Automatic Signals For Welded Track

This forum is dedicated to Riding Scale Railroading with propulsion using other than steam (Hydraulics, diesel engines, gas engines, electric motors, hybrid etc.)

Moderator: Harold_V

John Hasler
Posts: 1852
Joined: Tue Dec 06, 2016 4:05 pm
Location: Elmwood, Wisconsin

Re: Automatic Signals For Welded Track

Post by John Hasler »

Chuck writes:
> It just irks me that I can't narrow the error to 0.

Never possible in any system anyway. You just have to somehow establish a tolerable error rate.

A radical and seemingly complex approach: lots of cheap cameras and image processing. Security cameras are cheap these days as are computers. Just rope in a machine vision guy and it's a SMOP. Add some optical gates for ground truth.
User avatar
kcameron
Posts: 266
Joined: Wed Aug 17, 2011 4:23 pm
Location: Syracuse New York

Re: Automatic Signals For Welded Track

Post by kcameron »

Chuck,

A lot has changed in JMRI in the past 12 years. Your system comes pretty close to the LCC system. I've got one layout (HO) that uses LCC nodes (by RR-CirKits) to run ABS rules on power up. But it also has hooks so we can switch to CTC mode with the logic for CTC running in JMRI. Like the prototype a split between the field logic and the office logic. We are in the final stages of adding a physical US&S style CTC panel to match what I built in JMRI. The logic for CTC will remain in JMRI as it is quite sizeable. For scope, the layout uses 12 nodes for the layout and 9 nodes for all the I/O in the CTC panel.
-ken cameron
Syracuse Model Railroad Club http://www.SyracuseModelRr.org/
CNY Modelers http://www.cnymod.com/
Finger Lakes Live Steamers http://www.fingerlakeslivesteamers.org/
Member JMRI Developer Team http://www.jmri.org/
mailto: kcameron@twcny.rr.com
In the Upstate New York US area of the world
User avatar
ChuckHackett-844
Posts: 232
Joined: Wed May 03, 2017 3:54 pm
Location: Tampa, Florida

Re: Automatic Signals For Welded Track

Post by ChuckHackett-844 »

kcameron wrote: Wed Aug 11, 2021 9:29 am A lot has changed in JMRI in the past 12 years. Your system comes pretty close to the LCC system. I've got one layout (HO) that uses LCC nodes (by RR-CirKits) to run ABS rules on power up. But it also has hooks so we can switch to CTC mode with the logic for CTC running in JMRI. Like the prototype a split between the field logic and the office logic. We are in the final stages of adding a physical US&S style CTC panel to match what I built in JMRI. The logic for CTC will remain in JMRI as it is quite sizeable. For scope, the layout uses 12 nodes for the layout and 9 nodes for all the I/O in the CTC panel.
Yes, back when I looked at it, it looked like the LCC interface would be a nice fit but I just don't have time to learn enough about it to develop an interface to it.

I have recently implemented "Dispatcher Mode" which initially allows a dispatcher to hold signals with more capabilities to come.

I have had queries from customers about using an actual physical CTC panel or computer rendering of a CTC panel but no one has actually requested the support.

If I get a customer that really wants it I will get in touch with you, other JMRI folks, or a JMRI message board to see if I can join up with someone knowledgeable on LCC so that we could implement the connection. I think my data bus message structure would easily accommodate several levels of support - from JMRI as display only (field block logic) to no field block logic and JMRI in full control (no field block logic).

Let me know if you hear of anyone wanting JMRI to control ride-on railroads.
Regards,

Chuck Hackett, UP Northern 844, Mich-Cal Shay #2
Owner, MiniRail Solutions, LLC, RR Signal Systems (http://www.MiniRailSolutions.com)
"By the work, One knows the workman"
User avatar
kcameron
Posts: 266
Joined: Wed Aug 17, 2011 4:23 pm
Location: Syracuse New York

Re: Automatic Signals For Welded Track

Post by kcameron »

You should also look at how the MERG interfaces to JMRI. It is another CAN based system that might be closer to what you are doing. LCC has a lot of extra details to make it quite robust and those layers may well have been beyond what you felt needed. It did result in a fair bit of complexity in each node so there is no master, it is all strictly node to node with no concept of a boss.

