G Scale Model Train Forum banner
Cancel
Create thread New Forums

Arail development project

Jump to Latest
1 - 20 of 53 Posts

Brandon

· Registered
Joined
·
461 Posts
Discussion starter · #1 ·
Abstract: Arail (Yet to be decided name but "Another Railway", "Automated Railway" are some possible full names) is a development project underway by myself (and possibly others in the future) to create a new method of controlling a railway that combines different features than what have been traditionally offered in the hobby. The end goal is a railway that can fully run itself by the press of a single toggle button (to start and stop including bringing trains out of sheds and parking them inside as well) with other possible options of adding custom control of a railway by taking control of a one or more consist(s) to using predefined time tables and using real-time sensor data to know what is going on around the railway and for the software to react intelligently (eventually sophisticated AI) to events.

Mentality: The more prototypical a railroad is the more interesting it would be to some modelers. Prototypical railroads often operate quite differently (in terms of hardware) than what a sophisticated model railway does but with recent advances in micro computers it is now possible to have railway that can know down to an eighth of an inch where every train is at every moment and for it to think for itself and make choices on its own which can allow someone to do anything from run the entire railway by a push of a button to allowing people to become an engineer and be given all their orders by Central Control for what their train should do for the next hour, day, week or month and for modelers having to possibly contend with other computer controlled locomotives on the rails that could be higher or lower priority on the lines.


Goals: KISS (keep it simply stupid) is primarily the first goal. A well built API will allow countless different methods to control a railway. Open Standards and Software will allow anyone to get involved to help make this better for everyone.


Why not use XXXX: There are a lot of different software and hardware options in the hobby but they are primarily non-prototypical tools to simulate some prototypical behavior and have many limitations. If we take DCC and JMRI as an example, DCC was a great idea but it's 30 years old and is often slow to accept change which can sometimes be more in favor of what people making money want rather than what an individual modeler wants. Using DCC over wireless connections is limited and provides no method of sending location data back in a wireless controlled railway. DCC also lacks any method of locating a train to millimeters without very expensive equipment that needs regular calibration which takes fun out of being able to 'just run a train'. JMRI was built around dcc to help make some functions of modeling easier but as Arail is to take a whole new approach to a railway, a new foundation that's not surrounded with 30+ years fixes and patches it is a more optimal choice to start fresh (however Arail-Loco would support operation from a JMRI system).


Software in Arail: Software will consist of a software "client" (Arail-Loco) that runs on a mini computer of each locomotive (or a trailing car) and other Arail-Server applications that send commants to a locomotive.


Arail-Client:

Arail-Loco: This client software waits for a connection from a controlling piece of software and controls the physical hardware in the locomotive. XML can be sent containing more basic commands like: or 1 to other TBD instructions of how to reduce speed based on more realistic details such as slope at the current section of track and the number or types of cars (loaded vs unloaded coal cars) and so on. This will allow someone to use either a phone, tablet or JMRI to control a single train and also for more sophisticated server software (Arail-AIControl) to manage an entire railway if desired.



Arail-Servers:
Arail-ControlTest: This will be Arail-Test which will be a simple server that sends commands to a locomotiveHardware to run a short test of all available features of the locomotive or consist.

Arail-ControlBasic: This will be a non AI server controller that will run all desired trains on a railway and keep them from colliding but will not support any AI or more advanced layout functions (This will be used to create a server that exposes design flaws while allowing for me to enjoy the railway until I can develop a very complex server to do amazing things). This Arail server will work for the majority of modelers who 'just want to see their trains run' and want multiple trains running themselves at once including sharing mainlines, going different directions, and so on but aren't as interested in very realistic operation, time schedules, or don't want to spend any time "configuring" software other than measuring the distance from a sensor (ie: rfid sensor) to the front of the train and to the back of the train. In this server you can also take control of a locomotive and watch all the other trains either avoid your train or require you to avoid others depending on what operation mode you're in (obviously Arail-ControlBasic will make it hopefully impossible for any track to collide unless you _try_ to rundown another slower (ax in max speed) train that's fleeing to a siding or another line)

Arail-ControlAI: This likely be a full re-write of Arail-AutoBasic that you can configure things like "coal loading plant at X position in the track", configure what cars are in your consist so it simulates loading all coal cars, where passenger terminals are and how often you want trains to stop at them, and so on and press "go" and watch the railway learn how to possibly operate your particular railway to whatever goal is. Imagine deciding what trains you want to run that day, what time schedules you know are tough to make, and then adding a massive through consist of a coal load at the busiest "time of the day". I should note that 100% realism isn't the goal or even a realistic possibility but working towards it is a new area modelers haven't been able to do with multiple moving trains and one person before.



Software:
Licensing: First, the code will be OSS (Open Source Software) and released under the GPL license (GPL licensing basically makes it so everyone has a right to get it,, look at, modify and be able to talk about it, redistribute any changes they make and no company or person can take something that others wrote, in whole or part, and withhold anything they might do from everyone else either).

