White LED lumen testing

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
Having acquired quite a few different types of white LEDs over the last two years I've long wanted to see how they stacked up against each other. To that end, and since summer is traditionally my season for indoor projects, I ran a slew of tests. Besides the usual intensity and Vf measurements, I also built a device to enable me to estimate lumen output:

LED_lumens_tester.jpg


I made a cradle to hold my light meter out of silicon rubber for consistency of measurements. An old CD serves as the dial to rotate the LED. Starting at 0°, I first position the LED so that the light meter reading is as high as I can get it in order to obtain a hot spot reading. This is important because the relative value of all the other measurements to the highest reading is what will ultimately determine the accuracy. I record the reading, and then rotate the dial in 5° increments all the way to 180°, recording all the light meter readings. A few times during the test I block off the LED in order to obtain a background reading which needs to be subtracted from all the readings. The value is usually 0.1 to 0.2 lux, and while it may not sound like much, at the 17.5" distance between the LED and light meter it can affect the final result by as much as half a lumen.

After I obtain my results on the test jig running the LED at 20 mA I then take lux readings of the LED at one meter at different currents using the constant current source with the thumbwheel switches to change the current. I also record the LED's Vf. I then plug all the data into an Excel spreadsheet made for that particular LED and come up with the final results. Here is a .zip file containing all my spreadsheets so far for anyone who wants to see how I do my calculations.

This post contains my results for low-power LEDs. For those interested in power LEDs, I collected all of my power LED test results in this post.

Although I have tested many LEDs in this thread, I am not in a position to recommend specific LEDs for use in flashlights or other devices to people. I feel the need to mention this because I have received quite a few PMs asking me which LED would be good for such and such project. I feel it would be best to start a new thread asking for LED recommendations rather than asking me. I just don't have the time or desire or ability to help people determine which LED to use in any given project.

Before showing graphs of the results here is a list and short description of the LEDs I tested so far in order of date of purchase (all are from eBay with the exception of the Hebei LED and the old Nichia):

Nichia(?) 5600 mcd 20° 5mm white (purchased 2001 from Hosfelt Electronics)
A bought a pair of these for $3 each back in 2001 when white LEDs were something of a novelty. In fact, until I started buying on eBay I only had a handful of white and blue LEDs as they were just too pricey to buy in quantity. Anyway, based on the tapered shape and the warm color this LED is probably a Nichia. It actually greatly outperformed the 5600 mcd spec, and did surprising well for a 4-year old white LED, coming in at 28 lm/W efficiency at 20 mA. The tapered shape seems to avoid the ring of light at 60° off-axis as is the case with straight narrow viewing angle 5mm LEDs. As a result, the hot spot is brighter for a given lumen output, and very little light comes out the back.

SpectrumLEDs (noshankyhank at time of purchase) 6,000 mcd, 30° 5mm white (purchased May 2003)
The first white LED I purchased on eBay. Very blue tint.

ChiWing 5,000 mcd 3mm white (purchased July 2003)
Bluish color, wide angle beam.

ChiWing 8,000 mcd 20° 5mm white (purchased July 2003)
Slightly bluish-green, outperformed specs, fairly long-lived based on the one used in my PC as a power indicator. Very low Vf in the 3.00 volt area.

BestHongKong 10,000 mcd 20° 5mm white (purchased February 2004)
Slightly purple, outperformed specs, degrades almost instantly at drive currents above 60 mA.

SpectrumLEDs 8,000 mcd 20° 5 mm white (purchased November 2004)
Decent color but very dim for a late 2004 LED, seemed to degrade rapidly even at 20 mA.

warden_jp2002 26,000 mcd 20° white (purchased February 2005)
Slightly bluish tint, the highest output LED tested by me to date due to a slight wider beam for a 20° LED, no longer available on eBay.

Hebei L5B47VW5C 60° 5mm white (purchased March 2005)
Nice color, wide angle beam, purchased for testing for use in my taxi lighting project. Ultimately rejected for that purpose due to costs and output but still a quality LED, and probably very-long lived.

LCK 12,000 mcd 20° 5 mm white (purchased July 2005)
White with just a hint of blue, greatly outperformed specs, maybe these are LCK's 26,000 mcd LEDs which were put into a 12,000 mcd bag by accident.

BestHongKong 25,000 mcd 20° 5mm UWLC series white (purchased July 2005)
The UWLC series is designed for better color rendering and purer white. These LEDs excel on both counts, especially when rendering reds, and are neither too blue nor too yellow, coming in at a color temp of around 5000K. A slightly wider beam than most 20° LEDs, these come in near the low end of their 14,000 to 25,000 mcd range spec, average in the 16,000 to 17,000 mcd range. The wider beam gives them more output than would be expected for an LED with their intensity.

Light of Victory 35,000 mcd 20° 5mm white (purchased July 2005)
The second best color of any LED I've tested so far. Intensity is a close match to the warden_jp2002 26,000 mcd whites but slightly less in lumen output. These seem a match for the Nichia CS in terms of both lumens and intensity. Low Vf compared to most other white LEDs.

LCK 12,000 mcd 20° 5mm warm white (purchased July 2005)
The first warm white LED I've purchased. Color temp looks like 3200K to 3300K. As with the LCK 12,000 mcd white, it outperformed specs.

Jeled 40,000 mcd 20° 5mm white (purchased December 2005)
Very similar color to the Light of Victory 35,000 mcd, and identical lead frame design. Intensity is nearly 31,000 mcd at 20 mA, a new record for 5mm, 20° white LEDs. Vf is middle of the pack at 3.3V at 20 mA. I also managed to reach over 70,000 mcd at 80 mA, well above my previous records. These intensity records seem to be mostly because the beam is a bit more tightly focused rather than from increased output. Efficiency at 20 mA is a solid 70 lm/W, still less than the warden_jp2002 from early 2005.

BestHongKong 40,000 mcd 20° 5mm white (purchased January 2006)
Note that these are NOT the UWLC series pure white but are optimized for intensity instead. Part number is BEB1000003-100. They are slightly bluish white rather than pure white, maybe about 8000K, but with a very even beam. They are also not anywhere near 40,000 mcd at 20 mA due to a much wider than usual beam. A few of the ones with tighter beams were in the 25,000 mcd area but most were about 20,000 mcd. Vf is a bit on the high side at 3.4V to 3.45V at 20 mA. Thanks to the wide beam, output and efficiency are close to the best I've measured so far. In fact, at 5 mA these set a new efficiency record of nearly 97 lm/W. Although I didn't test at lower currents, they may well break 100 lm/W at 2 or 3 mA. Given the results of this test and the Jeled 40,000 mcd, there hasn't been that much improvement in overall white LED efficiency since early last year, although LEDs of top efficiency are now more widely available (remember that the record setting warden_jp2002 26,000 mcd LEDs last year were a very limited run, and it took until late in the year before comparable LEDs were widely available).

