Automated Motorised Traverser

[Deev]

After a bit of head-scratching regarding how to actually build the traverser using linear bearings. It turns out that I'll need about 10 linear bearings, and quite a bit of tube. As I don't know what diameter is big enough to give the required stiffness. It all seems complicated and expensive so I've bought a cheap pair of single extension drawer runners. The pair I've bought are 406mm and weirdly, roll both ways, so can be used straight out of the box without modification.

 

I've build a primitive rig to test the Stepper motor drive using a simple rack and pinion drive. Arduino code is here in case you're interested:

 

 

 

Given the crudeness of my rig and low cost of the components, it's repeatably accurate and good enough for a traverser.

 

 

Next job is to combine the DCC control code with the motor control code - A bit beyond my capabilities at the moment. Watch this space.

Edited December 8, 2012 by Deev

[MattWallace]

Great progress, can't wait to see the Arduino DCC code

[two tone green]

Very interesting. Thanks. Was exploring the possibility of one at the end of my O gauge layout . Food for thought.

[Deev]

Hi guys,

 

A major breakthrough this evening - I've worked out the Arduino DCC code, recoded it and spliced in the Stepper motor code to get the traverser mechanism DCC-controlled.

 

It's a bit rough and ready, and the wires aren't soldered in place, but it does work. I've set position or 'road' 1 to accessory decoder address 711, road 2 to 712 and so on to road 8 (although my mechanism only has 5 spaces, I'm planning on using 8 ). Using the Lenz handset I can move the traverser by selecting 711, 712 etc. It has a build in memory so it knows where it is.

 

See the demo:

 

http://www.youtube.com/watch?v=TjagvGhfOAg&feature=youtu.be

 

Code here (if you're planning to use it, beware it's scruffy with lots of obsolete commented out code):

 

 

 

Dave

Edited December 9, 2012 by Deev

[Gordon H]

From the appearance of the video, it looks as if there is currently no allowance for acceleration and deceleration of the mechanism.

Whilst that might be OK with the mock-up shown, once you have the much larger mass of a real traverser and its rolling stock attached it could become more of a problem. There is a risk of missing steps if the inertia is too high, plus lurching of the stock as the track moves instantaneously beneath it.

If you look at the video of my traverser back on Message #47, you will notice that the start and finish of a movement has an approximately 1/4" acceleration and deceleration phase to allow for these effects. This small amount has proved sufficient to prevent such problems arising. The overall movement is still performed by absolute step counting, so the same accuracy applies.

[tender]

The pair I've bought are 406mm and weirdly, roll both ways, so can be used straight out of the box without modification.I've build a primitive rig to test the Stepper motor drive using a simple rack and pinion drive.

Hi Dave

I've been following this thread with immense interest and hope at some point to build a automated traverser/fiddle yard. I did have a play with a Heljan Traverser with a view to modifying it to take a full train (Loco+4 coaches). I did manage to get it operating using DCC commands from RocRail computer software but mechanically I don't think it's up to the job.

Where did you get the draw runners from, I've looked at a few but like you suggest most have been single ended. Having a set set that rolls both ways is a good way to go I think.

 

Ray.

[Deev]

Hi Ray,

 

The drawer runners I bought came from eBay from a seller called 1ststopironmongeryandhardware.  Try here: http://www.ebay.co.uk/itm/190711596507?var=490090978785&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649 

 

Mine are the 406mm ones and it was just luck that they roll both ways, but that does mean that you get about 800mm+ of travel if you mount the runner in the centre.

 

How'd you manage to get your traverser to respond to DCC?  Is the Heljan unit DCC compatible?

 

Dave

[tender]

Thanks for the link Dave.

The Heljan Traverser is DCC ready but it took some time to get it working with RocRail as there wasn't any interface for it in the drivers. After a bit of toing and froing with the RocRail developers we did manage to get it working ok.

You can operate it directly with accessory commands from a hand controller but from what i remember the addressing is a bit weird.

I don't think the unit uses a stepper motor, it has a slotted vane in an opto switch so looks like it counts pulses for position. It seems to loose sync after a period of time requiring a reset to sync back again.

Ray.

[Deev]

Hi Ray,

 

That's interesting about the RocRail developers - I plan to use RocRail but can't find a full length traverser module in the manual.

 

Sounds like Heljan's traverser mechanism isn't up to the job - a reset every few times would be painful.  I wonder if this is the same system that they use for their intermodal gantry crane.  For info, I set my experimental device running all night and it was still accurate in the morning.