Language: At this point the server will be written in Perl as it can be easily changed and run on different platforms. There's a slight chance Arail-ControlAI could be C or C++ but unless performance becomes an issue I think perl/apache/mysql would provide for the fastest development and for others who can handle learning a little programming to maybe learn a new skill and do some specific things they like.
Data communication: XML/soap'ish

Hardware: Currently the first goal is to use a micro computer (simulating someone listening for commands and operating the controls on a locomotive), a wireless connection (as this is what prototypical railways use for communication), and a RFID reader (simulating the eyes of the engineer reading mile posts and knowing where he's at). IR, block detection, reed switches, and other sensors will likely not be supported because if a cheap, reliable, and simple method can be found to do everything we need, then by following the KISS mantra, adding support for other complex sensors that offer less data wouldn't make supporting and maintaining the project as easy or fun for us as we'd rather be outside running the railroad than inside sitting at a computer writing code.


Location Hardware: I've purchased several $10-$15RFID readers from ebay and actually have had better than expected results so I have no douvt RFID will be the key choice for tracking consists.

Mini Computer: Raspberry-pi is currently the best choice. For $25 you get a 700mhz 128MB computer with 2 usb slots and it uses about 1Watt of power. If it has the ability to run a full OS and Desktop + DVD videos it's more than enough for what we'd need on board a loco (or trailing car)

Motor Control: TBD Maybe Gertboard, Teensy, or a USB servo controller and a car or boat ESC? (This would provide a 5VBEC we'd need to run the raspberry-pi but I've not found 5 cell ESC's for trains that are under $50. Thoughts?

Lighting: Gertboard, pic, usb controller+servo relay? Or something else, Thoughts?
Smoke: Gertboard, pic, usb controller+servo relay? Or something else. Thoughts?

Sound: The raspberry-pi has stereo out. Since trains are technically mono sound you could run two different types of locomotives in a single consist and get two different train sounds.


Well, that's it for now...
 
Discussion starter · #2 ·
Arail-Loco XML...

Arail-Loco will allow you to either send an xml config to it when an Arail-Control connects to it (in case Arail-Loco is running in a trailing car and you connect it to different locomotives) or if you always use a particular Arail-Loco client with single engine it will in this case just operate as the last loaded XML file it got.

I'm working on the Arail-Loco client and would like to get input on what information people think is needed, wanted, or fits into a 'wouldn't ever effect anything but it would be fun to have kept with the locomotive/consist".

Here is what I have come up with so far. Please, please, please critique this. Just make sure you comment if it's "needed" "wanted" or "would be nice to have the information "saved" with the locomotive but it's purely for logging/data record and wouldn't affect operation in any way". Everything I have other than the and are need or wanted (Imagine Aral-ControlAI knowing which locomotives have people in them so when it puts consists together it knows to put the train with people in it at the front of the consist or send the unit to "staging" to have someone put a figure in it. I know this is a 'out there' idea but it's an idea that may spark an idea in someone else that could be really nice.

Code: 
              SD-45     SD-45             UP     1024     Test Loco     4     1     1     0                   20      1000      .3      3      1                       pin1       4       0                   pin2       4       0                   pin3       4       0                   pin4       4       0                   pin5       4       0                             audioLeft                             pin6                        pin7
 
#3 ·
I'm looking forwards to what you will come up with.

I can't say I agree with all of your reasoning, like not using DCC because it's 30 years old, and then you program in Perl, invented in 1987.... ha ha...

And there are ways to do bidirectional communication wirelessly...

And there are reasons that track power is used, because it eliminates the wireless hardware, saves money, and eliminates rf interference...

The main reason I responded is this paragraph:

Mentality: The more prototypical a railroad is the more interesting it would be to some modelers. Prototypical railroads often operate quite differently (in terms of hardware) than what a sophisticated model railway does but with recent advances in micro computers it is now possible to have railway that can know down to an eighth of an inch where every train is at every moment and for it to think for itself and make choices on its own which can allow someone to do anything from run the entire railway by a push of a button to allowing people to become an engineer and be given all their orders by Central Control for what their train should do for the next hour, day, week or month and for modelers having to possibly contend with other computer controlled locomotives on the rails that could be higher or lower priority on the lines.

In the first sentences you seem to embrace prototypical behavior... but then you talk about a train thinking for itself and making it's own choices... definitely not prototypical.

I kind of got lost after that because your second sentence has 134 words!

But, let's see what you come up with. Please keep us abreast of your developments, especially any low cost hardware that allows positioning to 1/8" and works wirelessly.

I'd love a system with the capability you describe (even though I will keep track power)...

Greg
 
Discussion starter · #4 ·
Like all people I have my own views on things, and they could be wrong or far off but in the world of discovery and science sometimes wrong ideas and good ideas turn out the opposite but you never know unless you try and follow it through. With this I'm sure things I may consider to be wrong or bad wouldn't be seen the same way by others and that's fine and vice verse, but don't shoot us down before we have a chance to stretch our winds and see what happens. :) I'm also not looking to have Arail ever get into 'this is better than that' by any means at all, I'm just trying to come up with a way to run a fully automated railway the way I've thought of that might work well.
DCC is great for many things but based on my own personal thoughts I feel DCC lacks some features that are critical (in my opinion which could be wrong or not the same as what others think) and I've come up with some other hardware options that I think would work better for what I'd like to do in my own yard.