Jeled 50,000 mcd 20° 5mm white (purchased January 2006)
Identical to their 40,000 mcd LED in design, except the color is slightly whiter. I tested three samples and averaged the results. In short, I broke all my previous records by a good margin. Intensity averages around 41,000 mcd at 20 mA and 55,000 mcd at 30 mA. I also reached average intensities over 90,000 mcd at 80 mA. One of the three with a somewhat tighter beam managed over 100,000 mcd at 80 mA. Vf is tending to the high side at 3.4V at 20 mA. Efficiency at 20 mA averaged 82 lm/W, well above previous records. At 5 mA all three samples were well above 100 lm/W, with an average of 110 lm/W. Just for kicks I tested at lower currents to see if efficiency improved further. It did, averaging nearly 125 lm/W at 1 mA, which is well into HID and HPS territory. Assuming a luminous efficacy of 330 lm/W, typical for white LEDs, energy to light conversion efficiency at 1 mA is around 40%. I also noticed an interesting phenomenom. After I test intensity versus current, I measure the intensity and Vf again at 20 mA just to see if the LED's characteristics changed from the excursion to 80 mA. Usually I find little if any change, and it's usually just a slight decrease in intensity from being abused. In this case, I not only found that intensity was higher after the test, but also that Vf was lower! This is very much what happens with a Luxeon. Because of this I redid the entire intensity versus current tests for all three LEDs after the initial "burn-in", for lack of a better word. Overall the changes bumped the average efficiency from about 75 lm/W at 20 mA to the 82 lm/W figure mentioned earlier. The changes also occurred in an LED that was run at 20 mA overnight.

Peak Snow 29 5mm white (acquired February 2006)
CPF member Pinter was kind enough to send me 5 samples of the Peak Snow 29 LEDs for testing. These LEDs have a wider than usual beam and a very even color. Color temperature on one sample looked like about 5500K with the rest falling around 6500K. Efficiency on these was impressive, averaging nearly 80 lm/W at 20 mA for the 5 samples. Even more interesting was that the samples ranged from 70 lm/W all the way up to a bit over 90 lm/W at the nominal 20 mA. Average intensity was 21,260 mcd and average output was 5.11 lumens, both figures about equal to the Nichia CS. This verifies the claim Peak makes which states that the Snow 29 LEDs match the Nichia.

Nichia CS bin B1U 15° 5mm white (acquired February 2006)
Along with the Peak Snow 29 LEDs Pinter also sent me 4 samples of the Nichia CS LEDs. At the nominal 20 mA operating current these came in at close to 75 lm/W, 4.83 lumens output, and 24,100 mcd intensity. The consistency between samples was very impressive as well-the worst measured 70.6 lm/W while the best was 76.0 lm/w, a variation of less than 8%. Where the Nichias really shined (pun intended) was at higher currents. The majority of LEDs I've tested increased little in output beyond about 70 or 80 mA, and so I usually stopped the testing at 80 mA. The 80 mA output was typically about 2.2 to 2.5 times the 20 mA output. On the other hand, the Nichias gave about 2.75 times their 20 mA output at 80 mA, and the curve showed no signs of flattening out. Therefore, I tested up to 100 mA, which is as high as my test jig goes (and the absolute maximum limit for 5mm LEDs). At 100 mA the Nichias were giving nearly 3.1 times the 20 mA output, or on average nearly 15 lumens! To put this into perspective remember that this isn't much less than Luxeons were giving a few years ago, and at 350 mA. Furthermore, the Nichias showed no signs of stress following their brief excursion to 100 mA. Their output at 20 mA once they cooled down was exactly the same as before. If you really need to overdrive a 5mm LED, then the Nichia is the best candidate.

MPJA 15,000 mcd 10mm white (acquired July 2006)
This is one of four different types of LEDs sent to me by CPF member milkyspit for testing. I wasn't expecting much based on the 15,000 mcd rating since this is way under the ratings for similar very narrow beam 10mm LEDs from other manufacturers. The five samples all exceeded their ratings by a large margin, coming in at 41.4, 47.8, 60.0, 62.9, and 81.9 cd at 20 mA. I tested the lumens on the best and worst ones, and averaged the results. Of course the intensity far exceeds anything I've tested to date, but this was expected given the tight focus. The efficiency averages a respectable 60.9 lm/W at 20 mA. Color is somewhat blue, with a color temperature of roughly 9000K. The beam is as smooth as can be expected from an LED with as tight a focus as these, but the bond wires and die are clearly visible. Measured beam angle is only about 8.5° but alot of light falls outside the main beam.

unknown 26,000 mcd 5mm white (acquired July 2006)
This is the second of four different types of LEDs sent to me by CPF member milkyspit for testing. After testing I'm reasonably sure that this isn't one of the warden_jp2002 26K LEDs which milkyspit thought it might be. Rather, the tapered shape causes me to think this is a Nichia. Based on the color and intensity, I'd say it's a Nichia CS 1BU rank T, and this is what I labeled the results as. The three samples came in at 15.5, 16.8, and 17.9 cd at 20 mA. I tested the middle one for lumens. Efficiency is 61.6 lm/W at 20 mA. Unlike the othe Nichia CS LEDs I tested, these don't do as well at higher currents. The output is more or less maxed out at 80 mA. These may in fact not be Nichias, but if so I have no idea what they are.

LS Diodes THC3 4-die 5mm white (acquired July 2006)
This is the third of four different types of LEDs sent to me by CPF member milkyspit for testing. This is the first quad-die LED I've ever tested so the results are quite interesting. The beam is very smooth and white, coming in at about 6000K. Half-intensity beam angle is fairly wide at 27.4° so the intensity is nothing to write home about, reaching only 8,300 mcd at 20 mA. Efficiency at 20 mA is 72.8 lm/W which is very respectable. Note though that since each die is only receiving 5 mA this number isn't as good as it appears at first glance. Where this LED shines is at higher currents. Once you exceed 40 mA it surpasses every 5mm white LED tested to date in efficiency. Even at 100 mA efficiency is still a very decent 50.1 lm/W. As expected, forward voltage is very low compared to other 5mm whites since each die is only getting 1/4 of the current.

SMJLED 4-die 5mm white (acquired July 2006)
This is the fourth of four different types of LEDs sent to me by CPF member milkyspit for testing. As with the LS Diodes THC3, the beam is very smooth and white. Color temperature is about 6500K to 7000K. Beam angle is fairly wide (31.6°) so the intensity is low, reaching only 7,100 mcd at 20 mA. Efficiency at 20 mA is 87.8 lm/W which is a new record. Once again, note that since each die is only receiving 5 mA this number isn't as good as it appears at first glance. Again, this LED shines at higher currents. Once past 20 mA it surpasses every 5mm white LED tested to date in efficiency. At 100 mA efficiency is 57.2 lm/W. As expected, forward voltage is very low compared to other 5mm whites since each die is only getting 1/4 of the current. Lumen output far exceeds any 5mm LED tested to date, reaching 17.67 lumens at 100 mA. While these numbers are great, remember that at 100 mA each die is only getting 25 mA. Just for comparison purposes four Jeled 50,000 mcd whites each running at 25 mA would output 26.5 lumens at an efficiency of 76.7 lm/W. Basically, this means that the dies used in the SMJLED (and the LS Diodes THC3) are decent, but not the best available.

Nichia NSPW310BS-E 60° 3mm white (purchased August 2006)
I purchased 200 of these on eBay to be used for lighting projects requiring very small LEDs. The beam is bluer in the middle, and most of the LEDs seem to have a color temperature in the 6500K to 7500K range. A few are significantly more blue or yellow than this however. Since these are wide angle LEDs intensity wasn't very high, coming in at around 2,600 mcd at 20 mA. Half-intensity beam angle measured 51°. As with the other Nichias I've tested, these do well at higher currents, giving nearly 3 times their 20 mA output when driven at 100 mA. They might do even better soldered to a PC board with heavy traces. Efficiency was low for a white LED these days, coming in at 36.2 lm/W at 20 mA but peaking in the 50 lm/W area when driven at 2 to 5 mA. Nevertheless, 36 lm/W is still 4 to 10 times higher than the efficiencies of the grain of wheat incandescents which LEDs like this might be used to replace. The NSPW310CS, which is the newer, brighter version of this LED, would do about 65 lm/W based on the relative typical intensities given by Nichia.