 

My traverser mechanism responds to accessory decoder commands, so calling '717' will move the traverser to lane 7 etc.  I've coding my Arduino to move whether the command sent is closed or thrown, so as to not confuse RocRail.  It's a workaround but should hopefully work.

 

In parallel to making the mechanical part of the traverser, I recommend getting all the electronic parts in stock as well to allow you to have a play.  

 

You'll need:

1) An Arduino (I use an genuine UNO, about £18 I think) 

 

 

2) Arduino motor shield (I used an Adafruit copy).  The motor shield that I bought is the double motor one that can control 2 x steppers and 2 x servos - mine came via eBay from China, cost £3 and took 6 weeks to arrive.  You can get them in the UK, but they're more expensive.

 

 

3) At least one stepper motor.  The motor I used is the 'Mercury Motor SM-42BYG011-25' which is a bipolar, 1.8deg, 12v stepper that met all my perceived perfomance requirements.  Mine was from www.proto-pic.co.uk.

 

 

4) PSUs - You'll also need a 12v 1.5A PSU for the motor shield and a 5v 200mA PSU with 5mm connector for the Arduino.

 

5) Drive mechanism - If you plan to use rack and pinion, you'll also need a 5mm pinion for the mercury motor (mine is a 12T 1:8 Losi one for R/C cars) and the rack itself (I used a few lengths of this: http://www.technobotsonline.com/standard-mod-1-plastic-rack.html).

 

6) Opto-isolator to interface the Arduino to the DCC signal.  For this you'll need: 1 x 6N137 IC, 2 x 10K resistor, 1 x 1K resisor, an LED, some wire and some prototyping board.  The diagram is shown on page 2 of this thread.

 

Also, you;ll need to download and install the software to program the Arduino.  Look here: http://arduino.cc/en/Main/Software.

 

Hope that helps

 

Dave

Edited December 30, 2012 by Deev

[tender]

Hi Dave.

Many thanks for the parts list, that will save a lot of shopping around. I'll try and get a Arduino in the next day or two to have a play with while i'm waiting for the other bits. Interesting to see that you intend to use RocRail. It takes a bit of getting used too, but the forum/developer backup is very good as i found with the Heljan Traverser. I keep looking at Train Controller but the price is off-putting.

 

Ray.

[Deev]

Hi Ray,

 

Good idea.  Familiarising yourself with the Arduino programming takes a little time depending on your background.  I found making the interface circuit and getting the Arduino to flash it's lights under DCC command an exciting step (for me anyway!).  As an aside I've ordered a load of 9g servos and the 'sensor shield' for my UNO to see if I can satisfactorily move points using the Arduino.

 

I've tinkered with RocRail on my last layout and managed to get it controlled by AndRoc which is the main appeal of it for me.  Since my experiment, tablet computers have come a long way so this is ultimately how I'd like to control my layout.

 

I had a look on the Rocrail website and found this: http://wiki.rocrail.net/doku.php?id=fj-fiddleyard-nl  and http://wiki.rocrail.net/doku.php?id=phg-fiddleyard-nl  and http://wiki.rocrail.net/doku.php?id=phg-new_fiddle_yard  Looks the hard-working dutch have already done what I want to do (although they call it a 'selection table'.  Peter Gilling's page is especially clever - he's solved a lot of the questions I had in my head, especially how to reduce the burden on block occupancy with so many tracks,  Take a look, it's worth it.

 

Agreed on the RR&Co. software - it's about 6 times more than I'd like to spend on software.

 

Dave

Edited January 1, 2013 by Deev

[MattWallace]

Hi guys,

 

Have either of you looked at using JMRI instead of RocRail? It's free and it works well.

 

It gives you throttles, decoder programming, turnout operation, routing and many more features including the ability to manage your layout from your phone/tablet (Android and iPhone currently supported!).

 

I'm not a developer on the project, however I'm using it in conjunction with MERG hardware and it works excellently (it works with most major USB/Serial -> DCC Command Station interfaces, not just MERG)!

 

Just a thought...

 

Matt

[Deev]

Hi Matt,

 

Yes, I've used JMRI already and found it better than RocRail in many respects.  On my last layout I managed to get to the stage of using it to automate train movements.

 

I found the DecoderPro software to be really excellent and this is my choice when I'm programming.  I may still choose to use it, although I've been concentrating on RocRail as:

1) it was first to support Android devices

2) I couldn't work out PanelPro.

 

The thing that I liked about JMRI was that there was potential to use USB devices on the PC (eg. sound, lights, servos etc.) but I didn't get round to getting it to work.

