LED Efficiencies

[spencer]

What is the maximum efficiency that current consumer available LED's can reach. I know that they are most efficient at lower currents but what is the magic number. Also, how efficient is a Cree at higher drive levels. I was amazed at how cool my P7 ran at about 200mA while still outputting a great amount of light.

1. What type of LED is most efficient 5mm or power?
2. What is the maximum efficiency reached by consumer LEDs.
3. How efficient is a power LED running at 350mA, 700mA, 1A?

[R33E8]

1. From what I've seen, power LED's are more efficient than 5mm LED's..

2. I think the maximum efficancy is around 150lm/w at a very low current from the Cree XR-E series (probably the R2 bin)

3. Here is a chart I found on KaiDomain which has the amount of lumen emitted at various current levels with a Cree Q5:

http://www.kaidomain.com/ProductDeta...ProductId=1805 ( I couldn't get it to paste correctly)

Apparently, they get almost 200lm/w at 30mA (acutal lm/w= 198.4)

[jirik_cz]

This excellent thread by jtr1962 http://www.candlepowerforums.com/vb/...ad.php?t=89607
has all answers for you.

[Erasmus]

3. Here is a chart I found on KaiDomain which has the amount of lumen emitted at various current levels with a Cree Q5:

URL ( I couldn't get it to paste correctly)

Apparently, they get almost 200lm/w at 30mA (acutal lm/w= 198.4)

That information is definitely false. The charts of Jtr1962 are reliable, better use these as a reference.

[Marduke]

IIRC, the magic number for a Cree XR-E is somewhere about 20mA.

[TexLite]

Heres the R2 graph from jtr1962's thread:

Cree 7090 XR-E bin R2 (acquired March 2008)

Read the whole thread if you have time,its immensely informative.

-Michael

[2xTrinity]

Also, I have tested and observed the efficiency "dip" below about 20mA on the power LEDs. It isn't an error in the graph. (with a precision bench supply I see a very substantial shift in both color, and output from 5mA to 10mA, well more than double the lumens).

In order to get high efficiency at super low outputs, it's neccessary to use PWM, and preferably set the "pulse" currents to 20mA. This is the reason I suspect the only lights on the market that go to small frctions of a lumen are PWM controlled, without filtering on the output side. A constant current driver is actually nothing but PWM with a filter on the output designed to "smooth out" pulses at a certain frequency.

One way I've considered (but never actually built) for a driver that varied current down to 20mA, then reduced output via PWM lower than that, woudl be to use two different frequencies for the modulation -- use a current regulator with high frequency, and then PWM at a lower frequency for super low outputs, so that the pulses get through the filter.

To be honest in very few applications would it be worth this trouble just to squeeze out that efficiency for such low overall output, as it's pretty seldom someone would want 1/10th of a lumen for weeks at a time...

[space]

What finally struck me after reading jtr1962's great thread on measurments, is that maximum efficiency for leds are reached when they run at the voltage that corresponds to the photon energy of the light emited by the LED.

For most current white LED's this is ~2.75V, the photon energy of the blue light emmited by the LED. (450nm)

So if one analyses the wavelengt emmited from a LED, the most efficient voltage can be calculated.


space

[TorchBoy]

maximum efficiency for leds are reached when they run at the voltage that corresponds to the photon energy of the light emited by the LED.

And under that voltage the efficacy drops away quickly?

[space]

Yes. All measurments I've seen have this property. I only have guesses for why, but there are probably others in this forum that could explain.

space

[2xTrinity]

And under that voltage the efficacy drops away quickly?

Correct.

A photon has a discrete amount of energy:

1240 eV / <wavelength in nm>

In an LED, in order to converting one electron (e) into a photon, you require at least the voltage (V) given by the formula above. For a 450nm blue LED, this means you need to supply at least 2.75V to force light emission at the dominant wavelength. LEDs however don't emit only at precisely one wavelentgh, they emit at a central wavelength +/- some amount. If fed a lower voltage, they will only be able to emit at the longest possible wavelengths where conversion will be less efficient. Color shifting is also apparent when underdriving LEDs to this extent, as the dominant wavelength emitted changes.

[TorchBoy]

That's absolutely fascinating. I guess it even explains why red LEDs have a low Vf compared to blue LEDs, but... why do green LEDs have the same high Vf that blue LEDs do? Or does it only look like they do due to inefficiencies in the conversion? I thought it was because of the doped semiconductor material that they were made from.