Facebook Page
White LED Tips

White LED Tips

[ Home ] [ Up ] [ Previous Page ] [ Next Page ]


White LEDs

LED and reflectorLight Emitting Diodes (LED) have recently become available that are both white and bright, so bright that they seriously compete with incandescent lamps in lighting applications. They are still pretty expensive ($3+ apiece) as compared to a GOW lamp but draw much less current and project a fairly well focused beam. The diode in the photo came with a neat little reflector that tends to sharpen the beam a little but doesn't seem to add much to the overall intensity.

When run within their ratings, they are more reliable than lamps as well. Red LEDs are now being used in automotive and truck tail lights and in red traffic signal lights. You will be able to detect them because they look like an array of point sources and they go on and off instantly as compared to conventional incandescent lamps.

Big Hauler with GOW HeadlightWhen used as a locomotive headlight, the effect is quite good. The Big Hauler on the left has a Radio Shack 12v 60 mA Grain-Of-Wheat bulb running at its ratings. The Shay on the right has the stock Bachmann bulb that came on the early production Shays, later Shays have a much more wimpy standard yellow LED as a headlight. You can see that the Big Hauler light is not very bright as compared to the Shay.

Big Hauler with White LED HeadlightThis photo shows the same two engines after a bright white LED has been installed in the Big Hauler. The LED is running at its rated 20 mA vs the 60 mA that the old GOW bulb took. Now compare the brightness of the two locos using the Shay as a reference to see how much brighter the white LED is.

The LED produces a tight bright beam that goes further down the track than most incandescent lamps. In a dark environment, it is bright enough to cast shadows more than 10 feet away. In the indoors environment, the LED might be too bright. These trains run at eye level and when the beam sweeps across my eyes, it is blinding.

The viewing angle and the brightness of LEDs usually vary inversely. A wider angle will result in lower brightness in any given direction as the total light output of the LED is spread over a larger area. For headlight applications, a 20° viewing angle is appropriate. This is the tightest angle typically available and produces the most intense beam in front of the loco. Intensity is rated in candlepower, a typical tight beam LED runs 4000 to 5600 mcd. The size of the diode package determines the minimum beamwidth, a larger package with a larger lens can focus a tighter beam. For most large scale headlight applications, a 5 mm diode package is adequate. There are 10 mm packages available that will produce a tighter beam, but a 10 mm (0.4") LED is really too large to mount on a loco.

This tight beam aspect of many of the white LEDs has a downside also. The tighter the LED beam, the dimmer it will appear when viewed off angle. Some of them have such a well formed beam that they are hardly visible at all at angles between 30 and 80°. At almost 90° they have a sidelobe and are visible again. A regular frosted diffuse LED, such as Bachmann and Aristo use, may appear brighter than a tight beam white LED in the daylight when viewed off angle because they spray light in all directions. However, the diffused wide angle LEDs have no hope of casting any sort of a beam in dim light or darkness. White LEDs with a different shaped (usually smaller) lens will have a wider viewing angle, but do not form as tightly a focused beam. They light up the general area in front of the loco instead of casting a beam down the track. There really is no free lunch.

white led spectrumLEDs are monochromatic (one color) devices. The color is determined by the bandgap of the semiconductor used to make them. Red, green, yellow and blue LEDs are fairly common. White light contains all colors and cannot be directly created by a single LED. The most common form of "white" LED really isn't white. Its a Gallium Nitride blue LED coated with a phosphor that, when excited by the blue LED light, emits a broad range spectrum that in addition to the blue emission, makes a fairly white light. The actual light has a blue cast and is similar in color to a mercury vapor street lamp. On the curve shown, the peak at the left is the shortest wavelength blue light from the LED. The lump of emission to the right is the longer wavelength emission of the phosphor. There are other types of "white" LEDs that are made from several different LED chips of different colors assembled into one package. These have not been particularly successful as they tend to change color depending on viewing angle and their color balance is not real good at best.

There is a claim that these white LED's have a limited life. After 1000 hours or so of operation, they tend to yellow and dim to some extent. Running the LEDs at more that their rated current will certainly accelerate this process.

The phosphor type of LED was developed by Nichia in Japan. I do not know if other manufacturers are making this type of LED now, but for some time, ALL the white LEDs of this process that were sold anywhere were made by Nichia.

