Building Signals, 8 Single Mast search light signals complete and painte

I was reluctant to buy anything from China, yet I had a good experience buying ATSF style cantilevers on (you-know-who) from a seller known as "we honest". 

https://www.shapeways.com/prod...ea?optionId=61718713

Rob Leese posted:

I was reluctant to buy anything from China, yet I had a good experience buying ATSF style cantilevers on (you-know-who) from a seller known as "we honest". 

Rob, why are you afraid to say eBay? It is not a dirty word, even here on the forum. Just say it, Voldemort!

Mike, Wehonest has a few things you might like. They do make a pretty decent cantilever signal bridge. What I buy from them is their searchlight targets.

This is a modified Lionel 450 signal bridge, with the Wehonest targets, 1/8" brass tube,  and 5mm bi-color LED's.

I'm using Bruce Chubb's  Searchlight Signal Driver cards to get 4 aspects, dark, red, green, and amber.

You can read more here.

Rob and Elliot, thanks for the replies.  i have a decent plan for wiring this, the signal positions and logic, but this project seems huge to me as I have a lot of unknowns for relays, signal driver approach, route checking logic, signals them selves, etc...  

Rob I ended up ordering some brass angle along with some really small wire (awg 30) and I am going to try my hand at making one, depending on how rough it comes out, I may check out the we honest one.  From the on-line pictures, I could not tell if it was cast or if it was brass?  Do you have close up of yours.

Elliot, I will have to follow up on the heads.  Curious on your signal driver circuit, once you buy the board from chub, and populate with components, what does it cost a piece to control one head.  Do you leave board uncut and wire back to it, or do you cut circuit board and locate it locally?

thanks,

Mike

Elliot, any idea on the cost per head for your signal drive boards once assembled?

I decided to try to build my own cantilevers.  I have made a few compromises on it when I drew up a set of plans based on some internet pictures.   I just finished one of the cantilevers, but with no signal heads yet.  The soldering is some thing I am working to get better at, much different than soldering wires.  I am sure there is a better technique to this, but I do not know it.  In the end I will use a polishing wheel to help me clear up the access.   The arcs made our of brass have a wave in them from being bent that I am not sure how to prevent and hoping they will become less apparent with flat black paint.  I did order some 30 gauge wire and hope to make some progress on signal heads tomorrow.  Here are the pictures I model it after and also the plans and progress thus far.  In the end, I plan to make two of these and also 3 track signal bridge.

I am also looking for information and pictures of how (or if) the sante fe had triple heads on cantilevers and if they mounted all heads above it or moved one below.

Thanks,

Mike

To answer one of your questions, try this link:  https://ogrforum.ogaugerr.com/t...67#63513879927057267

Hi Mike, sorry for not replying sooner. I almost missed your post the second time. One of the best ways to get my attention is to hit the "like" button in the lower left corner of my post. That way, I get an alert and an email and I can reply much sooner.

Chubb has a couple different boards that handle searchlight signals. The one I'm using does 12 LED's. There isn't a good way to cut them apart, because the traces on the back are slightly interwoven with their neighbors, so you can't just do a nice straight cut. The good news is, Cat5 cable is very cheap and convenient for going from the board to the signals in the field.

Furthermore, signals have a tendency to occur in groups of 4, which is perfect for a single cable. Example: single track main entering a passing siding, facing point, one mast two heads. These heads will indicate route and speed for an approaching train. Then there will be two masts with a single head, one for the siding and one for the main. Ta-da, 4 heads, one Cat5. Repeat at the other end of the siding. It's not always that neat and clean, but planning in groups helps you place your boards.

Here's a link to Chubb's page.

I'm using the one called SSD. You might be more interested in the RSSD.

Cost: The unpopulated cards are around $1.50. Components, I forget, but it isn't too bad, maybe another $2 per.

All of that said, you still need some kind of processor to tell the inputs what aspect to display.

Do you have a track plan I could look at?

 Thanks Rob,

I did not see the reference to triple heads on the same mast for the Santa fe style cantilever, did I miss it?

Thanks Elliot,

I have been working on my plan (still working through a rough draft, but the intent is there) for the signal system where currently I have relays providing block indication (3 aspect) and then plan to use an IC (maybe an Arduino) to check routes at each of my control points when a route request is made (logic listed at the bottom of page, needs to proof read still).  I have not added a stick circuit yet to prevent lining into routes from alternate control point, but between control points of east and west compromise will be CTC and outside of that ABS, with the exception to the hidden siding on track 1.

I have tried to add my power point as an image and it does not come out well, so I added the original power point as well which I use as a drawing program.

