Trains.com

Subscriber & Member Login

Login, or register today to interact in our online community, comment on articles, receive our newsletter, manage your account online and more!

model track geometry car ?

3892 views
49 replies
1 rating 2 rating 3 rating 4 rating 5 rating
  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
model track geometry car ?
Posted by gregc on Monday, July 26, 2021 6:35 AM

not sure if "track geometry" is an accurate description.   i'm curious what it would take to design/build a car that can be run over a model railroad to detect flaws in tracks.   

i'm working on a layout with a lot of hand built turnouts with lots of issues.   And would not be surprised if problems develop with use

without worrying about feasibility, what flaws would you want to detect?

here's what come to mind

  • track gauge
  • abrupt vertical rail transitions
  • abrupt horizontal rail transitions (especially on curves)
  • excessive gaps between rails (what's excessive)?

specific to hand built turnouts

  • non-tight points against stock rail
  • excessively narrow gap between points and stock rails (the DCC problem)
  • excessive gap near tip of frog
  • excessive vertical drop near tip of frog 
  • non-smooth movement over tip of frog (mis-aligned guard rail)

as for feasibility

  • i wonder if an accelerometer could detect jolts from misaligned track
  • i wonder if optical reflection could be used to detect position of car over rails.  the reflection when directly over the rail should be higher than if gauge is wider
  • i wonder if have a sensor car with a single truck would be better than a conventional car with 2 trucks.

no doubt there will be comments about feasibility or that's there's no need for such a device.   i look forward to more constructive comments.

greg - Philadelphia & Reading / Reading

  • Member since
    September 2003
  • 17,630 posts
Posted by Overmod on Monday, July 26, 2021 7:31 AM

Of course you could start with John Allen's car, with a BB in a curved track with tilt switches at each end.  Intermediate sensors or even a continuous element would add clinometer sensitivity, and an encoder on one axle would give you both position and speed data.  Make one crosswise as well as longitudinal and you have curve smoothness and cross-level depending on whether it's stopped or moving...

More seriously, you might adapt the three-axis accelerometer core in a phone to the necessary degree of position.  And a camera feed looking at the railhead would give you the 'stripe' from which offset would provide contact patch, and if you are good, perhaps focusably photomicrographically look at the state of corrosion or dirt, microarcing, state of cleaning effectiveness, etc either as a scan or discrete captures. Two properly calibrated, one to a rail, can give absolute and differential gauge.

A problem I see with this is the running precision and lack of play in the moving parts needed to get this to run smoothly.  The technology surely exists, but the cost likely becomes the critical factor, with handling the data close behind it.

  • Member since
    September 2002
  • 7,120 posts
Posted by ndbprr on Monday, July 26, 2021 7:44 AM

There was one in MR many years ago.  It was a clear plastic piece the size of a car with trucks that you could look through from the top to see what exactly was happening.

  • Member since
    September 2004
  • From: Dearborn Station
  • 21,722 posts
Posted by richhotrain on Monday, July 26, 2021 7:49 AM

I use a 6-axle diesel and a 2-10-4 steam engine to detect all of the track flaws that you mention.

Rich

Alton Junction

  • Member since
    May 2004
  • 6,872 posts
Posted by 7j43k on Monday, July 26, 2021 9:47 AM

For a track geometry car, the only thing I'd really be after is abrupt misalignments.  I would base it either on two cars, and their interaction (sort of an A-B car), or a single car with an added truck in the middle.  The former would likely be more sensitive.  I was thinking two "passenger" cars, painted and detailed to look the part.

An optical system is probably the best way, though that's tech I don't know much about.  The other way would be more a mechanical contact system--when it's not adequately aligned, the probe hits its surrounding contact system.

I have spent the occasional few minutes thinking on this, and even have a manila folder for it.  I would use it on a Free-mo setup to detect alignment flaws and humps and dips.  I was planning on including some kind of alarm system, either aural or visual or both.

For a home layout, where you can just take your time and nobody's looking, you could bypass the electronics and modelmaking, and just have two cars connected together, and a pointer on each one.  Shove the car(s) around, and watch the pointer.