 

Dave

[tender]

Hi guys,Have either of you looked at using JMRI instead of RocRail? It's free and it works well.It gives you throttles, decoder programming, turnout operation, routing and many more features including the ability to manage your layout from your phone/tablet (Android and iPhone currently supported!).I'm not a developer on the project, however I'm using it in conjunction with MERG hardware and it works excellently (it works with most major USB/Serial -> DCC Command Station interfaces, not just MERG)!Just a thought...Matt

I use DecoderPro but just cant get my head around the user interface for PanelPro.

 

Ray.

[MattWallace]

Hi Matt,

 

Yes, I've used JMRI already and found it better than RocRail in many respects.  On my last layout I managed to get to the stage of using it to automate train movements.

 

I found the DecoderPro software to be really excellent and this is my choice when I'm programming.  I may still choose to use it, although I've been concentrating on RocRail as:

1) it was first to support Android devices

2) I couldn't work out PanelPro.

 

The thing that I liked about JMRI was that there was potential to use USB devices on the PC (eg. sound, lights, servos etc.) but I didn't get round to getting it to work.

 

Dave

PanelPro can be a bit of a pain, however once you've got the basics down, it gets a little easier to master!

 

I've found that the way you need to do it is as follows:

 

1) Add an entry to the Roster for each Locomotive (ou can do this manually if you know the DCC address or you're working against the Main track instead of a programming one)

2) Work out the addresses for your devices (turnouts etc)

3) Enter the turnouts into the "turnouts" section under the "tables" menu

4) Open up the Layout Editor (this gives you a schematic of your layout instead of "just switches")

5) Place each turnout on the panel and associate it with an entry from the "Turnouts" table

6) Link the turnouts with track sections

7) Profit

 

The best guide that I've found so far is http://jmri.sourceforge.net/help/en/package/jmri/jmrit/display/LayoutEditor.shtml

 

Hope that's of some help

 

Matt

[tender]

Hi Ray,

 

 

I had a look on the Rocrail website and found this: http://wiki.rocrail.net/doku.php?id=fj-fiddleyard-nl  and http://wiki.rocrail.net/doku.php?id=phg-fiddleyard-nl  and http://wiki.rocrail.net/doku.php?id=phg-new_fiddle_yard  Looks the hard-working dutch have already done what I want to do (although they call it a 'selection table'.  Peter Gilling's page is especially clever - he's solved a lot of the questions I had in my head, especially how to reduce the burden on block occupancy with so many tracks,  Take a look, it's worth it.

 

 

Dave

Hi Dave.

Thanks for the links to the RocRail Fiddle yard. I've not been on the site for a bit and all this is new to me, someones been busy.

Thats one of the things i like about RocRail, its always under development with ideas from users being taken up by the development team.

 

Ray

[Deev]

Yes, they have, although the first post was made in 2003!

 

My favourite bit is the method by which they switch power to the track using a modified coach bogie and length of 3 rail track - novel.

 

Dave

[Deev]

After studying the ‘Dutch’ traverser in the link shown above, I’ve been really inspired by their method of getting power to the carriage from the table by using a six wheel bogie and 3 rail track.

 

 

It’s so clever in its simplicity and is completely fool proof whilst reducing the required number of block occupancy detectors massively.  It does have a few drawbacks in my opinion:

1) potential for losing signal due to dirty rails / wheels

2) doesn’t allow stock to be moved when not aligned on the table (although this is the whole point of it to be fair)

3) can’t be programmed

4) there is some transition where two tracks are energized simultaneously.

 

As I too plan to have 3 blocks to allow the trains to stop in the right position, having an 8 lane traverser would necessitate 24 block occupancy detectors, which would be rather expensive.  As such I’ve looked into using relays to switch power to the tracks instead.

 

I’ve managed to find a 8 channel relay module for the Arduino on the net that looks like it’ll do the job.  Having 4 sectors (ie. 3 detected blocks and 2 plain DCC blocks per lane) per lane would necessitate 4 of these modules, but it would allow me to switch all of the track sectors on the traverser on or off individually.  Can’t think of why I’d need to do that, but I like the idea.

 

 

As the whole setup is getting rather complex wiring wise, I’ve looked at moving all the electronics and the motor on to the carriage in order to reduce the number of wires going from the carriage to the table to 9 as well as allowing me to use a tensioned belt and pulley rather than rack and pinion as this seems to be the preferred way of the Dutch.  Diagram thus:

 

 

I can’t think of a better or cheaper way of doing this – unless anyone can think of a better way?

Edited January 2, 2013 by Deev

[Deev]

I started work on the traverser cradle design tonight - it's going to be a long process.

