Signal Project

[kvp]

How about the classic D-subminiature connectors? There are 37 amd 50 pin variants and you could mount them recessed (or even on small springs), so they connect when the boards are flush together. An even stronger alternative is the 36 pin centronics port (IEEE 1284). Both are easy to install and are used on many modular layouts in Europe.

[E6系]

Hello Mr kvp,

 

What wire gauges do the D-subminiature connectors accept?  I am using AWG12 for all DCC supply, except for the last 10cm to the rails.  Also, the connectors I use are designed to carry 12-14v all the time. :D

[kvp]

The common wire gauge is awg 20, but ribbon cable versions use smaller, while solder cup versions could use anything that can be soldered to the connector and that's around awg 12. Maximum current is around 5A per pin, so two ore more pins are needed for more. Voltage is up to 24V, but usually 12V signalling was used on standard serial connectors. High frequency transmission and noise shielding is especially good, but that only matters with DCC and other high frequency digital protocols. The general approach is usually to use more pins (next to each other, fed by a single wire) for high current lines and single pins for low current signal lines. Also the number of ground pins should be around the same as the number of power pins, even with common ground.

[cteno4]

The serial connectors are fine for regular 12v power at an amp or so, I've used them a lot with up to 16g wire if you roll your own. The rewired 20 or 22 g cables are fine for doing point control. We had planned on using db25 cables to go to each of the current jrm layout modules to control points, but we realized it was going to be more work than it was worth for the amount of point switching we normally do and find it more fun to have a point person throwing points and someone else on the throttle. But I tested cheap 15' 22g db25 cables to throw points and worked fine with both kato controller and with capacitor discharge.

 

I've also used the serial pins by themselves (no housing) and just insulate them with heat shrink. We used this on the first jrm layout where we built the layout on the fly and we needed the smallest and cheapest pins on cables to poke thru the 1" foam of the module tops. It was like 12 cents for a pair of m/f connectors. Worked well.

 

I like some of the auto connectors as they can handle heavy gauge wire and are robust, but downside is many don't like to come apart easily (not meant to be disconnected much) and many have no strain relief. We are using power poles for the current jrm 12v power with 16g. I've not become a huge fan on the power poles even though they have become the defacto us modular power connector as we have had a few failures with not that many connections and there is nothing for strain relief.

 

Cheers

 

Jeff

[E6系]

Hello Mr Jeff,

 

Thank you for your explanation however, the motorcycle connectors are just so easy to use.  Crimping tool is simple and a dab of solder seals the deal.  Also have much greater cross sectional area than the pins on the D-sub miniature connectors.

[mrp]

Hi All,

 

Not much of an update - I’m waiting patiently for the next batch of boards from China - so I thought I’d show you my attempts at painting and attaching the 3d printed heads.

 

 

 

The viewing angle for the LEDs turned out better than I expected.

 

[ToniBabelony]

Oh my oh my oh my, this looks very promising!

 

I can already see the level crossing warning light become a thing in combination with the Tomix moving bus system.

[mrp]

Hello All,

 

I little bit more progress…

 

As kvp accurately described earlier in the thread, my plan was also to drive turnouts using the same bus as the signals.  I just received the first batch of prototype turnout boards designed to do exactly that.

 

 

The board handles three turnouts a bit like the R-Y-G lights on a signal.  When the light is turned on or off, the board sends the appropriate polarity 30ms 12V pulse to the turnout.

 

It might be a bit hard to see, but here’s a short clip of a 3-way diverging signal - connected to a turnout board - connected to a 3-way regular signal - switching back and forth between the two routes.

 

 

Eventually, I should be able to string any number of signals and turnouts together - and, I guess, level crossings, lights, etc - and run them all from a single control interface.

 

[E6系]

Hello mrp.

 

Very impressed again.  If you have signal masts compatible with the connectors used by Digitrax then I would like to test.