If you're curious about why I'm not a fan of DCC I can go over those, but obviously that would be off topic and not appropriate to get into a discussion on here as this thread is about the development of yet another way to do railway control.

To note, you could run track power and DCC with Arail but that is a layer of complexity that isn't at the top of my todo list or part of what I believe would follow the KISS mantra. And of course no software is perfect so there will always be cut corners where needed until more time can be put into development or problem solving of ideas. I could note that I plan to have others use track power and use their locomotives while mine would run battery and all I'd need to do is put a car with the tracking hardware in the visiting locomotive and the software would know where it's at all the time and would route my trains around that visiting train. Or someone could still use DCC on their own layout with Arail and if they wanted to write modify the software they could also use Arail-ControlAI to control DCC trains but you'd still need a way of getting accurate location data back to Arail-ControlAI but that's a topic that would be better discussed once the basics of Arail are done and working. I'd love to design and build everything at once but this is a hobby for enjoyment and thus I can only do so much at a time. I'm writing this as GPL so someone like yourself who enjoys DCC could take everything I do and cut out the pieces I feel are important and add your own parts to do it as DCC and create a who new project called DCCArail which if it were better adopted than what I do you could pretty much wipe Arail off the map but I still could do what I wanted with what I wrote and no one could stop me from enjoying my railway the way I wanted. I hope this shows I'm not against any technology and I believe in letting all technologies have their chance, but I hope this project doesn't get harassed for being 'different'.



P.S. I'm not the greatest at English, its usage, or readability. I do my best and hope those who are interested in what I'm doing don't get too frustrated with my writing. I apologize in advance as it's not that of an English scholar.
P.S.S. Greg, it appears you have PM's turned off, if you're interested in receiving responses on the items you brought up please send me a PM on how I can reply outside of this thread.
 
#5 ·
Note that I was not espousing DCC, just saying your reasons to exclude it don't have a lot of "bite" so to speak.

Yes, if you look at my signature, you will see in BOLD red letters that I do indeed have my private messages off and to use regular email... can't make it more obvious without breaking forum rules for animated gifs ha ha!

Nothing wrong with your English, just very long sentences should normally be broken into shorter ones. It conveys the thought better and keeps the point centered.

The interesting part is where the ideas translate into hardware that makes sense and is available. Will be looking forwards to that. In the short term, you might very well adopt some DCC hardware to facilitate your higher level control system.

Regards, Greg
 
#6 ·
I've seen similar developments in the smaller scales, and VERY similar arguments against "old" technology like DCC. My thoughts are the same here. You need to (in my humble opinion) consider the user interface and automated control systems as separate from the actual train control technology. A properly designed system ought to be modular, and work with what people already have, rather than adding yet another incompatible control system into the mix. With DCC and JMRI, perhaps with a new form of train detection (not necessarily DCC) you could do much the same job, and not have to design a new system from scratch.

Remember that that 30+ year old technology has had 30+ years to work the bugs out. And it's still not perfect. Really think you can do better out of the gate?

I also respectfully submit that you do NOT need to know where every train is to within 1/8" at all times. Even the prototype doesn't know that. All you need is to know whether a train is in a given segment, and when it is approaching certain fixed locations. That makes the problem a good bit easier to solve.
 
Save
Reply Quote Like Like
#7 ·
Brandon

OK so I got lost in your technical jargon, but was intrigued with the hopes of using computing power for control and accurate knowledge of where things are. So here are some ideas that you may have the knowledge to develop. Model railroading has long been fascinating because the animation the moving trains give a model scene. It was possible all these years because of the low tech, simple KISS principle, the rails keep the flanged wheels on the track. What has been lacking is the next level of animation needed. The realistic animation of road traffic including cars, trucks, horse drawn vehicles, and their stopping, starting, independent turning, etc. Slot cars have been around along time but fall short on realism. I have seen battery powered vehicles following a buried magnet iron wire which look pretty good on a clean inside model railroad, but if some technology could be developed to control vehicles in our dirty garden railroad environment, it could be revolutionary. The next level of animation which would add alot would be the motion of living people and animals. Depiction of computer drawn beings in movies shows that we have the computer simulation down, what needs to be done next is to translate that to 3 dimensions and in miniature.

I'm not sure if your ideas could be applied to these needs, I'm just mentioning them as the next steps that could revolutionize model railroading.

The best of luck to you. Whatever you do I appreciate the amount of work involved and the technical expertise.

Terl
 
Discussion starter · #8 ·
Posted By Terl on 02 Jan 2012 02:32 AM
Brandon

OK so I got lost in your technical jargon, but was intrigued with the hopes of using computing power for control and accurate knowledge of where things are. So here are some ideas that you may have the knowledge to develop. Model railroading has long been fascinating because the animation the moving trains give a model scene. It was possible all these years because of the low tech, simple KISS principle, the rails keep the flanged wheels on the track. What has been lacking is the next level of animation needed. The realistic animation of road traffic including cars, trucks, horse drawn vehicles, and their stopping, starting, independent turning, etc. Slot cars have been around along time but fall short on realism. I have seen battery powered vehicles following a buried magnet iron wire which look pretty good on a clean inside model railroad, but if some technology could be developed to control vehicles in our dirty garden railroad environment, it could be revolutionary. The next level of animation which would add alot would be the motion of living people and animals. Depiction of computer drawn beings in movies shows that we have the computer simulation down, what needs to be done next is to translate that to 3 dimensions and in miniature.

