Wayne Roderick, 3rd Division, PNR, NMRA (life)
09/23/01 rev 01/16/07
A recent expansion project on the HO scale TETON SHORT LINE (TSL) called for three-color signals using the single lamp searchlight signal. Dual LEDs that can display green or red have been around for quite a few years. The NMRA Bulletin, May 91 had an article by Mark Hanslip using the dual LEDs in two color circuits. Others have demonstrated that the proper balance of red and green will display yellow. It's not perfect because the two LEDs are built side by side and so the nominal yellow color varies somewhat with the viewing angle.
The dual LEDs are available in two or three wire versions. I chose a two wire, T1 (about 1/8" diameter). The two wire version requires that the current be reversed to change color and of course this is easily done with an alternating current (AC) source. Unfortunately, getting the proper yellow hue is really not that easy!.
My goals were: High intensity light, approach lighting and a good quality of yellow. Starting with the popular CMOS 4049 buffer used by Mark in 1991 and again by Jeff Scherb, MR April 2001, I extended their simple circuitry to meet the goals. Jeff used the classic CMOS 50% duty cycle astable multivibrator (oscillator) published by Motorola in 1974 to get his AC signal for yellow. I was unable to get a satisfactory yellow, because my LEDs green light was much weaker than the red light and also for that for that reason the solid green was not intense enough. The CMOS 4049 could not source enough current to get the desired intensity. I needed about 30ma, so I replaced one of the hex buffer amps with an NPN transistor, 2N2222 or equal.
The yellow could be made very nice by reducing the red current. Easy enough to do by adding resistance in series with d7, but the solid red intensity suffered too much whem compared to my existing three color heads.
The answer is to adjust the duty cycle of the oscillator so green is displayed for more time than the red when yellow is desired. This is done with a shunt resistor steered by diode d4.
The TSL circuit also contains the priority logic for the three colors. While this could be left to the computer driver system as I do in my three-color heads, it just kind of fell in place because its quite impossible to display more than one color at a time with a searchlight signal.
The resulting circuit, complete with approach lighted green required five of the six inverting buffer amp's in the 4049 chip. The extra sixth one, u1c, was wired in parallel with u1b to share the load. Grounding the green input turns on Q1 via u1f and d5 to source current to the LED. The green current sinks to ground via u1b and u1c. Grounding the yellow input removes the oscillator inhibit via d1 and enables the green via d2. The oscillator will now switch the LED between green and red at a high rate and hopefullly our eyes will see only yellow. Grounding the red input will kill the yellow oscillator via d3 leaving it in red mode. u1b and u1c will now source about 20ma to the LED and inhibit the green transistor q1 via diode d6. The red LED current will sink to ground via d7 and u1e.
The LED current limiting resistance is two 150 ohm resistors very near the signal mast. I choose to do this, as I did with the three lamp signals, to minimize the electrical risks to the LEDs.
Yes, it looks like a lot of extra components, and there are many other ways to get the job done. I considered op-amps, CMOS & TTL, and even PICs but I had the 4049s, diodes and resistors in the junk box so that's the way I did it. The added resistors and generic silicon diodes are extremely cheap if bought in quantity from an electronic distributor rather than the corner Radio Shack. If you're making pc boards, you only have to deal with the additional complexity once while making your layout.
Select your LEDs and get about 20% extra count. Breadboard the circuit and select the ones that best match when yellow.. Trim the 47k resistor for the best yellow and use the discard LEDS for two color functions elsewhere such as turnout position indicators.
I'll get a pix up of the DIY searchlight signals when the ladders and gingerbread is all attached. Like my three-color signals, they will be of the "plug-in, bump and knock over, standup and plug back in" type construction.
Click on the thumbnail for a VGA ( 640x480) close up photo of the completed signal.
09/27/01 Last evening we hooked this circuit to a Sunrise Brand searchlight marketed by Ron Ferrel and it was gorgeous. Used the recommended 330 ohms limiting resistor. The signal uses a smaller, or possibly cut down LED in a very nice structure. Drop me an e-mail and I'll connect you with Ron if you're interested.
01/18/02 If you don't like all the parts count in the above circuit, try this one. It doesn't have built in color priority but works fine when driven from my computer interface. Don't let the transistors deter you. There are only twelve connections in the four, less than an IC and they are very cheap in quantity, probably cost less than an adequate H-bridge in an IC. Four of these circuits fit comfortably on a 22 pin proto- board. Nope, I don't waste my time laying out and etching pc boards for small quantities. I only needed four cards (16 circuits) for the expansion.
Three lead bi-color L EDs work fine with either circuit. Hook up the two anodes each with its series resistor to the output terminals, and ground the common cathode. You won't need the diode in the Green return path and obviously, you can discard the PNP transistors too.
12/30/02 Finally got around to putting the hardware together for our searchlight/target signals and decided I had room on the circuit card for the decoding. Remember, my computer outputs only two bits to control the four possible indications. This circuit uses only two driving transistors but does waste a bit of extra current. The power source for the lamps is out in the field just like my three-color heads. I like the consistency.
Teton Short Line Home Page