Train detection... ...DIY IR-DOT

[ShinyShoe]

OK so I want some train detection on my layout to automatically switch signals. but there could be a lot of detecting needed

 

I've read around, particularly because in my brain it would seem most sensible to detect by variation in current draw. But I think I've now worked out that thats not as good as I fancy...

 

As I read around I find three detection methods:

 

Current Draw - very little documented which suggests its probably not a goer!

- Advantage: no holes or whatever in the track

- Disadvantages: needs some form of isolation of blocks of track, only detects locos or carriages with resistors installed. if not you can only detect the front of the train not the back.

 

Magnetic - better documented - stick magnet to bottom of train. Put reed relay below ballast - relay senses magnet.

- Advantage: quite cheap to install relays, low tech.

- Disadvatage: Needs mags installing on locos and at carriages.

 

IR-DOT - well described - senses anything moving over (I have seen a light sensor described but as my layout is set at dusk I have rulled it out

- Advantage: No installation on stock or locos needed

- Disadvantage: Needs small hole to sense (although people say its not noticable), cost a heathcote module costs >£10.

 

My take:

The avoidance of needing to install anything on loco's/stock is my biggest aim so that I can have visitors. of

This is one of potentially 3 or 4 modular layouts - all would presumably best use the same detection method and so cost is important...

 

 

My layout is not massive but has a fair number of points (7 I think) and some quite complex routes (you can move from the up line to the down back to the up and onto at least three sidings - so I may need to have a lot of detection - haven't really figured out the details yet but detection in each 3 sidings, the entrance and exit of the up and down and possibly mid way along the down line before you can move back to the up or possibly on the section that moves to the up. - thats 8 sensors Thats going to cost £90 just for heathcote boards. Then I'll actually need to do something with my detection

 

Is there a cheaper solution that achieves my need? Can I build my own IR Dot? I guess the correct answer is yes... but I guess I mean can I build it cheaper than £90 for 8... What's the IRDot actually doing is it sending an IR light beam up and sensing when it gets bounced back?

 

Perhaps I could use this: http://docs-europe.electrocomponents.com/webdocs/0d1b/0900766b80d1bdd5.pdf with a latching relay (may need the latching relays anyway for the heathcotes - haven't studied what they need), or an IR LED emittor and reciever... probably through a transistor or something to manage the power needed by the relays...

 

Anyone done this?

[naugytrax]

For infra-red detection you could look into building HECTOR kits from MERG, which sell for 6 pounds 60. I think you have to join MERG before you can buy them, but if you need a lot of them that could be worthwhile. They seem to be equivalent to Heathcote Electronics' most basic device, the IRDOT-1.

 

Heathcote offers fancier units including some which incorporate latching relays, but of course the price goes up as you add more features. All the IRDOTs send up an IR beam and sense when it is reflected, as you guessed.

[SHMD]

I'm in the middle of doing exactly what you have just started.

 

I considered everything you have and eventually decided upon IR detection, VB10express for the control, and relays and transisters for the Points.

But here I differed, I decided to make my own.

 

Details :-

http://www.rmweb.co.uk/community/index.php?/topic/47043-printed-circuit-board-pcb-software/

 

 

I will have more time for a better reply later.

 

Kev.

[Kenton]

I have used IR-DOTs and they are great and well built units. I doubt if you could build the same cheaper. They can be a bit of a pfaff to set up as they do rely on the reflection of IR from the underside of the train and the built in delay is critical otherwise it will fire off with each coach/unit that passes.

 

Many years ago I used reeds and magnets - very simple to wire up but I found them very tempramental and running a loco round with a magnet seems to collect all sorts of debris.

 

Doing the same again I would again use IR-DOTs.

 

BTW you can remove the diode from the PC board to extend the leads - useful when the space under the point of detection is at a premium.

[Crosland]

Current detection can work just as well as IR or magnetic and is well documented. The other differences you did not list are:

 

Current: detects things anywhere in the whole block.

 

Magnetic and IR: Only detects when something passes a specific location.

 

RFID which, you didn't mention, falls into the latter category.

