Ok, here goes; my first showdown. I’d like to try to post one of these once in a while, mostly for fun, maybe as part of a giveaway.
This was not a scientific test, for sure. My integrating sphere is not even close to calibrated, but it can give me relative output results; good enough for a showdown.
So here goes…
On the left is a generic 10W LED module in a WW tint (~3000K). It’s arranged in 3s3p, for a total of nine 1.1W emitters.
In the middle we have the ol’ standby, the trusty XM-L T6 in a what I think is a 1A tint.
On the right is a Jishan 7W module in a cool white tint, probably very close to 6000K.
Right away we can see that the XM-L and the 7W have a tint advantage over the 10W and the XM-L has a power advantage over the 7W.
Both modules appear to have an emitter surface area advantage, meaning they should be running more efficiently (watts per mm2) than the XM-L.
I ran each at their max input current:
- 3A for the XM-L
- 1A for the 10W
350mA for the 7W
They were measured while still cold (within the first 30 seconds of power on), with a good heat sink and thermal compound (Fujik).
Well, give me your 1st-2nd-3rd guesses. I’ll post the actual outcome in a day or two.
Overall lumens from a ceiling bounce test will be in the following order since the XM-L and the 10W are running roughly around the same current draw or 12 watts of power consumption
1. 10W (Surface brightness will be lower but overall output will be higher due to the larger surface area)
2. XM-L (Higher surface brightness by some way but still lower overall light output)
3. 7W (Doesn’t stand a chance :D)
OK, so we have some guesses, and most of you got it right.
1st: XM-L T6 1A
2nd: Jishan 7W CW
3rd: Generic 10W WW
Good job Slim Pickens, texaspyro and PPTK!
I forgot to record the numbers, but I do recall the XM-L had a comfortable lead. The Jishan 7W had a slight edge over the generic 10W
If I had a Jishan 10W module, I think it would have tied or slightly edged the XM-L.
Thanks everyone!
The 10W lost the battle? I was afraid this was ganna happen, but also rather anticipated it. I bought several of the 10watt flood lights during wallbuys 50% off sale, with the intention that Id be modding them with all the spare XM-L’s Ive accumulated after all my emitter upgrades. I’ll run them “as-is” first to see how they do, but probably look forward to modding them all later. Maybe the 7k emitters will level the playing field some but I doubt it.
I bought three of them as well. I’m sure it’s the tint that brings down the output the most. It still looks bright because of the warm light. I have converted a few halogen work lights over using the WW modules (one 60W and two 10W) and they are very bright.
The 60W needed active cooling, so it’s no longer weatherproof.
I basically bought these for the enclosure. I plan to convert one to battery power using an XM-L T6 NW and a NANJG 105C.
I received my Lux meter today, and calibrated it to a bare XM-L T6 emitter. Measured the output at 700ma (25C) and got 3920lux. Divide that by 280 lumens from Cree to get a ratio of 14 lux/lumen. What an easy number to work with! Tested at the other known points (1.0A, 1.5A, and 2.0A) and each one worked out to within 5 lumens of spec! My integrating box seems to be integrating fairly well too; adjusting the entry angle has a very small effect on the lux readings.
Anyway, so on to the results. My measurements confirmed what I found with a simple light sensor and volt meter.
1st: XML-T6 1A - 917 lumens
2nd: Jishan 7W CW - 711 lumens
3rd: Generic 10W WW - 618 lumens
I find these results very close to what I expect from these emitters. the Jishan is rate to 100lm/W and a warm white (3000K) module with 90lm/W emitter under the phosphor would be in the 600 lumen ballpark.
Oh, and I couldn’t resist testing my TR-3T6 (modded to 5.6A, stock was 4.3A emitter current): 1564 lumens. Now I have a baseline for the XM-L U3 upgrade when the parts come in.
Well darnit! I just got my lux meter as well. After playing with it and experimenting with many different lights, it looks like I need to build an integrating sphere before I can tell the whole story. I think I remember how to paper machete. :bigsmile: Your method of calibration seems to leave very little room for error. I hope you dont mind that I calibrate mine in the same way once the day comes.
