Automation of 3 Train Shuttle (2 sidings each end of a single line)

[TimC]

Somebody will probably have worked this one out already so rather than reinvent the wheel I'd appreciate some ideas.

 

I want to automate 3 DMUs running along a single line at the back of a layout on DC.  I'm happy to use either a series of relays or a shuttle unit but I'd prefer not to have to build a MERG unit or other 'build yourself circuit board' controller.    

 

Here is a diagram of what I'm planning and a sequence of trains.

 

 

If possible I'd also like to introduce a delay in between each train arriving/departing.

 

I've not used any train detection system on model railways before so I'm open to advice here also.

 

Apologies if this has been covered before I've had a quick search but didn't find anything.

 

Thanks in advance.

 

[Edit - typos....grrr!]

[Titan]

Check out Heathcote electronics, I think they may have what you require:

 

http://www.heathcote-electronics.co.uk/automatic%20train%20control.html#SA2

 

I think it is possible to use them with reed switches and train mounted magnets, which might suit better as using all the infra red detectors you may need could be a bit expensive.

[TimC]

Heathcote do a shuttle unit which can handle a siding at one end but I've not found a unit/set up where there are sidings at both ends and the trains sequence around.

 

I'm sure I've seen a diagram of a suitable circuit using relays but as usual now I want it can I find it......  Anyone else seen it and remember where it is - I think it was an American Model Railroad related site.

[Kenton]

Someone on RMWeb had a tram layout doing just what you want, I can't remember who or what it was called. I beleive it was even shown at an exhibition or two. It can certainly be done with the Heathcote sets and I have used them for similar 1 to 2 shuttle ISTR that they had that as an example on their website. They are very helpful and no doubt would only be to glad to help design a 2 + 2 setup (well on condition/likelihood of a sale perhaps). It should be very easy to do using the points at one end to trip the arrival point on an arrival to let out the adjacent train after a delay. The point controls which track gets power from the central single track and in which direction.

[Gordon H]

Somebody will probably have worked this one out already so rather than reinvent the wheel I'd appreciate some ideas.

 

I want to automate 3 DMUs running along a single line at the back of a layout on DC.  I'm happy to use either a series of relays or a shuttle unit but I'd prefer not to have to build a MERG unit or other 'build yourself circuit board' controller.    

 

 

Odd (to me) that you apparently seem happy wiring up a bank of relays from scratch, yet worry about putting a purpose made unit together with all the components and instructions provided - is it because it might involve a PCB?

Anyway, your scheme happens to match exactly with the MERG ATC Configuration C-3 (see Figure 3 here: http://www.merg.org.uk/merg_resources/atc.php), so this particular wheel has already been invented   .

Kit wise, all you would need is an ATC, a CSR4 and a PD3 point motor driver (assuming solenoid operation) and a bit of wire.

The ATC comes ready equipped for running this track layout, and several others as well, all selected by a built-in switch.

 

Obviously a suitable power supply is required too, no matter which control scheme is adopted.

[Kenton]

Having read the merg solution it appears incredibly complicated (at least to me) probably due to all the 'additional' built in options and that is before the build it all yourself (can go wrong) is taken into account. The logic really is not that difficult that it requires a mini computer to control. With only 2 points and a bi-direction supply plus some form of detection in each siding to act as switch. Sure the ability to accelerate and decelerate is a nice feature (and probably more than simply reassuring if all 4 ends are visible to the "public" but might not be what is required if this is a FY to FY operating environment.

 

It does look like a good 'cover all options' solution and may be the way the OP would chose to go but there must be something simpler

 

BTW end track detection is not absolutely needed. If the points can be switched then so can frogs and so can isolating sections. Each siding then simply has an isolated section (any train entering a siding is killed at reaching the isolation the exiting train when selected by the point "relays" than has its isolating sections closed and thus releasing it. (these could even be driven direct from the track power supply as closing any section is only dependent on power in one direction (out of that siding) all other times it would be open (while occupied or while empty to receive the next shuttle from the other end). Such a system can also easily get around the obvious problem of only using short sections (a loco at the front of one end siding is at the end of the opposite siding. So 2 end stop sections are required for each siding, the 'long' section being powered until it is ready to receive a 'loco last' pushed train. (could be permanently wired if you consistently used the same train configurations)

 

Both systems fall over when using different traction units. Unlike DCC there is no solution to that, though I guess someone would say the PIC in the Merg option could be programmed for each different traction unit. I'm sure there is a DCC equivalent - also called a computer.

 

As for a delay - well that is going to be down to some timer circuit. Each extra requirement like that takes you down the same route of probably bespoke circuit. I don't believe the problem requires rocket science but it could soon seem to be that way.

