You guys do understand that I did create my baselines from “brand” name Ansi rated lights? That is almost all I and manxbuggy1 owned at the time when I made my light box. I’m saying lights by Olight, 4sevens, Eagletac, Fenix, Jetbeam, itp, Crelant are what I used to create the baselines.
Another thing to note that no light from any manufacture I’ve ever tested TWO different of the exact same series of light tested exactly the same. Meaning between brother and and I, we have or had multiples of several certain ansi rated lights. Thrunite TN31’s, Zebralight SC600, Eagletac TX25C2, 4Sevens Mini, Jetbeam BC10, Armytek Predator, etc. are just a few we have multiples of. And EVERY single time when compared them back to back, using the same battery type and charge, the two different lights will very slightly. I’ve never seen anything huge, but 5% or so is pretty uncommon.
BUT, when test the exact same individual quality light (not my brothers version of the same light) ALWAYS yields numbers with in 1% or so. So it’s not a variance in my testing method, but it is variance of the same lights produced by even quality “ansi” rated lights.
Every light from every manufacturer is going to have slightly different output period. And that should be perfectly understandable to everyone considering Cree themselves state their emitters can vary up to 7%.
So my point is that no member on this forum can ever get a testing method down to 1% accurately matching claimed ansi ratings. If the manufacturer can’t produce such accuracy, we obviously will never be able to properly produce such results.
So all we can do is get the best baseline possible based on as many quality and ansi rated lights as possible. And that is what I have done. And when I said with in 10% accuracy, I stated that is WORSE case scenario. Most lights tested with in a couple percent. I have tested many lights that matched with in 1 lumen.
Keeping in mind the 7% variance Cree allows, and human error will occur in the manufacturer even of quality lights like Fenix, Amrytek, Zebralight - having 90% of every ansi rated light I have tested to match with in that same 10%, and again the vast majority, much closer - I just don’t see how to do any better than that.
And keep in mind that other respected members like Tom E and DBCstm have tested tons and tons of lights, plenty of them ansi rated, with the same basic setup also with similar accuracy and results.
The way I see it - at a minimum we have created an easily repeatable method of testing lights that remain very consistant between many members accross the US. And those memebrs have likely tested hundreds upon hundreds individual lights as a whole. And actually quite a few of the EXACT same lights have tested between us, again producing very consistant numbers.
So, since no member here has easy access to a true integrated sphere AND would be willing and able to test hundreds of custom and modified lights - I think our testing method does the job we need it to do pretty darn good.
Ill let RMM, Tom E, and DBCstm correct me if im wrong. But they all tested TK61 to have 1200 lumen. Three different lights, three different persons, three different of your boxes. All pretty much at 1200.
That is 20% higher compared to selfbuilt`s baseline. Considering that many have Fenix as baseline, you are 20% higher compared to many of those. Unless TK61 reads a lot higher compared to other Fenix lights. But Selfbuilts measurement did not show that, nor have I seen a sign of that, and I have compared with other Fenix lights as well. If I should take my sample of TK61 and make sure my measurement was 20% too high compared to Fenix and Selfbuilt, my calibration would be so high that my others lights would read too high compared to ALL the other people who have done similar work. And they are ALL lower as far as I my observations go. I seriously doubt no one else then you guys will be that high reading on that light.
Again, Tom and DBCstm still have the highest numbers on K50.
Selfbuilt - 1450
Tom E and DBCstm - 1650 (roughly their average)
13,7% higher.
In the case with the light Tom E modded for JMpaul320. 8% higher.
If you were close to others peoples baseline, these large differences would not happen. And many people have done all the "same" calibration work, but they are more in line with each other. These numbers are not exactly within the few percent that you would expect people with similar baseline right? These are 8-20% higher. And note, im not cherry picking examples here. Im just picking lights that Ive personally been interested in, got comparable numbers, and maybe measured myself, so these numbers are fresh to me and also done by several people. The thing is, from memory, there a LOT of examples where the guys with your lumen box always have showed sign of having the highest numbers compared to others, so I can assure you, its not just these flashlights that show very high numbers.
When that is said, I have not given much thought to the type of lights differences are seen on. Could it be that especially these modern day powerful throwers have a tendency to read extra high in your lumen box/pipe thingy?
Maybe do some testing on that? This is your box right? And you all have quite similar boxes.
How is the conversion from light meter result to lumen done?
:popcorn: :-)
I like your persistance RaceR, measuring light is a very tricky business, I found that that to my disappointment too, but it is good that apart from the unavoidable measurement errors, at least the avoidable ones are pinpointed by you.
I stand accused of having some of that myself. But, in this case, you are chasing after a single fly in the middle of a stock yard.
Let me explain a few things on my TK61, and you’ll see what I mean.
When I got my TK61 I was intending to modify it, with only days before leaving on vacation and expecting to take the light with me. I’d never seen a light like this, the battery carrier set up to be able to stack a second carrier tripped me up. Right out of the box I tried taking a tail amperage, and shorted the carrier. This also did some damage to my driver. Pretty sure I said this when I first spoke of my results, but of course my memory is wack.