I got the idea for this design by watching the couplers move vertically as cars passed over some bad track.  It's amazing how bad track has to be before couplers disconnect.

 

Ed

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Monday, July 26, 2021 10:09 AM

don't rule out some sort of mechanical appendage(s)

greg - Philadelphia & Reading / Reading

  • Member since
    May 2004
  • 6,872 posts
Posted by 7j43k on Monday, July 26, 2021 10:15 AM

gregc

don't rule out some sort of mechanical appendage(s)

 

 

"...mechanical contact..."

 

Or are you talking about a "thing" that gets attached to a semi-random car, like a 50' flat?

 

Ed

  • Member since
    January 2017
  • From: Southern Florida Gulf Coast
  • 13,844 posts
Posted by SeeYou190 on Monday, July 26, 2021 10:26 AM

ndbprr
There was one in MR many years ago.  It was a clear plastic piece the size of a car with trucks that you could look through from the top to see what exactly was happening.

I have one, and it proved to be of very little help.

Mine had a six wheel truck on one end and a four wheel truck on the other. It never revealed anything beyond what I already knew by watching "normal" freight cars derail.

A basic NMRA gauge and a straightedge (steel rule) reveal so much more.

A car loaded with sensors might be the best idea, but I cannot imagine it would be affordable or practical to build.

-Kevin

Wink Happily modeling my STRATTON & GILLETTE RAILROAD. A Class A line located in a personal fantasy world of semi-plausible nonsense on Tuesday, August 3rd, 1954.

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Monday, July 26, 2021 10:31 AM

appendage angled down contacting the rail that could sense a vertical change.   would have a flange to maintain alignment

could something similar sense poor points alignment, a horizontal mis-alignment

how sensitive would it have to be?    of course it beeps.

 

we've all done this kinda thing using our fingers, moving cars across the track, running long locos.   i built one of those plexiglass cars.   but if there's a lot of track, having something that you can use a loco to push/pull around could help find things that are inconvenient to see.   and as I said, i've seen problems set it after some usage.

greg - Philadelphia & Reading / Reading

  • Member since
    February 2008
  • 1,790 posts
Posted by kasskaboose on Monday, July 26, 2021 3:12 PM

ndbprr

There was one in MR many years ago.  It was a clear plastic piece the size of a car with trucks that you could look through from the top to see what exactly was happening.

 

 
You always can buy one:
 
 
 
  • Member since
    January 2017
  • From: Southern Florida Gulf Coast
  • 13,844 posts
Posted by SeeYou190 on Monday, July 26, 2021 3:19 PM

kasskaboose
You always can buy one:

Hey, that is the one that I have, or one exactly like it..

I did not get mine from Kalmbach. If I am remembering correctly I bought it about 10-12 years ago.

-Kevin

Wink Happily modeling my STRATTON & GILLETTE RAILROAD. A Class A line located in a personal fantasy world of semi-plausible nonsense on Tuesday, August 3rd, 1954.

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Monday, July 26, 2021 3:33 PM

can we put a derailer on this thread before the posts start interfering with the main discussion?

greg - Philadelphia & Reading / Reading

  • Member since
    December 2002
  • From: US
  • 34 posts
Posted by fisherdm on Monday, July 26, 2021 6:06 PM

I've used a coupled pair of the clear plastic cars to aid in identifying flaws in my trackwork. One way to increase the "sensitivity" is to mount a long wire on a low stand in the middle of one car. Extend the wire to about the center of the other car. Any defect in the track will be magnified by the lever arm of the wire. The cars are also very light, so are quite prone to derailing - which in this case is what you want. That won't tell you the problem; that's what the NMRA track gage is for.

-Danial

  • Member since
    September 2004
  • From: Dearborn Station
  • 21,722 posts
Posted by richhotrain on Tuesday, July 27, 2021 6:05 AM

gregc

no doubt there will be comments about feasibility or that's there's no need for such a device.   i look forward to more constructive comments. 

gregc

can we put a derailer on this thread before the posts start interfering with the main discussion? 

greg, I have been following this thread with interest and my sense is that no one is trying to derail it. My guess is that it may be difficult to find forum members with the talent or time or interest to develop such a car.

