marklin-users.net community | Forum
»
General topics
»
Digital
»
How many reed switches for train position feed back
Joined: 09/11/2007(UTC) Posts: 116 Location: Melbourne, Australia
|
Hi
I have a blockage in my understanding of one principal in feeding back train position to the pc. Hopefully more experienced members can explain this process to me thanks. I am just about to embark on using software. I currently use Ecos but dont use the layout mapping so no feedback yet and no seperated sections of the layout. I was always going to use IR and or reeds for software control.
Whilst I understand when using a reed switch to send a (1) signal to the software that the train just passed over the position - then what? How does the software know the train is now in another position such as 500mm further down the track?
As I understand things - the reed is just in one spot - so without putting reeds every few cm's or inches - how does the software track it "continuously"?
(I am wondering if 2 reeds are used as in the old block system - where reed one signals the train has entered the "beginning" of say a 1 metre block and lights up the train icon in that block until the train triggers the second reed switch ?
I would really appreciate some insight into this subject and the principals employed. I could use IR across track methods or isolate 50mm of track by cutting one outside rail in the same way - but its still a short signal to the software and then the train has moved on.
Rgds
Greg
|
|
|
|
Joined: 09/11/2013(UTC) Posts: 336 Location: Toronto, Canada
|
I would like to understand this as well. My assumption was that the layout should be divided into blocks and a central station will be able to recognize which blocks have a lok on them based on current draw (i.e. no need for reed switches)? As it leaves one block, that block should then show as free, and then next should report the presence of the lok. Would love to hear if this is accurate and if there is a benefit to either method. |
My Layout Build | Märklin CS3+ | K-track | Merkur | Viessmann | LDT | iTrain | Modeling primarily DB EpIV-VI
|
|
|
|
Joined: 15/10/2006(UTC) Posts: 2,319 Location: Washington, Pacific Northwest
|
I don't have any experience (yet) with the digital and feedback train layout specifics, and am only speaking from the point of view of an electrical and computer engineer.
Reed switches are effectively 'contact switches' in the feedback system - a part with a magnet passes over them and they briefly close while the magnet passes. This change in switch state is detectable by the feedback system - so it knows something with a magnet just passed this switch.
The control software must knowwhere the switch is, in order to make some reasoning based on that feedback signal.
Electrically isolated track sections are not needed for this, as the reed switch feedback is independent of the track and power.
If you electrically isolate some track parts you can just as well detect that some rail vehicles are in that section - the benefit here is that you will know if anything, be it locomotive or car, is in the section, based on how you set things up. If current draw is used, only the loco would possibly be detected, or current drawing lighted cars. If the conductivity between the two ground rails is detected, any AC (ie conductive axles) cars are detectable. So you can determine if a train has fully left a section to know its clear, not just that the locomotive has passed a given point.
For software control, the ideal situation is to know where exactly a locomotive and all of its cars are located on the layout. You can do that visually but by sensors that is very expensive and complex.
In theory, in the world of digital decoders on locomotives we could get a locomotive broadcasting its presence back to a control system via an electrically isolated section, so if a control system detects there is a locomotive in an isolated block it could issue an identify yourself request, to which the locomotive responds - thus helping the control system conform where the locomotive has just entered. But that cannot be used to give you exact location within a block (without great expense and circuitry). Whether the Märklin system supports such query-feedback I don't know.
So for control systems to work with the simple feedback we have, you as the software control writer make some assumptions. You have routed a train to a particular section of the layout via setting switches and commanding the locomotive to drive. When the reed or load feedback detects a presence, you assume that the thing that just entered the block is indeed the locomotive you are trying to get to go there. Since you are in theory the only control system for the layout, the chances are very high your assumption is true. By creating rail sections that are of the right length with respect to train lengths, you can know that if a detection takes place at some point, that the head or all of the train must be inside a section, and can set the section it came from to free.
Its all a trade of using the simplest reliable presence detection combined with a layout design that allows making assumptions almost certainly to be true, and using that to turn signal states. One can use signal states to power the track just beyond them, so that a red signal will always have a depowered rail segment past it which would cause red-light runners to come to a stop. Thats added safety, and mimics some real world safety systems results. But thats just supplemental.
