New magic control system for model railroad ... is this a scam?

[lgrfbs]

Hi forum, it's been a while since I wrote here.

Found this product for model rail electronics, my stomach says is something "wrong" with this product.
At the risk of me becoming a "useful idiot" for a spammer so:
The product is Railmagic and according to the inventor himself,so his product should be able to position 16 locomotives on millimeter accuracy, with only two cables to the track and no other equipment as I understood it and that it should go on any digital system completely, from any manufacturer.

I think it's e-mail collection for further sale of email addresses that is the real goal.


Introduction to Railmagic - in stand-alone mode
https://youtu.be/JHCPrZNDM54

http://www.railmagic.com

Previous discussion on a Swedish model railway forum:
https://www.svensktmjforum.se/forum/index.php?topic=23001.0

So what do you think, has he come up with something that no one has thought of before?

[Fred27]

Sorry - I couldn't watch more than 30 seconds of that weird creepy guy. I kept expecting him to do some "real" magic and pull a string of hankerchiefs out of his sleeve, produce a rabbit or something.

[rrinker]

 Interesting. Since model trains is my other hobby, I am certainly interested in seeing if this is real or just more dodgy woo-woo.

First thing I see is, of course, the examples are always with trivial track arrangements. I am building a model railroad that fills up a 25x32 (feet, sorry, I'm in that backward country) and the biggest problem is that there's no way only 2 wires to the track can adequately provide power to that much track - the voltage drop in code 83 nickle silver will mean the train will darn near stop at the furthest point.

That may be a clue as to how it works. Since it appears tha track power needs to go through this Rail Magic device, it may be able to sense distance from the terminal track location where the wires connect to the rails. I don;t doubt that is impossible. At a low cost? Not sure about that, but all sorts of other sensor devices are getting cheaper as they appear in consumer objects, so I'm not quite ready to call bullshit on this.

However - I'd love to see how this would work on a larger more complex layout. Can it handle multiple track feeds? Perhaps if each had its own DCC booster - since both rails have to be gapped between such sections anyway, it would be the same as 2 or 3 of his simple demonstration tracks.

Second - what about multiple trains on the same track? He doesn't demo anything like that. But that's exactly what you'd want to do with some simple automation, run two trains around that loop, with one periodically taking the siding and stopping to allow the other to overtake it. Or alternate taking a lap around - one stays stopped, the other runs around, then stops, the first one now goes out and runs a lap, etc.

Summary, I think this could work as shown. But the usefulness of it beyond this simple demonstration - I don;t see it.

Edit: I just skipped ahead in the video to see it actually do something so I didn't have to watch and listen to him talk

[TheWelly888]

I doubt this would work at all - it's likely based on the resistance of the rails from the feeder to the train and this will change greatly in use (dirt on the rails!)

Sounds like woo-woo to me - and I'm interested in model trains myself.

The Mark 1 eyeball is by far the best for placing locos at the right place on the tracks.

[lgrfbs]

Sorry - I couldn't watch more than 30 seconds of that weird creepy guy. I kept expecting him to do some "real" magic and pull a string of hankerchiefs out of his sleeve, produce a rabbit or something.

Thank you, then I'm not alone in that particular feeling.
The guy has two more videos on his Youtube account if you feel like it. 
Obtain automatic braking in front of signals in seconds - https://youtu.be/TE94XejEyKE
Get computer-controlled in just one day - https://youtu.be/L-bUZ_UcKC0

Second - what about multiple trains on the same track?

Your question was up as a comment on the Youtube clip, attaching it here:

Phil Newman
1 month ago
Can this system cope with more than one loco on the same track?
 

Railmagic
1 month ago (edited)
Hi Phil,
It is designed to handle 16 locomotives in each power domain. The command station is one power domain, and each additional booster counts as a power domain. The current limitation of these is typically 3-5 amp, so therefore it is unlikely to have more than 16 running locomotives in a power domain. The number of standing still locomotives is unlimited.
Best regards, Ulrik

[Cyberdragon]

Then it's absolute BS.