I know of a layout or two that do use JMRI with their signals. But how long it works before a lightning storm takes out part of it is a recurring issue.
-ken cameron
Syracuse Model Railroad Club http://www.SyracuseModelRr.org/
CNY Modelers http://www.cnymod.com/
Finger Lakes Live Steamers http://www.fingerlakeslivesteamers.org/
Member JMRI Developer Team http://www.jmri.org/
mailto: kcameron@twcny.rr.com
In the Upstate New York US area of the world
John Hasler
Posts: 1852
Joined: Tue Dec 06, 2016 4:05 pm
Location: Elmwood, Wisconsin

Re: Automatic Signals For Welded Track

Post by John Hasler »

Pure theoretical speculation: sound should be very effective for train detection on welded track (I don't mean listening for the vibrations produced by locomotives, though you could do that too). Sound propagates *very* well in steel and a pair of wheels and axle should produce a distinct reflection. Like TDR but with ultrasound. Pressure waves are the obvious choice but other modes could be considered.

And you get track condition diagnostics for free.
User avatar
ChuckHackett-844
Posts: 232
Joined: Wed May 03, 2017 3:54 pm
Location: Tampa, Florida

Re: Automatic Signals For Welded Track

Post by ChuckHackett-844 »

John Hasler wrote: Thu Aug 12, 2021 7:13 am Pure theoretical speculation: sound should be very effective for train detection on welded track (I don't mean listening for the vibrations produced by locomotives, though you could do that too). Sound propagates *very* well in steel and a pair of wheels and axle should produce a distinct reflection. Like TDR but with ultrasound. Pressure waves are the obvious choice but other modes could be considered.

And you get track condition diagnostics for free.
Your theory is sound. Thoughts:
  • if using a passive sound (vibration, receiver only) detection, a stopped train would not be detected
  • if using 'active' detection (e.g.: like TDR with transmitter and receiver) the cost of the electronics rises significantly
  • difficult to determine direction the train is moving on short sections of track.
In thinking about it, the logic (and possibly the transducers) used in ultrasonic distance/thickness measurements might be applicable.

Thanks for the thought - the idea of 'sound TDR' had not occurred to me.
Regards,

Chuck Hackett, UP Northern 844, Mich-Cal Shay #2
Owner, MiniRail Solutions, LLC, RR Signal Systems (http://www.MiniRailSolutions.com)
"By the work, One knows the workman"
User avatar
ChuckHackett-844
Posts: 232
Joined: Wed May 03, 2017 3:54 pm
Location: Tampa, Florida

Re: Automatic Signals For Welded Track

Post by ChuckHackett-844 »

Thanks to all for the comments.

At the moment I'm thinking that the most practical method is 'wheel counting' using two mechanical/optical switches activated by wheels with direction determined by quadrature detection. I already have the reliable optical switches and interface electronics (used in my point detectors) so it's just a matter of designing the physical switch addition to the rails and some code in the current controllers. I can hopefully also minimize 'false occupancy' in the firmware (code). Note: 'false occupancy' is the most likely outcome of a miss-count but, thought annoying, is better than 'false clear' from a safety perspective.

One of the issues for automatic signals that use track circuit detection is that every rail joint must be bonded (wire jumper, same as prototype) because even aluminum rail joints can have high resistance over time. The bond wires have to be installed and maintained. I find that 99.5% of issues after a system is installed are poorly installed or broken bond wires.

In addition to implementing wheel counting for customers that have welded track it might also be attractive for customers with long blocks who do not want to have to bond every joint.