 

MERG has already been mentioned. I would also look at Rob Paisleys site for ideas http://home.cogeco.ca/~rpaisley4/CircuitIndex.html

 

Andrew Crosland

[SHMD]

If I could add to the above posts…

 

IR detection only detects, (if set up properly), an object passing – or stationary – directly over the sensor. (Loco, coach, mineral wagon, track rubber, sleeve cuff, detritus, etcetera…)

 

The Track current sense system will only detect the loco in the track circuit. Unless you fit pickups and resisters to every vehicle. (Even with the Loco only detection, I seriously considered this option.)

 

The Magnet/Reed Relay combination, again, would only detect the vehicle so fitted, but this just seemed just too low tech for me! Even so, I have bought some really amazingly powerful little magnets and reed relays in various sizes – just to play with of course.

I have found that, (my), IR detector design was immune to most lighting conditions from bright sun light to total darkness. It reliably sensed objects passing near it – 0cm to 7cm – in lighting all conditions.

I did find that it was triggered by pressing buttons, on a TV zapper, pointed within a 1 metre radius, (and in pretty much any direction!!! They put much more power through the IR LEDs than I do.), and by my old “filament” torch. (It was completely immune to LED type torches, LASER pointers, and bank note checkers!)

So from my original goal of it being completely impervious to ambient lighting conditions, I have – shall we say – moved the goal posts so that now I think of it as “having a high immunity” to ambient lighting conditions. (So now, the only down side that I can see, is that if I want to run the layout in darkened conditions then I will just have to make sure I use an LED based torch to admire my work!

I will be posting my complete design once I am happy that it works completely with the finished firmware. (It works brilliantly with the test software.)

It is not really intended to replace ‘commercial’ systems as it is customised for my own personal use. (No POTs or Relays on board as I achieve these functions elsewhere.)

With purchasing everything directly wholesale, and in China, I have got my board price down to less than £2.50 complete and that with two sensors. (~£1.25 per working IR detector.)

These prices, in no way, reflect commercial prices.

Things that got the price down were – Everything purchased in China – dirt cheap. No delivery costs – I was there. No manufacturing costs – I’m making everything. No profit margin – usually small anyway. No mark-up to cover failures and support – I’ll fix everything. And probably a whole host of other costs too – advertising, warehousing, admin…

In fact, come to think of it, commercial systems are bl**dy good value having done all of the above!

 

If anyone can find sections of my schematic useful, in their own projects, then nothing would please me more. (But please, I have not finished this project yet!)

 

Kev.

[ShinyShoe]

For infra-red detection you could look into building HECTOR kits from MERG, which sell for 6 pounds 60. I think you have to join MERG before you can buy them, but if you need a lot of them that could be worthwhile.

Yeh I found them later on last night. If I need 8 then 8 Heathcotes will cost nearly £90, So the MERG kit will be cheaper even with MERG membership.

 

Heathcote offers fancier units including some which incorporate latching relays, but of course the price goes up as you add more features.

Latching relays available on Ebay for about £1 each so limited advantage...

 

I considered everything you have and eventually decided upon IR detection, VB10express for the control, and relays and transisters for the Points.

But here I differed, I decided to make my own.

 

Details :-

http://www.rmweb.co....d-pcb-software/

 

Thanks Kev, not sure if that's a fantastic reply or the worst possible! I thought this was a fairly simple sensing task but it looks like a lot more electronics than I expected!! Fantatsic board designs BTW, I was just expecting to using some copper printed strip...

 

I could try getting a taxi to the electronics market but I think the bill would be quite high... ...I take it you are in China at the moment?

I have used IR-DOTs and they are great and well built units. I doubt if you could build the same cheaper. They can be a bit of a pfaff to set up as they do rely on the reflection of IR from the underside of the train and the built in delay is critical otherwise it will fire off with each coach/unit that passes.

So I don't need the timer to set the releant block's signal to danger - that should happen and continue to happen each time anything enters the block. But I guess I do need to use a timer for detection of exiting the block.

 

BTW you can remove the diode from the PC board to extend the leads - useful when the space under the point of detection is at a premium.

Good to know

Current detection can work just as well as IR or magnetic and is well documented.

Is the same method possible on DC and DCC? Is there a good reference?

 

The other differences you did not list are:

 

Current: detects things anywhere in the whole block.

 

Magnetic and IR: Only detects when something passes a specific location.

 

RFID which, you didn't mention, falls into the latter category.

