A very whacky idea for determining train position

[AndyID]

No, but it does have info on how many revolutions the train has driven since a particular point in time. The PC should be able to work out where exactly the thing is

 

Approximately, yes, but that's assuming the wheels don't slip. There's nothing to beat a bit of absolute encoding.

[peach james]

AndyID, the issue is simple- if you need sub 25mm accuracy for the entire layout, then it might be worth worrying about a different solution than current detection blocks & software solutions.  If 25 mm accuracy overall is sufficient, and higher accuracy is required for specific positions, then you use more detection breaks.  I am quite confident that you can use IR or Ultrasonic to determine a specific set of limits to sub mm accuracy, if it is required.  The question to ask is "why", because unless the loco slips a fair amount, it is going to be close enough for a station stop on a through train- probably more accurate (6 scale ft...) than a usual train would be.   If you need a more absolute detection, I know that using RR&Co 5.0C4, I can set a single point sensor to act as a stop, and every time that sensor changes state as the train approaches, the train will stop.  There is some latency in the system- in that the signal has to be encoded, transmitted, mutilated by the computer program, then retransmitted and acted on.  It still can lead to <mm accuracy in placement of a train.   What do you need better accuracy than mm for with regards to train location on a layout ?  Even loading/unloading operations should be designed to allow mm level misalignments.

 

The only place that I know of that has run into problems with wanting mm level precision is miniatur wonderland, where they have had large issues with the boats, as I understand.  Now, they don't have track to detect locations, and are in a dynamic environment (the water), so there are more possible negative interactions going on.  

 

http://www.miniatur-wunderland.com/exhibit/technology/ship-system/control/

 

Again, I'm not saying that it isn't possible to use resistance to a train to determine the physical location of the train on the track.  I'm just suggesting that perhaps, this is a solution looking for a problem, and that it may not be a sensible way to solve the problem that doesn't appear to be much of a problem in the layout design which is being contemplated.  I certainly would not recommend it as a way to having a layout to operate as a signalman.  On the other hand, if the originator wants to know more about how I have programmed Long Marton to be run in the way he seems to want to run, I'm more than willing to help out where I can.

 

James

[AndyID]

Good grief! I didn't think I was suggesting something that constitutes an existential threat.

 

If there is a fundamental reason why this can't work I would like to understand that. And who's to say this method can't be used to augment and enhance some existing method.

 

Far too much NIH methinks.

[kevinlms]

Good grief! I didn't think I was suggesting something that constitutes an existential threat.

 

If there is a fundamental reason why this can't work I would like to understand that. And who's to say this method can't be used to augment and enhance some existing method.

 

Far too much NIH methinks.

From what I read of your original post, your theory is based upon the idea that rail used for model railways has a resistance (which it does, albeit very small - indeed a few weeks ago, you quoted this statistic IIRC).

 

I also understand that you want to determine the position of a locomotive, based on accurately measuring this resistance, to within 6 inches or a foot. That is an incredibly low amount of resistance difference, that you're talking about. What level of tolerance would this system require? What would happen to your train position detection, if something went wrong, such as a minute speck of dirt? That would dramatically alter the circuit resistance.

 

Surely there has to be an easier way?

 

Good luck with your ideas, but I'm too far out of my depth to assist.

[melmerby]

 

I also understand that you want to determine the position of a locomotive, based on accurately measuring this resistance, to within 6 inches or a foot. That is an incredibly low amount of resistance difference, that you're talking about. What level of tolerance would this system require? What would happen to your train position detection, if something went wrong, such as a minute speck of dirt? That would dramatically alter the circuit resistance.

 

To measure a low resistance this accurately and with repeatability you need to use a 4-wire resistance measuring set up or something similar where the resistance of the feed wires (or anything else!) do not affect the readings.

 

With the 4-wire method you pass a current through the resistance from an accurate current source, say 1mA or even 10mA. The voltage drop across the resistance and only the resistance is measured.

From Ohms law the resistance can be ascertained. Laboratory grade instruments will give the reading in Ohms without needing to calculate it yourself.

As the measuring circuit is high resistance typically >10Mohm or even 100Mohm it's effect in shunting the resistance can be ignored or the measuring device might have a correction factor built in to compensate.

 

How you can apply such precision to the flaky world of model railways eludes me.

 

Keith

[melmerby]

Far too much NIH methinks.

Far too much re-inventing the wheel IMHO!

 

Keith

Edited August 20, 2017 by melmerby

[RAF96]

There is a device known as a Time Domain Reflectometer used for working out where a break in a cable is so as to facilitate making repairs.

 

A tad expensive for a model railway and you would have to work out how to get it to see your (several) travelling loco(s) as a break in the system, but you started this thread with the word 'whacky', so . . ..

 

Rob

[melmerby]

There is a device known as a Time Domain Reflectometer used for working out where a break in a cable is so as to facilitate making repairs.

 

A tad expensive for a model railway and you would have to work out how to get it to see your (several) travelling loco(s) as a break in the system, but you started this thread with the word 'whacky', so . . ..

 

Rob

Yes.

If an AC wave form is sent down an unterminated cable (ie. no load at t'other end) most of the signal will be reflected back.

Using this property the position of a break can be determined from the time difference between the sent and received waveforms.

 

Keith

[The Bigbee Line]

I'm following this thread and a admit that must of the technical stuff goes over my head.

 

I would ask a couple of questions to all parties:

 

• I the location of a train/locomotive being calculated from the resistance through the rails?  I presume out on one rail and back through the other.
•  
• If that is the case, how do you know where the front and the rear of the train are?
•  
• Likewise if another train/locomotive enters the same section, how will that appear?
•  
• If you have three train/locomotives in a section, will the middle one be hidden by the other two.