I'm not sure if your ideas could be applied to these needs, I'm just mentioning them as the next steps that could revolutionize model railroading.

The best of luck to you. Whatever you do I appreciate the amount of work involved and the technical expertise.

Terl I've thought about this for years... Having other objects moving on their own, carriages, cars, people, etc, is not simple and each would most likely require different mechanisms to work so there isn't a solution for all except hiring kids to play with them while you operate the railroad. :) The magnet+wire idea is popular and there are other tricks. You should take a look at Miniatur Wunderland in Germany, they have lots of videos online and some talk about how they do things. That's by far the most realistic modeling I've seen that encompasses all types of movement including trains. The idea of using AI to move people in a city isn't that compllex to do in AI but making the objects to move is most complicated part and although what will be Arail-ControlAI uses the same principles it's not closely related to what this project would be. VictorSpear on here could probably point you in the direction of small robotics that can act autonomously (carriages, cars, people and bikers) moving and stopping and in fact a central control wouldn't be needed, just sensors on each device to make sure where they're wanting to go isn't another object. Powering those would be trickier too but there's no reason you couldn't have cars and bikes driving around on their own using battery power. Also if each device can make its own choices and doesn't need to communicate with a server the cost and complexity to develop the system goes down very quickly. I think you can buy robotic kids for under $50 each that are basically car parts with a few sensors to make sure it doesn't run into a wall. Kids toys are even getting to this point where you can turn them on and they drive on their own. I was in Hallmark during some Christmas shopping and even they had the kids cars that you can draw a line with a pen and place the car over it and the car follows the line, I think it was $10 or something. I imagine in a few years some toy maker will have a toy that can be modified to follow hidden wires under things and when it reaches another trigger sensor it stops and can make choices if it wants to go one direction or another. Of course if you want to get the kit robots you can do that now but it's not 'toy' cheap.

Oh, one other thing, I have thought about using projectors for background scenery. If you take a ride on some of Disney's attractions and know the fact that some things shown are multiple projects in series it shows what can be done. Indiana Jones does this quite a bit as do some of the other dark rides like Pooh and the old People Mover with the Tron sequence... Now days the thing is using high powered projectors and 3D mapping to give things like the castles and It's A Small World a completely different look and feeling and it takes into account how to shade edges that aren't straight on so that depth can be removed or added. I do believe though that this has less appeal for modeling as we want to see real objects and not projections that wouldn't work as well during the day.
 
#9 ·
I've got to chime in here on this a bit myself. As Greg said, DCC has had 30 years to "work the bugs out", & DCC decoders (often pre-installed in locomotives) are available "off-the-shelf". This isn't the case for most proprietary control systems - the only ones which have been at least reasonably successful (aside from DCC) in large scale are DCS, Aristo's "Revolution", & RCS. If you're talking producing a new standard, you're looking at high developmental costs for any sort of mass production.
Image


Also, a lot of what you're referring to (in terms of prototypical operation) aren't applicable in many model railroad situations, particularly in large scale. I did notice your referral to the closest "real world" model railroad to what you're describing, "Miniature Wunderland" layout in Germany, which has a computerized control center that could make NASA Mission Control jealous.
Image




Note especially that the Minatur Wunderland layout is HO scale, & occupies a HUGE
Image
building. (Let's not even go into what it
Image
must have cost $$$$$$$ to build!).

Most Large Scale layouts (my own included) are positively tiny be comparison; the trains, being far larger than HO, also require correspondingly larger areas for equivalent track plans. My own railroad has a mainline run of @ 160 feet, with a trackage total of @ 300 feet if I include all my passing sidings, yard, & industrial trackage. Most of the time, I'm running narrow-gauge prototypes, usually slow, geared logging locomotives like Shays (typical prototype operating speed for a Shay: 10 ~ 12 MPH.
Image
). Operating at that prototypically slow speed, it takes a Shay an average of @ 5 minutes to make a full lap around my mainline. I am considering some form of automated operation for when we have visitors, but it will be limited to having 2 or at most 3 slow-moving trains, at least one operating in an "opposing" direction, pass one another automatically via sidings on a single-track mainline.

Another major large-scale consideration is that many of the layouts (mine included) are outdoors.
Image
The greatest consideration is that any electronics installed outdoors have to be in weather-proof housings. As well as moisture-proofing concerns, the electronics are also subjected to temperature extremes
Image
Image
that are not encountered in indoor model railroad situations. Detection schemes such as infrared and optical are usually unusable
Image
outdoors due to (real) variable lighting conditions. This also makes doing some of the (admittedly impressive
Image
) types of animation such as Miniatur Wunderland does impractical from a reliability standpoint.