Hadn't considered RFID - would have the nifty feature of being able to identify what was in a sector rather than just something. But for current layout would be OTT.

[noiseboy72]

A local miniature railway which cannot use track sections due to steel rail ties is considering using treadles to count axles.

 

Now how much are super light microswitches...

[Suzie]

Probably one of the cheapest methods is to use light dependant resistors. They are available from Ebay international sellers for less than £10 for a hundred including postage and with just the addition of a resistor to each one will drive most automation systems directly.

 

You will have to fiddle about a bit to get the lighting and resistance value right to make them work, but for less than 10p per channel it is worth the effort for a fixed installation. The LDRs just simply detect the shadow of the moving train.

[SHMD]

Hi ShinyShoe,

 

 

 

Thanks Kev, not sure if that's a fantastic reply or the worst possible! I thought this was a fairly simple sensing task but it looks like a lot more electronics than I expected!! Fantatsic board designs BTW, I was just expecting to using some copper printed strip...

 

I could try getting a taxi to the electronics market but I think the bill would be quite high... ...I take it you are in China at the moment?

 

 

Thanks but it wasn't my intention to put you off, rather the opposite really.

(I got back from China Saturday night - thanks.)

 

I was going to use strip board at first, (should definately have for the prototype anyway). I was just trying to show what was possible for very little outlay!

 

But as you only want 10 boards, or so, then Strip board should be fine. (There are no high frequency issues, just remember to add a couple of 100nF de-coupling capacitors.) After making 10 boards, you will be quite proficient at making them!

 

Don't be put off by the complexity of my design either, i added a lot of stuff to aid flexibility and most of it is just standard modular circuit design. (Like Lego really.)

 

Kev.

[ShinyShoe]

Probably one of the cheapest methods is to use light dependant resistors. They are available from Ebay international sellers for less than £10 for a hundred including postage and with just the addition of a resistor to each one will drive most automation systems directly.

I'm expecting to have my layout set at Dusk so not sure it will work unless I conveniently place a yard light etc in the right place....

[ShinyShoe]

 

 

Thanks but it wasn't my intention to put you off, rather the opposite really.

OK I'm going to hang fire doing anything with it for now... but I'm not excluding doing this..

[ShinyShoe]

Well I decided to have a go at designing a circuit that would maybe do the job.

 

Attached is the circuit diagram I've created along with a strip board layout. All yet to be tested but by the looks of things I can trigger a latching relay for about £2.50 a pop (building about 10) including strip board.

 

The intention not shown in the circuit diagram is to use the sensor of a train entering the next block to release the latching relay on the previous. That'll work for unidirectional track... still thinking about bidirectional!

 

I'm no electronics expert and have had to try and relearn some 20 year old electronics to come up with something that might work...

 

SHMD suggests adding capacitors - happy to do so but where?

 

[SHMD]

Hi, I will check the data sheet for your tranceiver, (TCRT5000), and see if it recommends any de-coupling capacitors, (for noise immunity).

Also, have you thought about how to trigger the circuit when a train isn't there, and conversly, how to manually reset it when you want.

But I will not be able to look at the data, (and your circuit), until tonight.

 

I'm very impressed with your artwork drawings, especially the stripboard one.

What program did you use?

 

Kev.

[ShinyShoe]

Thanks for the continued input into this design. Your help is really appreciated.

 

Also, have you thought about how to trigger the circuit when a train isn't there,

Why would I need to make the layout think a train was in a block when its not?

 

and conversly, how to manually reset it when you want.

Ah yes I've thought about that as I might park a train up in a siding and then need to release the block for use.

The 'unlatch' of the relay is done using a 12V feed on the otherside. Its not shown on the circuit diagram but is shown by the two green wires on the right of the strip board. That could come from a switch or another stripboard relay down the line.

 

 

I'm very impressed with your artwork drawings, especially the stripboard one.

What program did you use?

The schematic is just drawn in OpenOffice Drawing using some downloaded WMF outlines of components, and is pretty rough to be honest.

 

The Stripboard is drawn using a program I downloaded last night which is free and cross platform:

 

http://code.google.c...-layout-creator

 

Its not perfect. It obviously didn't have all the components I needed so I just used a 10Pin IC and 4 Pin IC to represent the surface mount relay and the IR Sensor. You can create new components using an XML file but I didn't have time to investigate. It can't do anything fancy like convert the layout to a schematic or check against a schematic which I believe some can. ... but its free and was pretty easy to use. It can do PCBs too I believe. Can also do list of components etc but I'd already created mine!