Why not try to develop such a car yourself? If you could accomplish such a feat, members and non-members alike would flock to your side. Seriously!

You mention such gadgets as an accelerometer, optical reflection, and a sensor car with a single truck. Why not take a shot at this yourself?

Rich

Alton Junction

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Tuesday, July 27, 2021 6:14 AM

richhotrain
Why not try to develop such a car yourself?

gregc
i'm curious what it would take to design/build a car that can be run over a model railroad to detect flaws in tracks.

what do you think i was asking?

greg - Philadelphia & Reading / Reading

  • Member since
    September 2004
  • From: Dearborn Station
  • 21,722 posts
Posted by richhotrain on Tuesday, July 27, 2021 6:38 AM

gregc
 
richhotrain
Why not try to develop such a car yourself? 
gregc
i'm curious what it would take to design/build a car that can be run over a model railroad to detect flaws in tracks. 

what do you think i was asking? 

Based upon your initial post on this thread and the knowledge and skill that you exhibit on this forum, I would think that you would be the ideal candidate to develop such a car.

I, for one, would love to see you do it. It would surely beat all of the trial and error testing that we all go through to find the source of derailments on our layouts.

Rich

Alton Junction

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Tuesday, July 27, 2021 7:17 AM

yes i'm interested in building such a car.  as I said

gregc
i'm working on a layout with a lot of hand built turnouts with lots of issues.   And would not be surprised if problems develop with use

i guess it's not obvious that one of the first things that better engineers do is bounce ideas off others, looking for concerns, flaws in their thinking and other ideas that may lead to better ideas.   (constructive reviews minimize trial and error)

if it's not clear, yes, i'm interested in hearing for others, even the less technical, about what such a car should check for and even crazy ideas for how to do it that could lead to more realizable ideas.   don't be shy.

some have already offered some very worthwhile suggestions

greg - Philadelphia & Reading / Reading

  • Member since
    November 2006
  • From: NW Pa Snow-belt.
  • 1,949 posts
Posted by ricktrains4824 on Tuesday, July 27, 2021 12:59 PM

Crazy ideas???

An accelerometer connected to an Arduinio?

Seems like the coding would be easy enough to detect flaws....

 

Ricky W.

HO scale Proto-freelancer.

My Railroad rules:

1: It's my railroad, my rules.

2: It's for having fun and enjoyment.

3: Any objections, consult above rules.

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Tuesday, July 27, 2021 3:02 PM

overmod suggested that in the 1st reply

have ordered mpu-6050.  need to see how sensitive it is.

wonder if it could sense a horizontal misalignment

greg - Philadelphia & Reading / Reading

  • Member since
    September 2004
  • From: Dearborn Station
  • 21,722 posts
Posted by richhotrain on Tuesday, July 27, 2021 6:09 PM

gregc

i guess it's not obvious that one of the first things that better engineers do is bounce ideas off others, looking for concerns, flaws in their thinking and other ideas that may lead to better ideas.   (constructive reviews minimize trial and error)

if it's not clear, yes, i'm interested in hearing for others, even the less technical, about what such a car should check for and even crazy ideas for how to do it that could lead to more realizable ideas.   don't be shy.

OK, I am certainly not an engineer, but I will take a shot at this.

I will skip over the issues that you raised in your initial post specific to hand built turnouts. 

But here is a list of issues that I have encountered with my own track work, some of which you mentioned in your initial post.

1. Track Gauge. This is one of the most significant causes of derailments and is often difficult to detect with an NMRA gauge. In fact, I don't find an NMRA gauge all that useful because in my opinion it doesn't easily detect parallel rails that are slightly out of gauge. What I would find helpful is a display that shows the spread between rails in decimals.

2. Vertical Rail Transitions. You mentioned "abrupt" transitions which usually result when connecting rails are joined at slightly different heights, but another problem is when the parallel rails are at different heights. How to measure?

3. Horizontal Rail Transitions. You mention abrupt transtions especially on curves. The most extreme example is outright kinks. Again, how to measure?

