result in a large number of individual tests and this number increases alarmingly with the number of levers involved within the release. It takes a significant time and since it is likely that you might have to stop or be otherwise distracted whilst performing the sequence it is highly desirable that the combinations are written down and marked off as the test is underway.
Rule 4- test the "free" move of conditional locking
The important thing to grasp about conditional locking is that locking should not always present. It must be tested:
Failure to test for the "free condition" would not detect the rightside failure resulting in the locking actually being a dead lock; whilst this may not be dangerous it will tie up the layout, cause delays and severely frustrate the Operating department and the passengers.
Rule 5- test for the counter condition holding
For every condition there must be a counter condition. The condition gives a degree of freedom and disables some locking that would otherwise be there, so the condition itself must not be allowed to change subsequently. For example, having allowed two levers to be reverse simultaneously only because a third happens to be in a certain position, it is now essential to stop that third lever itself being moved whilst the other two are both reverse.
Often this can be inherent in the locking, since the majority of conditional locking is reinforced by bothways locking. However sometimes it has to be provided explicitly and can be overlooked. IF THERE IS AN "or", "when", OR "if", ASK YOURSELF HOW DO YOU KNOW THAT THE THING THAT ALLOWED THE FREEDOM CANT BE SUBSEQUENTLY ALTERED.
If you see something like 21 locks (57 w 52) and 52 isnt dead locked by the relevant positions of 21 or 57 there has to be an explicit counter condition provided. The locking between 21 and 57 is only to be effective when 52 is normal and hence you must test that if 52 starts off being reverse, something prevents it being placed normal until either 21 or 57 are normalised. If this locking is missing this would be a wrongside failure.
Testing Locking Example
Consider a situation where disc signal 7 reads in two directions over facing points 8 and, if the points are normal, also needs to be released by the next disc signal 10. Its lever lead will say "8 or 10" which might be written on the Locking Table as: 7 by (10 w 9). This means 7 is released by 10 when 8 is normal and hence there is a 9 condition on the locking between 7 and 10 since there is no requirement to pull 10 if the points are set the other way. The need for the signal to lock the points bothways prevents this condition being destroyed whilst lever 7 is not normal. Hence there is no possibility of destroying the condition once the move has been set up.
To test the release:
pull 10, pull 7, try backlock on 10 [PROVES 7 BY 10],
replace 7, replace 10,
pull 8, pull 7, try backlock on 8 [PROVES 7 BY 8],
replace 7, replace 8.
Similarly if there is an opposite direction disc signal 9 that reads over the crossover points 8 in the other direction, the two disc signals need to lock each other but only when the points are reverse. Hence 7 locks (9 w 8R).
To test this lock and reciprocal and breakdown the condition:
pull 10, pull 7,
pull 9 (to check freedom also replace and re-pull 7, then both 7 and 9 in mid stroke),
replace 10, 9 and 7,
pull 7, try 9 [PROVES 7 LOCKS 9],
pull 9, try 7 [PROVES 9 LOCKS 7],
replace 9, check 7 free and replace.
Again 7 and 9 will both lock 8 so there is obviously no possibility of having cleared both disc signals and then moving the points.
However in some situations a counter
condition has to be provided specially and this is just one example. Trailing
points in rear of a signal would normally be locked bothways by it, however if there is an intervening facing
point the lock should not be applied if this facer is set the other way.
Suppose signal 12 at the end of an island platform loop with points 27
selecting which face is used by the incoming train and a trailing crossover 19
on the approach to the facing point.
with 27 reverse, 12 does not lock 19 [freedom with the condition broken down]
Details requiring careful checking.
The ports cut into the tappets have to be cut accurately so that they are opposite the relevant channel in which the locking is effected. The shape of the must exactly follow the nose of the dogs and if these surfaces are unable to fit closely then the result will be slack locking or in extreme cases a complete wrongside locking failure.
Any ports cut in a tappet will obviously move across the tray as the controlling lever is pulled. A problem arises if the motion which moves it away from the channel for whose locking it has been cut, takes it into another channel where there is other locking for that lever. This could mean that locking that ought to be held in position by the edge of the tappet is invalidly allowed to move and thus an unsafe wrongside failure of the locking would occur. These are known as foul notches or conflicting ports.
The tester must always be on guard to discover whether such conflicting ports exist. It is necessary to continuously apply pressure on the lever that is supposed to be locked throughout the entire travel of the lever under test as it is slowly moved. If at any point of the stroke of this lever there is a conflicts with a port cut for some other purpose, it will be revealed by the sudden loss of locking on the other lever.
ENSURE ELECTRIC LOCKS REINSTATED IF DISCONNECTED.
Although not strictly mechanical locking testing, it is often worth checking the setting of contacts on circuit controllers or separate contact boxes.
Recording the testing
Use a colour pencil (not red or green where these colours have been used to depict locking alterations) that is easily seen against the base diagram and is distinguishable from colours used by other testers if several are involved in directing the testing. Sign the drawings (Dog chart and Locking table) in the same colour to identify your markings.
There is no national standard regarding recording testing; you must come to a clear understanding with your fellow testers prior to commencing work.
Continue testing one lever until it is complete and then tick the lever number column itself.
When waiting for a margin to test between trains (or whilst travelling to / from site) update by ticking the entries for the other of the pair of levers that have been tested as a consequence of testing on the lower numbered levers. When all this ticking of the converses has been performed for a lever, put a cross tick on the tick in the lever number column.
Never allow anyone who you dont fully trust to act as ticker off on the Locking Table. Once there is the slightest possibility that incorrect ticks might have been applied, you are thoroughly lost and have wasted the testing previously performed. If a competent ticker is not available it is better to record in a note book the testing performed, and transfer this information subsequently.
Need for alterations
Determine whether the problem is one of design or implementation.
IF THE LOCKING NEEDS TO BE DISARRANGED (EITHER BECAUSE IT HAS BEEN INSTALLED INCORRECTLY, IS WORN ETC. OR BECAUSE THERE HAS BEEN A DESIGN CHANGE) YOU MUST PERSONALLY SUPERVISE THE REMEDIAL WORK. It is easy for a locking fitter to volunteer to do a couple of minutes work whilst you are doing something else but without personal observation you cant know exactly what he did and therefore what might need to be re-tested and more important what doesnt. Unless you have absolutely no doubt that any piece of locking was not disturbed you must re-test it and this can often be limited to the locking in a particular channel which passes a particular tappet. If however you are unable to be certain exactly what was disturbed, you will have to test the whole frame again.
Frequency of Testing
Some frame types are more subject to wear than others. Generally the more convoluted the linkage between the lever and the locking the more opportunity for loss of travel and slackness although where tappets are directly attached to the lever the locking itself can be subject to greater pressure. As an indication GWR 5 bar frames in good condition should be tested every 5 years unless locking is known to be getting slack, but frames such as the double twist often need to be checked annually, although the amount of usage is also a significant factor to consider.
Comments about this article should be addressed
to Peter Woodbridge