I have been working on and off over a few years on a large working interlocking. All the track is in, all the turnouts have working Armstrong levers, a scratchbuilt interlocking tower is almost complete, one of many semaphores is in and a full size tower board with track diagram and indicators lights is coming along (on the wall.) I have yet to add the levers for the signals, locks and derails (fake), etc., but in total I’ll have 40 levers. The miniature Armstrong levers use Tortoise switch machines with extra contacts to control the turnouts. (I will probably use memory wire for the semaphores.)
My model RR utilizes DCC (Easy DCC with radio control) and is HO. The prototype would have been a mechanical interlocking not unlike the current series of articles in Model Railroader just rerun. However, though I’m modeling a mechanical interlocking, I would like to use a computer program in the background to do the interlocking. i.e, not allow me to line up any conflicting signals and routes. (I suppose as a safety feature I could have the system kill the power on a conflicting route.)
Also, my tower board, in addition to working indicator lights will have a buzzer and bell to indicate track occupancy for the tower operator. Also, my semphores model the moterized kind with the blades dropping to a stop aspect with train occupancy.
What options do I have for computer software and components? I recently read Lionel Strang’s article in the December 1998 MR on using the IRDOT ONE from Heathcote Electronics with infrared technology. The detectors not only light up indicator lights but can be used to switch current. But what I don’t like
LION used simple dogs to prevent adjacent levers from being pulled.
LION is going to use Tortoise switch machines inside of this interlocking machine to lock any lever that is not to be moved. From there it becomes simple. You align the plant correctly or the levers remained locked.
LION will put the tortoise on its side, and will use a 1/16" welding rod moving up or down to lock or clear the levers. There should be enough contacts on the Tortoise machines to do this for you, but you may want to use relays in addition to or instead of since they could be in a more comfortable wiring location.
On Layout of LION if a northbound train is waiting at Dyckman Street, the home signal cannot be cleared unless the plant is pointing him to an available platform at 242nd Street. Lever 3, if unlocked, will clear the home signal at Dyckman Street, but will set the home signal at Botanic Garden to red. This way a following train can not use the same signal to run into standing trains and push them onto the floor.
LIONS do not use computers in the train room (except for music or radio stations). All of his logic for both interlocking and for block signals are via analog components. LIONS like relays that go CLICK. Him can see what they are doing are are easier to troubleshoot.
Interesting. Thanks for sharing. I’m using minature Armstrong type levers and they are connected to slide toggles with oblong holes drilled in them. I’d post photos but I find the way of adding photos here too fussy.
For the software, look at JMRI. It has all the logic capability needed for advanced interlocking control, and it works with many different layout control systems. And you can certainly set up and run that logic without displaying it on a computer screen. Here’s how one layout uses JMRI for something very similar to what you’ve described: http://jmri.org/community/examples/SPShasta.shtml
As the above link shows, C/MRI is still widely available and used, and it’s supported by JMRI. But so is Easy DCC. Have you looked into using that since you already have that in place?
I don’t know much about the abilities of Easy DCC when it comes to external inputs and outputs, but that’s where I’d start looking if I were in your position.
On the other hand, I use Digitrax and it has lots of possibilities for those inputs and outputs, from both Digitrax itself as well as from 3rd party LocoNet vendors such as RR CirKits, CML, Team Digital, etc.
With that in mind, if suitable I/O devices are not available for the Easy DCC system, another option to C/MRI would be to set up a stand-alone LocoNet (no command station is needed), and use that for your interlocking.
I guess these days it’s split - people who used C/MRI in the past still use it, although I think a lot use it with JMRI these days instead of writing their own code in BASIC. JMRI interfaces with multiple systems, including Easy DCC and C/MRI, and provides logic tables to set up the operation based on the rules you define for interlocking and so forth. You do NOT need to make an on-screen tower panel in JMRI, the inputs can come from hardware like a manual interlocking tower provided the tower levers feed some sort of input device.
