While preparing for the Year 2000 power grid collapse, which somehow never materialized,
I began experimenting with battery-powered room illuminators using super-bright
LEDs. I learned a couple of interesting lessons the hard way. First, I
found that LEDs tend to draw more current when they warm
up. When a bunch of them are connected in series, the effect is
amplified. The next thing I found is that LEDs tend to fail shorted,
not open. If you have a long string of them in series and one of them
shorts, the current goes way up in the rest of the string, and within a few
milliseconds, you have a whole string of dead LEDs.
It was at about this time I began experimenting with current limiters! It
turns out that it is fairly easy and very efficient to feed ten or twelve LEDs in
series from a 48-volt DC supply, using a current limiter consisting of an LM317
and a resistor of about 51 ohms. The current limiter compensates for the
effects of temperature variations, LED aging, power supply variations (which you should
expect from unregulated supplies), and accidental shorts across one or more of
the LEDs in the string. In other words the circuit is more
fool-resistant (but still not entirely foolproof).
This techniques allows for mix and match assortments of LEDS to be fed from the same source.
Ultra-bright LEDs in different colors have their own forward voltage characteristics, but most of
them have the same maximum recommended current, usually 20 milliamperes. This circuit lets
them share what they have in common.
Early results were
promising, because the LEDs marketed as "Super Bright" really are amazingly bright. A
few dozen of them, with various colors combined, produce a pleasantly warm light that would
be a welcome relief in the event of a blackout, and a lot safer than candles or gas
lanterns.
Don't think for a moment that I came up with this current limiter
idea myself. The use of the LM317 as a current limiter was known to ancient engineers, way back in the
20th century. This diagram is from the 1987 Motorola Linear and Interface Integrated Circuits databook,
which was sent to me by a Motorola distributor.
Editor's note and correction in very fine print:
When I wrote that paragraph several years ago, I had forgotten that I had a paid subscription to the Motorola
data sheet service. It wasn't free.
This diagram was lifted right out of
the LM317 data sheet. The circuit is
very simple. Use 51 ohms for R1 and you'll get a current limited to about 20 mA. Note
that in this arrangement, you supply power to the Input pin, and the load is connected to the
Adjust pin. (The 20 mA current comes through R1, not the Adjust pin.)
Most people try to run LEDs at a current of
no more than 20 milliamps. Traditionally, power is fed to one LED at a time from a
regulated five-volt supply with a series resistor of 150 ohms. This assumes that
the forward voltage on the LED is 2.0 volts, which is pretty typical of red LEDs. Other
colors, especially blue and white, have substantially higher forward voltage
at 20 mA.
This is a rough sketch of the circuit. Remind me to polish up my drawing someday. Note that
only four LEDs are shown in the drawing, but with a 48-volt supply you can feed ten or more LEDs.
48 volts is a reasonably safe voltage and 48-volt supplies are commercially available. (I built my
own.)
The circuit diagram is shown below. The number of LEDs is typically 10 or 12, depending
on the supply voltage amd the average forward voltage across each LED.
(Note 1) The differential voltage across the LM317 must not exceed 40 volts. If you use the LM1117,
the limit is 20 volts. The minimum voltage (to maintain 20 mA) is about 4.0 volts for the LM317 and only
slightly less for
the LM1117 "low-dropout" regulator,
which was a big disappointment.
Use this for R1 if you want variable operating
current.
Update: Here is some information
about mixing red,
green and blue to get white light from LEDs. But if you're building a reading lamp,
as opposed to a video display, mixing R, G and B is unnecessary and is no longer
economically feasible, since white LEDs are available at lower cost than ever before. On
the back cover of the latest Jameco catalog, there is
a 5000-mcd white LED featured at an attractive price. I ordered a few to check them out and
found that they are really bright and well worth the money. (The price gets a lot better if
you order a hundred of them. I'm thinkin' about it.) At 25 mA the forward voltage is
about 3.4 volts, and the color is very much like a cool white fluorescent light. After all,
that's exactly what it is.