Thanks,

Mike

Well Mike, it doesn't look like you need a bazillion heads, a couple dozen looks like it'll get the layout done. That would be just two Chubb SSD cards. Locate one near both end with your processor near the middle. The SSD inputs need a DC ground to be activated.

I am making more progress on this effort now.  I convinced myself it would be better to build signals first and then come back and take care of the signal logic.   After building a few of these, my standards have gone down on the amount of detail I am willing to add to the signals I plan to use.  After originally trying to model after a specific railroad has changed into a  just search light theme signal system (as opposed to the Santa fe or SP or BN, etc..)

I am also now convinced the search light signals are the easiest to build.  For the search light signal shield I ended up cutting circles out of styrene and then turned them on my power drill to get the as close to round as possible.  The brass brows had an internal diameter just hair tighter the 5mm, so I will need to sand each of the LED's just a bit to them to fit in.  I have a comparison with an atlas signal below.

For comparison, here is the side profile of a real search signal.

I finished one of my cantilever signals (still have 2nd one that needs to be built).  After talking to a few different people on 3 heads vs 2, and came to the conclusion Sante Fe only used 2 heads and I will make them work for all diverging routes. 

I decided to work on the two signal bridges I needed (3 track and 2 track).  Again, there are so many different images online for what is real, I found it very tough to come up with a standard.  I was really surprised that several of the signals bridges had full size signals (base and mast) mounted on them.  Pictured below is what I ended up with.  I really like the idea of the higher heads all being the same the height.  I used two plasticville bridges combined with brass and styrene.  I thought it came out nice.

I also completed my 2 track signal bridge.  After getting done, found my heads were not mounted symmetrically.  I still think it looks good and plan to use it.  I attached an image off of the internet of what I was using for a prototype.  I also used the remaining vertical structures from the plasticville signal bridge.

One cantilever signal to go and then locating ladders that look more scale for each of these structures.  The plastruct ones look too thick and hoping to find better ones.   Ideas on where I might purchase these ladders would be appreciated. 

Also thinking I will need to rethink some of my signal logic plans.  More fun in coming weeks

Thanks,

Mike

Making more progress on this now.  I have spent over a month trying to figure out how to light the signals.  Originally I planed to use  a 5mm RGB LED and attempt to blend the red and green for a yellow aspect.  The results were not great.  At some angles it looked yellow, other angles red yellow, other angles green yellow.  I also tried sanding the LED finish to have the colors refract more with in the LED, while this helped, it was still not a solution I was happy with.  I decided to start down the road of the atlas signals with three different color LED's.  I first experimented with 603 size LED's (1.6mmx.8mm) and was successful with them, there is a picture below for reference on their size.  Look for the white spec to the right of the red LED on masking tape.  Itty bitty!

I also decided to try the next size down which is 1mm x.6mm, really itty bitty.  While I could solder them separately, I could not solder them close enough together with my soldering iron with out having issues holding them.  Funny, as the first one I took out of the package i lost as it flew across the room.  These are the size of a piece of glitter.  Both of these types of LED's are surface mount and really designed for a pick and place robot at a board shop to mount/solder.

I spent a lot of time looking for the right colors as the yellows and greens never seem to look right (even from atlas).  I cannot just buy these items locally, so I have had many orders to various electronic supplies to find the the right set.  I used my Search light relay for a color reference and while successful at matching the yellow aspect, I have had to settle on on  incorrect green color.  

In the end I am using 585nm for yellow, 525nm for green and 640nm for red.  Interesting as I was doing this search, there manufactures that produce RGY LED's in a small package.  The colors looked off to me, but might be an option for others.

I was able to to solder the three LED's very close to each other and insert in my search light head.  I am happy with the results.  

Looking forward to finishing the 2nd Cantilever and getting them all painted and mounted on the layout.  I have also decided to attempt at building the single mast signals.  I may change my mind as I go down this path, but I am liking the idea of building them myself and accepting they will not be as nice as the atlas ones    It will probably be a few months before I finish the signals themselves and start integrating their functionality into the layout.

Mike

Hump Yard Mike posted:

 

 

In the end I am using 585nm for yellow, 525nm for green and 640nm for red.  Interesting as I was doing this search, there manufactures that produce RGY LED's in a small package.  The colors looked off to me, but might be an option for others.

 

Looking forward to finishing the 2nd Cantilever and getting them all painted and mounted on the layout.  I have also decided to attempt at building the single mast signals also.  I may change my mind as I go down this path, but I am liking the idea of building these all of them myself and accepting they will not be as nice as the atlas ones    It will probably be a few months before I finish the signals themselves and start integrating their functionality into the layout.