The other mixed mode way to do it is, as Steve mentioned, Digitrax Loconet. You do NOT need a Digitrax command station, or to change anything with your DCC system, to use Loconet signal and detection devices. There is at least one user on here, Elmer, who has NCE DCC to run his trains and uses Loconet for the detection and signalling. A large club.museum near me that was not too long ago featuered in MR does the same, and I was recently at the open house of another modeler near me who also mixes systems this way, with NCE to run the trains and Digitrax for the signals. One of the advantages of using Loconet is that, unlike C/MRI, there are multiple sources of hardware, everything from built up and ready to go to DIY projects using Aurdinos and everything in between, from many different vendors - you wouldn’t even have to use a Digitrax branded product at all. For example, I have a few boards I built up fron Hans DeLoof, called the LocoIO. These boards each have 16 bits of input or output - these would be something to use in your physical tower, with your levers activating switches onthe inputs, and outputs being used to ring the annunciator bell and light up the model diagram. Interface of the entire tower becomes a single 6 wire phone cord plus a power supply. If you don;t want to or can;t build your own boards, CML Electronics has similar I/O boards for LocoNet, as does RR-CirKits, as already assembled and tested devices ready to use. For just the signa
Thanks for the information so far everyone. It sounds very doable. When in college as a business major, one of the first computer projects I did was with basic, (of course later I learned Fortran.) I would rather keep the interlocking/signal system separate from the DCC but am open minded about it. So, generally, how is detection achieved for the indicator lights on the control panel, interlocking logic and annunciator bell/buzzer. Is it with resisters on the car axles, etc?
Yes, the usual way is still to put resistors across the wheels. Our club standard is a single 4.7K resistor on one axle, I do mine as a 10K on two axles of every car, that way is a car straddles a block it is detected in both. The detectors for DCC can go as low as 20K, so a single 10K is easily detected, and a pair of them is - hey look, 5K. It’s easy to do current detection with DCC since there’s always power in the rails - with DC there’s all sorts of workarounds with bleed circuits and stuff. Nothing like that needed for DCC.
IMO the best way to detect the current is with the transformer type detectors. This requires zero connection to the DCC system - the feed for each detection block goes through the core of a current transformer which detects the slight current flowing through the resistor wheels and activates the circuit. This is like the Chubb DCCOD, or the RR-CirKits BOD-8. You’ll need to gap one rail where you want to define the detection sections, and all feeders to that insulated section (just the one rail) should tie to a sub bus wire which passes through the sensor transformer and connects to your main bus.
With the detectors feeding some sort of input device, this generates a signal that your panel can use, or the computer can act on - say combine this block occupancy information with a contact from a switch motor to show the state of a specific turnout, and then use this logic to combine into an output to light up a signal.
Thanks, that’s informative. With my two-headed semaphore signals, in the interlockings, each head is controlled by a lever that controls two aspects. But with the electric circuitry, you explain, perhaps the circuitry can be a check like the real thing and not allow a signal aspect unless the plant is lined correctly. That’s very good, just like the real thing.
You might not need a whole lot, if your machine is like the one in the article and all mechanically interlocked. All you really need is maybe a detector on each approach block, and you can use one of several that can activate circuits without being wired to anything else. All you really need to know is when a train is approaching, and from what direction. The mechanical design of the levers will prevent lining conflicting routes. It depends if you will have any intermediate signals between interlockings. If you are, then perhaps the more complex design is warranted so you have the logic to generate those intermediate aspects. But if the whole purpose here is to just have a mechanical interlocking tower, you probbaly don’t need the computer or interface equipment. Contacts on the levers can operate the switch motors and motors for the semaphores, you just need the detection for annunciators. Electrically it should be pretty simple. It’s for the more modern type panels, with nothing but electrical switches and indicator lights, where we need to have logic of some sort to make sure no conflicting routes are created. Those mechanical lever towers, they physically prevented you from pulling two switch levers that would point oncoming trains right at one another.
Yes, but my full size tower board will have more indicator lights than just the indicator lights at the approaches to interlocking limits with lights lighting up and going out as the train progresses through the plant. (Some towers also had route levers with lights but my prototype didn’t do that.)
Also, my levers are not physically interlocked like in the Model Railroader article or on Lion’s layout, so some kind of circuitry will have to do that. They are hooked to Tortoise swtich machines via slide switches that throw the Tortoises and the switch points. (Of course, I have extra contacts on the switch machines.)
