Hi Everyone

First OMWB for me on RMweb after lurking around the forum for a while incognito!

So, I‘m in the process of building my layout after a hiatus from the hobby. As a kid, I had a couple of those chunky metal colour light signals on my layout and decided I wanted something a bit more realistic this time (managed to finally expand my budget beyond 90 pence per week!) I am planning on modelling a modern image terminus with a complex approach so theatre indicators were a necessity and of course I wanted them to work... but a quick search revealed nothing on the market from the usual signal manufacturers, so I decided to take the plunge myself and try and build one.

This proved to be easier said than done of course… but it was a rewarding project!

After a bit of research on good old RMweb I found reference to a couple of different approaches – some folks on this forum (with very advanced electronics knowledge!) have used LED dot matrix displays to great effect. However the skills needed to design the circuit are beyond my expertise, and sourcing a small enough display for 00 scale that uses white LEDs proved virtually impossible, so I abandoned that method.

I had heard the idea of using fiber optics discussed before. The simplicity of this approach means zero wiring at the signal head itself save the trailing optic fibers, and you can hide the guts under the baseboard. The “guts†would be a diode matrix connected to LEDs to power the array – with one input per character that lights the relevant LEDs to make each letter or number. The intention is to attach it to a DCC accessory decoder for computer controlled operation.

It became apparent early on that there were going to be a few problems. Fitting the fiber optics need down the signal post along with the other signal wires was one. And the thinner the fibers you use, the less light will get transmitted. So there is a compromise to be decided upon here.

After some testing, the 0.25mm fibers seemed to offer the best results. You can buy these fibers from specialist fiber optic shops or on ebay – either way they are pretty cheap. Generally these are bought in a multi-core cable and you strip away the plastic exterior with a hobby knife.

So here we go – along with the fibers, you will also need superglue, 10 and 40 thou black plasticard, clear stick-on inkjet media, warm or cool white LEDs, signal diodes (lots!), electronic stripboard, a pin vice drill and 0.25-0.3mm bit, and various widths of heat shrink tubing.

Building the display faces

Actually the trickiest part of this project came first. The display itself unfortunately required some compromise in model form. Real matrices use a lot of lights (64?) and there just isn’t space in the signal post to carry that many optics. Designing a diode matrix to operate that kind of variance would be quite a task as well. So I decided to go for the good old fashioned “8†display format – using 18 individual fibers. Note that this arrangement won’t display all characters (eg “Mâ€), but you can of course modify the layout of fibers depending on what you want to display.

I used 40 thou plasticard for the face of the indicator and cut it into 7mm squares.

Next I knocked up this template on the computer to use as a drilling guide. I printed this out on permanent clear adhesive film on a regular inkjet printer – and superglued this to the face of the face of the indicator. Printed out actual size it should measure 7mm square. (PM me if you want the actual PDF / Illustrator file to print out yourselves).

Here comes the really tricky bit. Angling the indicator so you can see the white-out area of the dots, you need to drill out the holes using a 0.25-0.3mm micro drill. If you’re wondering why I used clear film printed black, attached to black plasticard – I didn’t want any white showing through in operation, and if you angle the assembly when piloting the hold you can still see where to drill. The drilling is the trickiest bit of all. The problem is that the slightest movement in the wrong direction – or a slanted drilling angle – means that the fibers will point the wrong way and mess up your neatly formed characters. This was pretty frustrating and I probably had to make twice as many indicators as I needed, and then throw away the rejects. I spent a lot of time on this part and found that even on the best examples, I could not avoid a little bit of variance. However there is a way to correct this a little when you come to attach the fibers.

Note on this part: do not attempt after beer.. and keep a couple of extra drill bits in the same size as they are very fragile and will break from time to time.

Attaching the fibers is relatively easy. I threaded the fibers (1 foot long each) through the holes in the displays, leaving about 2-3mm poking through on the other side. (Remember, fiber is cheap so make them longer than you need and you can always make them shorter once mounted on the baseboard – not so easy the other way around!) Once all the fibers were threaded, I applied liberal amounts of superglue on the back of the display to cement the whole assembly in place.

The reason for leaving the fibers poking through is that you will be able to re-direct these by bending or gluing them on the other side to correct any slight variance in position that may become apparent once you test the display. If you cut them flush, you will have to throw the display away if it proves too wonky when tested.

Next – hold the fiber ends up against a light and make sure they are all transmitting light through the display face as they should. Adjust any by bending them and applying a very small amount of superglue to hold them in the new position if they are out of alignment. If all looks good, then the display assembly is complete apart from the finishing touches.

Next we come to the electronics part - the diode matrix.

This test matrix will need to be able to show 4 characters - “A B C D†. On my fiber arrangement simulated on the computer, these look like:

Not quite like the protptype, but not too bad a representation in theory. Next step is to work out how to build the matrix. I drew up a logic table which showed that 9 LEDs per indicator would be needed, and worked out how these could be divided into clusters of fibers to make each digit. This will give the logic needed for the diode matrix.

