I have used T-1 size LED's for signals of the position light and of the color light type with very good success. What I want to do now is use a single T-1 bipolar (red/green) LED and use diodes to limit an AC current flow to one direction (red) or the other (green), and then allow the full wave AC to give a yellow indication. Thus I will have a single light source that can be used as a red/yellow/green searchight signal.

The problem is that such LED's do not always make a good clear yellow when both the red and green LED's are "on". Has anyone experimented with various makes of such LED's for this kind of signal that produce a good yellow??

Thanks for any helpful hints.

John Johnston

-- John Johnston (johnjohn@vicon.net), May 14, 1999

Answers

John,

The general problem is that red LED's are much more efficent than green ones. I've never seen one that the red and green were balanced. If they were, the manufacturer would have to tone down the red to make it balance with the green. If someone does know of such an LED, I hope they will contribute their info to this forum.

In general, you have two choices. The easiest is to use different values of resistors for the red and the green. Use a higher value of resistance with the red to tone it down.

For people using a two lead LED where polarity needs to be reversed, do this:

------|<|---- resistor for red ----- > to bipolar LED. | | --|>|---- resistor for green -

I hope this crude diagram is clear. Obviously, I need to add this to my web page. I'll make a better diagram when I do.

Your other choice is to use a nonsymetrical waveform that has a duty cycle other than 50-50. This approach, while technically more sophisticated, is probably not for the average modeller.

-- Allan Gartner (wire4dcc@aol.com), May 14, 1999.

You can adjust the current through the red and green parts of the bi- color LED to yield a composite color, like yellow, as previously described. The three-lead type (almost always common cathode) allows you to light up both parts simultaneously with simple DC current drivers (read - transistor or contact switch and a resistor).

However, this is only 1/2 the problem. The simple low cost clear or frosted lenses don't mix the colors well. As a result, it still looks like two separate red and green lights spaced close together in the same housing. I have found only a few sources for a bi-color LED in the T-1 (3mm dia) package size. Most of them are in the T-1 3/4 (5mm) package. This is oversized for HO or smaller scale.

One thought that occurred to me is to mount a T-1 3/4 bi-color LED below the signal base and to use a light pipe to plumb light from the center of the lens up to the signal head. Drill a small hole into the center of the plastic lense, where Red + Green = Yellow, and insert a large diameter strand of optic fiber (left-over from a desk- art gift). This seems to work on the bench. The end of the fiber clearly shows Red, Yellow/Amber, or Green. I have not yet tried to build it into an HO scale searchlight signal.

A second method is to use the T-1 or T-1 3/4 bi-color LED and build the searchlight signal head around it. With careful filing, the physical size of the plastic case can be reduced and shaped. Just avoid cutting into the internal wires or LED die. Let the searchlight signal head mask off the front so that only the center of the lense, where the color blends well, is visible. Polish the front surface with 600 grit emery cloth.

Please let the rest of us know what you have tried and learned.

DonV

-- Don Vollrath (dvollrath@magnetek.com), May 17, 1999.

Don,

I totally agree that the lack of a complete mix is a very big factor in making the yellow less than desirable for most LED's. In fact I sent a few messages directly to Allan Gartner (above) on this very issue. [I should have used the "reply" button as I did this time to keep the message thread for all to see.]

What I am currently checking on is an LED "module" (with an intergral housing that puts the LED at right angles to .01 spaced pins, two in this case.) What appears hopeful for this module (Digi-Key P/N 67-1243-ND) is that the LED, unlike the great majortiy of such modules, is recessed into the plastic housing so that you cannot view it from other than head-on, and when viewed head-on the mixing of the red/green should be optimized by the lens. I am getting some of these in the next day or so and will let all know the verdict.

Your light pipe idea seems sound but I am to lazy to do all that work. However, the idea of roughing the surface to further "un-resolve" the two light sources seems good and simple - I will definitely try it and report.

Thanks for the good input!

Incidentally, I am trying to signal all of the main lines on a currently under construction HO layout. I already have detector/switching PCB's made that will operate position light and color light signals, but I want to get an easy and satisfactory result on the searchlight LED's before I make a similar detector/switching PCB for them. My interest in this was piqued by a recent article in the Signaling Special Interest Group newsletter of the NMRA, where a very elegant switching solution was shown for bidirectional 3 aspect signaling using position lights, color lights, and searchlights.

-- John Johnston (johnjohn@vicon.net), May 17, 1999.

I just tested the best candidate so far - P/N LTL-14CHJ, which is a 2 leaded "orange/green" bicolor T-1 LED from Lite-On, Inc. I passed this up originally because of the" orange" designation, but it sure looks red to me. This unit appears to have almost perfect color balancing, with no external weighting resistors necessary, and the color blending is at least acceptable. I am going to fool around with a few other "quick and dirty" blending enhancements such as a milky-white RTV lens overlay, and will report on my findings.

John

-- John Johnston (johnjohn@vicon.net), June 01, 1999.

I just recently scratch built a searchlight signal and put it in service on my layout. I used a bipolar, tristate red/green LED Jameco P/N 156945 for the lights. I can get a very good yellow light blend when a 330 ohm 1/4W resistor is used on the red diode and a 470 ohm 1/4W resistor on the green diode. The only drawback is that the yellow is good when looking straight on at the light. If you look to one side the light may appear more red or more green. I like using the tristate LED because I don't have to use an AC voltage source.

-- Bryan Kerstell (bkerstell@hotmail.com), January 25, 2002.

Yet another good tip. thanks The Jameco p/n 156945 is 3mm dia dual LED w/ 3 leads. I admit that driveing it w/ 2 DC transistor circuits does have good merits. My favorite is a 3mm, 2 leaded dual from LiteOn, Digikey p/n 160-1058- ND. There will soon be an article in Clear Block newsletter that describes a good way to drive a 2 leaded bi-color LED.

I'm interested in how you constructed your searchlight signal apparatus. More details & pictures please.

-- Don Vollrath (dvollrath@magnetek.com), January 25, 2002.

Here's how we expanded the signal system on the Teton Short Line. We've had 48 three-color heads in service for several years but decided to add variety with a recent expansion and chose to use searchlight signals. Two DIY circuits are shared that provide tailoring of color intensity and yellow balance, usable with two or three lead LEDs. http://www.ida.net/users/tetonsl/railroad/SgnlLED1.htm

-- Wayne Roderick (tetonsl@ida.net), January 25, 2002.