Mike

You are a little high on the green.  I looked for something closer to 500~510nm but I could not find smt in stock at Digikey or Mouser.  Only thru hole.  

I appreciate your efforts.  The LEDs I am currently using at work are 12 mil square.  That's 0.3mm square.  Makes glitter look big.  I'll keep looking.

Lou N

Lou,

If you are able to locate the green led in 1.6x.8mm, please let me know, very interested.

I spent a lot of time working on finding it with no success and I am about to start all the led builds, would love to have the correct green.

Cantilever Signal finally mounted in place.  This is starting to get exciting!

When looking for green led's most were not a darker green but a very light shade. Had to refine my search looking for "pure green" like these. Pure green smd

Thanks Nelson,

How do figure out the specific color (or wavelength) from the site you sent me?   I think I am looking for a more aqua or blue green color.  With out seeing the 510 that Lou describes above, I think he maybe close in describing the right wavelength.

Mike 

Nice work Mike,  thanks for sharing the details and photos.... I plan to kit-bash two Plasticville signal bridges together soon, so it was timely to find this thread with the photos.....   

Mike the closest smd to 510 I was able to find was a 515nm and was listed as "Bright Green".  It was described as a more amber-green. Bright Green smd this is listed as 515-520

Adjusting the resistor value will also shift the color slightly to how the eye views it.

I think creating the prototypical green color for signal heads modeling incandescent bulb lenses will be a challenge.  LEDs produce a relatively monochromatic or thin/narrow spectrum of light ... tens of nm wide.  Whereas a broad spectrum incandescent filament filtered by a green tinted lens emits a much wider/broader spectrum.  This research paper pre-LEDs is kind of interesting:

I suppose the logical approach would be to find a green LED wavelength that is centered at peak wavelength of the tinted green lens.  There's probably even modern research papers describing how the eye/brain responds to narrow monochromatic LED signal colors (red/yellow/green) vs. traditional lens color...and how it affects traffic accidents, etc.

It's interesting there is the term "Admiralty Green" to describe the signal color for green.  I've seen green LEDs called "Pure Green" or "Emerald Green" or even "Traffic Signal Green"... but never a green LED called "Admiralty Green".

I see this as kind of like trying to model incandescent headlights, passenger car lights, etc. using warm-white LEDs.  Again, the spectrum coming out of the LEDs is fairly narrow in spite of tricks to spread or broaden the spectrum of the underlying blue LED.  Hence the choices by some guys to use yellow LEDs because it renders (to their eye/brain) a better incandescent color - clearly a your-mileage-may-vary situation.

Nelson,

thanks for the link, the lead time looks to be pretty big, I will look to see if I can find it elsewhere.

Stan,

Thanks for your comments, I can see where you are headed.  The current greens led I have been using are no where close and believe could be much better if the right wavelength was available.  I need the size of an led to fit in the head, but you got me to thinking about your incandescent approach using a fiber optics path.  For now, I think i am committed to the led approach at this point.

Mike

I've never heard of any modeler crazy enough to try this, but since you seem to have the skill to assemble 0603 (1608 metric) LEDs I wonder if you could use 2 surface-mount LEDs (one blue, one green) to provide a broader spectrum of green.  This would work better with the fiber approach to provide a blended output so you don't see two distinct points-of-light so to speak.  OTOH if you've ever fussed with an RGB LED with its three closely spaced LEDs it is remarkable how the human eye/brain blends the 3 distinct wavelengths of light into a single "color" given the remarkably short height of the lens. 

How about one of the three-chip LED's and just blend the color to suit what you're looking for?

NelsonW posted:

When looking for green led's most were not a darker green but a very light shade. Had to refine my search looking for "pure green" like these. Pure green smd

The spec sheet says 525nm which is a bit away from 510nm which is really called cyan.  I can get 510nm T1 thru holes all day long but this SMT thing is quite a problem.

Lou N

gunrunnerjohn posted:

How about one of the three-chip LED's and just blend the color to suit what you're looking for?

I've got a PIC circuit for driving RGBs thru the visible spectrum.  But now it becomes having this circuit per LED.  Would probably require a digital pot to drive correctly in a digital format.  

Lou N

One more thing Mike,

Are soldering the LEDs with wire solder or solder paste?  Paste is much easier.  Contact me if you need some.

Lou N

Lou, I'd consider just tuning the resistors until I got the desired colors, no intelligence.