That last point - RELIABILITY (or lack thereof!
Image
) is a HUGE CONSIDERATION in any sort of model railroad automation. I have more than a bit of a clue on the subject, having formerly worked in the industrial electronics field (maintaining CNC machine tool equipment) - and the power levels large-scale locomotives require tend to lean more in that direction than HO equipment. Automating especially an outdoor large-scale railroad means dealing with many variables which are not encountered on an indoor model railroad; not only do we have to deal with rain, but also snow (Aristo-Craft sells a large-scale wedge snow plow which works reasonably well), leaves, dirt & animals
Image
as well. Just keeping track & roadbed maintained for reliable operation in that situation is enough of a job in itself; if they aren't reliable, adding automation to the mix will only make things worse. Just a bit of friendly, constructive criticism to let you know what you might be getting into.
Image
Tom
 
#11 ·
I'll start by picking on your choice of computer. I see a raspberry pi has 8 GPIO lines. I think this immediately invalidates your KISS plan. In a real product, there would be many more GPIOs, to say nothing of PWM, and interrupt inputs.
 
Discussion starter · #12 ·
Posted By Tom Lapointe on 18 Feb 2012 10:02 PM
I've got to chime in here on this a bit myself. As Greg said, DCC has had 30 years to "work the bugs out", & DCC decoders (often pre-installed in locomotives) are available "off-the-shelf". This isn't the case for most proprietary control systems - the only ones which have been at least reasonably successful (aside from DCC) in large scale are DCS, Aristo's "Revolution", & RCS. If you're talking producing a new standard, you're looking at high developmental costs for any sort of mass production.
Image


As embedded computers become less expensive while increasing in feature sets there is no need to develop any hardware as someone can use what's readily available. This means no proprietary hardware and functionality that is only limited by software. In the world of open development, hardware and software, there is no need for high development costs for any sort of mass production.


Also, a lot of what you're referring to (in terms of prototypical operation) aren't applicable in many model railroad situations, particularly in large scale. I did notice your referral to the closest "real world" model railroad to what you're describing, "Miniature Wunderland" layout in Germany, which has a computerized control center that could make NASA Mission Control jealous.
Image


Agreed, but getting 'closer' to real world does mean everything from using realist acceleration and braking performance based on locos real world specs and load to how communication works. For a technology person like myself, dcc track communication is a 'hack' imho and not how I would like my railroad to function (but there's no reason someone couldn't use what I develop to run a DCC setup), RF to me is a more 'realistic' approach and thus closer. However it is scale and won't ever be like full scale in all ways but a nice thing about being someone who's writing some or all of the code is I can add what I want to see. Also anyone who wants to add what their $.02 can download the code and do as they wish as well, I wouldn't stop them as that would slow down innovation. I've been part of OSS software development on dozens of projects for nearly 20 years, I never get tired of watching projects grow from what a group of people can do when it's a hobby.


Miniature Wunderland is amazing, if only they had a way to share what they've developed at a price that people could afford and they could support (having something that works is only half the product, writing documentation, keeping it updated, and having to not break other peoples stuff you can't test is the hard part). But as much as using any of their code would be, I can respect their choice to keep it in-house.




Note especially that the Minatur Wunderland layout is HO scale, & occupies a HUGE
Image
building. (Let's not even go into what it
Image
must have cost $$$$$$$ to build!).

Most Large Scale layouts (my own included) are positively tiny be comparison; the trains, being far larger than HO, also require correspondingly larger areas for equivalent track plans. My own railroad has a mainline run of @ 160 feet, with a trackage total of @ 300 feet if I include all my passing sidings, yard, & industrial trackage. Most of the time, I'm running narrow-gauge prototypes, usually slow, geared logging locomotives like Shays (typical prototype operating speed for a Shay: 10 ~ 12 MPH.
Image
). Operating at that prototypically slow speed, it takes a Shay an average of @ 5 minutes to make a full lap around my mainline. I am considering some form of automated operation for when we have visitors, but it will be limited to having 2 or at most 3 slow-moving trains, at least one operating in an "opposing" direction, pass one another automatically via sidings on a single-track mainline.

Another major large-scale consideration is that many of the layouts (mine included) are outdoors.
Image
The greatest consideration is that any electronics installed outdoors have to be in weather-proof housings. As well as moisture-proofing concerns, the electronics are also subjected to temperature extremes
Image
Image
that are not encountered in indoor model railroad situations. Detection schemes such as infrared and optical are usually unusable
Image
outdoors due to (real) variable lighting conditions. This also makes doing some of the (admittedly impressive
Image
) types of animation such as Miniatur Wunderland does impractical from a reliability standpoint.