No way would you be able to accurately measure position based on one resistance measurement alone. The power of different trains varies between different models, let alone gauges, let alone resistance in the wheels and pickups, motor brushes, exc. But 16 moving locos from one box? Hell no,  with all the trains on a parallel circuit their impedences would just combine down and vary all over the place into an unusable mess.

If the boxes somehow communicated with each other and could tell the resistance relative to each power connection then you mignt be able to do something but it would still not be "millimeter precision".

[lgrfbs]

After looking through all the videos, I think the magic box has access to:
1; LAN for the mobile phone to connect the control center through wifi in the router.
2; The NMRA-DCC signal from the control center (Rocos Z21/z21, Märklins Command Station 2) that goes through the box and then proceeds to the rails.

[Haenk]

Two ideas:
- purely timing based, this would explain why you need to parameter your train (i.e. higher weight, slower acceleration, longer braking distance). "Accelerate weight xyz for 5 seconds to speed b, reduce speed to c after d seconds, then brake after e seconds, calculated full stop after 37mm)
- inject timing signal onto rails (would probably require additional "refresh" hardware inside the train), like a delay line - does not seem likely as per setup shown

[Domagoj T]

Working on a model railroad is actually my day job, and I'm very close to calling BS on this. It might be possible that the magic box polls the trains (modern digital decoders have the bidirectional coms ability) and measure the response time (ping), but that would turn useless on any layout more complex than a simple oval.
On our layout we can achieve sub 10 mm stopping accuracy without issues, and even 5 mm is doable, but it involves a bit more than just two wires and too much computing power for the layout of our size to be used everywhere. We have more than 1500 meters of track covered by about a thousand feedback lines for train positions. All the locos we put on the track are measured on a calibrated track and the speed profile is stored in the control software. At most locations it's good enough to just detect the train entering a block and stop it x seconds later, but on stations we employ actual speed measurement and trains stop pretty much exactly where we want them to. The infrastructure behind all this is fairly substantial.

The demo video did not convince me. The setup used provides to many ways to misrepresent the situation.

[Kalvin]

Just to be curious: Do these model trains require some modifications in order to work with different controllers ie. would it be considered to be "ok" if the train needs some modifications to work with this new control system?

If the train can be modified to contain a WIFI module and sensors (A magnetic sensor and magnets placed under the track for speed and position estimation? A wheel rotation sensor/counter for speed and position estimation? IMU-module (gyroscope/acceleration/compass sensor to detect train movement for position estimation)? A sensor for counting the sleepers? A combination of different sensors mentioned) and the sensor data can be sent from the train to the controller over WIFI, then it may be possible to do something like this.

I do not know anything about the model trains etc., but wanted to perform a mental exercise on how I would approach this problem (without really knowing the domain).

[Domagoj T]

Modifications on the trains are matter of taste. Generally speaking most people don't mind modification on the interior (motor, gearbox, electronics) in order to improve the driving performance. On the other hand, there are some that will absolutely not accept any modification that is visible on the outside if the real loco doesn't have that, ie. its ok to modify the look as long as the result is more accurate and true to the real one. Then there are collectors that value the model only if it hasn't been altered in any way, preferably never taken out of the box. Then there are those that value the box as much as the loco itself (only when buying, when selling it usually doesn't matter  ).

Most electronics in these locos (called decoders) are so called multi-protocol, meaning they will work on two main modern protocols (DCC and MFX), but will also work on legacy digital and analog systems - no digital signal, just clean DC or AC in case of Marklin. Of course I'm talking about digital systems, there are still plenty of analog locos being produced, but this system in OP is digital.

Depending on manufacturer, there might not be enough room in locomotives to stuff in a whole lot of additional electronics. As you climb the price bracket of the model you lose space inside.