As for the paste, I had some, but what I had needed to be refrigerated and didn't last but three-four months, it seemed more trouble than it was worth.  What kind of paste to you use?  How do you melt it, with hot air?

gunrunnerjohn posted:

Lou, I'd consider just tuning the resistors until I got the desired colors, no intelligence.

As for the paste, I had some, but what I had needed to be refrigerated and didn't last but three-four months, it seemed more trouble than it was worth.  What kind of paste to you use?  How do you melt it, with hot air?

I have Amtech 62/36/2 and Indium 63/37. These are half full jars pulled off the line because they were close to expiration.  So far they keep well in the basement. I use a fine tip temp controlled iron. Seldom hot air. 

Want some?

Lou N

I'd try a small jar, got instructions?  I assume just cover the pad with a little and touch an iron to it, right?  I've watched the boards go through the IR oven, but I've never used the paste myself.  Of course, they put it on with the mask in one big swipe.  However, I might make prototypes go faster if you could just position the part and then solder it.

For hand work we use a syringe. 

Sent a private message for your address. 

Lou N

Lou,

i have never used paste solder before.  I would be interested in trying it.   Curious the method you would suggest with it?   My current method is to put the three LED next to each other (lens side down) on double sided tape where the anodes are just touching but the cathode sides are fanned just a bit to allow them to be soldered separately.  Apply A little bit of flux to the pads and get after it soldering all the anodes at once and then cathodes one at a time.  I use a fine tip on my hakko soldering station.

John,

as far as RGB led, this was my original plan with a 5mm LED.  I did work at tuning the yellow manually through varying the current on the green and red.  While I could produce a decent yellow head on, as one would move their view point, green or red could seen.  This is the reason I headed down the road of specific color LED's.  As I struggled with it, and looked to other's experiences with it, many others also had this as a common problem in trying to model search light aspects using the RGB LED.  Using this approach, it seemed like you would need an array of them to produce a convincing color.  As I researched more about the RGB led approach in general (not toy trains), many folks talked about the right approach being to  vary the duty cycle on each color and not changing the current. I am not sure if that approach produces better results as percieved by the brain.  Curious if you or others have information or experience on this for the single RGB led application.

i appreciate both of your responses.

Mike

I haven't done exhaustive experiments on these, it was just a different idea that popped into my head.   I have a strip with a controller that has the three-chip LED's, and I can get a pretty decent range of each color by tweaking the settings.

Hi Mike

 Your work on these signals is outstanding to say the least !!!!! I hope to see you at York

Thanks, Alex

Your signals are looking really good, Mike.  You have the patience and the skills for this, and they are really turning out nice.  

I think you said you were showing an Atlas head but does your model have the depth so that you can place the LEDs further away from the lens?  And are you actually using a lens or just have the LEDs exposed as the current photos show?

Placing the LEDs further back and using, say, 5mm clear acrylic/plastic rod should solve the issue of seeing two distinct points of light where the source of the light moves depending on which LED is on.

So given your determination to get this just right, are you able/willing to go the microcontroller route to control the LEDs?  Two ideas here.

1. So far the discussion seems to be about find the magic ratio of a blue LED and green LED to get the right shade of green.  But if you have a microcontroller driving the two LEDs, why not sweep the colors in real-time (faster than the eye can see).  In other words rather than having a single blended color for green you can sweep the green toward blue then toward green...hundreds of times per second (faster than the eye can see) so that you broaden the spectrum of the green.  Not clear if I'm explaining the concept but this is well within the capability of a, say, 50 cent microcontroller chip.

2. If I understand this type of signal head, you are modeling a moving lens over an incandescent lamp.  I haven't seen a video of it in operation but presumably as the lens moves the color dims and perhaps even goes dark until the next lens rotates into place?  If this is how they work you could also model this behavior with the same microcontroller chip.  This would be difficult to do at the discrete component level with resistor, capacitors, etc.

Stan,

Great suggestions, i agree with your thoughts on the bigger cavity to reflect out of.  As I worked through this, I decided to just use a tube for the whole assembly as one of the compromises to make it easier to assemble and not visible from the front.   I was not sure on how to create it with my limited tools and skills.  I have shown a close up of the back below, one with the shield on and one with it removed.

i do have plans to use a pic for a time delay in aspects and simplify the Home and Distant block aspect triggering.  Once all the signals are built, that will be the next hurdle.

