Measuring Lumens, how do you do it?

[turboBB]

@Tom E

I believe it was mentioned ANSI/NEMA FL-1 standards call for precise controlled environments for the calibration and measurements, but is it needed? How accurate do you need to be? What’s the effect of a temp swing betwen 68F and 74F for example? Thousands of dollars on equipment with controlled environments I’m thinking is a huge overkill for the precision we want, but without all the data, it’s a tough call. Of course the biggest joke is the variation from unit to unit on so called claimed ANSI rated lights. So if you measure to 1 lumen precision, but your lights will vary by 20 lumens piece to piece, what’s the point of that accuracy, unless you state all the facts - ex: here is the average #, and here is the expected range… Also, can your test sample size be big enough to spec a reasonable range?

Tom, great response. It seems there are some who believe that since CREE has tight tolerances with their emitter mfg process, that the differences from light to light would be in .0001% range but I've had samples of two lights where the differences in output were closer to the 10+% range (with max output @ ~3500lms). Of course this needs to take into consideration the potential differences beyond the emitter itself, however, suffice to say that I've been very happy with my PVC LMD set up.

The only thing I've altered since I first posted this is I now cut out paper templates to block out the gap between the light and edge of PVC. This minimizes any advantage for lights with large bezels since there is less light lost. Of course, the foam method that others use achieve the same purpose. I also didn't use the step down tubes but instead have the light meter clamped to the end with a black cloth surrounding it (to prevent from measuring ambient light). With this set up, I can measure from .1 lms to 10K+ with the main caveat being that I only have one reference light, a Xeno G10v2.

As for the temperature swing, it can actually be in neighborhood of +/- 5% but again, for our intents and purposes, that should be more than sufficient. Having a solid reference light and calibrating with it before each measurement will greatly minimize any differences from temp swings.

In the end, if you end up with a setup that produces consistent results between various lights, that's what matters more than being .0001% off vs. claimed ANSI lumens ratings.

mhanlen - The light meter can be any reputable one and the one Tom linked above should would perfectly fine. Again, it's the consistency that matters above all else.

Cheers,
Tim

[rdrfronty]

@turboBB

@Tom E

I believe it was mentioned ANSI/NEMA FL-1 standards call for precise controlled environments for the calibration and measurements, but is it needed? How accurate do you need to be? What’s the effect of a temp swing betwen 68F and 74F for example? Thousands of dollars on equipment with controlled environments I’m thinking is a huge overkill for the precision we want, but without all the data, it’s a tough call. Of course the biggest joke is the variation from unit to unit on so called claimed ANSI rated lights. So if you measure to 1 lumen precision, but your lights will vary by 20 lumens piece to piece, what’s the point of that accuracy, unless you state all the facts - ex: here is the average #, and here is the expected range… Also, can your test sample size be big enough to spec a reasonable range?

Tom, great response. It seems there are some who believe that since CREE has tight tolerances with their emitter mfg process, that the differences from light to light would be in .0001% range but I’ve had samples of two lights where the differences in output were closer to the 10+% range (with max output @ ~3500lms). Of course this needs to take into consideration the potential differences beyond the emitter itself, however, suffice to say that I’ve been very happy with my PVC LMD set up.

The only thing I’ve altered since I first posted this is I now cut out paper templates to block out the gap between the light and edge of PVC. This minimizes any advantage for lights with large bezels since there is less light lost. Of course, the foam method that others use achieve the same purpose. I also didn’t use the step down tubes but instead have the light meter clamped to the end with a black cloth surrounding it (to prevent from measuring ambient light). With this set up, I can measure from .1 lms to 10K+ with the main caveat being that I only have one reference light, a Xeno G10v2.

As for the temperature swing, it can actually be in neighborhood of +/- 5% but again, for our intents and purposes, that should be more than sufficient. Having a solid reference light and calibrating with it before each measurement will greatly minimize any differences from temp swings.

In the end, if you end up with a setup that produces consistent results between various lights, that’s what matters more than being .0001% off vs. claimed ANSI lumens ratings.

mhanlen - The light meter can be any reputable one and the one Tom linked above should would perfectly fine. Again, it’s the consistency that matters above all else.

Cheers,
Tim 

We’ll said. And I fully agree, based on my personal experience.

[Tom_E]

Thanks Tim!! I kind of went off rambling a bit, but hoping to make a couple of points.

[leaftye]

I'm about to try figuring this out tomorrow. I tried to find the 4 inch p-pipe that turbobb used in his pvc integrating pipe, but my hardware store didn't have it. I'm just going to use a white bucket I already have, plus some white corrugated polypropylene and a small sheet of glass. The corrugated sheet will serve as baffles. My lux meter doesn't have a remote sensor, so I'm thinking I'll attach the lux meter to the bucket with magnets to make checking and reattaching it easier and more consistent. Later on I'll get another lux meter with a remote sensor.

[mhanlen]

@turboBB

mhanlen - The light meter can be any reputable one and the one Tom linked above should would perfectly fine. Again, it’s the consistency that matters above all else.

Cheers,
Tim 

Ok, it’s ordered!