 

One of the things I need to establish is how close to pitch the tracks.  Obviously, the closer together, the more trains I can store in the same area.

 

Looks like 40mm is possible, with a strangely prototypical look about it.  This would mean an 8-lane traverser carriage would have a width of 330mm!  (maybe I ought to go bigger!?). Derailments may be an issue though, not sure what I could do about that, apart from catch rails on the throats.  Hmmm.  I'm pretty sure that none of my rolling stock sticks out more than either a Pendolino or a Mk3.  Need to remember to allow 70mm vertical clearance as well!

 

Edited January 3, 2013 by Deev

[MattWallace]

heh, an 8 lane traverser in just over a foot wide? Wish I'd known that a few months ago, I would have added one...

 

Never mind, one to log for the next layout...

[Deev]

Hi Matt,

 

Yes, sort of.  The carriage is 330mm or thereabouts, but it needs to move as well.  So it's 330mm with 160mm of travel either side.  I plan to build mine in a box frame, so it'll be 650mm between the outer frames.

 

Dave

Edited January 4, 2013 by Deev

[tender]

Hi Dave.

I now have a Arduino Uno, Motor Shield and the bits for the DCC interface. Stepper motor on order. Will build up over the next few days and see if I'm as successful as you in getting it working.

 

Ray.

[Deev]

Nice one Ray. If you build the circuit right, then I don't think you'll have much trouble. Best bet is to get the Arduino to switch its on board LED under DCC command first, then move on to the stepper control.

 

Dave

[Deev]

Not a lot of visible progress lately, but I've managed to get a few things done.

 

1) The first pair of relay boards have arrived, from Hong Kong.  Surprisingly weighty and although they were bought from the same place at the same time, they are subtley different and have different bolt hole patterns.  Hopefully they'll function the same.

 

 

2) As I'm building the traverser for my 'scenic-modular' Filton Abbeywood project (http://www.rmweb.co.uk/community/index.php/topic/27124-filton-abbeywood/page-2?) I've given some serious thought to designing a double deck traverser.  This gives me double the storage for the same space and given that I've done what I consider to be the hard bit, the electronics, it's just one more of the same.  

 

I reckon that the lower carriage could be supported from the bottom and the upper could be supported from the top.  The difference between the track beds (top to top) works out as a minimum of 82mm.  This sounds complicated, but it isn't really.  

 

I've done some sums as to what the maximum gradient would be to go from the 'scenic' level at the various radii curves that I need.  Filton Abbeywood has an island platform that's very wide, and limits my spacing somewhat.  As such, the curve radii work out at 560mm, 650mm and 725mm respectively.  The 'inner' two roads will be the lines to SWales and Parkway, the outer to BTM.  As the outer has a larger radius, it can climb much more than the inner roads.  Applying an additional 2% to the gradient to account for the curve means that the inner roads climb 36mm, the outer descends by 46mm, doing the opposite at the other end of the layout.  The effective gradient is 1:40, which hopefully shouldn't be too challenging for my fairly recent traction.  The diagram that shows a cross-section through the end of the layout helps to explain it.

 

 

 

[Deev]

I’ve spent the last few weeks working on the detailed design of the traverse.  I’ve settled on a double-deck arrangement with each deck being uni-directional.  This gives me double the storage for the same space in the garage.  The end-view of the arrangement is thus:

 

 

I’ve also spent more time researching the stepper drive mechanism.  From my drawings, the upper traverser carriage needs 280mm of travel and the lower needs 200mm total.  The diagram below shows the carriages in their positional extremes.  

 

 

I considered rack and pinion which has the advantage that it’s fairly easy to integrate into the design at the expense of weight, alignment and the fact that I’ve not been able to get a 300mm module 1 nylon rack any where.  Steel rack is expensive, heavy and feels like overkill to me.  As such I’ve looked again at belt drives, mainly on the numerous 3D printing forums.  Wrapping the belt around the motor is straightforward enough but the big problem is keeping tension in the belt.  There are many solutions that are quite fiddly and involved, but the one I like best is this one:

 

 

It’s so simple, cheap and can even be easily adjusted, but to be it seems as if there’s not much to stop the belt from turning.  I like the arrangement of tie-wrapping the ends of the belt together however, so I reckon it’s be possible to hook the end over an eye-bolt or a bolt-hook.  This would give accurate placement, require only a hole to be drilled in the baseboard and allow easy re-tensioning.  I’ve got all the bits I need in my parts box, I’ll mock up this arrangement to see if it’s feasible and check how much belt I need.