I'm a late comer to all this and thanks to a couple of members here for helping me when I ask some ridiculously stupid questions for help in building this. Any sort of calibration has not been carried out yet. This is a copy of many others here. I'm not sure what the differences would be but in Australia we have 90mm downpipe which is what this is made from I believe in the States you use 4'' which would be larger. Cost per bend was under a dollar.
Ok, it is actually nice to think of how this could be done. My ideas:
-companies probably will never sell enough numbers of a simple cheap and therefore unprecise (even if output-calibrated, my guess is that it will have an uncertainty of at least 10%) sphere, the only market would probably be BLF. So this will have to be a amateur effort.
-the size: I think that the entrance hole would have to be 80mm, to be useful for most flashlights. An exit hole of 40mm will fit most luxmeters. I would go for exactly the size of match's output hole and use his type of luxmeter sensor because that one is very easy to clamp into the hole in a fixed position repeatedly. In a well functional sphere the total of holes have to be less than 2% of the inner surface area, a small calculation learns that the minimum internal diameter (for convenience there's no reason to go bigger) should be 32mm (12.5 inch).
-I would go for the standard 40mm styrofoam ball, for sale all over the EU for about 14 euro w/o shipping, and cut out the 80mm and 40mm holes, make the baffle out of foamboard and glue it in. The inner surface needs carefully sanding with 1000 grit sandpaper to get rid of direct reflections. I would leave the holes and bits that stick out on the inner surface just like they are, their influence should be minor on the integrating properties of the sphere.
-I would not sell a luxmeter with the sphere, but leave that to the receiver of the sphere. Instead, I would sell a constant output flashlight with it that comes calibrated. It leaves the receiver of the sphere the option to mess with it, without immediately spoiling the calibration (like making extra holes, or do a coating). I would use a Uniquefire S10 for it, with the driver replaced by a 1x7135 lineair driver, and the led by a XM-L(2) between 5000K and 6000K, on a copper board soldered to the pill. The output is constant within 1 or 2%. Part costs are 20 dollar, but you will need a 14500 battery as well.
-to compensate the conversion factor variance of the sphere caused by the different shapes and sizes of the flashlights you would need an extra entrance hole to use the calibration light as a 'conversion factor correction light', but that would also require two extra baffles, making producing the sphere more difficult (it could be an option though). I'd just leave that out, and live with the 5-ish % variation.
-this type of sphere does not really need a frame or box to support it, just some tape to hold the two halves together, but that could be left to the receiver.
So in parts this sphere is less than 40 dollars, but producing them in quantities is no fun, so labor would be charged too, I would estimate labor on the sphere itself would be three hours, the flashlight mod another two hours (for me at least, Dale would do it in 10 minutes ;-) ), so in total 100 dollars before shipping does not sound strange to me.
But even better than producing the sphere itself, I think would be to share a standard build tutorial, like actually what I described above, or what Dale, jmpaul and manxbuggy did with their bended pipe design (which is probably not as integrating as a sphere but does give a repeatable measure of light output)
Well, the Chinese shipping fees seem incredibly low(pretty often free), so I have hard time believing that the shipping costs would be the thing to prevent this from happening. The sphere could be delivered in several pieces and then assembled at home. Or it could be of more compact shape, like the PVC pipe one. And I’m sure that most people would not mind paying few dollars for shipping if it comes to that.
Yes, tolerances and calibrations do matter, but nobody would buy one expecting it to be 100% accurate(that would require regular calibrations anyways). This could still easily become more accurate (if designed properly) than most of the home-built ones that people have, even if each unit would not be calibrated individually. Few lumens over or under does not matter that much.
Many people having the exactly same sphere would also make it easier to compare the results and make adjustments to the calibration if needed. We could even do ALL of the calibration here at the forum by collecting statistics from many Ansi FL-1 rated lights and making the conversion tables out of them ourselves. Home-calibration would become way easier and more accurate than what it would be with DIY sphere(since they are all unique).
Build two spheres with the same materials and type of luxmeter and the measurement variation can easily be 20%. In the end, synchronising the type and size of the sphere is not not the most important, the calibration is !
I would imagine that the calibration would be easier (less variables to account for), if you can directly compare to the results of other people with the same sphere(more data to compare and count averages from). But I’m a bit too tired to think this through right now, need to go to sleep 
The link is in my signature. … it’s match’s sphere design that was made with paper mache and a kids play ball
If you really want to have any accuracy, you need to realize that the "pipe thing", is not accurate and that the very small spheres are limiting also. In order for the sphere to be capable across a range of lights, from low lumen, to thousands of lumens, the sphere needs to be, a sphere for one thing and big enough to deal with all the lights.
That prohibits shipping. If you just want to go with something like the pipe, why not just do ceiling bounce and leave it at that?
I know, I'm full of crap, but really, why do you think every real IS out there is a sphere? Do you think they just said, no, make it a sphere, because we don't like pipes, or squares or rectangles? There's a scientific reason and there's a reason for the diameter too. Wishing it wasn't so, doesn't make it different from what it is.
Oh, to calibrate, you need several ANSI lights, not just one, and the same exact lights and batteries, with the same state of charge, need to be used.
OL’s right. I’m not an expert on the topic, but I completely fail to see how this PVC pipe business could do anything significantly better than a “calibrated bathroom” or lightbox or similar. Bigger is better, and stick with a sphere.
If we were really hot to trot I think the best thing would be “calibration lights” as has already been suggested. Not flashlights though. Why deal with batteries etc when you can have a small external PSU running an LED on a decent heatsink for around the same money.