The first lumens and lux readings I have on paper are with 2 IRLML2502 FET’s stacked on Q4 , I got 1176.45 at 30 seconds 172.5Kcd for Lux
Then I stacked an R025 resistor for 1200.6 at 30 seconds. 175Kcd for Lux.
Then I stacked a second R025 for 1825. 269.25Kcd.
Where did you get the quote that I said mine was making over 1200 lumens? See the fault? I do not know what it was making in lumens or lux right out of the box, I didn’t take those readings and damaged it before I could.
I have gotten a second TK61, fully stock, in hand for a lux reading of 162Kcd since then.
It’s the variables that sneak up behind ya everytime…
Please, I have already made up my mind so do not try to confuse me with the facts. 
Im sorry to confuse you with facts Richwouldnt, but it seems Im better at remember what Dale is writing then he is.
Nope, I dont see the fault. Here is the quote where you disagree with what you just said:
So, either you are wrong or either you are wrong.
I can quote Tom E as well.
Id say my memory was pretty right. You guys showed stock numbers that showed about 1200.
No matter which of your numbers are actually true. Your numbers are still measurements that's far beyond the others. And that is the point. Its not imporant if you guys average out out 17 or 20% higher in that exampe. Its still way too high. And its just one of many examples where your guys are way higher compared to everybody else.
Dale, dont get me wrong. Im not here to make you into some bad guys. Im considering to try and match my numbers to you guys. But ideally, I would want my numbers to match others as well.
Ive started this thread in order to get a discussion around the topic and some awareness around the issues. If it can get everybody closer in terms of numbers then nothing would be better. It will probably not happen, but at least I tried.
You share lumen and kcd numbers all the time, I would believe its in your best interest that your numbers are as good and comparable with others as possible. We are having these discussions because I want to increase the value of your post where you share numbers by making them more similar to other people numbers.
Im not trying to work against you here. This is just numbers. Its nothing personal. So please, lets stick to the numbers from now on.
My notes may not indicate to me the order I wrote them. I usually post when I do it, and the post itself serves as my memory in most cases.
Whatever.
I will help everybody out and not post numbers anymore, to avoid confusion and argument. This will keep my box working for me, which is what it was meant for to begin with. Thanks for clearing that up.
I will state, one last time, that I’ve been dealing with head injuries since immediately after receiving the TK61, so even deciphering my own notes is confusing. It’ll be so much easier not to have to.
Ciao
Ok, first thing is you DO NOT want use Fenix as a baseline. If a lightbox user does that - they will always be off from every other manufacturer. Is it possible that Fenix lights tend to be closer to true Ansi, don’t know. But I doubt they test any more accurately than Olight, Armytek, Zebralight, or 4Sevens.
And why I point this out, is because every manufacturer tends to be fairly consistant in one direction or another.
Now using my lightbox for example, the vast majority of these brands test as follows compared to their ansi ratings -
————————————————————
Fenix - usually 10-20% HIGH
Zebralight - usually 5-10% LOW
Eagletac - usually 5-10% HIGH
Olight - usually with in few percent + or -
4Sevens - usually very very close to ANSI
Armytek - usually 5 - 10% LOW
Thrunite - usually with in a few percent + or -
Jetbeam - usually a few percent HIGH
Maglite - usually very very close to ANSI
—————————————————————-
And when I say a light brand tends to test 5-10% high - I mean EVERY light I have ever tested by that manufacture tested high. Just the amount will vary a few percent. And same thing when I say they test very close or test 5-10% low - the entire brand is consistent in that direction.
So if I consistently test a throwers like TN31’s within about 5% of ansi, but test a Fenix TK61 at appx 1200 (and yes thats about what I tested manxbuggy1’s stock light at) which is about 20% high, why exactly should we use the Fenix as the “correct” baseline. And as a note, that 1200 vs 1000 ansi is about the most extreme I have seen on a light I have tested. The TK75 I tested stock was next in line to the worst at about 13% off.
BUT I never have issues like that with any other Brand. Again worse case is 5-10% across the board. I don’t recollect any brand have a single light sway away from their brand tendency to be high, low, or very close.
I get my baselines by using all the Brand name ansi lights and getting the closest possible median between them all.
And that leads to the question about my light box - yes that is my lightbox pictured above.
By baseline, I mean a mulitplier number. Because my lightbox obviously doesn’t have the volume of one 1m, my lightbox is going to have a lumen number way off to begin with. That will be true of any homemade light box. So to get the baseline, I start testing all the ansi rated lights I can get my hands on and record their numbers. With all those numbers, you then have to find out exactly what percentage they are off from their individual ansi rating. Once you have all those percentages, you then find the best and safest median point between all the numbers. And from that percentage, the “baseline” number is created. And that Baseline when all is said and done is simply a number that has to be multiplied (in all of my lightboxes and meter, the number is much smaller than 1) towards the test number which in turns gives you a good lumen number.
This is the same basic method that Selfbuilt uses too. Pretty much the only way to accurately create baselines.
