(Wait, I never mentioned sphere #3 ! 
. Well, I did make it, but although sphere #3 was supposed to be the simple one, during the build I got carried away and added all kinds of useful things. I have not done a build thread on it yet, but I will, one of these days.)
Another day, another integrating sphere. The aim of this one is to really be simple: see how much I can leave out without sacrificing too much on the performance. In this integrating sphere is much more thinking time involved than actual building time. In fact after figuring it out it was made and tested in 1.5 hours. I made it as a proof of concept so I starting out with a small 15cm diameter styrofoam ball just because I had one.
The inner diameter is 11cm, and because a rule of thumb is that less than 2% of the inner diameter must be occupied by holes (to maintain enough reflectivity for good integration), I made the entrance hole 30mm diameter. So that is the widest flashlight that it can measure. You can make this sphere with bigger size styrofoam balls and bigger holes, it will even be a better integrating sphere.
Now I cover the inside of the hole with aluminium-tape (the stuff you get in a plumbing shop), tin foil plus some (suitable for styrofoam) glue should work too. This is done because the detector will be next to the hole and it should not pick up light that enters the styrofoam layer directly from the light source.
I need a plastic ring to clamp the detector on a fixed spot on the outside of the sphere. I use a section of an old pill pot. you can use anything that is white, black or transparent (not coloured).
I taped it to the sphere with cellotape on the outside (cellotape sticks very wel to styrofoam). It can be anywhere near the top of the sphere, it just has to be in a fixed position so that the detector will be in the same position every time you attach it.
Stick the detector of your luxmeter onto the ring with a piece of tape and you are ready to go.
Why is this sphere made this way?
*With small spheres you run into the problem that your luxmeter is out of range with even weak flashlights. I have solved that before with a neutral density filter before the detector, now I did it by measuring through the styrofoam (idea stolen from texaspyro). The advantage is less light to measure and extra integration on top of the multiple reflections inside the sphere, the disadvantage is a possibly slight change of the spectrum of the measured light source.
*by covering the inside of the hole with aluminium, the detector just sees light that has already seen integration inside the sphere, direct light from the source going into the styrofoam is avoided.
*the detector is near the entrance hole so that light that goes into the styrofoam directly in front of it has undergone at least one reflection, this improves the integrating properties
*no baffle is needed, in fact the alu-foil works as the baffle
Does it work?
*With my cheap Tondaj luxmeter, the range of this particular sphere is 1 lumen to 17.000 lumen. Edit: if your sphere is bigger than this one, your range goes up which is not good for measuring low light levels. An adjustment to get the range down again is suggested in post #57.
*The integration (insensitivity to beam shape, just sensitive to the amount of light) of this sphere will be way better than the pipe measuring thing. Shining around a zoomed zoomie around the sphere in different angles gives a very constant reading. Moving around the luxmeter over the surface of the upper half of the sphere also gives a fairly constant reading.
*Because of the -relative to inner surface- large opening of this small sphere it is a bit sensitive to size, shape and reflectivity of the flashlight to be measured. For improved accuracy it would need a correction light source (like in my other spheres), but I chose not to add one. The error can be up to a few percent. A bigger sphere with -realtive to the inner surface- smaller hole will reduce the error (my 50cm sphere with 80mm hole hardly needs correcting).
*I measured three constant output flashlights to test the performance, a larger neutral white (XM-L2 4C) one and two identical flashlights, one with a 3000K XM-L2, one with a 6500K XM-L2. I used the neutral flashlight for calibration and used the found conversion factor to measure the two others. Then I compared it to the values of the same lights with my big integrating sphere (#2).
| Sphere #4, measured lux | Sphere #4, calculated lumen | Sphere #2, calculated lumen | deviation of Sphere #4 from #2 | |
| 3000K flashlight | 1920 | 159 | 172 | -7.5 % |
| 4000K flashlight | 4500 | 373 | 373 | - |
| 6500K flashlight | 2720 | 225 | 228 | -1.5 % |
That is quite a deviation from the other sphere. What is the cause? First, the bigger flashlight with reflective bezel causes an improved reflectivity inside the sphere and measures relatively high compared to the two small black ones. Another error is the different luxmeters: the cheap Tondaj used here and the expensive Mobilux in Sphere #2 (I made a thread on the differences between these luxmeters: https://budgetlightforum.com/t/-/28689 ). The Tondaj overrates blue, that is why the 6500K flashlight measures 9% higher with it compared to the 3000K one. Another possible effect could be the light-path through the styrofoam possibly causing some spectrum shift.
Conclusion
I could do numerous more tests on this sphere (I may dive more into it one day) but I think the point is made: this sphere, with the cheap luxmeter, works within 10 percent from my best sphere with all sorts of corrections and state-of-the-art luxmeter. I think it is very usable. (for the record: 10% off is not that much in light measurements)
For improved accuracy, you can carefully sand the inner surface with 100o grit sandpaper (I'm going to do that, quick fix), use a larger innersurface-hole ratio, a better luxmeter or add a correction light source (may do that as well). I do not recommend experimenting with coatings: no fun, poor result IMO.
Thanks for enduring another post on integrating spheres 
Your not alone.