Well, a flashlight because it is versatile, cheap, small and convenient to send by mail, and the output can be made repeatably constant within ~2%.
If I ever finish my sphere, trust it and manage to do a true calibration (if...), I may get to making some calibration lights for sending out.
For some of us here this light business is a hobby. Not only can we not measure lumens to any degree of accuracy but anything else we measure is only meaning full to the one measuring. Voltage, current, resistance and whatever else is only carried out with what our budget allows. Nothing that 99.9% of us use is calibrated but for the hobby side of things is good enough. The equipment that most of the testers here also fall into this category.
Even the best built lights from the same manufacturer of the same model and batch will all vary on output due to tolerances of the components used.
I'm having fun with my pipe dream, sorry light lumen measurer thingy and I'm sure most others are also. It is a great tool for checking mods etc. I will draw up some sort of scale with known lights I have against what the manufacturers ratings and be done with it.
Edit. Yes we could use the ceiling ceiling bounce method or the wifes small padded cell for measurements which I have done and may continue to do at times but wheres the fun in that when we can build a small cheap highly inaccurate pipe measuring device. Who stuck at the one position being missionary in the one place with the one person? I wont answer that. 
YES YES YES, thread subscribed.
Shipping a sphere is not that big of an issue. My 16” styrofoam sphere was shipped in a reasonable sized box because it was two halves nested together and weighs next to nothing. Also, note that those low shipping costs from China disappear once the package size gets even slightly large.
My sphere cost around $40. I don’t know what your time is worth, but there’s no way I’m gonna make up an inferior paper mache or plumbing pipe whatever for that kind of money.
I like the way you say that shipping is not an issue in the same breath as the statement about shipping from China becoming expensive in this size range. :~
Shipping an assembled, working, and calibrated unit without incurring damage is expensive. That’s why assembling our own from cheaply shipped components makes sense.
I definitely agree about not doing the job half way in order to save $20!
Doing a thread bump here because I’m helping with this: BLF Sphere Please read the thread to understand the goal with it- we’re not looking for perfect, but rather consistency with common calibration so folks can compare with each other equally. I have done a search here for “IS” and”Sphere” and went back to read all the numerous posts, and it seems this thread is the best for my questions.
At present I’m looking at styrofoam sphere sources but a few thoughts have crossed my mind. First is that Hannes started on an ABS sphere- do you think it would be a viable alternative to styro if the inside was coated with flat white paint? Any ideas on alternative cheap store-bought sphere materials? Would not want to do more than sand or paint these since there’s going to be dozens of these made.
Also what do you think of painting the inside instead of sanding? I have the proper spray equipment and paint which would make it easy and a lot faster than sanding. Remember dozens are being made and time matters!
Any other relevant thoughts on sourcing the sphere portion cheaply are appreciated.
Phil
I have done a lot of building and experimenting with home-made integrating spheres, and posted everything I found and thought out on BLF. The links can be found via my sigline.
I.e. based on my findings, I think one should stick to carefully sanded styrofoam of a minimum of 2cm wall thickness and no coating.
The spheres that you are going to build will perform the same everywhere, and with a common calibration, which is great, but still will not be a very accurate lumen-measurement devices because of some inherent errors of measuring light with a sphere: there’s measurement errors caused by the light source: the shape/size/reflectivity will influence the multiplier (the bigger sphere or the smaller the entrance hole, the less is this error), and errors are caused by the cheap luxmeter which sensitivity does not follow the defined luminosity curve (if flashlights with comparable tint are measured, which is also the tint of the calibration light, this error will be limited). Errors can easily be as high as 20-30%.
But don’t get me wrong, I think this ‘standard BLF sphere’ is a neat idea and a great initiative, it will make light measurements across BLF more consistent. But do not expect a miracle accuracy of within 1% or even 5% of a professional measurement, and measurements among BLF-members who obtain such a sphere will still vary.
Thanks Djozz, and I have read again all you’ve posted on the subject which has greatly helped me understand the matter.
It’s the logistics of it that I am looking at. How long does it take to do a decent job sanding a sphere like we’re considering? I can do paint coatings uniformly enough to not introduce a visually noticable difference between any two spheres. I could easily paint a hundred an hour for $0.25 each sphere total cost, and maybe less.
Nobody is being misled as to perfection or absolute calibration, the goal is good utility and good quality overall at an excellent price. Are there other considerations I’m missing as to why paint wouldn’t meet this criteria?
Phil
Having seen comments on the other thread I am also wondering what would be the best type of standard paint coating to consider? Barium Sulphate etc would take several coats, and thus would be beyond the scope of this project.
Ceiling paints are matte in finish, and made to be not directly reflective but instead to scatter light. They actually contain tiny broken quartz and/or silicon crystals which each direct light in a different direction thus scattering it when applied en-masse. This seems to be the same effect sanded Styrofoam gives: a uniform but non-directional reflecivity, just achieved in a different manner. This could also compensate for thinner sphere walls as it would reduce the amount of light reaching them.
Would such a finish be closer to the ideal surface?
Phil
Sanding the inside of a small polystyrene sphere like this from shiny to a good flat takes just a couple of minutes, way less work than whatever coating I’d think.
At least the 60%BaSO4/40%flatwhite latex paint mixture, that I used on two of my spheres, reduced the reflectivity of the spheres compared to bare styrofoam. It may or may not have a more even reflectivity across the visible light spectrum (which may be even more important than overall reflectivity), I could not measure that. But there is no reference of the reflection-spectrum of styrofoam and it may just prove to be very good…. (at least it is very white 
)