[ Top ]

LED Driver Circuits

RAM Products produces ready to use white LED headlight circuits, the RAM109 is a typical example. These include the LED and a small voltage regulator board. I bought one of these and was impressed by the intensity of the light, but the owner of the company warned me about using too high an input voltage, above 9 volts. Since I am planning to run it from 22 volts or more, I needed to make some changes. In the process, I believe that a simple modification to RAM's circuit will allow a higher input voltage range AND improve the life of the LED.

RAM109 circuitThis is the circuit diagram of a RAM109. It is simply a low current 5 volt regulator that drives the diode. The problem is that the diode wants to see less than 5 volts, so that when the regulator comes alive, it over drives the diode, up to 150 mA. I do not know the maximum current capability of the diodes that RAM uses, but I haven't found any diodes rated past 20 mA. Even the "ultra bright" LEDs in other colors are rated at 20 mA. A diode that is run at overcurrent will run brighter, but it won't last as long.

RAM IVThis is the response of the RAM109 regulator with a DC power input. Note that the peak current is 150 mA which might be 7.5 times the diode rating. This cannot be good for the diode.

20mA diode IVI purchased a generic white LED at Fry's. I don't know who manufactured this diode, but it is rated at 20 mA, the only current rating that I have seen for white LEDs. I was able to drive it up to 117 mA and it didn't burn out and it was REALLY bright. This diode makes its first light at about 1 mA and by 3 mA it is too bright to look at directly. Its appearance, turn on characteristics and brightness are roughly the same as the diode that came with the RAM109. This IV curve is log-normal up to about 20 mA. After that, it saturates, the voltage goes up but the current does not go up fast enough. This indicates that the diode is screaming in pain. The power dissipation and therefore the temperature of the LED is increasing rapidly.

simple current limiterThere are three ways to deal with this problem, however, all of them are fundamentally the same. An LED is primarily a current driven device so it wants to get its power from a current source. If the source voltage is high enough, a simple series resistor simulates an acceptable current source. If the LED is to run from a fairly high fairly constant voltage, such as the function output of a DCC decoder (about 20 volts), or an onboard battery (12-18 volts), or a Bachmann Shay or Climax flicker board (about 9 volts), or a Soundtraxx Sierra lighting output (about 6.8 volts) then the value of a suitable resistor can be read from this curve and the diode current will be acceptably close to 20 mA.

better regulatorIf the diode is to be run from a variable voltage source, such as track power, then a regulator of some kind will do a better job of controlling the intensity of the LED. This is the circuit of a regulator for a white LED. The only significant difference between this circuit and the RAM109 is the addition of a resistor to take up the voltage difference between the output of the regulator and the needs of the diode. The current in the diode is set by the value of this resistor. The diode current will be I=(5-Vf)/R where Vf is usually about 3.4 volts but it depends on the actual diode used. The 75 ohm resistor sets the current at a little more than 20 mA. An 82 ohm resistor will set the diode current at just below 20 mA. Users of the RAM circuit should probably consider adding a resistor in series with either wire to the LED.

better regulator IVThe 75 to 82 ohm resistor makes this circuit approximate a current source as the source voltage is well regulated and the diode current will be well controlled at a current level of about 20 mA. The input current to the circuit will be 5 mA or so higher than the LED current because the regulator itself draws a little current. Note that when the input voltage is high enough, the diode current is nearly constant and won't increase to an unsafe level. If you use a regular 7805 regulator in a TO-220 package (metal tab on the package) you can drive several diodes in parallel but each diode needs its own resistor. The regulator may require an additional heat sink. The TO-92 (plastic case) 78L05 regulator can handle only one diode and just barely at that.

lm317 current regulatorIf a constant current source is what is really desired, then an adjustable voltage regulator circuit, the LM317, can be wired as a current regulator. In this case, the current is set by the 62 ohm resistor. The current will be I=1.25/R. You can also use a LM317L (or the equivalent NTE1900) in a TO-92 package to drive a single diode (or more than one if they are wired in series). The LM317L will run a little hot, but it'll work up to 30 volts input.

If you do run diodes in series, it'll take another 3.4 volts or so of input voltage per extra diode to make the circuit work. This would be an appropriate solution in the case of onboard battery power, DCC or constant track power as there would always be enough voltage to operate the circuit. This configuration allows several diodes (of different types/colors if desired) to share the same 20 mA so that it also minimizes the total current draw.