[SHMD]

OK, no need for capacitors here, as your design only uses discrete components

I would recommend only building one, or perhaps two, prototypes first before committing to purchasing all the bits.

Also, with no built in delay you will have to mount your sensors very carefully as you will get a lot of chatter/firing on the relay as each coach/wagon passes by the sensor.

 

I would also like to know how you are going to power all your sensor boards.

 

I was thinking of the ‘manual triggering’ of the next Block as a way of route setting. Ie. When a train starts off from a Terminus Station or a Freight Yard.

 

How long is your intended mainline and how many ‘Blocks’ were you thinking of installing?

A Track Layout Plan would be really useful here. Especially as I know how good you are at CAD ;-)

(I think I will have a look at your software for my “Rough Diagrams” for work!)

 

Kev.

[SHMD]

I've just thought of one potential problem.

 

If your sensor is 'only just' seeing an object then it will 'only just' switch on. This will 'only just' switch on the BC182 and will operate it inbetween fully on and fully off. This will cause it to dissopate a LOT of heat.

 

To get around this potential problem will, again, require careful mounting of the sensor.

 

Kev.

[The Stationmaster]

OK, no need for capacitors here, as your design only uses discrete components

I would recommend only building one, or perhaps two, prototypes first before committing to purchasing all the bits.

Also, with no built in delay you will have to mount your sensors very carefully as you will get a lot of chatter/firing on the relay as each coach/wagon passes by the sensor.

Kev.

I'm no electronics expert (as I'll no doubt prove in a moment) but I do know a bit about what 'track occupied' is meant to do in a signalling system. Thus you need a stable state of either occupied or unoccupied so any detection system should work vaguely like the theory of an axle-counter and first object in sets it to occupied and last object out returns it to unoccupied.

 

On our models we can safely simplify this theory and ignore things like divided trains (except possibly in areas where we can't see the track) so what any sort of optical detector should do is change state when the first vehicle detected enters the section and either it or its fellow at the other end should revert the control for that section to an '=unoccupied' state when the last vehicle leaves the section.

 

In a very short unidirectional section I would think that a single detector could be used with some sort of 'smoothing' or timer circuit to avoid chatter?

 

Back to the electronics experts

[ShinyShoe]

Also, with no built in delay you will have to mount your sensors very carefully as you will get a lot of chatter/firing on the relay as each coach/wagon passes by the sensor.

Yes the relay may be refiring but as its latching the state should stay set as occupied.

 

I would also like to know how you are going to power all your sensor boards.

Until today I didn't know - a 12V supply from a controller potentially. However tidying up I found a 12V adaptor from something ??OLD PRINTER maybe. Its 12V and 4.5amp. Assuming its a regulated output then I might use that. If its not regulated then I'll look through my box of bits for one that is regulated or else acquire something suitable.

 

I was thinking of the ‘manual triggering’ of the next Block as a way of route setting. Ie. When a train starts off from a Terminus Station or a Freight Yard.

Should the signal box not have set the route? I guess if its a fairly set route then you might be able to do this but I'm expecting to have various route options...

 

How long is your intended mainline and how many ‘Blocks’ were you thinking of installing?

A Track Layout Plan would be really useful here. Especially as I know how good you are at CAD ;-)

Tiny! Occupancy for me is about triggering signals not flagging where a train is because the answer is in front of me! There is a track plan and a schematic for signalling. I need to update that to show where the occupancy sensors need to be. Expect something later today.

 

If your sensor is 'only just' seeing an object then it will 'only just' switch on. This will 'only just' switch on the BC182 and will operate it inbetween fully on and fully off. This will cause it to dissopate a LOT of heat.

Yip I gathered that from my reading up and revision of high school electronics. If a train only takes 15 seconds or so to pass over and then nothing passes for a minute I was figuring that would help...

 

first object in sets it to occupied and last object out returns it to unoccupied.

 

what any sort of optical detector should do is change state when the first vehicle detected enters the section and either it or its fellow at the other end should revert the control for that section to an '=unoccupied' state when the last vehicle leaves the section.