 

A few questions:

 

1. Are your turnout boards, which might better be known as "Signal Control Boards", or "SCBs",  fully automated?  I am thinking that an option for pushbutton control might be necessary.  For example, on a terminus platform the signal should be red, regardless of turnout position and occupancy, until several minutes to departure.  Then, if the track is clear and the route is set correctly, the controller will set the signal to clear or caution, as the case may be;

 

2. Would your SCBs be able to use feedback from DCC occupancy detectors rather than direct track connection?

 

3. It appears that your SCBs are able to report turnout position.  This could be a very useful accessory to a DCC layout.  To know how turnouts are set, most DCC systems set all turnouts to OPEN on startup and then change selected turnouts to THROWN.  However, I don't know of any standard equipment available to report whether the turnout motor successfully changed the turnout position.  Thus, if the turnout motor has failed, the DCC system will still believe that the circuit has been thrown.

[mrp]

Hello E6系,

 

Thanks for the comments.

 

1) The “SCBs” just receive a very simple data stream that controls the PWM setting for each individual light directly.  A chain of “SCBs” needs to plug into a controller board that knows what aspect to set each signal to - generally from a centralized automation system.  Currently I’m working on a serial controller board that uses the Dinamo protocol so that I can control the signals from the iTrain software.  In theory, you could also have simple DCC-based controller that would read DCC signal commands and drive a chain of “SCBs”, but I haven’t thought much about that yet as I haven’t done any work on DCC.

 

My approach for manual overrides would be to have some other type of on/off feedback to the central automation system and have the logic for how to respond (manual signal aspect, etc) programmed in to the automation system.

 

2) There isn’t any feedback directly to the “SCBs”.  I plan to have the central automation system monitor the occupancy detectors and set the signals (and block reservations) accordingly.

 

3) No, I don’t use any feedback from the “SCBs” or turnouts.  For my layout, I’ll be using a combination of Tomix TCS sensors, throttle current sensors, and possibly some IR proximity sensors I’ve been experimenting with.

 

Without a DCC-compatible controller for the “SCBs”, they can’t easily be used with the Digitrax system.  For the moment - if you already have the SMBK / SE8C setup - then I think it would be easier to just make some “SMBK-compatible” masts with Japan-style heads.

 

Coincidentally, the test SMBK designs I mentioned earlier in this thread just arrived back from the fab:

 

 

I don’t have an SE8C to test with, but I did try them out with 5V supply and an open-collector connection to ground through a 100Ω resistor (which is what the documentation says an SE8C does) and they work ok:

 

 

 

We can work out the best way to send you a couple to play with and see if they’re of any use.

 

[E6系]

  ... Look very nice.

 

You might find some of these links interesting.

 

http://www.rokko1go.jpn.org/main/photo/english/jp-signals.htm

 

http://www.isi.edu/~braden/japan_RR_signals/index.html

 

http://www.isi.edu/~braden/japan_RR_signals/new.index.html

 

http://www.jrtr.net/jrtr21/pdf/F44_Technology.pdf

 

http://en.wikipedia.org/wiki/Japanese_railway_signals

[mrp]

Hello E6系,

 

Thank you for the links.

 

In return, I have a link for you that I found quite interesting:

http://simgw.cosmos.ehime-u.ac.jp/~shimizu/n-usamodel/n-jpn-signal.html

 

I’ve been following the work on N-scale signals by Mr. T. Shimizu.  I think the above link is from his original research while at university.  He has a quite active blog, and has also posted many photos of the layout he built for developing and testing his N-scale signal designs and automation.

 

 

 

 

 

Quite incredible!  Mr. Shimizu even explained how he put in a raised floor - like you see in data centres - for the wiring.

 

 

Sometimes I wonder how many huge layouts there are like that in Japan - anonymously running away behind closed doors in unmarked buildings…

 

[kvp]

This is an absolutely huge layout! However when your signal boards and control equipment take up more space than than your train cases, then imho your focus might be off a bit. Anyhow, that's an interesting system and we can see how far we got when we compare his system with mrp's new one. Now if we could get the singalling system down to 4-6 wires, including block sensors, user input and everything, then we might start to arrive at a commercially viable 'all features included' system.

[JR 500系]

Amazing mrp san!

 

This is most promising and amazing project by far! Can we have some traffic signals for the moving bus system? It would definitely look good on the cross junction  :)

 

 

Still wondering if these would be made available off the shelves for amateur like myself to use~  :)

[mrp]

Hello kvp,

 

Yes, Mr. T Shimizu certainly takes his signals seriously!

 

Hello JR500 のぞみ,

 

Yes, I was also thinking about how best to do traffic signals for road intersections.  The problem is that all the real-life (and model) Japanese-style traffic signals have thin horizontal bars that would be almost impossible to shape from a circuit board.