[Titan]

Kenton has a point - if you want something really simple but a bit crude, you could buy a simple polarity reversing timer like this:

 

http://www.ebay.co.uk/itm/DC-MOTOR-REVERSE-POLARITY-CYCLIC-TIMER-SWITCH-TIME-REPEATER-900-960s-1A-12V-/281671426779?hash=item4194ece2db:g:HP8AAOxy0QBSO0z4

 

Put diodes on to every isolating section, when the train goes in it stops, reverse the current and it comes out again. Put a magnet on the bottom of the train and use it to operate a reed switch to change the points - probably via a relay as I don't think a reed switch would like solenoid motor current very much - feed the power to the tracks from the middle and each one would run in turn with whatever delay you have on the timer.

 

If the reed switch to switch the points is also supplied via the track supply and has a diode, this will also stop the magnet of the train unintentionally switching the points back again as the train comes out of the siding - it does not make a difference to the relay coil which way the current happens to be flowing through it, so it would then only switch the points when the train is on the way in to the siding.

 

Just make sure that all your DMU's have the power cars the same way round so they stop in the same place!

[DIW]

Check out Heathcote electronics, I think they may have what you require:

 

http://www.heathcote-electronics.co.uk/automatic%20train%20control.html#SA2

 

I think it is possible to use them with reed switches and train mounted magnets, which might suit better as using all the infra red detectors you may need could be a bit expensive.

 

I have just dug out the instructions for  Heathcote shuttle module SA1.1. It is stated therein that a siding can be added at one or both ends of the line, and that a IRDOT-P is required at the end of each siding. The IRDOT-Ps  would be wired to the points and to the SA1.1. the SA1.1 takes care of acceleration, deceleration and delay timing etc. in accordance with the settings you set and the detectors you (also) wire to the SA1.1.

 

I have no idea whether other variants of the SA1 are also capable of operating sidings.

 

Use the link supplied by Titan to see the prices of the parts!

 

Dave

[Gordon H]

Having read the merg solution it appears incredibly complicated (at least to me) probably due to all the 'additional' built in options and that is before the build it all yourself (can go wrong) is taken into account. The logic really is not that difficult that it requires a mini computer to control. With only 2 points and a bi-direction supply plus some form of detection in each siding to act as switch. Sure the ability to accelerate and decelerate is a nice feature (and probably more than simply reassuring if all 4 ends are visible to the "public" but might not be what is required if this is a FY to FY operating environment.

It does look like a good 'cover all options' solution and may be the way the OP would chose to go but there must be something simpler

 

Any perceived complication is indeed entirely down to the flexibility the system offers as standard. Putting anything together yourself can always go wrong, of course, but you would have to do that to a certain extent whatever the hardware, even if it is only wiring it all up correctly. At least with the MERG ATC there are no trackside detectors to find room for, fit and wire, so that aspect is a lot simpler. Just wire from the modules to the tracks.

The use of a microcontroller makes the whole exercise so much simpler to implement, maintain and upgrade when desired. You can even upload your own track and sequence configurations if desired.

With hard wired systems even minor changes or improvements are likely to be much more difficult to devise and implement.

 

 

BTW end track detection is not absolutely needed. If the points can be switched then so can frogs and so can isolating sections. Each siding then simply has an isolated section (any train entering a siding is killed at reaching the isolation the exiting train when selected by the point "relays" than has its isolating sections closed and thus releasing it. (these could even be driven direct from the track power supply as closing any section is only dependent on power in one direction (out of that siding) all other times it would be open (while occupied or while empty to receive the next shuttle from the other end). Such a system can also easily get around the obvious problem of only using short sections (a loco at the front of one end siding is at the end of the opposite siding. So 2 end stop sections are required for each siding, the 'long' section being powered until it is ready to receive a 'loco last' pushed train. (could be permanently wired if you consistently used the same train configurations)

 

 

 

This sounds just as complicated (to me) as the MERG solution.

 

Both systems fall over when using different traction units. Unlike DCC there is no solution to that, though I guess someone would say the PIC in the Merg option could be programmed for each different traction unit. I'm sure there is a DCC equivalent - also called a computer.

 

 

 

Not necessarily. If different traction characteristics cause real difficulties, there is another module which can be added to the system which gives individual control for each item. It is generally accepted that large discrepancies between mechanisms can a problematic for any such automated system .

 

 

As for a delay - well that is going to be down to some timer circuit. Each extra requirement like that takes you down the same route of probably bespoke circuit. I don't believe the problem requires rocket science but it could soon seem to be that way.

 

Delay is merely another parameter in the microcontroller operation. No need for extra circuitry at all. Read a value from a control Pot, use the value as the basis of a counter and away you go.

[Kenton]

I was knocking the Merg system, just commenting on the complexity (which may not be required by the OP). You cannot get away from the simple fact that a few wires is easy to test with a multimeter that probably everyone has. Start adding components to a circuit board and the potential faults multiply very rapidly. Many are aware of that and it puts all but the adventurous off doing it. Buying a unit pre-assembled helps. I just cannot get beyond the Merg solution being overly complicated, but then I do come from a DC train of thought.

[Gordon H]

I assume you mean "wasn't"...