For a while I was trying to do something similar to what you're describing, but for another purpose. I wanted the loco to detect if it crashed into something, or perhaps a wagon derailed and snagged something. There were a few issues I encountered. Axle rotation sensor can easily be done, but some locos have all the wheels driven, so you can't tell when it crashed, since the wheels will continue to spin (the motors are quite strong). Next I tried sleeper detector, but even with brand new tracks, right out of the box, I was getting inconsistent results between different production runs. When I put the loco on the track, the problem was only made worse by dust and various inconsistencies. Next on the list was a vibration sensor, but again, loco vibrates during a stall just as it does during normal operation. I gave up before getting to IMUs.
What there is on the market offered as a product is RFID reader that can be placed next to the track, and each loco gets a tag, but that does not offer position readout, only that that specific train passed a certain point, so not really a competitor to the OP thing. We have a few locos that use WIFI for live video feed, but we are not really happy with them. The range is a bit short (our layout is big, but still...) and we had trouble keeping the connection up. Short circuits are not uncommon on the layout, and each time the connection drops so we got tired of manually reconnecting the WIFI several times per day.

[CatalinaWOW]

As long as the model is for a train in the last 40 to 50 years you could add bar codes to the rolling stock just as the full scale guys did (at least in the US) and place readers around the layout.  Perfectly scale on the rolling stock and with some ingenuity the readers could be close to scale.  On earlier modeling periods the bar code could be on the bottom of the engines/cars.  Takes essentially zero space on the rolling stock.  This would give a location with a few millimeters as each reader passed, giving an update to dead reckoning systems.

[Cyberdragon]

As long as the model is for a train in the last 40 to 50 years you could add bar codes to the rolling stock just as the full scale guys did (at least in the US) and place readers around the layout.  Perfectly scale on the rolling stock and with some ingenuity the readers could be close to scale.  On earlier modeling periods the bar code could be on the bottom of the engines/cars.  Takes essentially zero space on the rolling stock.  This would give a location with a few millimeters as each reader passed, giving an update to dead reckoning systems.

Barcodes (KarTrak) were only used for 10 years (67-77). After that it went to RFID in the late 80s.

[CatalinaWOW]

As long as the model is for a train in the last 40 to 50 years you could add bar codes to the rolling stock just as the full scale guys did (at least in the US) and place readers around the layout.  Perfectly scale on the rolling stock and with some ingenuity the readers could be close to scale.  On earlier modeling periods the bar code could be on the bottom of the engines/cars.  Takes essentially zero space on the rolling stock.  This would give a location with a few millimeters as each reader passed, giving an update to dead reckoning systems.

Barcodes (KarTrak) were only used for 10 years (67-77). After that it went to RFID in the late 80s.

I am only a casual watcher of trains, but it seems to me that they didn't remove the tags when they switched to RFID.  I still see cars with the KarTraK tags today.  So you can legitimately model later eras with tags (or just move them underneath). 

[lgrfbs]

As I understand it, the locomotive or any other electronics on an existing track should not need to be rebuilt or modified.
This makes me very suspicious of how it would work.
 The only thing to do is to cut the cord to the rails and insert the magic box on the newly cut cords and connect a network cable between the magic box and the router.

Paj2 drew an electrical diagram.
(Attachment Link)

RM="magic box"
Z21=control center for the layout.

[HackedFridgeMagnet]

I know nothing about trains but wouldn't this be done by time domain reflectometry? It seems a complex but obvious way to do it.

[Domagoj T]

I know nothing about trains but wouldn't this be done by time domain reflectometry? It seems a complex but obvious way to do it.

How would that work on a layout where you have multiple current paths (some might be of identical lengths) to multiple locomotives? How would the system differentiate from different paths and different locomotives? Every locomotive is listening to all the instructions sent to the layout, but only react to the ones addressed to them, but electrically they are pretty much the same. Reflectometry might work on simple circuit but railroad layout if usually convoluted on purpose.

[HackedFridgeMagnet]

Well if it could work on simple systems then you could use multiple source points for more complicated systems.
Or even have multiple positions in the line that have known termination values.
Also if the processor knows the topology of the network it could keep a record of the position of the vehicles so that any instantaneously ambiguous results could be resolved.

Lots of ways but I really don't know what would be suitable or effective.

Working on a model railroad is actually my day job

Wow that would be a hard to get position.

