Planning for future automation

[AndrueC]

I have an established layout that is currently at the scenic stage and I'm thinking about adding automation. It's a fairly complex layout with two yards, each having four through sidings. There are four loops all of which can be run separately or in conjunction with other loops.

 

Three of the loops are connected by double slips so require constant turnout switching when joined but the outer two loops are connected via a diamond crossing so can just be left to run from one to the other and back.

 

I already have turnout motors installed and live frogs juiced powered from controllers (mostly - there's a couple still to do, lol). All sections of my track have droppers already so I'm thinking that I just need to isolate track sections and add detectors. Do I need multiple detectors for each run of track between turnouts or can I just have a single detector (in which case a lot of sections might already be isolated sufficiently because of the turnout frog)?

 

What I'd like is to be able to just fire the computer up and tell it to 'run stuff' while I watch.

 

 

Before I retired I was a computer programmer so I have no worries should that be required for the project.

Edited February 20 by AndrueC

[WIMorrison]

Essentially what you suggest is correct however (there is always a however) the placement and length of the feedbacks in the track and turnouts can be optimised for reliable running and operations.

 

if you plan to do automated shunting, or swapping of locos on trains then you will need a small entry feedback that always remains free to allow identification of a loco entering or leaving the block. If you plan to do it either end of a loop then you will need this short feedback at each end.

 

At stations, or where you want to stop trains then additional feedbacks are beneficial because they will reset the calculated position to actual position. In running blocks where a lower accuracy of stopping is desired then fewer feedbacks are required.

 

Turnouts do not require occupancy detection but it can be beneficial and should be considered, however (another one) I suspect that your use of frog juicers will prevent the monitoring of turnouts for occupation. If this is the case then you should feed the power to the turnouts and any unmonitored track through back to back diodes to ensure track voltage remains constant across the layout.

 

 

[AndrueC]

Luckily I'm using fixed rakes so no shunting required. It did occur to me that splitting the longer runs of track into multiple sections might allow true tail chasing (perhaps two trains on each of the outer loops). I could have phrased my description better as I'm actually only using frog juicers on the diamond. All other turnout frogs are powered from an appropriate motor controller.

Edited February 20 by AndrueC

[RAF96]

Depends upon your loco detection system as to if you need track sections or not.

I use reed switches and/or hall effect sensors situated in between the rails to trigger software events. Standard DCC bus with no track breaks anywhere.

Edited February 22 by RAF96
Typo

[AndrueC]

Thanks for the replies. I originally wanted to go RFID because as a computer programmer I want as much detailed information as possible. Unfortunately having planned out sections for half of my layout (the inner loop, the upper yard and its loop) I've realised that I'll probably end up with at least 60 sections by the time I'm done. That means I need to for the cheapest sensors for the most part although I accept that at either end of a siding there needs to be something more accurate for exact positioning.

 

So continuing my plan I think what I need is this:

 

• JMRI - that can run on my existing mail/video server.
  • Option 1: Run cables for monitor/keyboard/mouse into railway room. Minimal expense because the server runs headless normally so just needs holes drilling in a wall.
  • Option 2: Dedicate notepad using JMRI webserver. Additional cost but could be more convenient.
• Replace my NCE PowerCab - I'm thinking of going with MERG as that will probably be cheaper.
• Current sensors, IR sensors.

For sections/sensors I'm thinking that on the main runs there should be a sensor covering every couple of metres (my rakes are typically about 1 metre long) as that should allow some tail chasing and improve the resolution/timing for the computer.

 

Am I overlooking anything?

 

Current section plans (coloured track):

Edited February 21 by AndrueC

[ITG]

On 20/02/2024 at 13:30, AndrueC said:

(in which case a lot of sections might already be isolated sufficiently because of the turnout frog)?

I’m just finding my feet with automation via iTrain. I decided to scrap the then existing layout, and start again with automation, but that’s at least partly due to a bigger space becoming available. Nevertheless, with a scenic layout and underboard access required for at least some of the rewiring necessary, I wouldn’t have fancied modification!