That last point - RELIABILITY (or lack thereof!
Image
) is a HUGE CONSIDERATION in any sort of model railroad automation. I have more than a bit of a clue on the subject, having formerly worked in the industrial electronics field (maintaining CNC machine tool equipment) - and the power levels large-scale locomotives require tend to lean more in that direction than HO equipment. Automating especially an outdoor large-scale railroad means dealing with many variables which are not encountered on an indoor model railroad; not only do we have to deal with rain, but also snow (Aristo-Craft sells a large-scale wedge snow plow which works reasonably well), leaves, dirt & animals
Image
as well. Just keeping track & roadbed maintained for reliable operation in that situation is enough of a job in itself; if they aren't reliable, adding automation to the mix will only make things worse. Just a bit of friendly, constructive criticism to let you know what you might be getting into.
Image
Tom


Constructive criticism is always and highly appreciated. I do have an Aristo plow and plant o run during winter. Electronics outdoors are a huge concern, I can't tell you how many threads I've read about various failures from switches to ballasting that will affect the ability to run an automated layout. For me, I'm choosing to have no electronics outside of the train shed with the exception of signals one day. All switches will be pneumatic controlled but many will rely on typical solenoid but this gets abstracted to the software control. Due to potential quirks with ir, magnets and other sources I've elected to use RFID for tracking as some tags are waterproof (in glass) and will last for as long as the track (30+ years). In my design, a loco or a car somewhere in the consist would need enough room to fit a credit card sized computer, an extension board for controlling the loco (probably the size of an altoid can as well) and the RFID electronics (size of an altoid can as well). In G scale this is very doable in most locos and cars. The main constraint will be a battery which if you're using for loco power would be very large, otherwise a 2cell lipo would provide countless hours of use between recharges. This would allow as well for a loco to be track powered, with the "sensor car" reading and the track providing loco power. This is the simplest yet most versatile way of doing an automated railway I've come up with yet, and it could still evolve.


astrayelmgod: It's not simple as a functional goal but the hardware and hopefully software architecture (not number of lines of code) I'm hoping to keep as simple as possible. As for GPIO's, take a look at the "gertboard". It provides about 64 GPIO's, pwm, 2 motor drivers (the L6203 IC that can do up to 48V @ 5A and 4A RMS). Gertboart will be $10-$20 or so. And for people that use an RCS esc, they could by a $20 usb servo controller and use RCS's electronics to run a train as well. The software for automation won't care what you're using for driving various things except on setup you might need to specify it's an RCS receiver and what pins are connected to what outputs (lights, smoke, etc) so it turns the right things on.


I'll also note that I do plan on running consists of 3-5 SD-45's which obviously will be over 5A but I plan to have 8+ trailing cars for electronics and some will be for up to gertboard max of 5A and 1 or 2 trailing cars will use gertboard GPIO's to control additional IC's that can handle over 15A for those longer and more substantial load requirements. I'll also modify my SD-45's at some point to have everything inside, but only after I find a good way of accessing batteries in a SD-45 without a screw driver being involved.


Thanks for all the info, it does help me to second think choices and see what everyone things of my proposed ideas.
 
Discussion starter · #13 ·
Grrrr. mls put my comments to Lapointe inside of his quote box, so there's more there than just my reply to astrayelmgod. I should also note that gertboard is a product that's being designed to specifically increase the number of GPIO's by an individual who's closely associated with the R-pi project.
 
#14 ·
Posted By Brandon on 19 Feb 2012 09:16 AM
Grrrr. mls put my comments to Lapointe inside of his quote box, so there's more there than just my reply to astrayelmgod. {snip...}[/i] Brandon

I'm really not trying to be critical, but since that is where you typed your text (i.e. within the quote) just where did you expect it to be displayed?
Image
 
#15 ·
All I could find on the Gertboard were press releases. It isn't a product yet, so "$10-20" is at best a SWAG. Be that as it may, you still have a CPU board, an IO board, an external memory card for program storage (as there is none on the CPU board (SDRAM doesn't count)), some mechanical method for tying all that together, and a power supply. Suddenly, it isn't $25; it's probably three times that, at least. Plus, it is quite large. Yes, large. I see you favor SD-45s which is much larger than anything I can run on my layout, but even then you are talking about using trailing cars for electronics. Fine, but that needs some way to connect together, which all drives the costs up even more. And an awful lot of GRR'ers won't be able to use it because it just won't fit in their (small) steam locos. A solution that would have a wider potential market would have all the CPU, memory, IO, power control, and interface on one board the size of a business card. One board. One.
 
#16 ·
OK, another point; you're talking 800 feet of feet of track (more than double what I've got, nice if you have the space available.
Image
). You're talking running loco consists of 3 ~ 5 SD-45's
Image
- I would expect a single SD-45 to draw at least 2 (probably more like 5) amps by itself, if you're pulling a prototypically long enough train (I'd expect to see consists of at least 100 cars
Image
in that case!). The closest train I own to what you're talking is my USA Trains' New Haven "Merchants Limited" streamliner
Image
- which has a pair of USA Train's Alco PA's (comparable in size to the SD-45's; twin-motored, 6-axle locos.
Image
The train itself is only 5 cars long; but they are 5 very heavy aluminum passenger cars, weighing in @ 20 pounds apiece.
Image
Since our property is partially on a hillside, my mainline has a ruling grade of 3%; those PA's struggle
Image
to lug that 100 pound consist upgrade. (Note that in the photos below, I'm showing a 4 rather than 5 car consist; I've had trouble with stripping axle gears
Image
in the PA's due to the train's weight, so often omit one coach to make life a bit easier for them):

Image


Image


Image


Even running the 5 car consist, running on straight analog DC track power, I measured the total current consumption of the entire train (including the full lighted passenger cars) at 7.5 amps. (This was prior to installing DCC decoders - Digitrax DG-583S's in each PA, & a Phoenix P5 sound card in the lead PA).