And you are quite right that testing wires is easy with a meter that most people will have. That is why such a simple meter was specified for testing the ATC once it is built. The number of components involved is pretty minimal for the effects achieved. Beyond the PCBs themselves, everything is just wire to the track.

I am interested to know which aspects of the concept you see as being complex? Just the need to build the kits? I would think anyone who is happy to contemplate wiring up relays themselves would be equally happy putting a standard through-hole PCB design together.

Most users would never need to understand how the circuits operate anyway, which is no different to buying commercial equipment.

[TimC]

Thanks all for your input. I don't want to start a for/against argument over self build electronics, that was not the point of the thread. 

 

I should have probably said that I was trying to do this with very little expenditure.  I have a good stock of relays hence my preference and I was hoping that someone knew of the circuit that I had seen previously on someone's model railroad blog.  It was very much in 'simple' electrics territory that Titan refers to above.

 

Having re-read the Heathcote SA1 instructions and swapped e-mails with Clive, who was very helpful, I understand now how it can be done with his shuttle and IRDOT units (or reed switches).  Unfortunately, Its a bit more than I have to spend on what is essentially a bit of fun for a table top display unit for our club stand.  

 

I shall have a play over the next few weeks and if I get it to work I'll let you know.

[Kenton]

 

I am interested to know which aspects of the concept you see as being complex?

 

The PCBs themselves.

The concept itself is just too complex and covers too many options for the simple wire and section break solution (that I have not given in detail). This isn't an anti Merg post and if I wanted a complex solution to the simple problem might just have gone down that route. As it was I want for a Heathcote solution to my 1 end to 2 end plan. It worked (never been one to work or find relays that easy) Didn't know of the Merg opton back then but wouldn't rule it out. It has some nice features - if you want them but if you don't then a simpler option is there.

 

If cost is a real issue then neither the Merg DIY or Heathcote BIN options are going to meet that criteria. Though I'm at a loss as to how the delay can be achieved with just relays an no circuit board.

[TimC]

...... Though I'm at a loss as to how the delay can be achieved with just relays an no circuit board.

 

I have some timer relays that should be able to do that (and I guess they do have some circuitry).

[Gordon H]

The concept itself is just too complex and covers too many options for the simple wire and section break solution (that I have not given in detail). This isn't an anti Merg post and if I wanted a complex solution to the simple problem might just have gone down that route. As it was I want for a Heathcote solution to my 1 end to 2 end plan. It worked (never been one to work or find relays that easy) Didn't know of the Merg opton back then but wouldn't rule it out. It has some nice features - if you want them but if you don't then a simpler option is there.

 

I would be interested to see your proposal drawn out if possible. Only then will it be possible to compare complexities.

In particular, for the OP's proposed layout, how would your solution deal with the case where (say) only two of the three DMUs were available for service?

[Kenton]

 

I would be interested to see your proposal drawn out if possible.

 

So would I! don't you think if I had it on paper it would have been posted by now?

 

In particular, for the OP's proposed layout, how would your solution deal with the case where (say) only two of the three DMUs were available for service?

I do not see the requirement to cover all options (which the Merg certainly attempts to do - It does cover most of them by programming)

However the specific case of lame traction end could be accommodated by making the setup to a 1 to 2 end shuttle. (overriding one of the points for example) but I don't think that would be THAT much more difficult than reprogramming if such an option was also designed in. But with each additional "user requirement" it becomes much more like software - bloats and becomes more complex with more things to go wrong - which is where the Merg system ended up. It does so many things not all of which are required by everyone. That is not a fault, doesn't make it a bad solution in any way.

[Gordon H]

However the specific case of lame traction end could be accommodated by making the setup to a 1 to 2 end shuttle. (overriding one of the points for example) but I don't think that would be THAT much more difficult than reprogramming if such an option was also designed in.

 

Certainly, it ought to be relatively straightforward to conceive a hard-wired system to perform one specific fixed set of operations. That is why I think it would be an interesting exercise to understand how much extra complexity would be involved in making such a system re-configurable to a certain extent - i.e. at least to cover a few additional foreseeable situations that might crop up.

 

 

But with each additional "user requirement" it becomes much more like software - bloats and becomes more complex with more things to go wrong - which is where the Merg system ended up. It does so many things not all of which are required by everyone. That is not a fault, doesn't make it a bad solution in any way.

 

The MERG ATC system didn't 'end up' that way as such. It was specifically designed to be flexible in application, to cover as many of the member's previously expressed requirements for a shuttle system as possible, using a minimum of hardware. The internal operation is software controlled - quite deliberately - using a microcontroller as the only practical way to implement such flexibility at the flick of a switch. Users need not concern themselves how the operation is actually implemented, so long as it does the job - like most things you buy into these days.

I dread to think how many relays would be involved in implementing an equivalent hard-wired capability.

You could equally argue that a laptop computer does many things that most people never need - but the manufacturers always try to anticipate the needs of as wide a range of users as possible by including as many features as they can.