Jeled PLCC-2 2000 mcd white (purchased August 2006)
I purchased 100 of these PLCC2 surface mount LEDs on eBay to be used for small scale area lighting projects. I also purchased 50 each of the 4 available colored PLCC-2 LEDs (blue, green, red, orange). The beam is very even and color temperature seems to be in the 7000K to 7500K range. Since these are very wide angle LEDs intensity wasn't very high, coming in at around 950 mcd at 20 mA. Half-intensity beam angle measured 112°. Efficiency was unspectacular for a white LED these days, coming in at 43.4 lm/W at 20 mA but breaking 60 lm/W area when driven at 1 to 5 mA. I'll admit to being underwhelmed by these since they are speced at 2000 mcd typical. Had they done so with the same light distribution pattern they would have had efficiencies around 90 lm/W. Nevertheless, the mid 40s isn't horrible, and since I would mainly use these on something non-battery powered like a train layout efficiency isn't terribly important. These would make great miniature imitation metal halide parking lot lights, for example. Had they lived up to spec they would have been almost as efficient as the real ones! The orange ones would do a fair job simulating sodium vapor streetlights. As to why these aren't in the same league as the Jeled 50Ks, my guess is they're not using their best dice, and also the PLCC-2 package basically blocks any light past 90° which might be reflected off the sides or front of the encapsulant. This can be as much as an additional half a lumen, and that would have brought the efficiency up to the low 50s.

BestHongKong 20° 5mm warm white (purchased June 2006)
These are my second purchase of warm white LEDs. They have a hotspot with a color temp in the 3300K to 3500K range and a corona of somewhat lower color temp resembling the light from a sodium vapor street lamp. Overall I'd say the average color temperature is around 3000K. The intensity is around 14,500 mcd at 20 mA, Vf is a low 3.12V, beam angle is 11.9°, and overall efficiency is an unspectacular for these days 37 lm/W. Remember though that warm white LEDs tend to have somewhat lower efficiencies than their cool white cousins. Overall the output of this LED is much closer to the incandescent it seeks to replace than the greenish hued LCK warm whites I tested last year.

LCK 40,000 mcd 15° 5mm white (acquired January 2007)
This is one of four different types of LEDs sent to me by CPF member TMorita for testing. The three samples came in at 30.2, 30.9, and 31.3 cd at 20 mA. The average output at 20 mA was 4.91 lumens, average Vf was 3.26V, and average efficiency was a very respectable 75.2 lm/W. There wasn't much variation between samples in any of the parameters. The beam was somewhat blue, perhaps 8000K, but very smooth. Measured beam angle was 15.1°. These seemed to have less efficiency loss at higher currents than most other eBay LEDs I've tested so far. Even at 80 mA they managed 46.6 lm/W.

Jeled 50,000 mcd 25° 5mm white test #2 (acquired January 2007)
This is the second of four different types of LEDs sent to me by CPF member TMorita for testing. The three samples came in at 32.6, 36.2, and 30.1 cd at 20 mA. The average output at 20 mA was 5.01 lumens, average Vf was a very low 3.07V, and average efficiency was a very respectable 81.5 lm/W. There was a little more variation between samples than with the LCK 40Ks in terms of beam angle and intensity but overall output and Vf were close for all the samples. The beam was a little blue but somewhat blothy. Average color temperature looked around 6500K to 7000K. Measured beam angle was 13.5°. As with the LCK 40Ks, these seemed to have less efficiency loss at higher currents than most other eBay LEDs I've tested so far. At 80 mA they managed an even better 49.4 lm/W. It seems plausible that these and the LCK 40Ks may be using the same chip since the test results are similar. Since this is my second test of the Jeled 50Ks there appear to be some notable differences from my first test a year ago. Efficiency is about the same but Vf is much lower so total output is about half a lumen less. Efficiency at higher currents is much improved, as is total output. The original Jeled 50Ks only managed about 2.3 times their 20 mA output at 80 mA, or roughly 13 lumens. The newest batch manages about 2.8 times and 14 lumens while having about 21% greater efficiency. The only downside is that the efficiency of the newer Jeled 50Ks peaks at around 5 to 10 mA in the 90 lm/W area while the older ones averaged as high as 125 lm/W at 1 mA and close to 100 lm/W at 10 mA. You can see the differences below with a side-by-side comparison on the same chart (new in white, old in red). If I had to guess what the change was, I'd say a larger physical size, less efficient chip was used. This would account for the lower Vf and lower peak efficiency. However, by virtue of lower current density efficiency is actually better from about 20 mA and up. This is actually a very sensible decision since if anything 5mm white LEDs tend to be overdriven, not underdriven. Perhaps the larger chip was also used to help lumen maintenance. Lumen maintenance is probably getting more important for inexpensive Chinese LEDs since many are being used in Christmas lights where short lifetime would be very apparent. On another note, based on this test and the LCK 40Ks I'm somewhat disappointed that in a year 5mm LED efficiency has been stagnant. I was expecting 100+ lm/W at 20 mA by summer 2006.

Jeled_50000_Comparison.gif



LCK 35,000 mcd 2-die 30° 5mm white (acquired January 2007)
This is the third of four different types of LEDs sent to me by CPF member TMorita for testing. The three samples came in at 8.8, 8.0, and 9.0 cd at 20 mA. The average output at 20 mA was 3.50 lumens, average Vf was a very low 3.02V, and average efficiency was a decent 58.0 lm/W. There wasn't much variation between samples in any of the parameters. The beam was a pure white similar to the BestHongKong UWLC series with a color temperature of around 5000K, and very smooth. Measured beam angle was 21.3°. Even though these are two-chip LEDs they didn't do as well at higher currents as the LCK 40K and the second batch of Jeled 50Ks in terms of output. However, because they are dual die they will probably last longer at higher currents. Note that even at their rated 40 mA these don't come anywhere near their ratings of 35,000 mcd and 30° beam angle. However, this isn't unusual for eBay LEDs. The only way to know what you're getting is to buy it and test it. For what these cost they are still a decent value even if they don't live up to their billing.

Jeled 15,000 mcd 140° superflux white (acquired January 2007)
This is the fourth of four different types of LEDs sent to me by CPF member TMorita for testing. I only tested one of the three samples since the other two appeared very similar in output. The result was 2.9 cd and 3.46 lumens at 20 mA. Beam angle was 60.6°. Vf was 3.26V and overall efficiency was 53.1 lm/W. Color was very blue, perhaps as high as 9000K, and the beam was very smooth. Of course, these figures are nowhere near the claimed output of 15,000 mcd and a 140° beam angle. However, consider that an LED with those specs would have a total output of around 70 lumens so I wasn't expecting this part to meet specs. If it did, it would have had an efficiency in the 1000 lm/W area which would be physically impossible! Overall this is a decent LED which should give long life owing to the superior heat dissipation of the superflux or pirahna package.

New LS Diodes THC3 4-die 5mm white (acquired February 2007)
CPF member milkyspit sent me a new THC3 and a Quickar 5m100w for comparison testing. The beam is very smooth and white, coming in at about 6500K. Half-intensity beam angle is fairly wide at 29.4° so the intensity is only 6,400 mcd at 20 mA. Efficiency at 20 mA is 65.5 lm/W, a little lower than the original THC3 which was 72.8 lm/W. As with other 4-die LEDs this one does well at higher currents. Even at 100 mA efficiency is still a very decent 47.6 lm/W. As expected, forward voltage is very low compared to other 5mm whites since each die is only getting 1/4 of the current.

