SCMES Signalling System 2012
Designed and built by
The ground level track at Little Hay was until recently a relatively simple single-track layout around our 2-acre site for 7 ¼” and 5” gauge trains. It had a full three aspect signal system which was controlled by a BBC microcomputer. This has served the club well for many years.
Four years ago, a large programme to renew and considerably extend the track commenced. This included making a double track for much of the circuit with cross track route options and a place where the double lines have to safely merge into a single track again.
It was decided that the increased complexity of the new layout would overstretch the existing system which was twenty years old. There followed extended debate over the type of central controller to be employed. We considered a relay system, adapting a personal computer, making our own computerised controller and a professionally made industrial computer.
The relay solution was simple in principle but difficult to maintain and awkward to modify as the project developed. Modern PCs are not really designed for this purpose where the computing requirements are trivial but where the number of inputs and outputs are large. PCs are also not very reliable with their complex operating systems and hard drive storage.
We had the talent in the club to make our own computerised controller, but we feared that only the designer would ever fully understand it and there were health and safety concerns as well.
We finally decided to invest in a professionally made Programmable Logic Controller (PLC). This is the type of industrial computer that is used to operate everything from theme park rides to factory conveyor systems. It is ideal because it is electrically and mechanically robust, completely modular and extendable. It is designed to have a multitude of inputs and outputs (I/O), its memory is non-volatile, and its operating system is basic.
This was a significant decision for the club because PLCs are expensive, and it is difficult to master all the aspects of their application. It took us many months to gain this experience.
At the beginning there were many different views amongst members on the positioning of signals and the method of controlling routes. There was also concern about how the software could be tested with a system spread over two acres. This was solved by building a large mimic panel with all the
train detectors, points and signals on it and driven by the PLC. This proved worthwhile because it enabled interested members to debate all the layout issues and for the software to be developed.
By the start of 2011 we started onsite installation while keeping the existing system going. By April 2011 the new track was almost finished so we had a month’s shut down when the old signal system was removed and the installation of the new was completed and tested. By the end of April, it was quite usable but areas where improvements could be made soon became fairly obvious.
During the following ten months the system has gradually been improved in all areas of its design and this remains an ongoing process. Reliability of I/O terminal equipment is the main focus although the software is still being improved.
It was decided to employ two train on track detectors in each rail section; one just passed each signal and another 45 feet further down the track. This was based on a maximum train length of 45 feet. This arrangement enables the local signal to show red as soon as a train passes it and for the distant signal to go from red to yellow when the train reaches the next detector.
With fifteen signals we required 36 detectors allowing for some special use detectors. The detector design being used after a great deal of experimentation is a treadle system fitted to the common rail. The treadle operates an arm that moves in front of a solid-state sensor.
The three aspect signals use three bright LEDs for each colour. These are wired in parallel, but each has its own dropping resistor so that a failed LED does not affect the others.
We have five motorised points. These are controlled by signals from the PLC and the blade positions are sensed by solid state sensors. 12v motors drive a crank used to move the point blades. As trains are only permitted to travel around the track in one direction floating frogs are employed on all trailing points. This reduces the number of driven points considerably.
There are two methods of setting routes. They can be set by drivers at switch boxes just before the main routing points or by an operator in a signal box close to by. When set by drivers the points revert to a default setting once their train clears the junction.
The station Approach posed a greater challenge because trains run out of the controlled area when they reach the station, and it is important to distinguish between trains clearing the approach and those that stop in it. A wheel count system was installed and worked well with nearly all trains but there were exceptions. A laser beam along the curved is proving more reliable. (Laser head Laser Mirror)
A modern PLC was purchased and equipped with 80 digital inputs and 80 outputs all direct wired to the terminal equipment. All the logic is contained within the PLC written in ladder software language. A basic PLC has no means of seeing what is happening inside it but ours has three Ethernet ports. One of these feeds a Wi Fi hub which enables us to see the operational situation anywhere on site using a laptop or on a large display in the club house.
This graphic mimic display is active so that most inputs can be initiated from it. The access to this feature needs to be controlled but it is very useful for giving demonstrations and for testing software.
The 160 cable cores leaving the PLC are terminated on a wiring frame in the signal box. The multi-core cables from around the site are also terminated on this frame and connections are patched between the two sets of cables.
The main site cables radiate out to four distribution boxes around the site. Most of the terminal equipment is wired directly to these boxes.
All the site cabling is below ground drawn into ducting. All connections are made above ground.
The system works well but with 160 bits of terminal equipment around the track one failure per working day, a rate of 0.6%, still causes annoyance. It is mechanical moving parts that cause nearly all the trouble. The PLC cannot be blamed for any of our teething problems, it has worked perfectly.
The train detectors are the thing that cause the most failures because there are so many of them and they have a hard time. We are presently looking again at ways to improve reliability. Ideally to find a solution with no moving parts, which can be installed easily.