One of my reasons for adopting DCC was my requirement for some degree of automation of the layout, so that it could be operated single handedly in as near a prototype fashion as possible where using traditional cab control would need several operators.

There is a pre requisite of automation; the DCC system needs to know where the trains are. I use the term DCC system in the widest sense, and this needs to comprise 3 elements; the DCC control unit; a feedback detection system; and a computer that is connected to the DCC control bus to provide the intelligence. My LENZ system provides both a feedback bus to which detectors can be connected and a computer interface to Xpressnet to allow a PC to be linked into the system. In my case I’m using JMRI Panelpro as the control system on the PC.

I’ve been wrestling with the physical means of detecting trains for some time, and the re seem to be 3 main options.There is almost no guidance available on the subject, so I wanted to share my thoughts on the subject just in case they are of some use to others. The main options appear to be:

  • Block Occupancy detection requires the layout to be divided into individual blocks, each one isolated from the next, and the power to the block fed throgh the detector. The detector will sense any current being drawn by a train within the block
  • Infra red detection relies on a sensor module that produces an infra red beam and is able to switch state when a train is above the beam and the beam is reflected back on the sensor.
  • Magnetic detecion involves fixing magnets to the bottom of trains and these activating reed switches placed at strategic points on the track.

Needless to say there are pro’s and con’s of each, and I’m currently not convinced that a single method can meet the full requirement.

The key difference is that block detection provides constant detection of where the train is as it passes around the layout while the others only provide detection while the train is above the sensor.Most PC software seems to assume the existence of constant detection.

Block detection relies on detecting current being drawn from anywhere within the block and it will only report the block that the locomotive is in and ignore the rest of the train. This means that the train is recognised as soon as the locomotive enters a block, but the block becomes clear once the locomotive leaves, and not when the last car moves out of the block. The whole train can be detected if individual cars can all be made to draw current. Carriage lighting achieves this, as does the fitting of resistive wheelsets (metal wheels where a resistor is connected between the wheels to draw a small current). Of course if automation is to work successfully all trains must behave in the same way so if one train has carriage lighting others without this feature will need resistive wheels. While block detection will constantly report which blocks are occupied by trains it wont report the exact position of a train within the block and this makes triggering precise stopping difficult.

Magnetic detection will record the precise location of a train for stopping purposes as detection will only be triggered when the magnet under the train is directly on top of the reed switch. Magnetic detection is however incapable of detecting the whole train or providing continuous detection. The use of two detectors per block could enable sensing of a locomotive both entering and leaving a section, but not of the whole train clearing the section. One advantage, reed switches and magnets are very cheap, but it relies on every loco being fitted with a magnet and this might be a problem for visiting stock.

Infra red detection does detect the whole train, and a sensor at the start and end of each block could provide more accurate information than magnets. The disadvantage is the cost of these detectors at over £10 each. The use of 2 per block can soon get quite expensive. Of courseto save money the entry sensor for one block could be physically the same as the exit sensor from the previous one, and logic applied within the pc software keeping entry and exit detection separate. IR detection can also be made to provide continuous detection provided the space beween detectors is less than the length of the shortest train. In this way the train is always above at least one detector, but this increases the number of detectors needed and block size and position is dictated by train length and not appropriate start and finish of blocks for signalling purposes.

Infra red detection was my original method of choice. I’ve since changed my preference to block detection. The main driver is cost. The lenz system uses an LR101 module to link detection to the feedback bus. each LR101 can be fitted with 8 detectors. If an IR detector costs £10-£12 and an LR101 costs over £30 then the cost of detection adds up very quickly. Block detection using the Lenz system is equally expensive as Lenz block detectors also require attachment to the feedback bus via an LR101. Recently I discovered a module manufactured by an independent supplier but compatible with the Lenz system this combines 8 block detectors and a connection to the feedback bus on a single circuit for around £50 thus dispensing with the need for an LR101 and achieving detection for less than £7 per block.

I’m rapidly concluding that I probably need to adopt more than one method of detecion as there are points on the layout where I’ll need to trigger very precise stopping. I’ll use the infra red detectors to provide this (I have a number of IRDOT units already in stock awaiting fitting to the layout) and use the LDT RS-8-F modules for block detection. I’m currently revisiting the trackplan to determine the start and finish of each block before track is altered to provide isolation between blocks and power feeds are rerouted through the detectors.

Re wheeling rolling stock looks like its going to be a massive job, and currently I’m investigating how best to create resistive wheelsets. Current thinking is that as long as the front and back of every train can be sensed I dont need to bother with the stuff in the middle.