4. Lack of (Sufficient) Easement Into Curve. There must be an easy way to measure curvature. A simple example is the use of sectonal track. Observe the abruptness of that first piece of curved sectional track adjoining the last piece of straight sectional track. So, when using flex track, how abrupt is that transition from straight to curved?

5. Transition From Track to Turnout. This transition is often abrupt and the result is a severe horizontal rail transition. Once again, how to measure?

Just a few thoughts. Speaking personally, these are my Top 5 causes of derailments on my layouts, past and present.

Rich

Alton Junction

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Tuesday, July 27, 2021 7:17 PM

thanks for the thoughts.

not sure how to measure gauge.  posted my thoughts on optical reflection.   was thinking of using a pair of axles with n-gauge wheels, close together replacing a single HO axle.    one sprung right and the other left and some way (optical alignment ??) sensing too wide/narrow.

which is more critical too tight or too wide?

been thinking about accelerometer and think it may be possible to detect any abrupt rail transition, both vertical and horizontal.  large gaps.    may also detect concave kinks and maybe misaligned tracks such as track and turnouts.    think this is promising if acceleromter is sufficiently sensitive.

hadn't thought about easements.    an accelerometer may be able to detect an abrupt change in curvature (i.e. from straight to curve) and may be able to recognize a lack of even minimal easement.

i'm surprised to think that an accelerometer could do so much.

thinking that there could be different cars with different purposes: gauge car, accelerometer car.   thinking of 4-wheel bobber caboose frames

 

greg - Philadelphia & Reading / Reading

  • Member since
    September 2004
  • From: Dearborn Station
  • 21,722 posts
Posted by richhotrain on Tuesday, July 27, 2021 8:52 PM

gregc

not sure how to measure gauge.  

which is more critical too tight or too wide?

I never really had much of a problem with out of gauge track until I built my most recent layout. On my old layouts, most derailments were caused by vertical or horizontal (kinks on curves) misalignments. But on my newest layout, I did something stupid and that was to remove too many ties on soldered curves. That led to out of gauge curves that were too wide. So, I solved the problem by cutting out and re-laying soldered curves with much fewer ties removed.

Rich

Alton Junction

  • Member since
    August 2019
  • 39 posts
Posted by hgodling on Tuesday, July 27, 2021 10:18 PM

Looking at the NMRA standard the tolerance for gauge in HO is only +/-0.006 in. It is hard to find a device that can measure that accurately, let alone while moving and getting it to fit on a car. I think the wobble in the mounting of the trucks would exceed any error you are looking for in the track. 

To judge the easement you need the slope of the acceleration. This is very hard to get from an accel. Accels tend to be very noisy (even the very expensive $1k ones). Phone use filters to make it work, but anything we would be interested would be very short, and likely filtered out along with the noise. 

I have thought about putting a magnet on a flat car so I can mount a digital level to measure the grade. 

I think it would be hard to get a car that would be more useful then your NMRA gauge and running your finger along the track. However, even if it doesn't work real well, it sounds like it could be a fun project and could make for an interesting conversation piece when showing off the layout. 

  • Member since
    May 2004
  • 6,872 posts
Posted by 7j43k on Tuesday, July 27, 2021 10:49 PM

I suspect making a "gage checker" is doable, using Greg's spring-loaded individual wheels.

A problem shows up, though, when the gage checker goes through a frog.  It's especially bad when it approaches from the pointy end, as the wheel that approaches the frog will be spring-loaded to take the other route.  

And you have a big mess.

 

While all the other information described is nice, my main goal would be vertical changes.  That is simple to do, though the way-cool alarm system and flashing lights is a bit more work.

One POSSIBLE super simple method would be to connect the two test cars with couplers that have had their knuckles filed down to a lesser than normal height, say, arbitrarily, half the normal knuckle height.

If the car(s) separate, you gotta fix it!

Ed

  • Member since
    September 2003
  • 17,630 posts
Posted by Overmod on Wednesday, July 28, 2021 8:23 AM

I can't imagine measuring gauge effectively in a 'model' context any way but optically, probably reading the railheads or point structure with a magnified camera input and using optical-mouse approaches for edge detection and discrimination.  

A problem that remains is the absolute reference or baseline that the car would use.  Inertial is nonstarting; accelerometers even remotely practical in this application, as noted, are useful only for short-period differential.