5 of my 7 10W CW floodlights have been received from Wallbuys. For the most part, Im rather impressed with the output, especially considering the technology of the 2 year old emitters. The tint is warmer than T6 1A (which was a nice surprise) but they definitely are not warm white. The emitters are screwed down plus a dab of fujik between the thick cast aluminum host. The housings could easily dissipate twice the energy without overheating. At about $6ea, they were an absolute steal. I was prepared to convert them all to XM-L and wire them in series with this power supply. Now that I have the power supply, I might buy another one and run a pair of XML’s in each housing. The arrangement would allow a 2 mode operation with 7 emitters series wired to each power supply. Activate power supply 1 and each housing illuminates 1 emitter. Activate the second supply and the second emitter illuminates in each housing. Alright, maybe Im getting a bit carried away. Fun stuff! I suspect the stock housings will provide more than enough lighting for my needs.
When you get your U3’s, can you please test one to see how it compares directly with your “XML-T6 1A – 917 lumens” result?
I forgot to test the Wallbuys floodlights. Tomorrow. BTW, they are underdriven. I’m working on a review and will include a simple mod.
My ‘sphere’ is a cardboard box with two coats of flat white ceiling paint inside. Not a sphere, but works well. I’ll test every different bare emitter I get and post the results… somewhere.
You have a simple mod to increase the output on the Wallbuys floodlights? That would be great. I assumed they drew 12-13 watts based on previous reviews of other 10W floodlights. How wrong I was. I just tested a couple floodlights plugged into AC mains with a P4400 Kill-A-Watt and got 7.1 and 7.3 watts respectively. Im glad that you posted because I was getting ready to use them as is. Much looking forward to your review!
Which Lux meter did you get? I have the trusty ol CT-1330B. Im amazed at the repeatability and stable output readings for such a budget meter.
Question: since you calibrated your meter specific to your lightbox, how does that impact your readings for ANSI lumen measurements? Would they be one in the same?
Yeah, it’s just a resistor mod, but I forgot the value I used. I’ll be opening the unit up to review again and will know then.
Lux meter is LX-1010B, a basic model of eBay for $15 or so. Does the job, and seems to be fairly consistent. I’ll probably get another one at some point that goes up to 200,000 lux for throw measurements.
There are three important aspects to an integrating sphere:
it must integrate the light as even as possible, regardless of light source beam shape (that’s the job of the flat white paint), and
the sensor needs to be shielded from direct light from the light source (a small baffle between the emitter hole and the sensor port)
calibrate the lux to lumen ratio using at least one known light source (I used a bare Cree emitter with known specifications)
I seem to have the first two addressed adequately. I need to do more calibration with some different light sources (XM-L T6 3C, XP-G, ANSI FL1 rated light, anything with a known spec)
I’d venture a guess that my readings are close to ANSI lumen measurements. As I verify it with different calibrated light sources, my confidence in it will improve.
I remember that Don was using a milk carton for his integrating “sphere” with good results. Its to bad that we dont have a means to easily calibrate our findings more precisely with one another. Although your method probably comes close (like within 5), I bet there are to many unknown variables to contend with (like production variances, etc.) to get it down to a 1 or 2 variation. Not that its a goal, but it would be fun to have direct comparisons with each others mods. At least we now have a means of quantifying our own results.
Thanks for the tips on building the enclosure. I found a few descent threads that explain the concept in greater detail.
My remaining floodlights arrived this morning… about 37 days after ordering. lol Looking forward to your review with mod tips. Did they appear to get a lot brighter after the mod?
make sure the lightmeter has a photopic filter as the emission spectra of those different LEDs will no doubt be quite different. you can check this by using a color LED - the filter will make output from a color LED very low.
The integrating sphere is only a small part of systems used to measure LEDs
I would say an actual low cost calibrated system would be without 5-15% accurate, with humidity and temperature adjustments.
Here’s a sneak-peek
Before emitter current: ~600mA
Before lumens (bare emitter): 450
Before lumens (OTF): 417
After emitter current: ~920mA
After lumens: 564 OTF
That’s an increase of about 150 lumens, just over 25%. Not bad, and it is noticeable. I also did a 10W Warm White emitter swap which I prefer over the cooler emitter.
One diode on the driver gets quite hot especially after the mod, so I added a small heat sink to it and drilled a few holes in the plastic enclosure that holds it within the aluminum case.
When the review is up I’ll post all of the mod details in the first reply, hopefully tonight. Pics and data have been captured already
I was afraid this was ganna happen, but also rather anticipated it. I bought several of the 