Automatic Signalling (G Gauge in the garden)

[Jamesd]

Hi all 

 

I have been building a garden railway G gauge for a year and I am about to move on to Singling using 2 aspect colour light signals.  I am looking for some suggestions and ideas on how to automate it.  The Railway is used for Battery RC locos and Live Steam and has no track power ( I hate track cleaning )

 

The track is a single line with a loop at each end and a spur off one of the loops to a station area.

 

My aim is to break each loop in to 2 or 3 blocks (One loop is much larger than the other)  and the single line between as a single bidirectional block.

 

My ideal outcome is that the remote drivers can follow the automated singling on the loop and the Station is manually signalled from a control box. 

 

I have considered reed switches, Hall sensors and infrared detection but I am open to all ideas and feedback with either relays or a PI and or Arduino to do the logic.

 

Looking forward to ideas and suggestions?

 

Many thanks 

 

James

 

 

 

[Nigelcliffe]

Automatic signalling seems to imply track detection of occupancy.  Which means track circuits of some sort for the occupancy to work.  Which appears to be a non-starter for reasons you've outlined.

 

Why not step back a little and look at the prototype ?  In the past, it might have been a token/staff system, or tickets issued for trains following each other, with the token going to the last train in the group.   
The token is carried by the train.   Token inserted into ground frame or passed to signalbox to unlock the signals/track.  Track and signals then set for next movement, appropriate token withdrawn and carried on train to next place. 

 

Going a little more modern, radio based token system have been used on some lines (eg. NW Scotland).  Might be able to implement something using that approach;  a token carried in each loco indicates position as it approaches the signals.   Then the signals need only have local power supply (battery?) plus their radio token mechanism.  

 

 

Infrared and outside generally doesn't mix well due to the sunshine !

 

 

- Nigel

[AndyID]

I think reed switches would be your best bet although you'll have to weatherproof them and their connections. RTV should protect them for some time.

 

If you decide on a maximum train length it's not too difficult to use logic to determine when a section is occupied. Bi-directional operation can get a bit tricky.

[cliff park]

Reed switches are surprisingly robust, and of course the glass tube is hermetically sealed. Use heatshrink for the connections, and a larger gauge heatshrink to enclose the glass tube, against mechanical breakage, and they should be pretty bombproof. Also remember it is possible using different positions for reeds and magnets to identify individual locomotives etc.

[Jamesd]

Many thanks for all the feedback on my post :-)

 

The radio idea from Nigel would be fun bit looking at the parts needed quite expensive compared to reeds and magnets (so maybe one for another time).

 

Im going to either go with reed switches or 24v AC track circuits.

 

If any body is happy to advise on how to build the logic?  im thinking either a PI based solution or using relays? Be great to understand what others have done.

 

Thanks agin for all the replys.

 

James

[Nigelcliffe]

Potential pitfalls of the two processes, which will come down to wiring lengths...

 

24v AC track circuits.  Probably reliable (subject to wheel and track state!), but harder to interface to any logic circuits.  I'd look at opto-isolators to get things into any logic system (PI or Arduino). 

Reed switches.   Easier to interface (arguably just using the voltages on the processor logic), but runs risk of voltage drop and loss of reliability over distance (depends on wire quality, and more importantly quality of joints in wires).   Higher voltages over the switches will help with voltage loss, but then has same interfacing issue as the 24vAC idea. 

 

*if* you can make the signalling logic work (see no reason why not), then a PI with JMRI on it could do all the signalling logic/interlocking stuff without needing to write code.  This could be tried out on a normal computer in a simulation mode to prove enough worked before investing in actual hardware and wiring.  The GPIO pins on the PI are available as input/outputs to JMRI, so provided 24 pins is enough for your system, that's your hardware interface. 

 

  

[AndyID]

13 hours ago, cliff park said:

Reed switches are surprisingly robust, and of course the glass tube is hermetically sealed. Use heatshrink for the connections, and a larger gauge heatshrink to enclose the glass tube, against mechanical breakage, and they should be pretty bombproof. Also remember it is possible using different positions for reeds and magnets to identify individual locomotives etc.

 

Heatshrink is good but it won't exclude moisture. The leads on reed switches will corrode as will solder joints in wet conditions (the fuel gauge sender in my boat has confirmed this.)

[AndyID]

11 hours ago, Nigelcliffe said:

Easier to interface (arguably just using the voltages on the processor logic), but runs risk of voltage drop and loss of reliability over distance (depends on wire quality, and more importantly quality of joints in wires).  

 

 

Unless the signalling currents are huge voltage drop isn't going to be a problem but voltages induced in long wires connected directly to the inputs of any processor are quite capable of destroying the processor. It would be best to use opto-isolators on all inputs from the layout.

[ColinK]

If you haven’t already joined, I suggest you sign up to G scale Central (Its free) as its a similar forum to RMweb, but specifically for G scale with lots of people who have experience of garden railways.