My system would also allow the two methods (track circuit and wheel counting) to be mixed in a system - e.g.: track circuit for short, easily bonded sections and wheel counting for long sections.
Regards,

Chuck Hackett, UP Northern 844, Mich-Cal Shay #2
Owner, MiniRail Solutions, LLC, RR Signal Systems (http://www.MiniRailSolutions.com)
"By the work, One knows the workman"
John Hasler
Posts: 1852
Joined: Tue Dec 06, 2016 4:05 pm
Location: Elmwood, Wisconsin

Re: Automatic Signals For Welded Track

Post by John Hasler »

The electronics would be cheap. I've got a $20 (retail price) ARM-based SBC here on my desk that could do it all easily and run idle most of the time. The transducers would probably be the most expensive parts but I doubt that they would be unreasonably so (and ones for passive listening would not be cheaper).

I don't see the problem with direction determination. In any case the mandatory requirement is train detection, since that is all that conventional systems do.

Lots of transducers out there for NDT but instrumentation stuff is pricey. Probably makes more sense to fab up a magnetostrictive transducer. Just a nickel puck and a coil.

Of course, this only works for welded rail.
User avatar
ChuckHackett-844
Posts: 232
Joined: Wed May 03, 2017 3:54 pm
Location: Tampa, Florida

Re: Automatic Signals For Welded Track

Post by ChuckHackett-844 »

John Hasler wrote: Thu Aug 12, 2021 9:14 am The electronics would be cheap. I've got a $20 (retail price) ARM-based SBC here on my desk that could do it all easily and run idle most of the time. The transducers would probably be the most expensive parts but I doubt that they would be unreasonably so (and ones for passive listening would not be cheaper).
The system I sell already has a processor on it and the inputs needed for the sensors as well as signal drivers. The firmware already has all the block and interlock logic so that's not an issue.

My processors are more expensive than an SBC/Arduino because the quantity I make is too samll so no economy of scale. I looked at using Arduino but, by the time you add the lightning protection and other items to make it actually usable for a significantly sized railroad it is not economical.
John Hasler wrote: Thu Aug 12, 2021 9:14 am I don't see the problem with direction determination. In any case the mandatory requirement is train detection, since that is all that conventional systems do.
?? to do wheel counting you have to know direction (i.e.: train goes half in and then backs out). Yes, if you are talking about ultrasound through the rails but, I fear the transducers and electronics would be too costly for a railroad to, for example, detect 100 track sections like one of my customers.

BTW: Full-size railroads, and I, detect direction between detected track sections by observing the change in track occupancy ... 'this one was occupied and the neighboring tracks were not and then this neighbor went occupied' ... not foolproof but adequate.
Regards,

Chuck Hackett, UP Northern 844, Mich-Cal Shay #2
Owner, MiniRail Solutions, LLC, RR Signal Systems (http://www.MiniRailSolutions.com)
"By the work, One knows the workman"
John Hasler
Posts: 1852
Joined: Tue Dec 06, 2016 4:05 pm
Location: Elmwood, Wisconsin

Re: Automatic Signals For Welded Track

Post by John Hasler »

Look at the Teensy SBCs from Sparkfun. Adafruit also has some that might be interesting. RaspberryPi also has quite a few boards.

The ultrasound system (useful only for welded track, of course) needs a dedicated $20 DSP for signal processing. It would provide input to your system. There is no electrical connection to the track in the ultrasound system. I think that the range might be such that the physical blocks could be very long. The latter would need to be determined by experiment, though.
User avatar
Atkinson_Railroad
Posts: 226
Joined: Mon Jun 08, 2015 6:27 pm
Location: Michigan
Contact:

Re: Automatic Signals For Welded Track

Post by Atkinson_Railroad »

John Hasler's suggestion about sound is a good one.

A search using terms... Acoustic Detection or Acoustic Monitoring can provoke additional ideas.

Sound is a very good source for determining the status of many different kinds of mechanical arrangements.

A rolling wheel upon a rail is a mechanical arrangement.

The creative ways in which one detects a sound and converts it to some other useful signal source
becomes a whole new frontier for discussion.

I'm reminded of the time a master machinist asked what I was looking at while
machining a part running a mill. He said, "There's nothing to look at, you go by sound."

Additionally... I recently replaced brakes on my vehicle.

What motivated me to get new brakes?

The Sound they were making ; )

John
Post Reply