For an LED that is rated at 20mA, is it safe to PWM it at 40mA half the time? Will that make it nearly as bright as if continuously supplied with 20mA?


Thanks Olin and tcrosley for your answers. I didn't want to distract from the question by adding the details of what I am doing, but here it is:

I am working with an LED strip from http://www.environmentallights.com/LED-Strip-Light-Double-Density-4-Wire-Red-Green-Blue-by-the-5-meter-reel_P3846.aspx. They do not have a datasheet per se. I think the LEDs are connected as shown in this image http://www.ladyada.net/wiki/_media/products/ledstrip/astripsch.png.

Based on what I read about LEDs in general, 20mA per color seemed safe. I tested it - using 12V, potentiometer and 3 inches long strip, that is 3 leds in series each color - for short periods, few minutes, and 20mA per color seems to have worked fine, good enough brightness and no overheating or color intensity variation.

I will be working with up to 2 feet length strips, that is 160mA per color and 480mA to get White. I may have up to 10 strands to control. That will add up to about 5Amps. I will be using the Propeller microcontroller from Parallax to flash the strips in various patterns and colors. Will use mosfets to drive the LEDs.

I was thinking of supplying about 200mA instead of the 160mA needed per color, but then I will PWM the leds such that only one color is on at a time; at a frequency of 1KHz. So to get white, each color will be on for one third the time, that is 0.3 msec. At any moment only one color is on, only 200mA is drawn. I imagine the leds will be less bright, hopefully not by much.

So, for 10 strands, I will deal with only 2A instead of 5A.

Please let me know if what I am thinking is off.



The datasheet will have this information. Some LEDs that can only handle 20mA continuous are spec'd for very brief pulses of a much higher current. You would have to specify the PWM frequency to be able to look up the answer to this question in a datasheet.

For one particular product, the CREE CLV1A RGB LED, you may drive red at 50 mA continuous or with a 200 mA pulse (pulse width ≤0.1 msec, duty ≤1/10), and blue and green at half that current. It's all in the datasheet. You can also see that there is a diminishing return as the relative luminuous intensity vs. forward current is not linear at high currents.

I think you should not worry about the brightness loss from driving your LEDs at a 50% duty cycle at the maximum continuously rated current. IIRC the eye has a non-linear response to brightness and they will appear somewhat brighter than half as bright, and you won't destroy your LEDs if you screw up the PWM.


If you are driving the LED from a microcontroller with a PWM output, and setting the maximum current above the safe continuous operating value, you need to make sure there are no failure modes where the PWM could stop with the LED driven on. I have seen this happen with a multiplexed, multi-digit seven segment display where something caused the software to fail and one digit remained overly bright until it failed.

  • 3
    \$\begingroup\$ There are many relatively easy ways to design a circuit which will prevent an LED row from being driven with too high a duty cycle. On the other hand, I've seen a lot of displays that look like some LEDs were destroyed when a row got 'stuck' on. I would certainly include 'failsafe' circuitry in any design I wanted to produce. \$\endgroup\$
    – supercat
    Feb 1 '12 at 21:42

Your spec "rated at" is unclear. Rated for continuous operation or maximum pulse current?

What you are looking for is the maximum allowed instantaneous current. This is found in the datasheet along with the maximum average current. A LED rated for 20 mA average current can most likely be run at 40 mA at 50% duty cycle as long as the frequency is fast enough, probably a few 100 Hz. Ordinarly T1-3/4 LEDs can generally take this. Some high performance lighting LEDs are run so close to the limit that their maximum pulse current and average continuous current are fairly close. As always, of course, read the datasheet.

LED brightness is pretty much proportional to current except that there is often a falloff in this relationship at high currents. Usually this is small, but again, read the datasheet. High performance and high power lighting LEDs tend to have more falloff because their intended operating point is already somewhat into this region.

In any case, efficiency (light out per electrical power in) falls off with current because the voltage accross the LED increases with current. Even if the light output is always proportional to current, the voltage will increase with current a bit and therefore the input power increases with current more than the light output. For most ordinary 20 mA indicator LEDs, 40 mA half the time at a few 100 Hz is for practical purposes almost the same as 20 mA continuous. This is less true for higher power and higher performance LEDs. Again, read the datasheet.

  • 1
    \$\begingroup\$ This is opposite of what I had been taught by my professor in lasers. He took a 20mA LED and put Amps into it without damage. The pulse was VERY VERY low duty cycle, this would cause a large burst of light then a long cooloff instead of a long slow heating cyclo allowing the system to operate at a higher efficiency. I would like to do further testing to show this but I cant afford the laser power supplies he was using. \$\endgroup\$
    – Kortuk
    Feb 2 '12 at 11:38
  • \$\begingroup\$ He also took the LED and changed its color by intentionally overcurrenting to the point of damage. That was just fun to watch. \$\endgroup\$
    – Kortuk
    Feb 2 '12 at 11:39
  • 1
    \$\begingroup\$ @Kortuk: I'm not sure what point exactly you are making. Are you saying that a short high current pulse results in more overall efficiency from electrical power to average light than a steady current? Resistive losses alone should guarantee this is not true. Most datasheets for LEDs targeted at efficiency or high light output will show a falloff of light/current at high current. I've certainly never seen one that shows higher light/current at high current. Simple indicator LEDs are intended to be run at low levels, so sometime they show a straight line, but never a increasing one. \$\endgroup\$ Feb 2 '12 at 13:27
  • \$\begingroup\$ As a way to get more light from a cheaper LED and that you can severely over-current. I think I misread your second paragraph, my mistake. \$\endgroup\$
    – Kortuk
    Feb 2 '12 at 15:40

I've done that in some multiplexed arrays of high efficiency LEDs (used as a graphic display). because it was the only way to get the rated light output from the leds with mux.

I've used 2A at 0.5% duty cycle (LEDs were rated at 20mA).

Displays are working almost 24/7 since 2007... so, it surely is a bad practice, but leds doesn't seem to care... :-)

  • 1
    \$\begingroup\$ And what are they displaying? :) \$\endgroup\$
    – clabacchio
    Feb 2 '12 at 9:07
  • 1
    \$\begingroup\$ Texts, some simple B/W (well... Black/Green) graphic. Nothing fancy... they're a sort of really big Nokia "old style" phone displays. \$\endgroup\$
    – Axeman
    Feb 2 '12 at 18:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.