Quickar 5m100w 4-die 5mm white (acquired February 2007)
This LED is similar to, but less expensive than, the LS Diodes THC3. In a head-to-head comparison with the new THC3 it did very well. The beam is very smooth and white, quite similar to the THC3, and with a similar color temperature of roughly 6500K. Half-intensity beam angle is fairly wide at 33.9° so the intensity is a mere 5,600 mcd at 20 mA. Efficiency at 20 mA is 63.2 lm/W, a little lower than the new THC3. At 100 mA efficiency is a very decent 46.4 lm/W, just a hair under what the new THC3 does. Once again, forward voltage is very low compared to other 5mm whites since each die is only getting 1/4 of the current.

SuperbrightLEDs 7,500 mcd 90° superflux (acquired February 2007)
CPF member TMorita sent me this LED along with a bunch of others for testing. The result was 2,600 mcd and 3.83 lumens at 20 mA. Beam angle was 72.0°. Vf was a very low 3.04V and overall efficiency was a decent 63.0 lm/W. Color was pure white, about 6500K, and the beam was very smooth. This is a decent LED which should give long life owing to the superior heat dissipation of the superflux or pirahna package.

LVEHK 10,000 mcd 20° 3mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 3.0, 3.0, and 3.6 cd at 20 mA, way less than spec. Average beam angle was 29.7°, somewhat wider than claimed, which explains the lower intensity. Color was somewhat blue, around 8000K. Average efficiency of the three samples was only 25.9 lm/W, average output was 1.67 lumens, and average Vf was 3.22V. Although efficiency of these isn't that great, it is much better than the 2 to 4 lm/W of the tiny incandescents which they might be used to replace.

SpectrumLEDs 2,000 to 4,000 mcd 180° 4.8mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. These are the 4.8mm "top hat" package which is designed for area lighting applications. The results were 1.75, 1.84, and 1.81 cd at 20 mA, around the low end of the spec. Average beam angle was 105.6°. Color temperature was 6500K, which is pure white to most people. Even though the intensity was low because of the wide beam angle average efficiency of the three samples was an excellent 79.2 lm/W. Average output was 5.10 lumens, and average Vf was 3.22V. Consistency between samples was impressive-outputs were 5.06, 5.15, and 5.10 lumens. Overall these are great LEDs for any application required flood lighting, or perhaps to focus with a reflector.

SuperbrightLEDs 18,000 mcd 15° 5mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 24.6, 23.3, and 24.2 cd at 20 mA, higher than spec. Average beam angle was 15.3°. Color temperature was a slightly but not offensively blue 7000K. The beam was smooth. Average efficiency of the three samples was a decent 66.0 lm/W. Average output was 3.99 lumens, and average Vf was a very low 3.02V. Consistency between samples was impressive-outputs were 4.12, 3.76, and 4.09 lumens. Overall these LEDs are solid performers.

SuperbrightLEDs 18,000 mcd 30° 5mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 13.1, 13.2, and 12.3 cd at 20 mA, way less than spec. Average beam angle was 20.2°, also less than spec. Color temperature was a match to the 15° version of this LED, coming in at 7000K. The beam was smooth. Average efficiency of the three samples was a solid but not overly impressive for these days 50.7 lm/W. Average output was 3.19 lumens, and average Vf was 3.14V. Consistency between samples was even more impressive than the 15° version-outputs were 3.10, 3.24, and 3.22 lumens. Overall these LEDs are decent even though they didn't meet their specs.

BestHongKong 35,000 mcd 20° 5mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 29.0, 29.1, and 29.5 cd at 20 mA, slightly less than spec. Average beam angle was 14.7°. Color temperature averaged around 7000K but the beam was a little blotchy, with a blue center and yellowish corona. Average efficiency of the three samples was a decent 65.8 lm/W. Average output was 4.53 lumens, and average Vf was a somewhat high 3.44V which hurt the efficiency numbers. Consistency between samples was quite impressive-outputs were 4.53, 4.64, and 4.41 lumens. The high Vf and poor scaling of output with current (output at 80 mA was only 2.06 times the output at 20 mA) indicates that these LEDs likely use small but efficient dice.

LVEHK 55,000 mcd 20° 5mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 28.6, 28.4, and 29.7 cd at 20 mA. Note however that the 55,000 mcd spec is at 25 mA. Average intensity of the three samples at 25 mA was 34.6 cd, still somewhat below spec. Average beam angle was 14.0°. Color temperature averaged around 7000K but the beam was a little blotchy, with a blue center and yellowish corona. Average efficiency of the three samples was a decent 64.1 lm/W. Average output was 4.15 lumens, and average Vf was 3.23V. Consistency between samples was decent-outputs were 4.30, 4.20, and 3.94 lumens. Output scales very well with current (output at 80 mA is 2.53 times the output at 20 mA), and efficiency numbers increase by only about 25% at lower currents compared to at 20 mA, indicating that these LEDs use large dice. While these LEDs are decent, and comparable in performance to the BestHongKong 35,000 mcd, I was disappointed that they didn't even come close to their brightness and beam angle specs.

LVEHK 65,000 mcd 40° 8mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The beam pattern was somewhat unusual with a slightly darker spot in the center, and the brightest part of the beam around the dark spot. The results (dead center of the beam) were 6.8, 5.2, and 7.5 cd at 20 mA, way less than spec. About 5° off-axis these numbers were around 15% higher. Average beam angle was 31.1°. Color temperature was quite blue, about 9000K. Average efficiency of the three samples was a not too impressive 43.2 lm/W. Average output was 3.06 lumens, and average Vf was a very high 3.55V which definitely hurt the efficiency numbers. Consistency between samples was good-outputs were 3.16, 2.91, and 3.13 lumens. Despite the high Vf indicating a small die output scales very well with current (output at 80 mA was 2.65 times the output at 20 mA), perhaps because the larger lead frame of the 8mm package conducts away heat better.

LVEHK 140,000 mcd 12° 10mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 63.5, 59.8, and 59.8 cd at 20 mA, well below spec. Average beam angle was 7.5°. Color temperature averaged around 7000K to 7500K but the beam was reasonably smooth, at least as smooth as can be expected from LEDs as tightly focused as these. Average efficiency of the three samples was a decent 67.6 lm/W. Average output was 4.44 lumens, and average Vf was 3.29V. Consistency between samples was decent-outputs were 4.59, 4.39, and 4.35 lumens. Output scales very well with current (output at 80 mA is 2.57 times the output at 20 mA), and efficiency numbers increase by only about 25% at lower currents compared to at 20 mA, indicating that these LEDs use large dice. These are decently performing LEDs overall.

BestHongKong 135,000 mcd 12° 10mm white (acquired February 2007)
TMorita sent me 3 samples of this LED for testing. The results were 166.2, 129.0, and 144.9 cd at 20 mA, all over or very close to spec. Average beam angle was an extremely narrow 6.5°, almost all hotspot with virtually no sidespill. Color temperature averaged a blotchy, slightly yellow 5000K but the blotchy beam quality is to be expected from LEDs as tightly focused as these. Average efficiency of the three samples was an excellent 79.6 lm/W. Average output was 5.14 lumens, and average Vf was 3.23V. Consistency between samples wasn't that great-outputs were 5.54, 4.67, and 5.18 lumens. Output scales very poorly with current (output at 80 mA was only 2.16 times the output at 20 mA), indicating that these LEDs likely use very small but very efficient dice. The much higher efficiency numbers at low currents (in the 115 lm/W area at 1 to 2 mA) further support this conclusion. A single one of these LEDs can make a very impressive flashlight with plenty of throw.

