Measured 3 Fenix HL55's In My Styrofoam Sphere - Which is Right?

Original Post here (describing my current problem), but post #2 is were I measured all 3 HL55’s in one measurement session.

So I’ve been messing with my D.I.Y. round foam integrating sphere again and needed to “re-calibrate” it to my “known Lumen value” light (I made a slight change at the entrance hole). This “known” light is a Fenix HL-55 headlamp which I do not own, but borrow from a co-worker (who uses it regularly). Well at work I got them to buy an HL-55 for use working underground (which no one even uses, but that’s beside the point other than to say it is in “like new” condition). So when I needed to re-calibrate this time around I just borrowed the new HL-55 from work. Using that light as my “calibration light” I noticed that all my previously measured “calibration check lights” were measuring significantly higher. So I borrowed my co-workers HL-55 and compared it directly with the one from work (both measured during the same measurement session and swapping batteries between them to cross-check consistency) and did discover that the HL-55 from work measures pretty consistently 6% lower than my co-workers. Now that I’ve typed this out I suppose 6% is within reason for “bin tolerance”? So which values should I hold as “true”? Should I average them? My co-worker loved his HL-55 so much that he bought a 2nd one, should I borrow both of his to see how all 3 compare and perhaps average the 3? (I’m leaning toward this.)

Any “budget” well tested constant regulated lights out there I should buy myself for calibrations / checks? Most seem a bit pricey for the purpose (i.e. Zebralights, Fenix. . .). One of these days I think I will buy known bin emitters and see how I fit the Cree datasheets (and guess I should buy a couple).

Here’s my “Lumen Calc” spreadsheet and the values calculated with multipliers from both headlamps (see the “HOLD” values). I also turned around and calc’d multipliers for each light based on that light’s “previously established / expected value”. Just note that those previously established values were based on my co-worker’s HL-55, so of course they will match that one better.

My co-worker’s HL-55 is listed as “Matt Steele’s” and the one at work is listed as “U-Gnd”. Thoughts?

-Garry

Ok, I measured all 3 Fenix HL55’s at one time (all on fresh charged high-drain cells) and here are the results. The two HL55’s in the first post are #1 and #3 in this spreadsheet. This additional “#2” measured even lower than the lower one above, so now I have a 9% spread in my values! Is 9% within the range of a Cree bin for an XM-L2?

So what value(s) would you calibrate a sphere too? Average all 3? Throw out #1 and average #2 and #3?

-Garry

Bump. Anybody?

-Garry

According to Cree’s data sheet for the XM-L2 page 3 states:

“Notes:
• Cree maintains a tolerance of ±7% on flux and power measurements, . . .”

Fenix states the HL55 uses an XM-L2 U2 Neutral white emitter:

I have more than a 7% spread in outputs.

-Garry

Perhaps it is the individual components of each driver, the combined tolerances of these components would then vary the output. ?

Hi Garry, there is no way that with help from Cree datasheets and whatever flashlight manufacturers measure for their lights you will get a calibration that you can trust to be within 1 or 2% of the (holy) REAL lumen. For that you need to buy a calibrated lightsource for dear money and even then for any under 1500 dollar luxmeter the calibration will only be true for the 2700 Kelvin tungsten lamp spectrum.

The best you can do is live with the uncertainty of a calibration that may be off, but invest in making that personal calibration consistent. So choose a well regulated flashlight that you know gives a very consistent constant output, and has a rated output that fits in what you feel is about correct, and keep that one exclusively for checking your calibration.

I suppose that could be an explanation. I now understand exactly the danger of calibrating to only one light! Perhaps I should average the lowest and highest performers?

-Garry

For ease of use, I would just choose one, you may be lucky to precisely choose the one that is correct. :slight_smile:

…but you will never know for sure :smiling_imp:

djozz, I understand that. I know I already have error tolerance in other areas (luxmeter, sphere construction, etc. . .), I’m just trying to limit how much error I’m introducing from my calibration. I don’t feel like I can claim my numbers are within 10% if my calibration already shows 9% variance.

-Garry

Keep in mind that whatever people think, hardly any output or throw number posted here on BLF can be trusted to be within 10% of the truth.

You are actually a bit special that you care enough to want it right, and the result is that you come across all the pittfalls that others just ignore.

Yep, I agree. (Though I am holding selfbuilt’s lumen ratings for the HL55 as my calibration!)

I’m a bit of a perfectionist. I probably am making a mountain out of a molehill.

Thanks for the input!

-Garry

I’m leaning towards averaging the lowest and highest values and calling that good enough. Or maybe I’ll apply a least squares average to all 3. :slight_smile:

-Garry

Ok, computing the average of the 3 lights and then re-applying that average multiplier (an average of Low, Med & High modes only) yields the following results:

Note that the average multiplier computed on just the “U-Gnd” HL55 is very close to the average of all 3 lights. So in the end, all my previously established lumen values will now raise (i.e. my lights got brighter without me doing any mods to them! :slight_smile: ).

-Garry

As I always say, a man with one watch knows what time it is. A man with two watches is never sure.