The most efficient use of an LED?

Just had a quick question for yall. Probly applies to most other electronics as well, but oh well. Would the most efficient use of an LED, be when it was producing light, but little to no heat? I was reading one of my other threads and another member was telling me that if I ran a 900 lumen P7 at 900 lumens output, that only about 1/8 of the energy would leave as light, the other greater part would leave as heat. Well, what if you backed it off some. To say 500 lumens output. You would save battery, (at the cost of lumens output of course) but wouldn't you also gain efficiency cause you were not generating as much heat? Or maybe, you could go even lower and run it at like 300 lumens and generate even less heat. Thoughts? Opinions?

It was my impression that an LED is at it's most "luminously efficacious" (to clobber a term) when it is running at that current and forward voltage at which it just begins to light, and the luminous efficacy goes down as you approach the maximum current the LED is designed to safely handle.

Yes. The LED will output more lumens per watt as the current is reduced. See the threads of interest sticky at the top of this forum. "White LED lumen testing" as performed by jtr1962 shows lumen per watt graphs. As you will see, the white LEDs tested are most efficient at low currents and drops as current increases. Of course, operating a power LED at very low current is not very cost effective.

BennyLava said:

Just had a quick question for yall. Probly applies to most other electronics as well, but oh well. Would the most efficient use of an LED, be when it was producing light, but little to no heat? I was reading one of my other threads and another member was telling me that if I ran a 900 lumen P7 at 900 lumens output, that only about 1/8 of the energy would leave as light, the other greater part would leave as heat. Well, what if you backed it off some. To say 500 lumens output. You would save battery, (at the cost of lumens output of course) but wouldn't you also gain efficiency cause you were not generating as much heat? Or maybe, you could go even lower and run it at like 300 lumens and generate even less heat. Thoughts? Opinions?

Click to expand...

Quoting the efficiency as "[light:heat] ratio" doesn't really make sense as there is no definitive cut off between visible light and radiated heat (IR); the eye's response to light intensity varies depending on the wavelength, gradually tailing off to zero towards the red/IR end of the spectrum.

Likewise, quoting the "[radiated power: power consumption] ratio" doesn't take into account radiated heat (IR) or UV light etc.

The best way is to quote luminous efficiency in lumens/watt which is directly related to the human eye response.

Power LEDs will produce maximum efficiency at the lower end of the drive level range but well above the point where it just lights up. The most efficient currently available power LED is most likely the Cree XP-G R5 bin.

I know the XR-E Q5 produces the most lumens per watt at 20 mA of drive, over 150 lumens per watt. The XP-G? Have no idea but I read something interesting about the levels on the Quark lights.

206 lumens out the front 1.3 hours runtime
85 lumens out the front 5 hours runtime
22 lumens out the front 24 hours runtime
4 lumens out the front 168 hours runtime (7 days)

So I did some math and came up with "lumen-hours" on 2AA cells

Max 268 lumen-hours LED current 700mA
High 425 lumen-hours LED current 250mA
Med 528 lumen-hours LED current 50mA
Low 672 lumen-hours LED current 10mA

The moonlight mode is 1mA but gives 30 days of runtime so the regulator throws off the LED efficiency rating somewhat.

I'm going to put my two settings at Max 206 lumens OTF and Medium 22 lumens out the front so when the Eneloops sag from the 1.5 amp current draw, I can switch to 22 lumen medium to get another hour of usable light.

I was looking at one of those "7 up" devices at Cutter. 7 XP-G R5 LEDs on a 40mm star would be a good test of lower level of drive but still get high output. Seven R5's at 50mA would give 110 lumens (stealing 4sevens OTF readings) and very long runtimes. Seven R5's at 250mA or around 5 watts total would be 600 lumens out the front? Maximum drive would be 350mA to all seven of the LEDs for around 1,000 lumens at the LED. The total power draw at 350mA is 7.6 watts which is easily cooled by a Mag head.

1,000 LED lumens at 7.6 watts sounds great. Running 7 of the LEDs at 10mA for around 30 lumens at .2 watts gives great battery life. An alkaline D cell has 18,000mAH capacity and a 3D has 81 watt hours usable. Say the regulator is only 66% efficient at that level, it still gives over 10 days of 30 lumen light.

It is winter time, the new XP-G is out so some serious mods should start to show in the next few months.

The fantasic White LED lumen testing-thread by jtr1962, shows that almost all LED he tested have highest efficacy when driven with a voltage equal to or just above the band gap energy for the LED. For most white LED's, in which the LED emits 450nm wavelength blue light, it translates to ~2.75V. But at that voltage only roughly 1/10th of the rated output at normal drive current is reached.

space

jtr1962 has done a lot of testing of different LEDs and his results are in both graph and chart form. The latest LEDs like the P7 are towards the back of the thread.
https://www.candlepowerforums.com/threads/89607

Efficiency is a funny thing. It depends on what you are measuring against. Comparing cost of the LED to amount of light would point to using one LED while comparing efficacy would have you spend more money for more LED's to run at lower power. Comparing cost of the controller to useful wattage at the controller often means running multiple LED's in series instead of one LED. After looking at a few datasheets, it seems that many of the LED's will generate half the lumens of full power at 1/3 the wattage of full power, so two LED's instead of one LED could allow for a 50% longer run time in a flashlight. Now we just have to wait for the prices to drop enough that the actual LED die price is the smallest part of the cost.