Nichia NFSW036L, bin T, tint C1 white (acquired February 2007)
CPF member TMorita sent me this LED along with a bunch of others for testing. The result was 3,340 mcd and 4.02 lumens at 20 mA. Beam angle was 51.3°. Vf was an extremely low 2.72V and overall efficiency was a decent 73.9 lm/W. Color was similar to a cool-white fluorescent, about 4500K, and the beam was very smooth. The LED's output scaled very well with current, giving 3.57 times the output at 100 mA as it did at 20 mA. Efficiency at 100 mA was a decent 49.7 lm/W. However, all this is to be expected since this LED is designed to operate at 150 mA with a heat sink. This is also why the Vf was very low. Although my tester couldn't go above 100 mA by looking at the relative outputs in the spec sheet for this LED I estimated output at 150 mA to be around 20 lumens. This is right on spec.

SuperbrightLEDs RL5-W45-360 360° 5mm white (acquired February 2007)
CPF member TMorita sent me this LED along with a bunch of others for testing. The very wide but very low output presented special measurement difficulties. Even at 17.5" distance of my testing jig the intensity hitting my light meter was only in the 2 lux area, and the scattered background light accounted for about one-third of this. Measuring the intensity at 1 meter proved impossible at 20 mA so I did it at 100 mA and used the intensity ratio measured at 17.5" to calculate my results. The result was a mere 280 mcd. Calculated output was 4.18 lumens at 20 mA, close enough to the 4500 mlm (4.5 lm) spec. 50% beam angle was nonexistent since the LED more or less emitted in a complete 360° circle. Vf was 3.16V and overall efficiency was a decent 66.2 lm/W. Color was slightly blue, about 7000K. These are very unusual LEDs in that it appears they have their phosphor deposited on the outer surface of the package. They appear yellow unlit but once lit they are white. Any application where an LED seeks to mimic a light bulb in terms of intensity distribution would be a good one for these.

SuperbrightLEDs RL8-W110-360 360° 8mm white (acquired February 2007)
CPF member TMorita sent me this LED along with a bunch of others for testing. This is an 8mm version of the 360° LED described above. Again, the very wide but very low output presented special measurement difficulties as described for the RL5-W45-360. The intensity at 20 mA was only 280 mcd. Calculated output was 4.48 lumens at 20 mA. At 60 mA, which is what these LEDs are speced at, output is 10.54 lumens, slightly above the 9000 mlm (9.0 lm) spec. 50% beam angle was nonexistent since the LED more or less emitted in a complete 360° circle. At 20 mA Vf was a very low 2.85V and overall efficiency was an excellent 78.6.2 lm/W. Corresponding numbers at 60 mA were 3.12V and 56.3 lm/W. Color was cool-white fluorescent white, about 4500K. These are very unusual LEDs in that it appears they have their phosphor deposited on the outer surface of the package. They appear yellow unlit but once lit they are white. Any application where an LED seeks to mimic a light bulb in terms of intensity distribution would be a good one for these.

ISP 22,000 mcd 25° 5mm white (acquired March 2007)
Three of these were sent to me by CPF member TMorita for testing, and are special order using one large single die for operation at 100 mA. They are not the stock 2-die ones rated at 40 mA. The three samples came in at 6.4, 6.6, and 7.1 cd at 20 mA. The average output at 20 mA was 3.37 lumens, average Vf was a very low 2.94V, and average efficiency was a decent 57.3 lm/W. There wasn't much variation between samples in any of the parameters, especially the forward voltage. The beam was slightly blue, with a color temperature of around 7000K, and very smooth. Measured beam angle was 25.7°, which slightly exceeds the advertised 25°. At higher currents these did very well, giving slightly over 13 lumens at 100 mA. At their rated 100 mA these manage 25,900 mcd, exceeding their 22,000 mcd rating. Overall these are decent LEDs, and probably quite long lasting at higher currents due to their large-die construction.

LCK 55,000 mcd 15° 5mm white (acquired March 2007)
TMorita sent me 3 samples of this LED for testing. The results were 27.4, 27.4, and 27.9 cd at 20 mA. Average beam angle was only 10.6°, so the less than rated intensity wasn't due to a wider beam angle as is sometimes the case. Color temperature averaged around 7000K but the beam was a little blotchy, with a blue center and yellowish corona. Average efficiency of the three samples was a decent 59.8 lm/W. Average output was 3.79 lumens, and average Vf was 3.17V. Consistency between samples was quite good-outputs were 3.80, 3.73, and 3.84 lumens. Output scales very well with current (output at 80 mA is 2.65 times the output at 20 mA), and efficiency numbers increase by only about 25% at lower currents compared to at 20 mA, indicating that these LEDs use large dice. These LEDs are decent but I was disappointed that they didn't even come close to their brightness and beam angle specs.

Jeled 55,000 mcd 15° 5mm white (acquired March 2007)
CPF member Minjin sent me two of these for testing. The two samples came in at 35.2 and 32.4 cd at 20 mA. The average output at 20 mA was 5.11 lumens, average Vf was a very low 3.05V, and average efficiency was a very respectable 83.9 lm/W. This means that the Jeled 55Ks squeak by for a new efficiency record for 5mm single die LEDs at 20 mA. The beam was a little blue but somewhat blothy. Average color temperature looked around 6500K. Measured beam angle was 14.1°. These have less efficiency loss at higher currents than most other eBay LEDs I've tested so far. At 80 mA they managed 50.9 lm/W. Even at 100 mA they are a decent 42.1 lm/W (and over 15 lumens!). Peak efficiency numbers increase by only about 20% at lower currents compared to at 20 mA, indicating that these LEDs use large dice. While I'm glad to have finally measured a new efficiency record, I remain disappointed that in over a year 5mm LED efficiency has more or less been stagnant. Since 5mm LED efficiency usually precedes the next jump in power LED efficiency by up to a year, and there are planned increases in power LED efficiency for 2008, I remain hopeful that we'll have 5mm LEDs of 100+ lm/W at 20 mA within the next few months.

Hebei 520MW7C 20° 5mm white (acquired June 2007)
CPF member LEDude sent me 10 samples of this LED for testing. I picked three of the ten at random to test. The results were 15.6, 14.1, and 15.9 cd at 20 mA. Average beam angle was 19.3°, very close to spec. Color temperature averaged around 7000K-7500K and the beam was a very smooth. Average efficiency of the three samples was a decent 65.1 lm/W. Average output was 3.98 lumens, and average Vf was a low 3.06V. Consistency between samples was quite good-outputs were 4.04, 4.01, and 3.88 lumens. Output scales fairly well with current (output at 80 mA is 2.42 times the output at 20 mA). It's been over two years since I last tested Hebei LEDs. These LEDs are quite an improvement in efficiency over their previous offerings.

Hebei 530MW7C 30° 5mm white (acquired June 2007)
CPF member LEDude sent me 10 samples of this LED for testing. I picked three of the ten at random to test. The results were 12.0, 12.3, and 11.4 cd at 20 mA. Average beam angle was 22.1°, somewhat under spec. Color temperature averaged around 7000K with a very smooth beam. Average efficiency of the three samples was a decent 58.9 lm/W. Average output was 3.55 lumens, and average Vf was a very low 3.01V. Consistency between samples was quite good-outputs were 3.57, 3.49, and 3.60 lumens. Output scales fairly well with current (output at 80 mA is 2.41 times the output at 20 mA). Overall, performance is very similar to the 520MW7C but that's to be expected since they likely use the same die, but moved further towards the front of the epoxy package for a wider beam angle.