Another comment you made I found somewhat flabbergasting - you say your system will require 8-plus
Image
trailing cars for electronics?!?!
Image
On my USA streamliner, all electronics are totally enclosed in the Alco PA's. The lead PA has the most in it, with the entire Phoenix sound system (the P5 sound card, heavy-duty 3" speaker with a large magnet, volume control up / down switch, and serial programming jack) totally enclosed within just the fuel tank. (Making the sound system an entire easily removable module). On both PA's the Digitrax decoders are installed centered on the loco chassis' floor, with huge amounts of room to spare. Sorry to "bust your bubble"
Image
, but Arail sounds like a giant step backwards.
Image


Image
Tom
 
Discussion starter · #17 ·
Posted By astrayelmgod on 19 Feb 2012 10:16 AM
All I could find on the Gertboard were press releases. It isn't a product yet, so "$10-20" is at best a SWAG. Be that as it may, you still have a CPU board, an IO board, an external memory card for program storage (as there is none on the CPU board (SDRAM doesn't count)), some mechanical method for tying all that together, and a power supply. Suddenly, it isn't $25; it's probably three times that, at least. Plus, it is quite large. Yes, large. I see you favor SD-45s which is much larger than anything I can run on my layout, but even then you are talking about using trailing cars for electronics. Fine, but that needs some way to connect together, which all drives the costs up even more. And an awful lot of GRR'ers won't be able to use it because it just won't fit in their (small) steam locos. A solution that would have a wider potential market would have all the CPU, memory, IO, power control, and interface on one board the size of a business card. One board. One.
Yes, $10-$20 is projected cost, but these guys know their IC's and costs quite well. For 'tying it all together' a $2 usb hub will do that (unless you're just using the r-pi for location tracking (no power control) then the 2 usb ports on the r-pi are all you need, no usb hub. Power supply is a sub $5 BEC used for model airplanes/cars/boats (since I plan to use 4-5 cell LIPO's, this works very well). Last I checked a 4AH lipo was about $40 and being an electric flyer for many years, I've seen the price drops about 30% per year. My goal was to have a solution sub $100 per consist. I have yet to decide how I'll wire things up, but I do know I'll be simulating MU and airbrake connections for use to control separate motive power, individual lighting control, smoke, and sound between a trailing car and loco(s) and daisy chain it through them. I won't personally be using the current connector (forget the type) used by Aristocraft because it simply isn't a large enough gauge wire to handle 10+ amps over it's 18 gauge wire without potential problems, so I'll re-wire most all my locomotives except my RS-3's and some fa-1's that only have one headlight and I'm not a huge fan of smoke so most FA1's won't get support for smoke. BTW, from my tests of my locos, fa1's use .6-.8amps (no load, and I have about a dozen fa1's from REA to present day aristo versions) and .7 to 1.1amps (16 car load, mixed car type and axle type) for motor control (no need to add lights or smoke to this number as these will be run off the l6203 motor IC). SD-45's/-9's use 1.3A-1.7A no load and 1.9A-2.4A for a 23 car load. Thus my _theory_ is the gert would be usable by many people but of course not all.

I also am aware that what I'm doing won't work for many large scalers but you do have to remember that my #1 goal isn't to create something for everyone or replace anything on the market but to create a way that I can personally run my railroad how I want to and this is my proposed method, mainly because nothing on the market allows location tracking down to the accuracy I want, accuracy far beyond what most modelers would every want in a garden railroad. The biggest limitations I saw in large scale is software and hardware and everyone is at the mercy of the DCC organization or another companies proprietary device. No company gives modelers the ability to modify the software in devices to add features they want. Hardware in DCC and even the revolution are very limited when compared to a full computer with a usb port. I've worked in the embedded linux field for many years and unless you're selling millions of units it's almost always cheaper and faster to throw more hardware at the problem than work in specialized SDK's that are less robust than commonly used environments. If aristo had a usb port for communication with off the shelf devices and drivers and they opened their software for others to expand I'd not be doing this project. Also if I were a electronics guru I'd consider making my own DCC decoder that did support RFID or other location tracking but I don't, I have to rely on the skills I do have and that I enjoy using since this is a hobby.

Even if no one ever used what I did I still think sharing what I am doing here might help someone else on a related type of project. I also think that over the next 100 years model trains will need to evolve or they will become a thing of the past. I really like the idea of watching a coal train move every minute or so to simulate a full train being loaded with coal (and maybe one day I will make a working loader as well) and trains stopping at stations based on time schedules. I also like the challenge of making everything work so I don't have to touch a control and I can see it all work as I had spent years building and tinkering with things until I can do that one day. :)
 
Discussion starter · #18 ·
Posted By Tom Lapointe on 19 Feb 2012 11:28 AM


Even running the 5 car consist, running on straight analog DC track power, I measured the total current consumption of the entire train (including the full lighted passenger cars) at 7.5 amps. (This was prior to installing DCC decoders - Digitrax DG-583S's in each PA, & a Phoenix P5 sound card in the lead PA).