Here is a diagram of the logic table. Note: each colour represents a different “cluster†of fibers (some LEDs have only single fibers, some have several). The columns in the table on the right show which clusters or fibers need to be lit to display that character.

Now to make the circuit boards. I used standard stripboard, cheap ‘n’ cheerful signal diodes, and warm white 5mm leds (since some prototype photos from my chosen period seemed to show warm white lit indicators). For newer layouts I guess a standard cool white LED would be more prototypical.

Translating the logic from the table above into a circuit board was fun. Once I had that part cracked, I knocked up a template to print out and overlay over the PCB to reduce errors in placing the components. Some hacking of the stripboard is required – the LEDs are oriented “with†the tracks, so you need to break the tracks in between the anode and cathode to avoid a short circuit. You also need to bridge between the LED anodes with wire (or be lazy like me and use lots of solder) since these will be connected together (the cathodes provide the switching since it will be DCC controlled).

The yellow lines running vertically in the circuit diagram show the input tracks for each character, and at the bottom there is a terminal attached to the decoder output. Really all this circuit does is bridge (using diodes) between the yellow line input for that character to the blue lines corresponding to the LEDs that need to be lit. Note the orientation of the diodes must be correct for this circuit to work. The black lines on the diode symbols on the plan should match up with the black band on the signal diodes themselves.

The common positive is shown running along the top. Copper strips on the board run vertically.

Once all is soldered up, check that the LEDs light up as they should when you apply the negative lead to the relevant input. If so, congratulations since the most difficult bits are now completed!

Finishing touches

Making the hood for the theatre requires 10 thou plasticard, which can be taken from a 24mm x 8mm strip, with the parts that will be the two vertical sides chamfered as per the real thing – this strip is then glued and folded around the display assembly.

Attaching the display to the signal post can be accomplished via your preferred method. It is probably easiest to attach above the signal head, but my examples are offset to the side of the signal heads, above a PLS. Note that these fibers can be bent around a pretty small radius, but they do like to spring back!

Note: actually “folding†the fibers is not recommended however these newer fibers are not glass, so they don’t break if you do fold them. The light carrying properties may be compromised a little if you do this – it is your judgement call if you think it is needed for cosmetic reasons (i.e. to conceal the fibers a little better at the back of the signals). Personally I don’t think the bunch of fibers is too noticeable since they are clear. You could also paint the fibers grey and it would not affect their light carrying capability.

I attached my displays to the signal assembly with superglue, and had to hold them nice and straight for a minute or two while the glue set. Here’s a couple of photos of the finished assembly (couldn't resist testing out the PLS as well!)

Next you need to identify the fibers and bunch them together according to which LED they will fit to. Label the fibers/ bunches 1-9 along with the LEDs and attach your signal to the baseboard.

Our final hurdle is how to attach the display assemblies to the LEDs. After many lost hours spent trying to drill micro holes in LEDs, I came across the method on the internet of nesting different thicknesses of heat shrink tubing. This seems to work very well and is fast and cheap to accomplish. Better yet, it is much harder for the fibers to “jump†out of the LED assembly even if they’re not permanently glued in (mine are not as I may need to remove the signal for maintenance at a later date).

I bought a variety pack of heat shrink in various sizes and fitted the largest size over each LED. I then “nested†two smaller sizes inside each other, and placed them inside the larger tube over the LED. I shrank them all with a lighter which gives the result shown on the board picture earlier in this blog. Apologies but I am unsure of the size of tubes I used – this took a bit of experimenting.

Connecting up the fibers to the relevant LEDs means just inserting the fibers into the top of the shrink assemblies until they contact the LED. You can superglue clusters of fibers together and number them with tape for easier re-connection once you mount the assembly on your layout.

Here is the picture again of the whole assembly: the green board is my accessory decoder.

Now we are nearly ready to test. Just the finishing touches to apply:

The real theatres seem to have an orange tinted glass on the front, probably to reduce glare I guess. Using Lee brand filters meant for stage lighting, you can simulate this. From a Lee swatchbook I selected Filter 238 – “CSI to Tungstenâ€, which is a light orange. I cut this out to the correct size and mounted it to the face of the matrix display. To increase the size of the dots and hide the imperfections and the fibers themselves, I underlaid the orange filter with another square of diffuser material – this time filter 414 “Highlight†which did a good job of softening the light.

These filters are specialist items but you can get the swatchbook from stage lighting retailers for free or for a small charge. They are also sold on Ebay from time to time. The swatches are small but more than enough to make 30+ theatres from the sheet you get as a sample.

Ok, hard work over. Now time to connect up the fibers and test the signal!

Not perfect by any means but quite pleased with the result.. now to the production line to turn out the other 9!