Hebei 550MW7C 50° 5mm white (acquired June 2007)
CPF member LEDude sent me 10 samples of this LED for testing. I picked three of the ten at random to test. The results were 5.1, 4.5, and 4.9 cd at 20 mA. Average beam angle was 47.9°,very close to spec. Color temperature averaged around 6500K with a very smooth beam. Average efficiency of the three samples was a decent 66.5 lm/W. Average output was 4.14 lumens, and average Vf was a low 3.12V. Consistency between samples was quite good-outputs were 4.36, 4.07, and 4.00 lumens. Output scales fairly well with current (output at 80 mA is 2.51 times the output at 20 mA). Performance is very similar to the 520MW7C and 5307C as expected. All three of these LEDs are quite decent performers given the low cost, and quite well made.

Jeled 210000 mcd 10mm warm white (acquired December 2007)
CPF member knabsol sent me 3 samples of this LED for testing. The results were 5.1, 5.5, and 5.6 cd at 20 mA. Average beam angle was 26.7°, less than the specified 40°. Color temperature was very low, perhaps 2500K, and with a very smooth beam. In all honesty these looked more like sodium vapor lamps than anything else. Average efficiency of the three samples was only 38.1 lm/W. Average output was 2.20 lumens, and average Vf was a low 2.88V (expected due to the 5-die construction). Consistency between samples was not so good-outputs were 1.81, 2.27, and 2.53 lumens. Output scales very well with current as expected (output at 100 mA is 3.47 times the output at 20 mA). Even at 100 mA the intensity of these averages only 18.7 cd, well below the spec of 210 cd. Not sure what to make of these really. The large case size seems to make the beam smoother rather than focus it.

Jeled 255000 mcd 10mm white (acquired December 2007)
CPF member knabsol sent me 3 samples of this LED for testing. The results were 8.9, 9.9, and 11.3 cd at 20 mA. Average beam angle was 24.2°, less than the specified 30°. Color temperature was around 7500K, and the beam was very smooth. Average efficiency of the three samples was a decent 60.6 lm/W. Average output was 3.48 lumens, and average Vf was a low 2.87V (expected due to the 5-die construction). Consistency between samples was OK-outputs were 3.30, 3.31, and 3.84 lumens. Output scales very well with current as expected (output at 100 mA is 3.70 times the output at 20 mA). At 100 mA the intensity of these averages only 37.2 cd, well below the spec of 255 cd. As with their warm white cousins, the large case size seems to make the beam smoother rather than focus it. In both cases performance seemed disappointing given the 5-die construction.

Oatley Electronics 80000 mcd 10mm white (acquired December 2007)
CPF member lctorana sent me 3 samples of this LED for testing. The results were 5.7, 6.1, and 7.1 cd at 20 mA. Average beam angle was 38.8°. Color temperature was around 7500K, and the beam was very smooth. Average efficiency of the three samples was a very decent 72.4 lm/W. Average output was 4.26 lumens, and average Vf was a low 2.94V. (expected due to the multi-die construction). Consistency between samples was excellent-outputs were 4.06, 4.31, and 4.41 lumens. Output scales very well with current as expected (output at 100 mA is 3.60 times the output at 20 mA). At 100 mA the intensity of these averages only 22.7 cd, well below the spec of 80 cd. These are supposed to be able to output 25 lumens at 150 mA. My test jig only went to 100 mA, and the average output at that current was 15.34 lumens. At 150 mA output would probably be around 20 or 21 lumens, not much below the 25 lumen spec. Overall, these are pretty decent LEDs. It seems in all cases that the large 10 mm case size seems to make the beam smoother rather than focus it. For maximum focusing it appears that the 8mm case is best based on my tests so far.

Deal Extreme 20mm white (acquired March 2008)
CPF member nein166 loaned me one of these for testing. This thing is huge! The die area and leads are much larger than on normal indicator type LEDs. Results at 20 mA are 6.6 cd, 4.23 lumens, and 70.7 lm/W. Beam angle is 39.1°, and Vf is a very low 2.99V. Output scales well with current, and at 100 mA is nearly 3 times that of 20 mA. Beam is extremely smooth and around 7000K. These make great area lights. I would imagine that they would have a very long lifetime even at 100 mA due to the superior thermal path.

BestHongKong 25000 mcd 5mm 5-die white (acquired March 2008)
CPF member TMorita sent me 3 samples of this LED for testing. These are BestHongKong's 5-die whites which have 5 dies and a larger lead frame to allow operation at 100 mA. The results were 6.0, 6.5, and 6.9 cd at 20 mA. I assume the 25,000 mcd rating is at 100 mA, so the corresponding results at 100 mA are 24.3, 25.8, and 28.1. The average is 26.1 cd, slightly above spec. Average beam angle was 34.7°, less than the specified 50°. Color temperature was around 7000K, and the beam was very smooth. Average efficiency of the three samples was a decent 72.8 lm/W at 20 mA and 52.9 lm/W at 100 mA. I was expecting somewhat better than that due to the 5-die construction. Apparently they're not using the most efficient dies for these. Average output was 4.07 lumens at 20 mA, and average Vf was a low 2.87V (expected due to the 5-die construction). Corresponding figures at 100 mA were 16.41 lumens and 3.10V. The 16.41 lumens comes reasonably close to the specified 20 lumens at 100 mA. Consistency between samples was excellent-outputs were 3.91, 4.14, and 4.16 lumens at 20 mA. Output scales very well with current as expected (output at 100 mA is 4.03 times the output at 20 mA). Overall these appear to be a decent, well-constructed LED capable of sustained operation at higher currents.


LED Tech 14000 mcd 5mm 100 mA white (acquired March 2008)
CPF member TMorita sent me 3 samples of this LED for testing. These are LED Tech's 100 mA whites which have a larger lead frame to allow operation at 100 mA, and probably also multiple dies. The results were 2.8, 2.8, and 3.2 cd at 20 mA. I assume the 14,000 mcd rating is at 100 mA, so the corresponding results at 100 mA are 11.0, 11.2, and 12.3. The average is 11.5 cd, somewhat below spec. Average beam angle was 77.1°, nearly equal to the specified 80°. Color temperature was around 7000K, and the beam was very smooth. Average efficiency of the three samples was a very decent 86.8 lm/W at 20 mA and an excellent 63.8 lm/W at 100 mA. These are better numbers than the very similar BestHongKong LEDs because they're apparently using better dies. Average output was 5.12 lumens at 20 mA, and average Vf was a low 2.95V (expected due to the multi-die construction). Corresponding figures at 100 mA were 20.06 lumens and 3.15V. Consistency between samples was OK-outputs were 4.71, 5.05, and 5.60 lumens at 20 mA. Output scales very well with current as expected (output at 100 mA is 3.92 times the output at 20 mA). Overall these appear to be a decent, well-constructed LED capable of sustained operation at higher currents.


Nichia NSPS500GS-K1 5mm white (acquired March 2008)
CPF member TMorita sent me 3 samples of this LED for testing. These are Nichia's latest and greatest 5mm whites. The results were 36.0, 39.2, and 39.8 cd at 20 mA. I think these are speced for 44,000 mcd, so the average of 38.3 cd is only slightly out of spec. Average beam angle was 13.4°, a little less than the specified 15°. Color temperature was around 7000K, and the beam was very smooth. Average efficiency of the three samples was a excellent 96.7 lm/W at 20 mA, so these take the new efficiency crown. Even at 100 mA efficiency manages to remain at nearly 50 lm/W. Average output was 6.15 lumens at 20 mA, and average Vf was 3.18V. Corresponding figures at 100 mA were 21.43 lumens and 4.34V. Given the rather steep increase in Vf with current, these don't appear to be multi-die as some here have speculated. Consistency between samples was OK-outputs were 5.75, 6.19, and 6.52 lumens at 20 mA. Output scales very well with current despite the apparent single-die construction (output at 100 mA is 3.48 times the output at 20 mA). This isn't much worse than the two types of multidie LEDs just tested. It looks like Nichia has another winner here, and we have a new efficiency champ.