 

 

I really want to stick with designs that can be easily routed from PCBs so that I can just solder on a couple of LEDs and they’re ready to use (rather than very complex life-like plastic models that you need to glue and then connect with super-fine enamel wire).

 

But any section of PCB needs to be wider than 2mm to 2.5mm, so that limits a bit what you can do.

 

I guess I could make some designs to test that just have the general shape of a traffic light but not much close-up detail.  Something like this:

 

 

It would only need to be about 4cm tall, so it might look ok from a distance.

 

I’m still in the early prototype stage on a lot of this - but yes, in future, if everything works well and is relatively inexpensive to make, then it would be fun to try and package it up so that others could try it out.

 

[kvp]

Yes, the european type would be much easier, since the japanese 3 color railway signals look just like a normal european road traffic signal, while the 2 color looks like a pedestrian one.

 

On the other hand, you could mount the signals on solid rods that also carry the power. I'm thinking about longer board to board connector pins, that can be used for such job. Two pins (without their plastic brackets) for the pedestrian lights and 3 pins (two above on both sides of the pcb and 1 arched below on one side) for the road lights. You could see through the pins (rods in this case), they are strong and easy to solder and paint. Also easy to mount 90 degrees, since one signal could be sourface mounted, parallel to the board, while the other could be through hole mounted. (because pedestrian signals are usually rotated 90 degrees from the main lights) This would mean you need 2 small signal boards and a single mast with wire routes and pads and/or holes for the pins.

[mrp]

Hello kvp,

 

Do you have a link for these board connectors?  I’m not sure I follow your description.  Or maybe you could post a rough sketch?

 

It might be a little difficult working with such small sizes if the signal is made up of multiple parts.  Here’s a photo of a 3-way signal with 2mm pin connector, one with a 2.54mm connector, and a couple of bits from Kato Dio Town traffic light kit:

 

 

The pedestrian walk signal works out at about 3mm x 4mm, and the main signal about 3mm x 10mm - although I guess you could make a long thin board containing a number of heads, and then “depanelize” them with a Dremel.

 

It would be great to come up with a design that doesn’t involve any complicated assembly process.

 

[kvp]

http://lomex.hu/getpic/43-07-73.jpg/0/tuskesor-40-p-dual-dphss403782372-tt-254mm-pin-064x064-tord/rnd076649670

 

These board to board connectors could be used to hold the small panels and to supply the power. Just remove the black insulation and you have strong and easy to solder rods. They can be surface or through hole mounted. For the 3 light board, you can use them doubled, one on the front and one on the back of the signal board and that means up to 4 rods per small signal board or 3 leds (more with multiplexing). The lower two have to be bent to follow the arched shape of the prototype. Not too elegant, but easy to assemble.

 

Actually, i've built a few simple 2 light signals many years ago by taking two of these, soldering two smd leds between the rods on one end, then gluing a target board on top and painting the whole thing. (i was driving them with bipolar power, like Tomix and Lego signals) The resistor was under the 2 pin socket glued into a hole.

 

ps: i also use these long pins to make screw terminal to tomix plug adapters

[mrp]

Hi All,

 

I’ve been working on the control interface for signals, accessories and blocks.

 

After a few false starts I’ve finally managed to get a basic prototype working, so I thought I’d post a few pics.

 

The system is made up of one or more MiniBase boards that talk to the PC automation software and can be linked together using CAN bus - plus a number of plug-in daughterboards for different functionality.  To start with I’ve made PWM driver boards for driving the tracks (PimDriver) and Tomix TCS interface boards (PinTcs) for plugging in TCS sensors.

 

MiniBase board:

 

 

PimDriver (PWM) board:

 

 

PimTCS board:

 

 

And here’s how you stack them:

 

 

To test it, I wired up a basic single track loop with a two track station.  It has:

- 4 independent control blocks

- 4 TCS sensors

- 2 automatic points

- 5 signals

 

Here’s a couple of very poor quality videos of it automatically driving two trains.

 

 

 

One nice thing is that the PimDriver board supports Tomix CL Lighting, so the train lights stay on while stopped at the station.

 

The MiniBase is connected to iTrain software running on a MacBook Pro using a FTDI USB/Serial converter.

