Signal system and signals for an N-TRAK layout?

[Sokramiketes]

Just so you know, once you've spent a lot of money and installed a fully operational ABS signal system on the layout, operators will still talk to the public and ignore the signals.  Just come watch the HO Midwest Mod-U-Trak guys sometime.

Mod-U-Trak uses DCC, but the signal system is independent of track power completely.  Bruce Chubb DCC Optomized detectors feed Integrated Signal System's signal drivers.  There's no reason you couldn't use the DC detectors on Ntrak though.

For a modular layout, you want ABS signalling.  You want to minimize wiring between module joints.  You do not want to be running wires back from each module to a central control point.  And, you want to be able to swap modules in and out without having to reprogram the system.  With ABS, the signal on one module only needs to talk to the next signal down the line in each direction.  ISS boards are good for this, for MiNi we're going with TracTronic's boards which are basically the same operationally.

You're going to need three fat wires (bus wires) per mainline, and a couple skinny wires (phone cabling) to carry the signal information.

The three bus wires:
1) DC Wire 1
2) DC Wire 2
3) DC Wire 2 detected

1 and 2 are continuous with no breaks.  3 is broken at the block boundaries.  3 is fed from 2, and the feed is passed through the detector.  With the three wires, you can use a single detector to cover multiple modules, as 3 will connect to each of those modules and feed the signal through a single detector before being tied into 2.  This is paramount for being able to swap modules in and out seemlessly.

Sidings that are not detected are fed straight from 1 and 2.  While detected mainline is fed from 1 and 3.

We're running 12 bus wires around MiNi (2 track mainline).  For modules at a block boundry or interlocking you need to break out most of the wires.  Or, for simple pass-through modules without signalling, you can break out just the track feeders 1 and 3.

Here's a signalled module using terminal blocks to break out from the main 12-wire harness:


The 'Orange and Yellow' and 'White and Blue' are paired (1 and 2 per above), while the 'orange with yellow tracer' and the 'blue with white tracer' are the detected (3).  Other wires are a 12V supply for accessories, a ground, a 9-12v supply for tortoises, a 5v for signalling boards, and an a/c supply for yellow searchlight indications off bi-color LED's. 

For a pass through module you need all 12 wires in the harness to pass through, but only break out for feeders for the mainlines.




All that gets you is the detection, then you have to wire up all the signalling boards.  But, telephone cable or CAT5 will work there, which is easy to pass through a module joint.  That's our next step, I've been building TracTronics searchlight driver boards. 

[Walkercolt]

That dozen wires gets to be a deal killer on a lay-out our size. The sheer amount of wire is frightening....all those connections, and the exspense. The voltage drop over 300+ feet of wire means we ain't gonna be using "recycled" 22-26 ga. telephone "bundle". There's an answer out-there...but I haven't found it yet. Something based on DCC signaling(but that will work with DC) seems to be the most practical so far.

[Sokramiketes]

That dozen wires gets to be a deal killer on a lay-out our size. The sheer amount of wire is frightening....all those connections, and the exspense. The voltage drop over 300+ feet of wire means we ain't gonna be using "recycled" 22-26 ga. telephone "bundle". There's an answer out-there...but I haven't found it yet. Something based on DCC signaling(but that will work with DC) seems to be the most practical so far.

I don't see a way around that, as you still need to set up detected blocks, and that's the minimum number to do two main tracks... with N trak you'll have another 3 wires.  DCC signalling still needs that input from detectors.

Really, your only option is to fake it with time delay stand alone signals if you don't want to run all that wiring.  Something like the Kato signals.

[Walkercolt]

Mike, you're confirming what I've found out so far, and it ain't too good. I'll keep looking, but 3600 feet of 16 ga.wire... I'm agin' it, if you know what I mean. : One semi-good thing, wanting semiphores means I can just detect the engines, and then the semiphore drops to yellow "restricted" and I can "time-out" going to red, then to green. But the detectors just take too darn much wiring to "feed" to a "signal controller" that can translate to CAT5 or RJ-11 or RJ-12(Loco-Net), that I can see. Somewhere in the back of my mind is the idea some kind of hex-driver or encoder could cut down the wiring, but then I'm building circuit boards again. (Heavy sigh!) :-[

[mmyers]

The only way to reduce the wiring between modules is to put the circuit boards (signal controllers) out on the modules along with the detection circuits and the signals. That puts all the electrics for a section (block) on one module. Each block gets connected together by a single data cable. There will be some unused sections of the controllers because there will be parts of each circuit board that won't be needed at each location. That's because we don't want to run wires past the end of the module. Everything has to be on the module and they all have to talk to each other.

Team Digital's SIC24 Controller fits the bill for a signal controller. Mix and match the detection type. Detectors just need to take the SIC24's sensor inputs low. There's 8 sensor inputs. Enough for both directions on 4 lines. 24 outputs will handle three aspects in each direction on 4 lines. No computer is needed, but programming is required for each board. That's the unfortunate part. Once programmed, each module will now be dedicated to a block on the layout so the logic works every time.

If 25+ aspects are needed or 9+ sensors are needed, another SIC24 will be needed. That runs the cost up but retains the single data cable connection.
If you are running DCC already, I presume that there is some type of control bus being run now. SIC24's are compatible with Digitrax loconet and NCE. They can also be stand alone.

[Walkercolt]

If we don't assemble the layout the same way everytime(we don't) and use all the modules(we don't) then I need a board on every module(and power to it, etc. etc etc....). Makes things a bit difficult. The thought of troubleshooting the wiring at a show, gives me daymares as well as nightmares. Thanks everybody for your thoughts and inputs...they have been helpful, not the way I wanted, but very helpful. Kinda like finding out your engine block's cracked from people who've had the same problem. Helpful, but not what I was hoping for. :-[ :'(

[John]

Put your signals and detectors on club owned corners .. you need four of those anyway .. problem solved.