Cree C503B-WAN-CCACB231 5mm white (acquired January 2009) NEW!
CPF member Holepuncher sent me 3 samples of this LED for testing. These are Cree's brightest commercially available 5mm whites. The results were 26.9, 26.2, and 25.4 cd at 20 mA. These are speced for 23,500 to 32,900 mcd according to the datasheet, so the average of 26.17 cd is well within spec, although on the low side. Average beam angle was 15.1°, dead on the specified value of 15°. Color temperature was around 6500K, and the beam was extremely smooth. Average efficiency of the three samples was an excellent 79.2 lm/W at 20 mA. At 100 mA efficiency drops to 31.5 lm/W. Average output was 4.93 lumens at 20 mA, and average Vf was a low 3.11V. Corresponding figures at 100 mA were 12.36 lumens and 3.92V. Consistency between samples was great-outputs were 4.67, 5.16, and 4.97 lumens at 20 mA. Output doesn't scale as well with current as some other recent 5mm LEDs I've seen, but it's not horrible, either. Overall these appear to be great LEDs, and could be even better if they used Cree's best dice. Just doing some rough assumptions regarding phosphor conversion efficiency of 80% and package efficiency of 80%, I've calculated that these are using blue dice with an output of about 24 mW @ 20 mA. Cree's best dice produce 33 to 35 mW. This would produce a white LED of 110 to 115 lm/W, perhaps even as high as 125 lm/W in a wide-angle package.


cece718 6000 mcd flat top 3mm white (acquired January 2009)
I purchased 100 of these for use in model railroading lighting applications. These are the flat top 3mm LEDs sold on eBay by seller cece718. Color temperature is around 7000K, and the beam is very smooth. I knew these couldn't come anywhere close to their 6000 mcd spec given their wide angle. I measured 800 mcd at 20 mA. Output was 3.49 lumens and efficiency was an OK for nowadays 54.1 lm/W. Beam angle was a very wide 129°. Output doesn't increase much with current, and levels off at 5.91 lumens at 60 mA. This isn't surprising given the poorer thermal properties of the 3mm package, and in service I doubt I would run these LEDs over 10 mA. Overall, these LEDs aren't bad given their price. In fact, I was pleasantly surprised they performed as well as they did. It was only a few years ago that only the best LEDs managed 50 or 60 lm/W. Nowadays most commodity LEDs seem to reach that mark.




Osram LW E6SG PLCC4 white (acquired May 2009) NEW!
I purchased 200 of these for use in model railroading lighting and LED night light applications. These are surface mount PLCC-4 style LEDs. Color temperature is around 5500K, somewhat warmer than usual for small LEDs and the beam is very smooth. Intensity was 1010 mcd at 20 mA. Output was 3.15 lumens and efficiency was an OK for nowadays 48.8 lm/W. Beam angle was a very wide 119°. Output scales well with current, and was 8.07 lumens at 100 mA. The claim to fame of these LEDs is a special silicon encapsulant which gives a lifetime of 50,000 hours at 25 mA and 25°C. I put them in 9 LED night lights so far in order to attempt to validate this claim. The LEDs which were previously in these night lights faded to practically nothing after 3 years of continuous use.

I'll be adding to this list as I acquire more white LEDs...

Here are the graphs for efficiency versus current:

Nichia_5600.gif

SpectrumLEDs_6000.gif

ChiWing_5000_3mm.gif

ChiWing_8000.gif

BHK_10000.gif

SpectrumLEDs_8000.gif

Hebei_L5B47VW5C.gif

warden_jp2002_26000.gif

LCK_12000.gif

BHK_UWLC_25000.gif

LoV_35000.gif

LCK_12000WW.gif

Jeled_40000.gif

BHK_40000.gif

Jeled_50000.gif

Peak_Snow_29.gif

Nichia_CS_B1U.gif

MPJA_15000_10mm.gif

Nichia_CSB1U-T.gif

LS_Diodes_THC3.gif

SMJLED.gif

Nichia_NSPW310BS.gif

Jeled_PLCC2.gif

BHK_5mm_warm_white.gif

LCK_40000.gif

Jeled_50000_2.gif

LCK_35000_2_die.gif

Jeled_15000_superflux.gif

LS_Diodes_new_THC3.gif

Quickar_5m100w.gif

SuperbrightLEDs_7500_superflux.gif

LVEHK_10000_3mm.gif

SpectrumLEDs_2000_4_8mm.gif

SuperbrightLEDs_18000_15.gif

SuperbrightLEDs_18000_30.gif

BHK_35000.gif

LVEHK_55000.gif

LVEHK_65000_8mm.gif

LVEHK_140000_10mm.gif

BHK_135000_10mm.gif

Nichia_NFSW036L.gif

SuperbrightLEDs_RL5_W45_360.gif

SuperbrightLEDs_RL8_W110_360.gif

ISP_22000.gif

LCK_55000.gif

Jeled_55000.gif

Hebei_520MW7C.gif

Hebei_530MW7C.gif

Hebei_550MW7C.gif

Jeled_210000WW_10mm.gif

Jeled_255000_10mm.gif

Oatley_80000_10mm.gif

DX_20mm_white.gif

BHK_25000_5-die.gif

LEDTech_14000_100mA.gif

Nichia_NSPW500GS-K1.gif

Cree_503B_WAN_CCACB231.gif

cece718_flat_top_3mm.gif

Osram_LW_E6SG_PLCC4.gif




[/SIZE]
Here is a series of head-to-head comparisons of all the LEDs for a number of different parameters:

Vf_Comparison.gif

Lumen_Comparison.gif

Efficiency_Comparison.gif

Intensity_Comparison.gif


Note how there is more or less (with the exception of the SpectrumLEDs 8000 mcd white) a steady improvement in efficiency from 2003 until the present. Most of the LEDs clustered together in the efficiency graphs were purchased at roughly the same time.

Here is a chart summarizing the results at 20 mA test current:

White_LED_Comparison_Chart.gif


As to the accuracy of my results, for a sanity check the Light of Victory 35,000 mcd seem a close match to the Nichia CS in terms of intensity and lumens, and the Nichias are supposed to have efficiencies in the 60 lm/W area at 20 mA. This is exactly where the Light of Victory 35,000 mcd falls. The warden_jp2002 actually is around 74 lm/W at 20 mA. I'm guessing someone got a supply a Cree's XT-24 chips for those. Nothing else could give efficiencies that high. Also note how these same LEDs break 90 lm/W at 5 mA. (Nothing thus far has beat these LEDs, which were available for a fairly short time on eBay.) - Note: The warden_jp2002 LEDs were finally beaten by the Jeled 50000 mcd ones, but they held the record for nearly a year.

Update 04/10/2008: The Nichia NSPW500GS-K1 takes the crown now at 96.7 lm/W-tantalizingly close to the magic 100 lm/W mark.
 
Last edited:

chimo

Flashlight Enthusiast
Joined
Sep 16, 2004
Messages
1,905
Location
Ottawa, Canada
jtr1962, great work! I d/l'd the zip file but I got errors in it. Could you check it out? I would like to compare you data to mine. Thanks,

Pual
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
How are you generating a lumens value from the set of lux measurements? Are you doing a series of spherical caps, and summing up the intensity/area for each cap to get a total lumen output?