[Domagoj T]

I know too little about reflectometry and nothing about this device, so I'll give this device a tiny bit of benefit of doubt, just in case it turns out to actually work, since it's interesting and relevant for my job. But this sounds like patent trolling. You don't need a functional device to patent a concept. Perhaps he just wants to sell the patent.

Wow that would be a hard to get position.

It certainly is a small market. 

[HackedFridgeMagnet]

Unfortunately there are plenty scammers on youtube.

ps. I have no idea if reflectometry is even viable on train tracks, it's only a guess.

[Haenk]

After having re-watched those videos, it still seems to be a (not quite simple) time/speed/distance calculation, considering further parameters (train weight etc.).
He precalculates/predicts train positions, no more (and no less). I assume this works quite well with digitally controlled trains, but is far from being failproof.
Adding sensors, as might be the case in one video, could help. Not sure, if they are optional, but I can't image how this should otherwise work in a complex layout.

[rrinker]

 Domagoj T hit on what my next idea might be - using the Railcomm feedback in some sort of TDR fashion, or at least use the cutout interval for something. However, he also got right to the root of the problem - get beyond a simple oval, and have multiple paths for the trains to be on, and it all goes out the window as there is no unique single point time of flight measurement for an specific path, and the opposite, more then one path would have the exact same distance from the measurement point, so how, without additional sensors, would you possibly know which location was the actual occupied one?

 Barcodes have been done, or at least attempted, by sticking them to the underside of the loco or rolling stock and putting the readers between the rails.

 RFID has been used and can be made to work pretty well, using the very short range readers so they only detect equipment with tags on the track the reader is on, and not an adjacent track. There are semi-commercial systems for this available. Coupled (no pun intended) with the train tracking capability of some of the advanced automation applications, this would work well to feed in what equipment started at what point, and then the subsequent activation of more standard block detection plus knowledge of selected routes would allow the software to place any piece of rolling stock anywhere on the layout with great precision.  Me, I tend to limit the automation on my layout to the scenery rather than have it run trains. Playing with the trains is my semi-escape from the computer controlled everything life.

 Also - your day job is playing with a model railroad? How do I get a job like that? I mostly work remotely anyway, so I've been little affected work-wise with the whole virus thing, so i can't just sneak off to the basement all day.

[Domagoj T]

Also - your day job is playing with a model railroad? How do I get a job like that? I mostly work remotely anyway, so I've been little affected work-wise with the whole virus thing, so i can't just sneak off to the basement all day.

When I started this job (2011) there was quite a bit of playing, but with time my responsibilities shifted towards the back end. My defacto job is finding solutions to everyday problems we encounter while building and operating the layout (mostly electronics and mechanical devices), but I do occasionally get to play with trains and build stuff.
As to how do you get a job like that, well in my case it started with a chance encounter. You know, the good old "being in the right place at the right time". Knowing (at that time not very well) the owner who was about to start the project and was in need of help, helped too.
If you are serious about this, there might be something around your place. There aren't many commercial model railroad layouts/exhibits out there, and I guess clubs don't have staff on payroll, but there are quite a few businesses that provide model building services (trains/architectural). If you're up for a career change, perhaps look into that?

[lgrfbs]

One I'll think about, is at larger layouts, you have the supply of "new" electricity to the rails at regular intervals, this to remedy the powerloss in the rails at the layout.
So if one were to measure a reflection on a power district, then surely the whole zone would be seen as an entrance and no reflection would occur.
Do I think right here or would it look as if you threw a whole fist with stone in the lake there will be ripples on the water for every stone that breaks the water surface.
Is that what it will look like for the electronics if you have several equivalent connections to the rails?

[rrinker]

  In the case of multiple feeds to the same section, I think it would be more like throwing a handful of peddles in the lake. Perhaps there is affordable equipment good enough that if there was a deivce at every feed location, or at least some of them, it could in a way triangulate. On a really large layout there will be multipel power sources as well, with insulated gaps between sections on different power sources. Now, if each section were trivial enough, one of these devices in each section could work for that section and not interfere with the one in the next section.