I’m having a problem locating 3mm Bi-Color Red Green common anode LEDs for my HO searchlight signals. I have some that I bought about 20 years ago and they work great but all the newer LEDs have a hot spot on Red. My older LEDs work very good as a three color signal producing a nice amber/yellow with both the red and green on simultaneously.
I have purchased three batches over the last two years from different venders and all have the hot spot. I took a picture of one and I don’t think I’ve ever had this much trouble taking a picture. I set the camera with the lens fully closed and used two light filters and this is the best I could do.
If you look closely you can see the hot spot is very tiny but very visible.
Are you using different resistors values with the red and green side? Because you typically need a larger value resistor for red, equal current to both sides generally does not equal equal brightness, both due to the different chemistry and the perception of the human eye.
Dick Bronson has probably the best THREE color LEDs - he had them made with specific colors to match proper railroad red, green, and yellow.
Honestly I don’t see the hot spot - I see a small section at the very top that is darker, probably because it is in the shadow of the green part of the LED. Green is probably similar, witht he slightly darker spot 180 degrees opposite that for when red is lit - it just may be less noticeable because the green is more diffuse than the red in appearance. Once enclosed in a signal head and facing the track so it’s not directly end on visible I can’t image you even see that. But try a reduced current for the red side - maybe 1ma for red and 2ma for green.
I couldn’t get a good picture of the hot spot, I gave up after ten attempts. The hot spot is very dominate from almost 0 to 10ma. Very bright compared to the overall brightness. The super bright spot is coming from the junction against the side of the LED and appears to be about .02” in the defused .118” lens.
As I sweep from side to side the itty bitty spot looks bad, maybe something like 8 to 10 times brighter than the defused light.
The LEDs I bought back in the 90s don’t have the hot spot.
Do you have a link to Dick Bronson’s site? That sounds like the best solution.
Tony’s Trains has them in stock. I have a pretty good electronics inventory including six colors of 402 LEDs and a few 0603 RGBs so I’m going to give it a shot with what I have on hand.
I’ve come up with a soldering jig to help me solder itty bitty stuff. I came up with enough stuff to build up a few heads, I even have red, green, yellow and black Litz wire in stock.
Now that the big day is over and the little ones are somewhat contented and not wanting grandpa to put together or fix stuff 24/7 I’m going to try making a three color 3mm signal head. If it’s doable I have a lot of time and I enjoy doing this kind of stuff.
My wife gave me a fantastic tool for Christmas.
Now when things don’t go well I don’t have to go to the gara
I agree with Randy about the need for different resistors for the different colors. I built an Oregon Rail Supply signal bridge, but when it came time for the lights I couldn’t get them right until I realized that they needed to be in series, not parallel, because the internal resistance of the LEDs themselves would not let current flow through both red and green lights in a parallel circuit.
The newer LEDs work OK but the red hot spot really spoils the red signal. It isn’t a question of current draw with the red its a flaw in manufacturing. The red hot spot follows your eye when you move your head, the bi-color LEDs made in the 90s don’t have that problem.
I’m going to try to make my own three color lamp using 0402 SMT LEDs, if I can’t cut it soldering the itty bitty chips then I’ll go with the three color LED from RR Circuits. I’ve been using 0402 chip LEDs for yellow behind the bi-color LEDs and that wasn’t too bad even before I came up with the soldering jig to help my shaking hands. With the jig I do much better.
Maybe if you used some fairly fine sandpaper to rough up LED case a bit? I assume this is a diffused one and not a water clear. The water clear ones I’ve seen have very definitely visible spots right at the actual emitting junction. That’s probably the difference with modern ones - the point contact is more of a point than an area, so the light truly is coming from more of a point. Improved manufacturing techniques do not always equal a better product.
All of the LEDs I’ve tried are defused, but I tried roughing several up with some 220 grit sandpaper as you suggested and it really helped, thanks a bunch!
If I mount the LED with the hotspot on the top under the shield it is barely visible at direct viewing and not at all at about 15° up. Upping the current from 2ma to 3ma helped after roughing the lens up. Problem solved!!! Thank you Randy, the Forum Guru to the rescue again!