There is a risk in your assumption about sections (ie blocks) already being isolated due to frogs being isolated. Many current sensor units actually will run multiple feedbacks, eg Yamorc unit runs 16. But (I’m pretty sure) all the 16 in this instance need to be the same ‘polarity’, ie either the ‘red’ or ‘black’ wire. The problem with assuming your isolated turnout frogs are ok, is, with turnouts being different ways round on the layout, some of those currently existing isolators will be on red, some black. So not quite as simple as at first seems.

My suggestion would be to use your layout plan to draw up a ‘virgin’ electrical-cum-track plan diagram to find the ideal configuration of blocks, feedbacks, breaks etc. Of course, you’d need to do some research and study to learn what is optimum in your chosen software.Then see how much difference there is between the ideal and the compromise-based-on existing plan options.

Ian

Edited February 22 by ITG
Added

[AndrueC]

Thanks @ITG. To be honest I'm still uncertain whether I want to go with current sensing. My layout is at an advanced stage and the thought of cutting track concerns me. I might yet decide to go with spot detection instead although of course that means yet more wires under the board. It also raises some questions of efficacy. My layout's running is straightforward though - just tail chasing - so I'm hopeful that it will be more amenable to spot detection. Once you know a train has gone into a block you'll know it'll eventually come out. The only exceptions are the sidings and I'm going to have spot detection there anyway so that the computer can 'reset' it's knowledge of position when a train parks up.

[ITG]

30 minutes ago, AndrueC said:

Thanks @ITG. To be honest I'm still uncertain whether I want to go with current sensing. My layout is at an advanced stage and the thought of cutting track concerns me. I might yet decide to go with spot detection instead although of course that means yet more wires under the board. It also raises some questions of efficacy. My layout's running is straightforward though - just tail chasing - so I'm hopeful that it will be more amenable to spot detection. Once you know a train has gone into a block you'll know it'll eventually come out. The only exceptions are the sidings and I'm going to have spot detection there anyway so that the computer can 'reset' it's knowledge of position when a train parks up.

On the now dismantled layout, I did use a few (just 3) Heathcote Electronics infra red IRDOT sensors to simply sense when a hidden storage rod was occupied, just to light an LED panel light. Worked ok, but a little fiddly to install under track between sleepers, and some tweaking necessary dependant on surrounding surface proximity and light level. These could link to automation software.
As you say, this type of option wouldn’t be without more wiring, but then pretty much all options will.

Ian

[WIMorrison]

I think that the option with least wiring will be current sensing, and certainly that will provide reliable results that won’t get affected by external influences - and they can be more cost effective.

 

installing current sensing is no more hassle on a finished layout than it is if done when building.

 

superglue the rail chairs either side of where the cut will be and after it has set dremel a gap using a fine cutting disc. Very simple and very effective and doesn’t affect performance or reliability.

[5BarVT]

Still using current sensing, there are some designs that use a transformer design rather than diode drop.  Although mine are all on the same polarity, they don’t have to be as there is no electrical connection between the rail and the sensing circuit.  Will still require fiddling around under the board to separate wires and route the right ones through the toroid.  Thats where it’s easier if the boards separate so they can be upended for working on and ‘too late now’ if they don’t.

I use Rr-Cirkits Watchman available from Coastal DCC, supplemented by MERG DTC2 where a multiple of 8 is not cost effective.

Train detection in turnout areas is an interesting topic (!).  For the commercial systems (iTrain and Traincontroller) you don’t need it to make the programme work, for JMRI you would need to check.  (I think I have seen a JMRI discussion that hinted that train detection sections need to be abutting for its logic to work properly.). When comparing, also beware terminology differences - in TC and iTrain a Block is the track between points where the train will stop and may have one or more train detection Sections. In JMRI, Sections are the track between the points which may contain one or more Blocks!

 

I’ll be back with more generic thoughts later.

 

Paul.

[5BarVT]

P.S. Meant to say that ITG’s idea of drawing up a wiring plan from your layout (hopefully knowing how your DCC bus and droppers are routed) would be a good feasibility study.  That would give you an idea of how much effort and cost will be involved.

 

If you discover that JMRI does not need continuous train detection, and you go with diode drop detection, Iain’s comment about back to back diodes for undetected areas means either a pair of diodes for each junction area, or, a second bus around the junction areas.
 

Paul.

Edited February 23 by 5BarVT
Hit <Save> not <Return> !