In addition, there will also be green signal indicator lights on the board that will light up when a signal is thrown (that is a no brainer, as far as circuitry).
Also, though I mentioned it before, the prototype for my semaphore signals had small electric motors behind the signal heads and when the train “knocked down” the signal, the signal went back to its most restrictive aspect (just like searchlight and color signals that go to red as the train passes.)
The semaphores in the Model Railroader article were acutuated with mechanical rods up the mast (probably on the model too), with the operator changing the aspect back to stop with the same lever in the tower. These were older style interlocking semaphores and rare by the 1950s, but a scant few were still around.
Well in that case, forget my last post and go back to the one before. You’ll have the contacts on the levers, and then you’ll need some legic to lock out the conflicts and so forth. For semaphores, there are some circuits that use servos to drive then that even simulate the bounce when they drop. Looks pretty cool. That all happens behind the scenes, you sned commands to the circuit to say “display stop aspect” and it moves the servo and whatnot. Or “display proceed” and ir raises the semaphore.
Another thing about the computer logic, you can tie as many things together as you need. Say Block 5 occupancy is needed (maybe along with additional data like turnout position) to determine the aspect to display on signals 6E, 5W, 4E, and 3W. No problem. Plus it can ALSO light up an indicator for block 5 on the mimic board. Just need the one input fromt he block detector, and as many conditionals as are needed to fit the logic of the plant.
This works both ways - such as “If Block 5 occupied and switch 51 Normal, Set Signal 3W STOP” but also for the interlocking, so you can;t change the route under a moving train. Such as “Lever 4 to Normal, and Block 5 Not Occupied, Switch 54 to Normal”
Normally on a lever machine this movement would be mechanically blocked. In this case you’re going to have to logically block it, which means the levers may not indicate the current state of the plant. On something like a USS machine, you have the Code Start button which actually attempts to apply the changes, and rejects invalid requests or conflicts, and the indicators no longer match the switch handle positions. But each signal and switch lever has indicators with it, which indicate the current state, not necessarily the desired state pointed to by the switch levers.
Searchlights ARE semiphore signals. Little itty bitty semiphores in a single cabinet. This is why they are being replaced out in the real world with three lamp heads. Semiphores and searchlights only worked with the same interlocking plant from the same manufacturer. It was built as a system. Now that the big interlocking frames are going away, replaced by a computer screen, the need for mechanical signaling is also gone. But if you build a mode railroad, then you do need to be true to the era that you model.
US&S plants used compressed air to move the switch motors, the levers in their plants had a neutral position to save (I presume) air. Thus they had the lever capacity to display three aspects from one lever. PRR used US&S signals exclusively and used position lights to display a signal, so that was electrical, but there may have been a mechanical component in the equipment cabinet, for clearly air pressure alone can not iluminate light bulbs.
On a GRS machine, such as I have modeled, one lever is used for each set of switch points.
If you could move the lever, it was unlocked and was safe to move
It would only move half way, in doing so it would lock all other levers that would be in conflict with the new setting. It would send an electrical signal to the switch motor whcih would move, and then on the overwind, would generate a retrun signal that proved that the switch was aligned and locked.
Once that signal came back to the machine the lever could finish its stroke releasing levers that were no longer in conflict with this alignment. The alignement was locked
Thanks for the commentary Lion. (Incidentally, there were some other manufacturers of Armstrong lever plants if you go back far enough, including a US&S predecessor.) I am building Armstrong lever plants with semaphores because that is what I remember and what my prototype (Wabash RR) used. And in later years, in addition to the mechanical locking, there was some electrical circuitry used too.
Yes, I have the knowhow and skill to build a real locking mechanism with dogs and tappets but don’t want to spend that time that I can use on other projects in my basement empire, including scratch-building. And yup, the Wabash’s supplier was US&S, whether it be for its semaphores, searchlight or color light signals. I find the added benefit of using semaphores is that if the signal is not facing the train operator, the aspect can still be seen without the use of signal repeaters on the fascia.