Not directed to Stan, and a bit of rant.  Between work, a young family, and limited time, trying to find the time balance on what is reasonable to tackle on my own, what is easier to just buy, what is not as critical to be modeled, all while being able to see progress on the layout itself ( tend to be impatient if visible progress is not obvious) is a constant battle for me.  I am curious who else adopt this approach and struggles with trying to Create their vision of perfection from their imagination and then downgrades as the execution process starts.  This is opposed to doing what ever it takes to accomplish the vision.  I think this comes down to being high railer (not true scale) and accepting none perfection such as tight radius, lobster claws, short passenger cars, third rails, huge rails, none fixed pilots, etc... 

Mike

Hump Yard Mike posted:

Lou,

i have never used paste solder before.  I would be interested in trying it.   Curious the method you would suggest with it?   My current method is to put the three LED next to each other (lens side down) on double sided tape where the anodes are just touching but the cathode sides are fanned just a bit to allow them to be soldered separately.  Apply A little bit of flux to the pads and get after it soldering all the anodes at once and then cathodes one at a time.  I use a fine tip on my hakko soldering station.

John,

as far as RGB led, this was my original plan with a 5mm LED.  I did work at tuning the yellow manually through varying the current on the green and red.  While I could produce a decent yellow head on, as one would move their view point, green or red could seen.  This is the reason I headed down the road of specific color LED's.  As I struggled with it, and looked to other's experiences with it, many others also had this as a common problem in trying to model search light aspects using the RGB LED.  Using this approach, it seemed like you would need an array of them to produce a convincing color.  As I researched more about the RGB led approach in general (not toy trains), many folks talked about the right approach being to  vary the duty cycle on each color and not changing the current. I am not sure if that approach produces better results as percieved by the brain.  Curious if you or others have information or experience on this for the single RGB led application.

i appreciate both of your responses.

Mike

 

Mike,

Contact me at: louis1951@windstream.net or text at 216-990-2016 if you want to try solder paste. 

Lou N

Alex and Tom,

Thanks for the nice comments.  

Alex,

My dad and I will see you at York.

Making progress on the 2nd cantilever, shown below next to the first one.

As of today I have completed the signal construction.  Show below is my 2nd cantilever.

For the two signal bridges and cantilevers, I still need to add ladders and wire conduit paths.  I am excited to get these last few items wrapped up and then get them painted.

For the single mast signals (8 of them), I built them all myself from scratch out of brass except the grates which I used the lobster cage brass kits for.  The lobster cage kits are not cheap, and I trashed most of the material anyways. I wish I new a source to buy the etched brass from as stock sheets.  The shield are made of styrene.  My first one is shown below.  

After I completed the first one, I bent the remain rails and plat form supports for the remain 7 signals off of a pattern.

Here are the unpainted brass structures.

Here they are painted

Creating the light inserts and building the lamp boards will be the next effort.

I wanted my signals to match the approach I took in making the signal bridges and cantilevers. In the end I thought they came out nice, but continue to appreciate the value in atlas signals.  The atlas ones  have much more detail and cost the same or less when compared to producing such a small run of signals.  The little things seem to add up, like the big role of chicken wire I use for the ladders, styrene sheets, pricey grates (pieced out lobster cages), etc...

Looking good Mike. Nothing like seeing the progress your making with your signals.

Thanks for nice comments SGM.

A quick update on my progress. I finished construction of the rest of the signals (have not gotten to the LED's yet).  

 Just about wrapped up with my signal wiring plan (attached) and about to start the arduino project for clearing the routes.  I ordered a couple of components for shift registers to help with the 40+ I/O I need.  This is all new to me, but I am sure old hat for many of you.   I would like to check the box that I can have the shift registers work as opposed to buying the larger arduino mega.

In the end, while this will be greatly slimmed down CTC version of the real thing, it will provide route specific aspects and not allow conflicting routes to be cleared.

Notice the east and west control points servicing the town of "Compromise" (aren't all of our layouts), .  Hidden junction only has the double headed search lights for facing point moves as the trailing points are covered up in the mountain.  Also, the logic here cheats a little because of this, but still will not allow a conflicting route to be cleared.

My plan with system as a whole is to have physical relays steer the home and distant block indications for both west/east directions and then have the signal requests (EZ and WZ representing east and west signal requests) and switch indications (reverse, SWR and normal, SWN) come into the arduino with logic equations controlling relay outputs for east and west clear relays (I call them ECLR and WCLR)  to clear the signals.    The logic equations I am going to use are written out on the power point.  

If I can get this to work in short order, I can then focus on the circuit logic at each signal.  Hoping to have some progress to report on this soon.

In being transparent, I under estimated the time and effort needed thus far for this project.  I thought I would be completed by now, but in fairness, in the beginning I never planed to build my own signals and populate them with LED's.   It has been quite the adventure!

Mike