Don't try to parallel diodes from a current regulator, the diodes will not share current properly unless you add a balancing resistor in series with each diode. In that case, you might as well use the 7805 constant voltage regulator.

lm317 current regulator IVThe LM317 current regulator has just a little less overhead voltage than the circuit with the 7805 in it. It also doesn't care what the LED voltage is, it'll drive 20 mA into any single diode or a stack of various diodes. When running a single diode, there is not a lot of difference in performance between the two circuits, use whichever one you have the parts for.

rectifierIf the circuits are to be run from track power or DCC then a circuit called a full wave bridge rectifier needs to be used to convert the track power to a constant polarity. This circuit outputs a fixed polarity DC voltage that the any of the three circuits above can use. The voltage drop of the rectifier increases the input voltage that it takes to run the circuit by about 1.5 volts.

half_wave_rectifierIf you want directional lighting and also want to use a regulator, then use this half wave rectifier circuit. It will allow the LED to come on in one direction only. By using any of the regulator or current limiter approaches from PWC track power, the LED will come on at 100% full intensity at the first application of track power, well before any locomotive will start to move. If you do use PWC, then the resistor current limiter approach is highly effective and it the least complicated approach, just use an 820 ohm resistor in series with the diode and the half wave rectifier circuit.

If you simply want a directional headlight and are using regular track power and are just using a resistor to limit the diode current, then you may not need a rectifier at all. An LED is a diode and it will pass current in one direction only without any help from other circuits, within limits.

LED VfThe LED generates light when current passes through it in the "forward" direction. This curve shows the forward IV characteristics of four sample diodes. The yellow diode is an Aristo yellow LED that came out of a Center Cab switcher. The red and green LEDs are just two unknown heritage diodes that I got in a grab bag of diodes at an electronics swap meet. The only thing that I know about these is that they are red and green and they do make light. The white LED is a 5600 mcd 5 mm diode from The LED Light. The important thing to note is that the red, green and yellow diodes are all Gallium Arsenide devices and all have about the same forward voltage drop. All their red, green and yellow cousins will be about the same. The white LED is a Gallium Nitride device and has a considerably higher forward voltage drop. All other white LEDs of the same technology will be about the same.

LED VrWhen operating as a reversing headlight directly from track power, the LED on the "rearward" end of the loco will be off. The diode is said to be reversed biased and no current should flow. However all diodes have an avalanche or breakdown voltage. This is the highest voltage that the diode can stand before it starts to conduct in the reverse direction. Reverse conduction is very hard on any diode not designed for it. Whatever current that does flow will do so with high voltage impressed across the diode and the dissipation will be very high and could result in rapid thermal destruction of the diode.

For example, if this particular red diode were used in a reversing headlight application, the first time that it saw more than 20 volts, it would conduct large currents in the reverse direction and probably blow up in a short second. Discerning observers may note that the reverse IV of the red diode has a slight negative resistance component. This really happens on this particular diode, the Vr does go down a little with increasing Ir.

Most of the white LEDs are rated for a Vr (reverse voltage) of just 5 volts. This is not high enough to cause the diode to block reverse currents and if it really is that low, the LED will need another rectifier diode in series with it to block reverse currents. I tested one white LED and found that at 20 volts and below, it blocked current adequately well and that the leakage at 30 volts would result in only 1.5 mW of dissipation, an acceptable value. All white LEDs may not be the same. A manufacturer does not rate the reverse voltage at such a low value without a reason, some of the diodes may indeed have very high leakage at lower reverse voltage and might need an external blocking diode. Note that just adding two 20 volt diodes in series does not make a 40 volt diode. Their leakages may be significantly different and the low leakage diode will see most of the voltage.

If you need to locate a source of regulators or need data sheets to determine the pinouts you can try these links.

Regulator IC Data Sheets
Regulator Type National Semiconductor
Data Sheets
NTE Electronics
Data Sheets
Radio Shack
Part Numbers
Adjustable Regulator
TO-220 case
LM317 NTE956 900-4516
Adjustable Regulator
TO-92 case
LM317L NTE1900 N/A
5 Volt Fixed Regulator
TO-220 case
7805 NTE960 900-4492
5 Volt Fixed Regulator
TO-92 case
78L05 NTE977 900-4491

[ Top ]

White LED Sources and Information

I've been scanning the net for suppliers and manufacturers of white LED's and so far I've found several. This list will grow as I find more.