 

In a very short unidirectional section I would think that a single detector could be used with some sort of 'smoothing' or timer circuit to avoid chatter?

Stationmaster - I believe thats the conventional approach detect the train entering and wait for Xseconds and then assume it left. To me thats wrong! The train could have stopped and so the block is still occupied. So I'd prefer to sense in and sense out rather than time it.

 

However the state that SHMD describes doesn't mean the block is flickering between occupied or unoccupied in my build because its using a latching relay. So as the front of the loco crosses the sensor it detects the train and the relay latches to occupied. The loco finishes passing and the sensor will detect the gap between trains and so stops operating the relay - but its a latching relay so it doesn't unlatch so the state stays occupied. As the wagon passes over the sensor detects further occupancy and will tell the relay to latch but as its already latched thats just surplus electrical activity which SHMD describes as chatter.

 

The only way to unlatch the relay is to send an entirely different 'unlatch' signal to it which I'd suggest should come from the next sensor although as I type I now realise I may have a slight flaw in my plan because I was going to unlatch when the next block became occupied rather than when the next block finished being flagged as occupied (which may be different to being flagged as unoccupied)...

 

It is possible that a very long train would be longer than a block. In that case you might get some slightly odd oscillation betwee occupied and unoccupied on my current setup.

 

I was going to suggest using the relay (NC terminals) to switch off the IR sensor/IR LED when the block is showing occupied. That would reduce chatter to the relay and mean that the transistor was only getting hot very very briefly. However, it would mean I couldn't develop a bit of circuitry to detect when a block is not being flagged as occupied...

 

Thats probably not articulated very well...will need to have a think...

[The Stationmaster]

Stationmaster - I believe thats the conventional approach detect the train entering and wait for Xseconds and then assume it left. To me thats wrong! The train could have stopped and so the block is still occupied. So I'd prefer to sense in and sense out rather than time it.

Isn't that exactly what I said? Sorry if this wasn't as clear as I thought it was -

'so what any sort of optical detector should do is change state when the first vehicle detected enters the section and either it or its fellow at the other end should revert the control for that section to an '=unoccupied' state when the last vehicle leaves the section.'

 

Going beyond that the logic (although not all of the circuitry) of the whole thing is relatively easy, viz

1. Section 'not occupied'

2. Sensor detects 'something' entering the section and triggers 'occupied' in the detection circuit (and a latch be it relay or solid state) maintains it at occupied once it has been triggered to 'occupied'.

What happens next depends on the sort of operations going on in the section, thus in a simple uni-directional section

3. Detector at far end of section detects 'occupied' and then detects 'clear' once the last vehicle passes (this implies some sort of 'smoothing' or timing to ensure that gaps between vehicles are not detected as meaning 'clear'.

4. Detection of a 'something' occupying then clearing the second sensor resets the detection circuit to 'not occupied'.

 

Obviously it gets more complex if trains can enter and leave at the same end of the section and even more so - possibly - if they can enter or leave at either end in any permutation.

 

Overall the only thing which would not then be detected without continuous sensor coverage is a divided train or a vehicle left in the section but if sensors are suitably cheap and invisible they could perhaps be duplicated or the equivalent of the good old VOL (Vehicle On Line) switch could be used for station platforms where vehicles are left when the train they were detached from departs.

[ShinyShoe]

As promised here are some track plans:

 

1. Schematic of track showing signalling and point numbering (catch points are not shown for clarity). Black lines show the wiring for LED route indicators via a number of latching relays)

​

2. Schematic of track showing block occupancy

 

3.Scale Track Plan

 

 

4. Matrix showing (red boxes) where danger signal to be shown following block occupancy

 

5. Matrix showing (red boxes) where danger signal to be shown based on point setting

[ShinyShoe]

Whats not shown on the diagrams is that the lines are bi-directional (why make life easy!!)

 

As its a combination of a timesaver and a inglenook shunting puzzle there could be wagons dropped on the main lines to run around etc. That'd be block 1 and 5 I guess. So I would suspect a simple latched switch will be needed on those blocks on the control panel to mark them as still occupied as VOL as Stationmaster suggests.

[SHMD]

Wow! You’re not a technical author are you Shoeshine?

 

Your two sidings have a trap point each and then only one signal and after the two traps! (In other words you can obey the signal and still derail!)

