This website has been archived from TrainWeb.org/girr to TrainWeb.US/girr.
White LED Tips
Contents
- White LEDs
- LED Driver Circuits
- White LED Sources and Information
- Modifying the Color of a White LED
- Example White LED Headlight
Installations
28 Oct 01
White LEDs
Light 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.
When 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.
This 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.
LEDs 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.
This 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.
This 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.
I 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.
There 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.
If 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.
The 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.
If 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.
The 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.
If 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.
If 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.
The 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.
When 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 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.
- Nichia developed the white LED and may be the only manufacturer
- LEDtronics sells all kinds of LEDs and LED assemblies
- Optotech may make their own diodes
- Bivar has announced white LEDs.
- Sharp has also announced white LEDs
- The LED Light sells the Nichia diode and some nifty reflectors
- Radio Shack sells white LEDs through their catalog only. There are 3 mm and 5 mm versions
- Hosfelt sells white LEDs
- For the Love of LEDs provides a history of LEDs
- Fiber Optic Products, Inc. sells LEDs and fiber optics supplies.
[ 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.
These 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.
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.
| 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 |
This page has been accessed times since 11 June 00.
© 2000-2002 George Schreyer
Created June 11, 2000
Last Updated June 23, 2002