[mmyers]

The only problem is using them in the same order each time if they are to work like a true signal system.
Block occupancy is relatively easy, because all that's needed is a detector and a signal in the same place. Having  a yellow aspect two blocks back requires a network or a ton of wires.
We do block occupancy with DBD2's mounted in a box. Each detection box gets plugged in line between DCC booster and the block (power district). The DBD2's can handle our NTRAK towers with either light bulbs or led's with no extra circuitry required. A single 4 wire phone wire between the DBD2's and the towers is the only extra wire to run. The towers just hang on the skyboards.
Our trouble has been combining analog detection and DCC detection. No real trouble having both types present. In fact the DBD2's will detect on an analog block --sometimes. They are not reliable enough on analog to be the only detection. On DCC, they don't miss. Our total cost to add them so far is $300. That's for 24 detection zones and includes all of the hardware as well. We already owned the signal towers.
The real bear is combining the outputs with that of our Lindsay detection so that only one tower is needed for each block. I cooked a couple of transistors on the Lindsay detection units on the first try. The Lindsays and the DBD2's are both common anode outputs. The DBD2's came through with no damage.  We were testing an optocoupled circuit at Bedford but I inadvertently allowed a couple of jumpers to short and fried a voltage regulator as well as melting my breadboard. The circuit did work until the smoke started, but I need a more reliable test bed before asking for club funds to build a bunch of them.

Martin

[Walkercolt]

Martin, the DC lines are the one's we're trying to signal(of course), and "interfacing" DC with a DCC control circuit is a problem, and we use 6 corners, because we have some very nice curved modules, and the 6' curve on "Route 66". Several places can offer us boards and so on for what we want, but the wiring becomes the undoing. Every connection is a potential problem, and a "troubleshooting manual" would look like Webster's Unabridged. The reason for signaling is so we can run several trains at once(in different blocks) on the DC lines to provide more trains for the public to view. Over-running blocks is a problem when the blocks have to be 80-100' long, and around the corners. We can't see the end of the train in the next block always(rarely in fact). Just like the prototype, signaling is the "answer", but thousands of dollars in wiring and equiptment(and those darn connections to deal with) just isn't where we can go. I can understand Bruce Chubb saying for every thousand dollars on top of the layout, he had $20,000 underneath the layout, and he's an electrical/electronic engineer. On a stationary layout, I would be far less worried about the wiring and such, but dragging modules half-way cross-country in a Ryder can cause "fool-proof" wiring problems. I don't want to think about broken circuit boards from rough handling/transit. When I was in the Air Force, they told us a B-52H had about 2000 miles of wiring inside. They probably were conservitive. I look inside my PC at all the wires, then look inside an Apple G-8 with fiber optics running from the hard drives, DVD drive, etc. Not hard to understand the Apple engineers thinking at all.

[bradb]

I'm not sure this is applicable at all, but I suspect it might be.

I was swilling beers and discussing the latest tech trends with a friend of mine last night and he told me about his latest interest, the Arduino (http://www.arduino.cc/).  I'm always interested in new gadgets, the smaller the better.  And when he mentioned the I/O capabilities this board has, I thought immediately about possible applications to the hobby of model railroading.

It turns out that http://www.siliconrailway.com/ has a modified version of this board called LEDuino, which is a version of this open source hardware that includes additional functionality, including (optional) DCC support.

Despite the low-cost of this board (about $40 w/ DCC) I suspect this is probably too expensive for some sort of signal system, but I do like the possibilities...

Brad.

[Sokramiketes]

Martin, the DC lines are the one's we're trying to signal(of course), and "interfacing" DC with a DCC control circuit is a problem,

Just depends on which detector you use with the DC/DCC question.  The thing driving the signals can be "DCC" without having to use DCC on the rails.

The Digitrax signal system, for example, can be used on a DC layout... you have to run loconet around the layout to tie into all the signal controllers, but those controllers take input from block detectors that can tie into DC or DCC fed rails.   

[wcfn100]

This seems like that should real easy to to do with DCC and only use the loconet wiring.


Each dectector should be able to communicate with whatever signal it's wants to via the signal's address.  Or another way to look at it, a signal could be assigned to a block of detector addresses and as long as one detector in the block shows occupied, the signal stays red.  And this can all be done with only loconet wiring between the modules.




Jason

[Walkercolt]

Jason, N-TRAK standards, and most of our club memembers(me included) aren't all DCC. DCC on N-TRAK must be optional, with lines running plain DC available. The guys at Digitrax are going to get back to me on their protocols for their signaling system(which don't seem that complicated). But the detector wiring still is very intense on each module. I can essentially duplicate the prototype signaling system, with about 16 wire pairs around the whole layout(for two tracks). That is NOT happening. I can cook-up circuity on each module that would reduce the wiring to a few wires, but I can't do it cheaply. ???

[wcfn100]

Jason, N-TRAK standards, and most of our club memembers(me included) aren't all DCC.

The DCC would just be for the signals in this case.  It's independent of the track power. 

But the detector wiring still is very intense on each module.

With a DCC 'like' sytem you would only need one 'loconet wire' to run through each module.  And not one per track, just one no matter how many main lines you have.

After thinking about it, it seems really easy logically, so it must be cost prohibitive as I know nothing about the cost of electronics.


Jason

[Walkercolt]

Jason, so far the cost is prohibitive, but I am talking to the guys at Digitrax, but I really don't have a lot of hope, unless they want to use us as "guinea pigs". I can hope for that at least.