Another comment you made I found somewhat flabbergasting - you say your system will require 8-plus
Image
trailing cars for electronics?!?!
Image
On my USA streamliner, all electronics are totally enclosed in the Alco PA's. The lead PA has the most in it, with the entire Phoenix sound system (the P5 sound card, heavy-duty 3" speaker with a large magnet, volume control up / down switch, and serial programming jack) totally enclosed within just the fuel tank. (Making the sound system an entire easily removable module). On both PA's the Digitrax decoders are installed centered on the loco chassis' floor, with huge amounts of room to spare. Sorry to "bust your bubble"
Image
, but Arail sounds like a giant step backwards.
Image


Image
Tom




Tom, I went over the current usage of locos in my last post reply, but bascially it's lower than you're thinking (not by much) and the motor controller is on a separate circuit than the lights and since the lights will be a relay to a 5v 20+A BEC I doubt I'd ever max that for lighting, smoke, or other accessories. :) But this also requires you separate out the lighting circuit(s) and smoke circuits from the motor circuits on a loco. I'm guessing I'll have a 2 wire connection for the motors a 6-12 pin connection for lights, smoke, and sound. I have yet to do research to find the needed gauge of wire to run everything and what connectors will support the wattage needed. I'm hoping that I can fit these connections in bundles that appear to be MU and/or airbrake lines.

The "8 trailing cars" comment was the number of the trailing cars I'd make, but only 1 is needed per consist. having 8 trailing cars would allow me to run 8 different consists on the track at once.


Arail will be a step backwards in many areas, I will fully admit that, especially size requirement. But there are always trade offs with different solutions. Honestly if DCC supported RFID or any other locating down to less than an inch (or even a foot) of accuracy I'd use it, but my railroad would require over 50 blocks and that's just a scary thought for long-term reliability. And since knowing where each locomotive is, is the most important thing in an automated railroad that missing feature in DCC, revolution, mts, etc are what force me to find another way to control consists, and the best thing I've come up with is what I've talked about here. I'll spend hundreds or thousands of hours on this project over the years so obviously it would be cheaper for me to get a second job with that time if it were just a matter of buying more expensive hardware if it existed.


You do have to admit that this isn't a unrealistic method and that it would work for some people, especially those who already stick a battery in a trialing box car or coal hopper... A brand new, sub $100 solution that will give all the same features found in the revolution and most in DCC feature sets but that the modeler will have full access to the software, hardware, the ability to expand hardware, and location support with only a need of about 8 cubic inch requirement isn't really that bad at all. It also doesn't lock someone into a specific handset. For those that want to use a pda, tablet, computer, or anything else they can. They're only limited by what they do, not what a for profit company says their customers can do.

Anyway, :)
 
#19 ·
"...#1 goal isn't to create something for everyone or replace anything on the market... "

OK, that wasn't my first impression.

I guess it's safe to say that I don't share your aversion to creating new hardware. Considering the vast quantity of custom software you envision, complete custom hardware would be much, much less than 0.001% of the effort.

"No company gives modelers the ability to modify the software in devices to add features they want. "

Yes, take my word for it; there are several really, really good reasons for that....

Well, as I said before, I look forward to seeing what you do with this.
 
Discussion starter · #20 ·
I just finalized my raspberry-pi order this morning as well as a pololu 18v7 usb motor controller so in about two weeks I'll have actual hardware to play with so development will continue again.

Over the past few months I've done a lot of research about prototypical railroad operations, model operating sessions, hardware, software and I've come to the conclusion that trying to model too much prototypical data in G scale just isn't too feasible with G scale products so to make ARail simpler some of the prototypical ideas I had (fully realistic braking based on which cars are attached, slope of the track, etc) isn't likely to make it into ARail for some time if ever. I think that's not something 90% of people would ever take the time to setup for a garden railroad. Also as useful as RailML is I don't think hardly anyone would use that either so for now it'll be removed from the design.

I hope to report by the middle of June to have the following done:

ARail client features:
Bootable SD card image for others to download and play with
Forward, backwards movement with adjustable momentum setting
Automatic (on movement) and manual Whistle, Bell, and air brake release sounds working (No engine sounds yet, still haven't found a good source of or way to play repeating sounds in perl without odd sound pauses in a test script)
RFID tag reading and feedback
Bonus:
Control of lights (might just be LED's depending on if I've taken time to redo a locos incandescent lights)

ARail Web Server features:
Basic web interface to control locomotives
Support for movement functions and selecting momentum rate
Manually request sound signals to be played.
 
1 - 20 of 53 Posts
About this Discussion
52 Replies
10 Participants
Last post:
Martan
G Scale Model Train Forum
posts
460K
members
32K
Since
2007
A forum community dedicated to G scale model train owners and enthusiasts. Come join the discussion about collections, displays, models, styles, scales, reviews, accessories, classifieds, and more!
Full Forum Listing
Explore Our Forums
Live Steam General G Scale Discussions Beginner's Forum Model Making Track, Trestles, Bridges and Roadbed