 

These might seem stupid questions, but to be quite relevant.  Can the answers be dumbed down to a level that I can understand.

 

 

Thanks

 

Ernie

 

p.s.  I did suggest using a third rail earlier, you could also use studs as in the marklin stud contact, but as a means of detection rather that power supply...

[melmerby]

 

p.s.  I did suggest using a third rail earlier, you could also use studs as in the marklin stud contact, but as a means of detection rather that power supply...

As did I even earlier, but using resistance wire so that the actual position down the wire could be determined.

Each block could have it's own resistance (of varying lengths) and the position of the contact, hence a known point on the train, could be determined as a proprtion of the overall resistance.

If you have one end of the resistance at 0v and the other at 12v (say) you would just be measuring the voltage at any point along it and hence the position.

Practically there are several ways of doing this and it is used for such things as X-Y plotters to determine tha carriage position.

 

Keith

[SHMD]

I like the idea but getting down to practicalities...

You would have to change how you "wire" up a layout.

 

Each rail length, instead of being connected to a "bus", would have to have each rail end connected to the next rail end (across the fish plates) with soldered copper wire via short loops - I suppose as per prototype - but completely against model railway wiring best practice.

 

This would also be incompatible with DCC and DC plus not allowing layouts with very long runs because of the voltage drop. You would also have to make complex layouts, with multiple routes, wired so that each selectable route is an "allowed route" with known parameters. All other routes/sidings/sheds/coach lighting/etc would have to be isolated.

 

You may, however, be able to apply this to a particular "test" layout to check feasibility before committing more resources to a more "prototype" model railway project.

 

 

Kev.

[The Bigbee Line]

The technology of this thread is way above my skills.  So going back to the OP's title regarding a 'whacky idea', I decided to try it myself...

 

Thinking long and hard I thought an analogue train indicator could be devised.  I decided to use simple 'flag' system, coupled to an analogue scale that would allow easy location of any item of rolling stock at a glance.

 

The prototype has been fitted to one of the sidings on my O gauge layout. The yellow metal calibrated scale is both Metric and Imperial to cater for all tastes.  A simple flag / pointer is attached to each wagon.  The flags can be numbered for each item of stock.  Flags can be of different colours to denote different types of vehicle.  If modelling in Scale 7 a more refined flag is available...

 

The heart of the system is the 'reader' I use a Master Reading System otherwise known as the Mrs.  She never fails me with cries of '.....Look Bob it's a Black Five.. " to break the monotony of shunting.  The position of stock can be downloaded to a piece of paper to allow easy set up for the next session.

 

The only downside is the slight parallax errors depending on where I get the Mrs to sit and how many G&T's I need to ply her with to entice her into the workshop....

 

 

 

 

 

 

[AndyID]

Far too much re-inventing the wheel IMHO!

 

Keith

 

Quite right. Fred Flintstone took it as far as it needed to go (IMHO)

[Junctionmad]

The idea of using the resistance between a fixed point and another movable point to determine position is of course nothing new.

 

The problem is " real life " , under tightly controlled situations it's doable , but with model railways you have large changes in resistance, the difficulty/complexity in profiling each possible track route combined with all possible combinations of motor , including any other track picks ( Carraige lighting etc )

 

In this regard , it's a entirely possible but practically improbable system !!

 

Of course a solution already exists , it's called railcom , and it will report loco address and speed , given the info is resolved almost immediately the loco hits the block section , a simple speed time computation allows you to hence predict position

[Junctionmad]

 

The prototype has been fitted to one of the sidings on my O gauge layout. The yellow metal calibrated scale is both Metric and Imperial to cater for all tastes.  A simple flag / pointer is attached to each wagon.  The flags can be numbered for each item of stock.  Flags can be of different colours to denote different types of vehicle.  If modelling in Scale 7 a more refined flag is available...

 

hmm, so many issues, besides its mechanical complexity,  theres the issue of it fitting in the structure gauge, at first glance it would seem to be a CDD, ( canopy demolition device).  Then we have various legibility issues ( without my  glasses on ) not very " accessible " solution in this modem day.  rolled out to its full extent , it would seem to consume significant numbers of steel measuring tapes and result in a blizzard of coloured flags, that my brain could have serious problems with. 

 

Thanks , but Ill think ill stick to the conventional tried and tested old time solutions, like several networked raspberry pis, a slew of DCC block detectors and railcom equipped DCC chips. 

[raymw]

Could be an interesting experiment. Keeping it purely as Andy initially described. It's not a case of solving a problem that has already been solved, but in thinking 'outside the box', to see what is discovered on the way to perhaps an alternative better solution (the journey is the destination).

 

Without adding any extra sensors, sections, or other components. merely relying on current and voltage levels in the system, then I think I'd build a relatively simple neural network, and let it more or less self-train on the layout. Maybe  one network per section (if you want sections) and a supervising arbitrator. If after a few months I'd proved that there were too many variables changing over time - dirt build up on track/whatever, that made the noise far outweigh the steady state values, then a relatively simple audio digitizing system, using a  couple of microphones would easily handle the problem - Ferranti digitisers, from about  50 years ago could resolve to 0.01mm without much difficulty.

 

Plenty of possibilities, if you want to add more sensors. A trailing loco wheel set, or a separate wagon. Wheels insulated, apart from a very short section of the circumference. Count the shorting pulses would get you to within more or less 3 times the wheel diameter, more shorts to get more accuracy. Need not be dead shorts, of course, if you worry about motor stalling.

 

If it was easy, everybody would do it - https://www.goodreads.com/quotes/573771-i-hear-you-say-why-always-why-you-see-things Too many box openers...