I attempted to take a picture but that’s beyond my expertise in picture taking. This is with the lens closed and three neutral density filters.
Comparing the pictures, the hot spot was not visible in either (I think direct view is just way too much for the camera to resolve without some even more extreme settings), however in the new picture, the red is more continuous without the flat spot at one point like the first pic.
The difference could be the first picture was a bare LED the second picture the LED is in a NJI housing. Both pictures have the three neutral density filters and the lens set to minimum.
I’m not a photographer I’m a retired electronic techie that love his trains, my pictures prove that.
I’m no photography expert either, all I know is that for any chance of the camera even showing what you see with the naked eye is going to need to be darkened a whole lot more. If the hot spot is noticeably brighter than the rest of the glow in the diffused casing, then the light level has to be reduced so much that the regular glow is pretty much blocked out, allowing only the bright spot to be picked up by the camera sensor. Smallest aperature, fastest speed, and lots of filters, I guess - anything to reduce the amount of light getting through. Probably beyond the basics which is where I run out of ideas. This is a pretty specialized thing to get a photo of something so specific because you are trying to photograph a light source itself, rather than a subject illuminated by a light source. Like trying to take a picture of the sun - hey, you wouldn’t happen to have a welding helmet, do you? That might be a dark enough filter. But the plain type, not those fancy new auto-dark ones that go clear so you don’t have to lift the helmet and then automatically darken when you strike the arc. The LED probably wouldn’t trigger the darkening. If you’re still interested in trying to picture it. I have a ton of cheap bi-color LEDs that came in various kits, I’m going to hook a few up and see if they have the same issue.
BTW did I read that right, you have the bi-color LED with a dedicated yellow LED stuck behind it to make the yellow? How does that fit in the signal housing?
I bevel the edge of the bottom of the LED by about 20° with sandpaper then use a drop of CA to attach it. I solder a single solid strand of #28 stranded wire between the Anodes then use four #40 Litz wires down the inside of the support tube. I paint the LEDs black with Crafters Acrylics, everything but the lens. The bi-color LED and SMT LED looks like the back end of the signal head.
This TrainCat bridge has Mel (early to mid 90s) bi-color LEDs with the yellow SMT, even looking close its hard to see the yellow SMT. I painted the signal heads black to prevent light leaks first then airbrushed the entire bridge.
This is a Mel signal. The cabinet is a Mel resin casting (NJI) with a K&S pole. The signal head is K&S tubing wi
Well that’s certainly an interesting DIY way to get 3 colors. Thankfully I don;t need to do that sort of thing (I think I’d spend the money and get Dick Bronson’s tri-color LEDs if I did), since my prototype used Type G 3 light signals, so I need discrete sets of 3 LEDs, never 2 or 3 colors in the same hole. So long as I don;t screw it up and put the wrong color LED in the wrong position.
I dumped a whole bag of LEDs (some day I will get ALL my parts sorted out into the small drawers I have) and every single one was a water clear, not a diffused one to be found. ANd of the first half dozen I tried, all were a single color (very obvious because the die is visible through the water clear case). As to which is what color - only way to tell is poer them up. With very low current (I set my power supply to current limit at 1ma), a green one looked like, well, a blob of green light - souldn;t really see the point of emission, the whole thing lit up green. But a red one - if viewed straight on, it;s a tiny point of red, with a slight bit of halow around it if not truly perpendicular to the eye. From more of an angle, it’s a solid red blob. So I’m guessing the hot spot appearance has to do with the way the point contact is made - red is a natural color, it;s just chemistry, but some other colors are really UV lighting a phosphor - except green normally isn’t, green is a discrete color like red, and should have the same issue. White and some blue shouldn;t do that, because they rely on the phosphor, although there still may be a hot spot with such LEDs, it would be in the UV and not visible. That’s why I had the idea of making the diffused ones more diffuse with sandpaper. The water clear ones are like magnifiers, so the point of light is very obvious. It seemed weird that this would happen in a diffused package, but my thought was that it wasn’t diffusing enough. Glad it worked out. I should see what happens if I rough up one of my water clear ones.