I would have thought that one trap point, (to protect the main line from the sidings contents running away), and positioned after the (ground) signal S5 would be more appropriate for shunting wagons here.

 

Also, I wouldn’t bother making the sidings a block, but would rather have them be signalled for “1 engine in steam” instead. Ie. when shunting the siding(s) they become part of the mainline but when not they are isolated from the mainline regardless of their contents.

It is the mainline that is to be “protected”.

 

But I’m a no-body when it comes to signalling, sorry.

 

But, none of the above helps when all you want is for a ‘model’ train to ‘somehow’ automatically set the signals correctly.

In other words it is very easy, here, to confuse – the usually conflicting needs – of running a model railway and signalling it proto-typically.

 

Kev.

(PS. That old PSU should be OK but remember to put in-line a small enough fuse – try 0.25Amps at first – on the output. [Remember – the bigger the PSU capacity, (ie Amps), the more trouble it can cause, (damage/fire), when a fault develops].)

[ShinyShoe]

Wow! You’re not a technical author are you Shoeshine?

Certainly not!! I like drawing pretty pictures thats all!!

Your two sidings have a trap point each and then only one signal and after the two traps! (In other words you can obey the signal and still derail!)

Hoping stationmaster is still looking at this one as my interpretation of trap points has always been different - that said I'm not in anyway an expert I only started fiddling with this stuff a year ago. My understanding was that if I drop a wagon on a siding and its breaks failed the trap points were there to stop it running onto the main line. Its an unmanned wagon so it can't tell what the signal says!

 

 

Or do you mean the signal could be green and the traps set to trap and take the back part of the train out? That was the purpose of the red CP5A/5B when they are open...?

The reason for doing it that way (as when people wee reviewing the layout they said the trap point should be on the combined siding was using commercially available trap points it happened to give me half a wagon's working length extra on my sidings to do it that way and the aim was to keep the plank properly small!

 

I would have thought that one trap point, (to protect the main line from the sidings contents running away), and positioned after the (ground) signal S5 would be more appropriate for shunting wagons here.

To do any major shunting will need to use the main down line as a head shunt anyway.

Also, I wouldn’t bother making the sidings a block, but would rather have them be signalled for “1 engine in steam” instead. Ie. when shunting the siding(s) they become part of the mainline but when not they are isolated from the mainline regardless of their contents.

It is the mainline that is to be “protected”.

 

I think thats what I concluded when I drew it out yesterday as the Block 2 and 5 being occupied didn't affect the other lines.

 

(PS. That old PSU should be OK but remember to put in-line a small enough fuse – try 0.25Amps at first – on the output. [Remember – the bigger the PSU capacity, (ie Amps), the more trouble it can cause, (damage/fire), when a fault develops].)

Good I just tested it and its regulated :-)

[The Stationmaster]

Hoping stationmaster is still looking at this one as my interpretation of trap points has always been different - that said I'm not in anyway an expert I only started fiddling with this stuff a year ago. My understanding was that if I drop a wagon on a siding and its breaks failed the trap points were there to stop it running onto the main line. Its an unmanned wagon so it can't tell what the signal says!

 

Or do you mean the signal could be green and the traps set to trap and take the back part of the train out? That was the purpose of the red CP5A/5B when they are open...?

The reason for doing it that way (as when people wee reviewing the layout they said the trap point should be on the combined siding was using commercially available trap points it happened to give me half a wagon's working length extra on my sidings to do it that way and the aim was to keep the plank properly small!

A signal normally protects the trap points - i.e. it is there to stop a movement running towards them when they are not set for that movement. So if the signal is in advance of the trap points (i.e. it is beyond them in the direction of travel towards both trap point and signal) then you have unprotected points and you are inviting derailment - with a nice friendly green light.

 

However although it is not unusual in past practice (but certainly frowned on in new work for some years now) you can have a single signal applying to both lines out of the sidings and thus in rear of the two trap points and all you need do is move the dividing point between blocks 5 & 6 - which in reality is what would happen anyway as the trap points would be in the equivalent of you Block 5 for some (but not all) locking purposes.

 

If you haven't yet laid the track there is a very simple answer to getting in the two traps and that is to use what I call a half double slip - i.e, at one end it is just like a double slip but the other end is an ordinary turnout - as illustrated below -