All looks fine to me 
Cool, this is going to keep my free time busy for the next few days
Tips from my experience (and I’m certainly no expert):
Not sure about how your luxmeter mounts, but the way mine is it’s easy for the readings to change based on how tightly it’s held against the foam ball (mine held with velcro strap which can be seen here. So what I’ve done recently is borrowed the Fenix headlamp I originally used for calibration and used it to calculate “known values” to hold on my own “Calibration Check Light” which is simply a Convoy M1 XP-G2 with 6x7135 AMC’s. It’s low & medium modes are very steady, high can vary greatly so I don’t hold the high number (I would suggest using lower modes for calibrating & not trust extreme outputs - even the Fenix light varied in “Turbo”). I now plan to reserve this light for only sphere calibrations / checks. My multiplier does change, but not drastic. I often compute a multiplier from the average of a few steady output lights. I’m finding medium modes (i.e. like 200 lumens) seem better than trying to hold really low values (1 to 20 or so lumens).
Create a spreadsheet for your lumen calcs and run your computed multiplier back against all measurements so you can see how your “check lights” hold up and how consistent they are. I’d be glad to email you a copy of my latest Excel spreadsheet if you want. There’s a screenshot of my old one at the O.P. of the link given above. Record your measurements at startup & @ 30 Seconds (and I make note how stable the readings are). Use a timer (I use a kitchen timer or my phone and set it to 33 seconds which gives me 3 seconds to get situated after tapping “start”).
Should go without saying, but always test on full batteries and make note which batteries you used and voltage (helps when your confused over low or different readings taken a month or year prior!).
Measure highest modes first while batteries are fresh and give some the batteries some “rest time” between measurements.
Most importantly - be extremely consistent with where you align the head / lens of your lights during measurements. I eyeball (best I can) that face of my light is always aligned with the bottom level of my entrance hole.
Cover you sphere with a large shopping bag (or whatever) when not in use to keep dust off and out of it.
I still have questions about the use of mine & values I am calculating (for instance some lights in high modes seem to give me much lower values than I expect).
You should buy or borrow a good “known value” light (like many Fenix lights) to get your calibration nailed down correctly.
-Garry
Good tips, thanks mate 
In this design it is better to align the face of the light with the top level of the entrance hole, in that position there is no direct light going from the flashlight to the area where the sensor is, which should give better integration.
But it is insightful to compare the readings of both positions, if the ratio of both positions is the same for a flooder and a thrower, it is apparently not so critical what the best position is (but as Garry said, decide on a consistent position).
Great idea and even better ease of making it. Kudos to you Djozz for the guide.
Planing on making one with a 200mm ball, with wall thickness of 23mm. Now as my biggest lights are Convoy L6 and Couroi D01, i would need a 76-77mm hole to be able to measure them and similar sized lights.
Will i get away with a 77-80mm hole on the ball im planing on using, or there is just no way and i should try and find a ball with bigger diameter?
Waking up some sleeping beauty but this cheap integrating sphere is really something nice and as i am making one (well ... actually several as i found a source for 20cm and 30cm polystyren balls not far from where i live) i have some questions about possible upgrades.
Do you think that adding some kind of diffusion layer (frosted glass, dc-fix, ...) inside the ball at the entrance hole would help reduce the reflectivity variation from one flashlight to the other ?
Of course it won't be perfect because some of the light will be reflected from the diffusing thing back to the flashlight where it will be reflected again and it will reduce the light entering the sphere too but i am wondering if the overall result could be better
Has someone already tried that ? what to expect ?
Mmmm maybe the only way to know for sure would be to add a ref light and once you have one installed you don't need that much to reduce the reflectivity variations ;-)
I was watching a doku about LED light (danger of the blue part. . . .and other effects on the body ). When this nice integrating sphere came up.
And i was thinking the 40cm sphere i was planing is bulky 
I have 50cm at the office, it’s just enough for torches 
When I’ll have a load of money and a big house I’ll buy this one :
We should chip in a few dollars each and buy this for djozz. I’m sure he’d fit it into his unit somewhere. 
We’re not using the bath very often, I can break it out and place the sphere there, squeezing ourselves past it to reach the shower. But we can’t afford to shower anyway after I bought that thing.
Best integrating bathroom in the world.
Too much information, but I do the ceiling bounce test in the bathroom quite often at night, so it makes sense to me.
If we can get him a even bigger one the space problem goes away because he could move directly into it.
A bit cold in the winter but during the day he could use the calibration quartz tungsten halogen lamp to heat it.
Preventing my son sticking his dirty fingerprints on the wall will be a daytime job. “Get your f#lthy fingers off my wall, you’re ruining my multiplier”
Here’s mine, 20cm in diameter :
I made the entry aperture the size of a C8 bezel and coated it with some black electrical tape.
I have some quite good measurements, but I don’t have many stock flashlights unfortunately… I tend to mod any new light that comes in my collection 
Anyway, I average my multiplier with a dozen stock lights, and then apply it to my measurements.
I’m glad to have tested it, but now I kind of regret to not have bought the 30cm version (11€ vs 5€…)
Well, I guess I’ll buy it to measure some bigger lights.
Looks good X3 !
Your calibration leans on the published numbers of Olight, I’m not sure how they compare to real lumens. But the multiplier at least does not show enormous deviations so that looks trustworthy.
What bother me is my Eagletac D25A 219C… I have a multiplier of about 4 with this one
EDIT : I don’t know how Olight round up their numbers, but I do know that they own an integrating sphere (maybe 1m in diameter) as can be seen on the office/factory tour that Marshall did a few years ago
Made from 4 inch , schedule 80 PVC “P” trap and 90’s and with a .05 (or 1/2 of actual reading on meter) comes in about 3% under what the Maukka calibration lights show.
The cut out on the side of the top opening , is so that headlamps or other 90 degree lights can be centered for testing. Also have different sized foam cut outs , from 3/4 inch opening up to almost 4 inch to block any light from escaping the opening.
Below are the pictures of the light tube and under those , the Readings Muakka had on his integrating sphere and my actual readings from my light tube set up with those same lights.
Hi Guys! I read all comments on this beautiful thread. I got curious and I registered, so I am new here!
Is there a welcome cake or something? =D
I was wondering what power could such integrating spheres sustain. what if I plug in a 10W halogen lamp?
Cheers!
Lorenzo
Welcome to BLF lelea!
We have no cake but we do have Milla, if the Raccoon chimes in that is.
Not sure about the halogen inside a styrofoam sphere, 10W sounds still ok for a single measurement but do not do a runtime experiment with that. I have coated a few spheres with latex paint mixed with bariumsulphate, I think those can handle some more heat than bare styrofoam.