A home-made integrating sphere.

I have since long been unsatisfied with my milk-carton light box.
It was easy enough to use and showed me the light's output directly in lumens - I thought. That was based upon measurement of a Zebralight H50 which should give 66 lumens. When I found out how to do a reasonable calibration (described below) it showed that all my numbers were about 12% too low. The H50 gives more than 66 lm (about 74 lm) on a freshly charged Eneloop. First when the battery is down to the more flat part of the voltage curve it gives 66 lm. But the worst thing about the light box was that the measured number had a great variation depending on the angle I hold the light over the box.

Then I started the project of building a sphere out of two old aluminum lamp screens from floor lamps.

The construction of the baffle inside was a problem until I stumbled upon this article (mainly page 4).

Another problem was the painting inside the sphere. Should it be Alkyde or water based paint. I ended up with a water based foundation paint that I brushed on. For the top layer I sprayed lightly several times with a spot sealer ment for walls and ceilings, called "Polyfilla". It is white and total matte when it dryes.

Sphere before painting, notice the sensor knob behind the baffle.

After painting.

The calibration is easy when you know the area of the hole. The principle is to hit the hole with a uniform beam of parallel light (the sun or a throwers hot spot at several meters distance). With a luxmeter the intensity just in front of the hole is measured. This number is lumens pr. square meter that hits into the sphere. Multiply with the hole's area in sq.meters and the result is the lumens number that the sphere meter should show. To read lumens directly I removed the sensors opaque dome and used only the sensor diode with its green filter (the knob on the sphere picture). This gives a great sensitivity. A few layers of scotch tape across the sensor opening lovered the sensitivity to the number I wanted. The method is also described in this post.

The meter shows 50 lux (x100) which translates to 50 lumens.

I am satisfied with the result as the measured numbers are insensitive to the angle of the light beam into the sphere.

Well done! Actually excellent job!

Thanks for the credit, Budgeteer, hope this can be an inspiration to make your own sphere.

Indeed, trying to find a ball made of styrofoam. Optionally going to a wooden box. Eventually a Styrofoam fish trasport box will do.

Would like to go your style and tweak it to read lumens directly, opposed the silly math variant which gives me nausea everytime i look the calculations needed along with calibration with another photometer...

I will probably play with "filters" to dampen the photo sensor sensitivity to the desired level. Will have to send 3 different lights along with fully charged cells to a friend who can do lumen measurements on pro level to have real lumen values to calibrate the sphere with.

Looking for something like this: http://www.raysownbrand.com/img/container.jpg


Very nicely put together too. Which goes to show that someone who designed measuring instruments for a living will do better than me.

I'm impressed.

Thanks Don, but i'm afraid I would make a very poor living making spheres. The first picture is two month old despite the fact that I have plenty of time as a pensionist. I guess lack of speed comes with age. But then again there was some hurdles to be jumped along the way and I like to think thoroughly over problems to not destroy possibilities, a process that also gets slower...

The sphere has produced some intriguing results:

My 4 pc. of XP-G R4 stars neutral bin 4B from KD is in fact R3, (which is also evident from Cree's survey - the R4 is non-existing in neutral).

One of my XM-L T6 stars (KD or DX) in fact lies in the U2 bin which is also logically possible as the U2 bin has first recently been announced.

Some plug-ins with R5 are so weak so I suspect they are in fact R4. Needs further investigation.

Dare I mention it? Our common C303 measurements could indicate that your measurements are 5-10% on the low side...No, forget it for the moment .

But you only need one of them. And it looks really good. Nobody has ever successfully accused me of neat work.

I'd have thought my box over-read rather than under read. Useful to have a good comparison - it is useful that we both measured the same light.


Outsanding work sir. I've been in the process of making one for a few weeks now, but always find an excuse or reason not to mess with it. Thanks for the inspiration to get back to finishing that particular project A good I.S. would definately come in handy for one of my current maglite builds...