There isn't yet the ability to put programmed controllers INTO the locomotives such that the locomotive can detect when it gets to a certain point (by an RFID chip it can detect) and thus stop itself. I can see that coming down the road, but I don't think we are there yet. For now, a processor off track manages it all. |
|
 1 user liked this useful post by Minok
|
|
|
Joined: 04/04/2013(UTC) Posts: 1,291 Location: Port Moody, BC
|
Since in your original post you say you want to provide this feedback to a PC, I am assuming you are planning on using some form of model railroad software such as Rocrail or other commercially available packages. Therefore, if that is the case it really depends on what you want to do. In Rocrail, for example, you can have multiple reed switches or other feedback sensors in a "block". Preferred is 2 per block where the first one is "Enter" and the second one is "In". You can get away with one (in a short block) where it becomes an "Enter to In" function that the computer uses to calculate the time for what you want it to do. Obviously you can add more within a block to provide further feedback but it does become expensive.
The problem that I have with my equipment and layout is that you cannot place the magnet at the same place on every locomotive. Therefore, if you want consistent feedback for every locomotive, to stop exactly at a signal for example, the reed switches and magnets most likely will not work the way you want them to. In that case the conductivity method as mentioned by Minok would be more appropriate since any AC axle will trigger the feedback. Therefore, in an "In" stop situation the first axle will cause the train to stop.
So it really depends on where on the layout and what function you want to do there as to what type of sensor may be most appropriate. I have a mixture of reed switches, contact tracks, and conductivity sections and all have their individual pros and cons.
Cheers Peter
|
|
|
|
Joined: 03/08/2011(UTC) Posts: 1,571
|
Hi,
To answer your specific question;
By telling the PC software on startup that train A is in block 1 and heading in direction X it will then keep track of the train. I.e when block 2 is being occupied it will assume this is train A and so forth.
The more blocks and the more feedback you have the stronger system you have. However if you have a faulty turnout and there is a mitchmatch between reality and what the computer knows, you will of course head towards trouble and potential collision.... Just like in the real world.
Anyway this is the fundamental difference between CS and PC software. A CS will never know that train A has just passed block 2, it will just initiate the action every time something pass the feedback sensor in block 2.
Railcom+ is a bit different. Here the decoder for individual locomotives will feedback their ID to the track and sensors placed along the track and will then be able to identify what train has actually passed the sensor. It's probably the safest system, but the sensors does require more wiring than e.g. a simple contact track with 1 wire running to a S88 feedback module and not all systems support Railcom.
It would have been nice if Marklin had designed something similar e.g that a signal could functioning as a ID listener and report back to CS. Maybe in the future, as they seems to be heading towards more and more layout automation with the CS.
Brgds Lasse |
Digital 11m2 layout / C (M&K) tracks / Era IV / CS3 60226 / Train Controller Gold 9 with 4D sound. Mainly Danish and German Locomotives. |
 1 user liked this useful post by Danlake
|
|
|
Joined: 09/11/2007(UTC) Posts: 116 Location: Melbourne, Australia
|
Thanks all for your comments. My attraction to digital was always forever progressing technology and simplicity - such as using 'virtual blocks' and set on the pc software. I think reading the advice - that the use of reed switches for virtual block entry and current draw sections of a few inches for exit and stopping will be best. FOr long virtual blocks I can always add more reeds or IR sensors (that are triggered for the whole length of a train). The comment on Railcom is important perhaps - I have ESU Ecos which is railcom enabled - but need loco decoders which are also railcom enabled - so does anyone know if MFX decoders are railcom enabled as 95% of all my running stock is MFX driven? (BTW - I am studying 'Games on Track" system in Denmark - transmitters on/in each loco broadcasting loco position and name every inch of the way via a ceiling receiver/satellite and feeding the signals to a PC or command station including Ecos and CS and others or a combination of both) In this system the work instead of track blcoks, sensors and decoders is wiring each transmitter to the + & - leads anywhere on the loco. See:- www.gamesontrack.comI am keen to know if MFX decoders are railcom compliant please. Regards Greg
|
 1 user liked this useful post by Gregzim
|
|
|
Joined: 30/08/2002(UTC) Posts: 1,289
|
Originally Posted by: Gregzim  Hi
I have a blockage in my understanding of one principal in feeding back train position to the pc. Hopefully more experienced members can explain this process to me thanks. I am just about to embark on using software. I currently use Ecos but dont use the layout mapping so no feedback yet and no seperated sections of the layout. I was always going to use IR and or reeds for software control.