I suspect there may be a use for making the car capable of 'telemetering' data or storing it quickly on board, and only processing to find flaws 'later'.  Although there is a sort of elegant satisfaction (to me at least) in the prospect of a scale Sperry car that goes around the layout making tiny colored marks on the rails corresponding to type and severity of defects...

  • Member since
    January 2009
  • From: Maryland
  • 11,128 posts
Posted by ATLANTIC CENTRAL on Wednesday, July 28, 2021 9:31 AM

I measured the "straighness" of my trackwork with a lazer after the fact when asked about getting track perfectly straight:

 

Since I have never had these issues Greg is referring to, even when I was only 15 and hand laying/building my own turnouts, I would have no need for his proposed tool.

Sheldon

    

  • Member since
    May 2004
  • 6,872 posts
Posted by 7j43k on Wednesday, July 28, 2021 9:58 AM

It does seem that there's no need for this after-the-fact tool if the layout builder did it right the first time.  And if he didn't, the time to have caught it is during initial construction, when there's not any scenery in place.

I think there is a possible use for one in modular setups, where the joints may need fine tuning, or someone brings a module that has awful trackwork.  Just run it around the mainline and see if it squawks.  Loudly.

There ARE egos big enough, by the way, to get offended when it is pointed out that their module has awful trackwork.  "Really?  YOU'RE offended?  WE have to deal with this thing."

 

 

Ed

  • Member since
    January 2009
  • From: Maryland
  • 11,128 posts
Posted by ATLANTIC CENTRAL on Wednesday, July 28, 2021 10:01 AM

7j43k

It does seem that there's no need for this after-the-fact tool if the layout builder did it right the first time.  And if he didn't, the time to have caught it is during initial construction, when there's not any scenery in place.

I think there is a possible use for one in modular setups, where the joints may need fine tuning, or someone brings a module that has awful trackwork.  Just run it around the mainline and see if it squawks.  Loudly.

There ARE egos big enough, by the way, to get offended when it is pointed out that their module has awful trackwork.  "Really?  YOU'RE offended?  WE have to deal with this thing."

 

 

Ed

 

Just one reason why I have decided that clubs or modular groups are no longer my cup of tea. I will build my railroad, then invite those who I desire, and who are interested, to come play.

Sheldon

    

  • Member since
    July 2009
  • From: somerset, nj
  • 3,751 posts
Posted by gregc on Wednesday, July 28, 2021 10:51 AM

gregc
i'm working on a layout with a lot of hand built turnouts with lots of issues.   And would not be surprised if problems develop with use

  

especially interested in alignment of points.  (how?)

greg - Philadelphia & Reading / Reading

  • Member since
    January 2009
  • From: Maryland
  • 11,128 posts
Posted by ATLANTIC CENTRAL on Wednesday, July 28, 2021 3:45 PM

gregc

 

 
gregc
i'm working on a layout with a lot of hand built turnouts with lots of issues.   And would not be surprised if problems develop with use

 

  

especially interested in alignment of points.  (how?)

 

OK, now I have some questions.

 

First, what do you mean by alignment of points? Length? Gap on open side? Gauge?

Second, what is the method of turnout construction? Fastracks? Why are turnouts in place without ties? Is that paper under all of this with the track plan?

What is under the paper? Does the paper stay? (seems like a bad idea to me)

I don't a have a detailed track plan to understand, but just in these two photos I see track geometry that makes little sense and seems less than ideal.

When I need or want to, I have been hand laying track and turnouts since the early 70's, and everyone has their own methods and were taught different approaches, but I see a lot going on here I would not be doing.

There are a few S curve alignments the prototype would not even tolerate in a yard?

Is that commercial track? Then why are we hand laying turnouts?

Sheldon

    

Subscriber & Member Login

Login, or register today to interact in our online community, comment on articles, receive our newsletter, manage your account online and more!

Users Online

Search the Community

ADVERTISEMENT
ADVERTISEMENT
ADVERTISEMENT
Model Railroader Newsletter See all
Sign up for our FREE e-newsletter and get model railroad news in your inbox!