As you can see, most of the power is consumed by the
series resistor(s), and more power is lost because the 5-volt regulator is less than 100% efficient.
Check the manufacturer's specifications for the devices you buy, and you may find that the maximum continuous
current rating is 30 to 50 mA; however, the "typical" current is almost always listed as 20 mA.
This is the biggest advantage in the use of the series-fed circuit: If you put three or four
different colors of LEDs in series and feed 20 mA through them, there will be a different forward voltage
on each diode, ranging from 2.0 to 3.8 volts. With the LEDs fed in series, the forward voltage of
each device is not a concern. As long as the forward current is appropriate, the voltages won't matter.
Other advantages:
You don't need a regulated power supply.
The system has higher efficiency since there is only one current-limiting device.
If one or two LEDs in the string are shorted, the others are undisturbed.
After you build this current limiter circuit and find that it works, you'll naturally want
to see how high the current can go without toasting the LEDs. The specifications
for the LEDs in my breadboard experiment included an absolute maximum current rating
of 50 mA. Of course, nobody drives them that
hard… intentionally.
But I found that as the current exceeds 40 mA, very little additional
brightness can be squeezed out, and after only a few seconds, I was greatly concerned
about the life expectancy of the LEDs, so I never completed the (destructive) test
for the ultimate LED current.
As you can see, I also have a collection of
analog panel meters.
Recently I was asked where I buy LEDs for this kind of experiment. I made my
most recent bulk LED purchase
from Marktech, which is a Toshiba
(and Cotco) LED distributor in New York. If you are buying 100 to 1000 LEDs
at a time, this is a good place to shop. For small quantities and a wide
variety of LED colors, get the catalog
from Hosfelt Electronics. Their
prices are a little higher than wholesale but still pretty good.
Only very recently have LED "bulbs" shown up on store shelves, and unless you shop on the internet, the
prices are still quite high, and most of the products I've seen are spotlights and accent lights rather
than omnidirectional "bulbs." Drop-in replacements for ordinary light bulbs are hard to find, but
there is a company
called Luminosity LED offering
such a device already. Again, the internet is the marketplace for merchandise on the leading edge
of technology.
If I were to undertake this project from scratch today, I'd probably go a different
route and concentrate
on 12-volt compact
fluorescent
lamps. A lot of information about efficient lighting schemes
of all types can be found on this
page.
Cheap,
super-efficient LED lights on the horizon. Incandescent tungsten-filament light bulbs face a global
switch-off as governments push for energy efficient fluorescent lamps to become the standard. But the light
could soon go out on those lamps too, now that UK materials scientists have discovered a cheaper way to produce LED
bulbs, which are three times as efficient as fluorescent lamps.
LED There Be Light. Torraca is a small
village of 1,200 people in Italy. It is also the first place in the world to be totally illuminated by light-emitting
diodes (LEDs). Representing a sea change, much like when electric lamps first graced London's Holborn Viaduct back
in 1878, some 700 streetlights (each containing 54 LEDS) now line Torraca's arteries — and locales around
the world, from Beijing's Bird's Nest Olympic Stadium to the Raleigh Convention Center's Shimmer Wall in North Carolina,
have begun to use LEDs to light up the night.
Do
LEDs Disrupt our Biological Clocks? Epidemiological studies have linked circadian disruptions to
health problems, such as cancer, cardiovascular disease, and obesity, and scientists are trying to determine
if light at night — and blue-rich LEDs — are a cause.
Most people try to run LEDs at a current of
no more than 20 milliamps. Traditionally, power is fed to one LED at a time from a
regulated five-volt supply with a series resistor of 150 ohms. This assumes that
the forward voltage on the LED is 2.0 volts, which is pretty typical of red LEDs. Other
colors, especially blue and white, have substantially higher forward voltage
at 20 mA.
As you can see, most of the power is consumed by the
series resistor(s), and more power is lost because the 5-volt regulator is less than 100% efficient.