Your ‘Mel’ signals have inspired me to resurect a very old scratchbuilding project. I started to make my own searchlight signals quite some time ago, but for various reasons I dropped it. I have the signal heads already made and I have all the ladders, platforms and signal boxes to finish the project. Thanks for the kick in the bum!
I built up an Arduino Sketch that will drive the common Anode LEDs. It is a simple program and works on a truth table so it won’t do any fancy stuff, just drive LEDs from green to yellow to red for each signal. A switched ground occupancy detector operates the MEGA.
No detection and the signal is green, one detector, next block, and its red, advance detector, second block and its yellow.
Green = advance block high + signal block high = Green on
else green off
Yellow = advance block low + signal block high = Yellow on
else yellow off
Red= signal block low = Red on
Super cheap to build, one MEGA can drive 16 signal heads. I’m using two Arduino MEGA expansion boards for distribution. 64 wires per MEGA. I’m using the standard Arduino connectors and carrying the signal ground from the expansion board so that a four pin connector can connect each signal. Less than $20 per 16 signals.
You could probably run the entire layout off a single Arduino Mega - if you used something like the Microchip MCP23x17 port expanders. They’re about $2 each, and give 16 bits of inport or output, driven with either SPI or I2C, and you can but up to 8 of them on each SPI or I2C line. Unlike shoft registers they are directly addressable, so you don’t have signals flashing, or going blank while all the new data is shifted in. Like most things, there are libraries for these, so it’s easy to write data to them, or read them - a pair set up for input would give you connections for 32 detectors with just 4 pins of the Arduino used. Another 4 pins and you could have up to 128 LEDs driven. Outputs of the 23x17 are low current, ok for LEDs, but you could do the same with your lighting controller and have the 23x17 outputs drive the high current drivers to control light bulbs. Heck, I think with a bunch of 23x17s and the drivers, one Mega could run both the signals AND your structure lighting. There’s enough code space, enough pins to run multiple I2C or SPI buses, and it should be plenty fast enough. I’m using those 23x17s on my psuedo-CMRI control nodes, and the cpNode from MRCS uses them for their IO expanders.
If it helps anything, in a previous life I addressed the issues with older bicolor LEDs driven to produce a range of orange/yellow in AC (the specific application being safety repeater boards for nuclear powerplants in ITU gr.oup R10)
The “AC” drive to the LED has to be asymmetrical to control both the apparent color and brightness issues in ‘yellow’. We did this in time domain in order to retain perceived isoluminance when either diode core was lighted with straight DC through appropriate resistance. Doing this at the desirable high frequency is much easier and cheaper than it used to be!
The LED face, whether domed or lensed, needs to be made ‘diffusing’ after the manner of an averaging photo lightmeter dome. As I recall, something like Testor’s Dullcote applied over solvent etch of the plastic face of the diode encapsulation would serve; this might be easier and certainly less time-consuming than trying to fine-sand especially once a signal head had been assembled.
Thanks Randy, now I have another goodie to put on my plate.
Right now the MEGA does what I need, I only have a total of 14 blocks. I do appreciate your input, I wouldn’t be getting anywhere without it.
Presently I have not only a full plate but an overflowing plate. I have so many projects on my bench I can’t even see my bench. Almost all of them are Arduino related one way or another.
Thanks for getting me going with the NANOs, I’ve found stuff to do with them too. My favorite is the UNO mainly because I can install them in my Mel card shelf easily. I did install a NANO on location in the bottom of my communications tower module to realistically flash the tower beacon. With low room lighting it looks like the real thing with its slow b
Just for kicks I made a full size drawing of a Mel made 3mm three color LED using 0603 micro LEDs.
I’m planning to make a couple to see if its worth the effort to make them.
I have hundreds of diffused 3mm LEDs so using my 5” side flush cutting wire cutter I wacked of the upper lens at the diode junction leaving me with a 3mm diameter lens for the 0603 LEDs. I glued the front of the lens to a 3/16” square piece of basswood using Amazing Goop, it can be easily removed from Goop without damaging the lens. After sanding the cut end of the lens it left me with a 2mm long ready to go lens.
I don’t have any 0603 LEDs in stock so they are on order.