[ Top ]

Modifying the Color of a White LED

The typical white LED has a kind of a blue cast, an inappropriate color for nearly any kind of locomotive headlight. This can be modified to some extent at a small loss in intensity. Recently a post was placed on an internet bulletin board concerning painting the LEDs to change their color. I couldn't find the post again, but I did remember that the author recommended using Testor's Turn Signal Amber to paint the diode lens. I decided to try it out with that paint and a couple of others.

painted LEDsThese four diodes are wired in series and are running exactly the same current. The one on the left is unpainted, the next one is painted with Testor's Insignia Yellow, the next with Tamiya X-24 Clear Yellow and the rightmost one is painted with Testor's Turn Signal Amber. This photo was taken at much less than 20 mA because at full intensity, the LEDs drove my camera nuts. The unpainted one is characteristically blue. The Insignia Yellow is partially opaque so that it did more to block the light than anything else. The Tamiya Clear Yellow seemed did remove most of the blue cast and seemed to do a better job than the Testor's Turn Signal Amber. The Tamiya paint also flowed better. Both the Tamiya Clear Yellow and Testor's Turn Signal Amber are translucent paints and do impact the intensity a little. The Testor's paint is an oil base, the Tamiya is an acrylic.

The beam cast by the Testor's treated diode onto a white sheet of paper shows a loss of intensity as compared to the Tamiya painted diode. The Tamiya treated diode casts a yellowish beam onto white paper and has considerably better uniformity than the Testor's treated diode. The intensity of the Tamiya diode seems to be about the same as the untreated diode. The diode painted with Insignia Yellow produced a significantly dimmed and nonuniform beam without much color improvement.

Don't overdo the paint. All it takes is a very thin film to cut the hard blue color. Too much paint will result in either a yellow or green beam. Either dry brush the lens or airbrush the lens.

center cab with one painted headlight

All you need to do is dab a little paint on the lens. One coat seems adequate to shift the color back toward the yellow more like a real headlight.

This center cab switcher has had the right headlight painted with the Tamiya Clear Yellow. The difference in color and intensity is not huge, but the blue cast is reduced. The visual effect by eye is larger than the camera shows, probably due to saturation of the camera's imager by the intense LEDs.

In operation in darkness, the painted LEDs cast a whiter looking beam. With the Tamiya paint, the intensity is not significantly reduced, it still lights up objects 20 feet ahead. Looking directly at the beam reveals a very slight greenish cast. This greenish color does not show in the objects illuminated by the beam.

It is possible to paint LEDs of other colors too. I put an 8 degree beamwidth, 8000 mcd red LED in the rear of an Aristo Doodlebug and used it as a FRED. The beam was so tight that the LED was only poorly visible in daylight unless you were looking straight down its beam. In that case, it was much too bright. I painted that one with Tamiya X-27 Clear Red. This did two things. First, it diffused the beam from the LED and caused the LED to be clearly visible from any angle in daylight without materially reducing the intensity in the beam. Second, since this LED has a water clear lens, it made the LED look red even when it was off. This simulated the red lens of a FRED better.

[ Top ]

Example White LED Headlight Installations

Where I have to get into an engine for any reason, I am going to change out the headlight. These are links to the circuit implementations that have been used to install a White LED headlight.

White LED Installation Descriptions
Engine LED Type Circuit Type Link Author
Bachmann Big Hauler 5 mm, 20 mA Resistor Current Limiter Big Hauler Tips George Schreyer
Aristo C-16 5 mm, 20 mA Resistor Current Limiter C-16 Tips George Schreyer
Aristo FA-1 5 mm, 20 mA Resistor Current Limiter from Soundtraxx Sierra FA Tips George Schreyer
Aristo Center Cab 5 mm, 20 mA Resistor Current Limiter Aristo Center Cab Tips George Schreyer
Aristo L'il Critter and USA GP-9 5 mm, 20 mA Resistor Current Limiter Super bright white LED's Jon Foster
Aristo SD45 5 mm, 20 mA Resistor Current Limiter Headlights
Sierra Driven Ditch Lights
SD45 Tips George Schreyer

[ Top ] [ Home ] [ Up ] [ Previous Page ] [ Next Page ]

This page has been accessed hit counter times since 11 June 00.

© 2000-2002 George Schreyer
Created June 11, 2000
Last Updated June 23, 2002