 

Here’s a screen capture of iTrain running the automation.

 

 

[kvp]

Super cool and looks rather high quality! So if i understand correctly, you have one serial bus to the pc, another for debug text output, your original signal/accessory bus with the special pwm protocol and two can buses. Each board has a dedicated icsp header for programming the microcontrollers and a serial debug output (needed for software developement). I assume the can buses and/or the i2c bus and the power also goes through the board to board connectors on the sides. Which bus do you use to communicate between the boards (both in hardware and software) and what protocol do you use between the base board and the computer (especially what software protocol)?

 

For me this system looks a bit overcomplicated, but it's probably because i don't know the full capabilities of the system. I would assume a simple loconet like bus would be sufficient for quite a large layout and that would mean only a single serial connection to a computer through a dumb adapter board and then the turnout, sensor, pwm and signal/accessory driver boards would be connected with a single 3 wire cable between them. (power, ground, signal)

 

I would really like to know the design decisions behind your prototype and the full capabilites of your system.

[The Next Station Is...]

I like the look of this and the stacking board idea like the Arduino. Keep us posted with how you get on!

[mrp]

Thanks for all the comments.

 

Hi kvp,

 

I hope it’s not too complicated!  When you plug a stack together - and connect all the DC feeders and sensors - it’s all fairly compact.  The boards are only 5cm x 5cm.

 

 

I was thinking that each station area on my layout would have it’s own MiniBase board (with however many stacked PWM and sensor boards are needed) and they’d be chained together using the CAN bus.  Each MiniBase board has two CAN connectors so you can chain them.

 

I guess I could have made free-standing boards for sensors, PWM outputs, etc - but having them as daughterboards means I can just have one set of power supply circuitry and CAN interface on the base.  Plus it makes it really easy to prototype new plug-in boards since they don’t have to deal with power or comms.

 

The stacking connectors for the modules carry the I2C bus SDA and SCL, +12V, +5V, +3.3V and a PWM synchronization clock.  The inter-board wiring for the CAN bus has +12, CANH, CANL and GND.

 

I’m currently using the Dinamo protocol to connect to the PC - mainly because it’s supported by iTrain and is specifically designed for mixing DCC and DC block control.  Also, Dinamo supports OC32 Aspects - which makes programming multi-way signals a lot easier (instead of having to configure a table of on/off relay outputs).

 

I’ve never found an English version - but here’s the Dutch version of the spec:

http://www.vpeb.nl/files/1413/2770/6881/dinamo_interface_specificatie3_1.pdf

I just run it through Google Translate and then try and guess the meaning…

 

The commands I’ve implemented are:

- Unlink >> ( 011100B ) ( bbbbbbb ) ( 0000UZ0 )

- OM32 Output >> (0011MMM) (mmuuuuu) (commando) (parameter)

- Speed >> analog with polarity (010001B) (bbbbbbb) (PSSSSSS) [(AAAAAAA)]

- Kickstart >> (011110B) (bbbbbbb) (value)

- Link >> ( 011101B ) ( bbbbbbb ) ( 0000IPS ) ( sssssss )

- Pulse Solenoid >> (0010CMM) (MMMMMMM) [time]

 

[ToniBabelony]

I'm very very happy to see this project is still going strong! It's a very exciting and relatively simple system for what it can do. I can't begin to imagine the possibilities for small and large layouts!

[cteno4]

Mrp,

 

Brilliant! Nice work!

 

Jeff

[kvp]

Very interesting! Now i understand why i didn't find the power connectors, the power is in the CAN bus (which is very similar to the loconet i mentioned) and the stacked boards communicate with I2C. You said the board to board connectors carry the PWM synch. How do you synch between two stations? Imho the same problem is there for DCC mode too. Afaik both loconet and xpressnet has the RAIL +/- signals to keep boosters in synch.

 

The protocol you are using is interesting and i think you probably also implemented switch (0b10) and switch response (0b11) for the TCS sensors. I don't really understand the whole addressing system yet, but it must be very similar to what DCC uses, except it also has block addresses. Imho it looks like you are pretty much made your system software compatible with the Dinamo system from www.vpeb.nl but added the signal/accessory protocol conversion on an emulated oc32 layer. That's a pretty good idea!