Whilst I understand when using a reed switch to send a (1) signal to the software that the train just passed over the position - then what? How does the software know the train is now in another position such as 500mm further down the track?
As I understand things - the reed is just in one spot - so without putting reeds every few cm's or inches - how does the software track it "continuously"?
(I am wondering if 2 reeds are used as in the old block system - where reed one signals the train has entered the "beginning" of say a 1 metre block and lights up the train icon in that block until the train triggers the second reed switch ?
I would really appreciate some insight into this subject and the principals employed. I could use IR across track methods or isolate 50mm of track by cutting one outside rail in the same way - but its still a short signal to the software and then the train has moved on.
Rgds
Greg Hi Greg! Are you using M, K or C tracks?  Peter |
|
|
|
|
Joined: 15/10/2006(UTC) Posts: 2,319 Location: Washington, Pacific Northwest
|
Originally Posted by: Danlake  Hi, It would have been nice if Marklin had designed something similar e.g that a signal could functioning as a ID listener and report back to CS. Maybe in the future, as they seems to be heading towards more and more layout automation with the CS.
Yes, that would. I can envision RFID chips on each loco and car, the size of grains of sand we now have and used in 'chipping' pets with IDs. They are passive and require no power thus. Then you can mount a reader in track side signals and other infrastructure, that is powered, and it interrogates the items that pass it to get IDs, which it can then pass back by wire to a PC controller. Thats how real rail systems also work - they bolt bigger RFID systems to the cars/containers, and can get train readouts as they pass a sensor at the yards. The only trick is managing the power of the transceivers to make sure they ONLY read the RFID tags on the one side they are facing, and not 2 or 3 tracks over, but directly in front... and that may be the problem on H0, the scale.. and why such a system isn't feasible. |
|
 1 user liked this useful post by Minok
|
|
|
Joined: 03/08/2011(UTC) Posts: 1,571
|
Originally Posted by: Gregzim  Thanks all for your comments.
I am keen to know if MFX decoders are railcom compliant please.
Regards
Greg Hi Greg, Railcom is designed for dcc protocols only. Railcom + is the newest system which allows automatically registration of the decoder (similar to the mFx protocol). MFx decoders are not multi protocol. Some Esu decoders are (can be used on both dcc and mfx). See also link: http://www.locgeek.com/2...com-plus-what-are-these/Reed switches are getting mixed reviews. The disadvantages is of course that it will only detect the loco or car that has an magnet while contact tracks or IF sensor will detect as long as something is physically on the sensor. Some users also claim it to be less reliable. If you have C or K track I would recommend making up your own contact tracks by isolating one of the outer rails. Brgds Lasse |
Digital 11m2 layout / C (M&K) tracks / Era IV / CS3 60226 / Train Controller Gold 9 with 4D sound. Mainly Danish and German Locomotives. |
|
|
|
Joined: 09/11/2007(UTC) Posts: 116 Location: Melbourne, Australia
|
Thanks for all the comments.
I have C track so will use the outside rail and virtual blocks with a slow/stop feed back at each end and pc software as all the differnet systems, protocols, products and bus's can all be by passed by simply switching to computer control. That way it makes no difference what I run or what system or with what decoders and I can see the image of each loco moving around on the screen.
Phew - at least thats that decision done :) (ALL I have to do now is spend weeks learning it !!)
Cheers
Greg
|
 1 user liked this useful post by Gregzim
|
|
|
Joined: 10/02/2006(UTC) Posts: 3,997
|
|
|
|
|
|
marklin-users.net community | Forum
»
General topics
»
Digital
»
How many reed switches for train position feed back
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.