I was thinking of doing this a while ago, but never got around to it - I wanted to see how accurate my lumens coefficient was, and how close the lumileds datasheet was in reagards to the beam pattern of a luxeon high done.
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
Yes evan9162, that's exactly how I'm doing it. For example, I take the relative lux readings for 0° and 5°, average them, multiply by the area in steradians, and then multiply by the intensity in candelas. After that, I'll do the same with the 5° and 10° values, etc, and add this to my running total. It's interesting that for narrow viewing angle (i.e. 15° or 20°) LEDs the traditional methods of calculating lumens in the main beam grossly underestimates the output. One can see why by examining the radiation pattern of one of these LEDs. Some light falls outside the main beam but this decreases in intensity rather quickly. Still, since the spherical area is larger as the angle increases, this spill still makes significant contributions to the output. After that you usually have a bright ring about 60° off axis. This shows up clearly in the graphs in my spreadsheets. While the intensity might only be 2% or 3% of the maximum intensity, it is spread over a large area and contributes a good 15% to 20% of the total output. Finally, a great deal of light actually makes it out the back because of reflection off the front lens. This could easily add another 10% to 15% to total output. I was honestly amazed by this myself although it makes sense. For example, using the usual way of figuring lumens, and assuming a 20° beam angle, I might have guestimated about 2.3 lumens for the Light of Victory 35,000 mcd LED based on it's peak intensity reading of ~24,000 mcd at 20 mA (the 35,000 mcd spec is at 30 mA according to the manufacturer). I usually multiplied this by 1.25 to account for spill outside the main beam, giving me a final estimate of 2.9 lumens. The actual measurements come out to 3.75 to 3.9 based on a few samples, or over 30% more. I guess I'll revise my adhoc mutliplier to about 1.65 to give a more realistic estimate.

I'll also add that since the dial is just screwed to the tester and easily removeable, I might make another one up complete with a heat sink and mounting area to test Luxeons, Lamina BL-2000s, and other power LEDs. My constant current source can supply up to 2 amps if need be. I can verify my methodology if I get lumen measurements in the mid 30s with my Q-bin Luxeons.
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
jtr,

Sounds cool. I think I may finally do this myself someday. In another thread, I used the method, combined with the typical beam distribution from the luxeon data sheet to come up with a "Lumen coefficient". You would simply take a 0-degree lux measurement of a HD luxeon at 1m, and multiply by the coefficient to get a lumens estimate. I'm hoping to both a) verify that this coefficient is usable and b) see how well the beam distribution in the datasheet holds true.
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I remember that thread, Peter. It looks like you're getting a little less output than I did because the light coming out the back of the LED didn't make it to the fraen lens. Still, it looks like we're reasonably close which makes me feel better. I did a double take after I entered the data for the warden LEDs. I was saying no way can they be this good but I guess they are.
 

idleprocess

Flashaholic
Joined
Feb 29, 2004
Messages
7,197
Location
decamped
Nice!

I've often thought about doing something like that, but lack the math background (and mesurement apparatus) to do something like that.

The calculations in the LEDDB are rather lacking for accurate mcd * beam angle : lumens calculations.

I'll have to look over your spreadsheets. Most manufacturers include radiation patterns, so maybe I can borrow your work for more accurate numbers from spec sheets...
 

NewBie

*Retired*
Joined
Feb 18, 2004
Messages
4,944
Location
Oregon- United States of America
A fella should also consider de-rating the lumens due to heat produced in real life. Especially if using as a "calibration" source.

jtr1962- Have you thought about running a luxeon at 20/40/60mA to see where they fall?
 

idleprocess

Flashaholic
Joined
Feb 29, 2004
Messages
7,197
Location
decamped
Hm. Looks like I'm going to have to take a different route - but at least I got off my butt and have a somewhat better understanding of optics.

So deviously simple-(sounding)!

1 lm == 1 cd / steridian

65.542173315° = 1.0000000000116693 sr (got tired of monkeying with digits in the beam angle calulator here)

4Pi steridians (sr) / sphere
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I made a jig for testing power LEDs. Here are my results for a Q3J Luxeon, a Lamina BL-2000, a Seoul Semiconductor W32182 bin RSX0I star, and a Cree 7090 XR-E bin P4:

Lumileds_Luxeon_White_Q3J.gif

Lamina_BL2000.gif

Seoul_Semiconductor_W32182_RSX01.gif

Cree_7090_XR-E_bin_P4.gif


Here are the relevant spreadsheets.

I have 5 Q bin Luxeons. Four of them seem about equal in brightness and the fifth is about 10% dimmer. Assuming then that the dim one is at the low end of the Q-bin (say 31 lumens), then the other ones should test in the mid 30s. My results give me 37.1 lumens, a little past the middle of the Q-bin. I think this more or less verifies the accuracy of my methods. At worst my results are less than 10% too high.
 
Last edited:

ViReN

Flashlight Enthusiast
Joined
Apr 3, 2004
Messages
3,078
Location
CPFReviews.com
Wow .. Amazing Data /ubbthreads/images/graemlins/smile.gif Thanks for doing the test jtr1962

Any Chance you will test the Nichia NSPW500CS ... C0 / A0 Tint, U Bins...

Also, It would be interesting to see the BS LED's too...
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I'll probably get 5 of the Nichia CS to test from Grumpy's group buy if he has any more. Other than that I don't have any Nichias except possibly two really old ones from around 2001. I purchased them from Hosfelt Electronics but they have the tapered shape typical of Nichias. At the time they were rated at 5600 mcd, 20° viewing angle, which goes to show how old they are. Assuming a similar light distribution pattern to other 20° white LEDs I've tested so far I would imagine they would put out roughly 0.9 to 1.0 lumens at 20 mA, for an overall efficiency in the 15 lm/W area. Maybe I'll run a test on them just to verify my estimates.

If anyone wants to send me LEDs for testing PM me for my address and send a stamped, self-addressed envelope if you want the LEDs returned. For now I'm sticking to testing white LEDs. I've tested a few colored ones, but finding a true hotspot reading at 1 meter is somewhat difficult thanks to the splotchy beam patterns. Also, because of these beam patterns I'm not sure how accurate my methodology would be. Most white LEDs have a fairly even beam, making the results more valid.
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I added the LCK 12,000 mcd warm white, which arrived today, to my list in the first post and updated the graphs accordingly. You might need to refresh your browser to see the updated graphs.
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I added an old (c.2001) white LED, probably a Nichia, to the first post and updated the graphs. I also made the comparison charts easier to read by using heavier lines.
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I've finished testing any colored LEDs I have worth testing (the dim indicator LEDs would be pointless). Here are the results in the order in which they were purchased:

SpectrumLEDs_4000_red_3mm.gif

SpectrumLEDs_5000_ocean_blue.gif

ChiWing_3000_amber.gif

ChiWing_5000_red.gif

ChiWing_5000_blue.gif

BHK_12000_green.gif

ChiWing_4000_ocean_green.gif

ChiWing_3000_orange.gif

Hebei_5T47VB5C_blue.gif

ChiWing_4000_green_3mm.gif

LoV_11000_red.gif


The blue LEDs are surprisingly efficient in terms of lumens per watt, especially the Hebei blue LED. The likely reason for this is that the dominant wavelength is probably somewhat higher than 470 nm. The rest of the colored LEDs fall more or less where one would expect. Except for blues and reds, colored LED efficiency has not been advancing as fast as white LED efficiency, especially for amber, orange, and yellow green.

The relevant spreadsheets were added to the .zip file linked to in the first post of this thread.
 
Last edited:

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
I added the BestHongKong 40,000 mcd and Jeled 40,000 mcd to the list in the first post and updated the graphs accordingly. You might need to refresh your browser to see the updated graphs. The relevant spreadsheets were added to the .zip file linked to in the first post of this thread.
 
Top