Never, never, use just one light brand to create a baseline. That is unless someone can prove without a doubt that brand is more accurate than every other quality brand out there. And thats just not going to happen considering most quality brand light companies use integrated sphere to test and rate their own lights in. Obviously there are variables even in their professional spheres or in the way they do their testing.
Thanks for the explanation and list rdrfronty! :) That is some great data!
That explains a lot. Lots of people seem to use Fenix as reference. Or certainly have Fenix lights much closer to their baseline.
I’ve definitely noticed a big difference between Zebralight’s lumen scale and EagleTac’s lumen scale. They’re like 40% different from each other. But ET tends to list both “emitter lumens” and “OTF lumens”, and I think their “emitter lumens” scale is fairly close to ZL’s scale.
So, when I test EagleTac lights on my ZebraLight-calibrated light box, the values come out way higher than ET’s spec… at least, if I base it on their OTF lumen specs. But they’re pretty close if I use ET’s emitter lumen specs instead. Similarly, when people use an ET-calibrated light box, ZL measurements will look rather low.
I’ve stuck with ZL’s scale though, because they usually seem to at least be pretty internally consistent. ET, however, seems to frequently publish specs which barely even resemble their products.
In any case, I agree with the approach of just measuring a lot of products from reputable manufacturers and then plotting a correction curve which gets the closest average match to all the data points. Kind of hard to do otherwise when everyone is using a different standard “lumen”.
I used Olight, Thrunite, and nitecore
out of the 12+/- lights I used the only bogus one that wasnt within a few % (for any of the modes for that matter), was the nitecore EAX hammer… it was all over the place… all the other lights were within reason when I compared corrective factors for each light and each mode.
It seems to me that the only way we will have standards that all will agree as fair is to do as djozz suggested. A couple of lights in different classes need to be circulated with cells in them and tested by each tester with a constant charge level then passed on to the next tester and so on. This would be a slow and moderately costly process but then we could find a happy medium and adjust our formulas accordingly.
From reading the full thread this has already been done by a number of the people who are doing the testing and reporting results. The only way to make this work would be to mandate it which is not going to happen for a volunteer group of enthusiasts who are spread around the world. Are YOU willing to pay the shipping, import duty and VAT charges to send a group of test lights to the EEC for calibration purposes? IMO not a practical suggestion. From the evidence presented by rdrfronty in post #88 even the manufacturers cannot all agree on their Lumens measurements, supposedly done to ANSI standards. If they, with their resources, cannot do so then how can you expect a group of hobbyists to do much better without bankrupting themselves?
I havent read through the whole thread, and dont do lumen readings, so take what I say with a grain of salt...
I have the LX-1330B light meter, and find it accurate enough for my readings...
Nitecore EA4 - Manufacture rated 20kcd (for CW, tested light was WW). Tested at 17.4kcd. About 13% lower than rated. How much difference would cool and warm white make?
Energizer high intensity 2xAA - Manufacture rated 90 meters, I calculated 85 meters. About 5% lower than rated
Energizer ultimate lithium 2xAA - Manufacture rated 85 meters, I calculated 90 meters. About 6% higher than rated
Now most of the lights I have are rated in chinese scales, so I didnt bother comparing those. I have a Jetbeam rrt-0, Supbeam k40 and Supbeam x40 on the way I can take readings and see how they compare, but the only one I would really trust to be close is the Jetbeam.
But like Dale said, I didnt buy my meter to post accurate results, I bought it to have a baseline and to see the improvements of mods. The meter I bought spits out numbers within tolerance to stock lights that would be near accurately true, and it spits out the same numbers consistently.
is that so wrong that it isn’t worth commenting it?
Yes.
First of all, it is better to calibrate from a wide variety of lights. If you only use one, you may be very far off for any number of others. And you will have no way of knowing this.
Furthermore, this gets back to my earlier point…having ACCURATE numbers that differ from ANSI, or other testers, is better than having a (wrong) number that happens to agree with someone else’s (still wrong) number.
I was merely pointing out the only way that everyone would be happy with, not suggesting someone do it.
If your luxmeter follows the V(lambda) curve correctly, you only need one calibrated constant lightsource for calibration and that's it. If it doesn't follow the curve well (like every cheap luxmeter around), it is best to have several reference lights, in different colour temperatures, and have different correction factors for different light colours.
BTW, my next little project will be a 3x7135 constant output light to send around.
Regardless of strict absolute accuracy, I find even a basic light box to be very useful for relative measurements… like, for example, today I finally measured a SK-68. On a ZL-calibrated scale it got 39 lumens on an Eneloop on the narrowest setting, 72 on the widest setting, and 101 with the bezel removed to run in “mule” mode. So, even lower than the 120-140 lumens I was expecting. But it still has pretty decent throw… 5550 cd (148m) on Eneloop, 12400 cd (222m) on Li-Ion.
The throw there might look a little high, but I tested an atypical SK-68 with improved optics and a smaller hotspot.
If it is not “25” ,then I am going back to School!