Integrating sphere #4 (the fast and cheap one)

Thank you Djozz =)

did they end up collecting money for a larger bathroom with a larger integrating sphere for you?

Unfortunately I have to run long measurements and it has to last “forever”. The point is also to be able to change the lamp when it dies without messing with alignement.
I will think about painting with latex. maybe it’s worth doing a test! In that case is it even necessary to have the foam sphere? I could paint any other material, right?

How should I think about efficiency of integrating spheres in terms of input-output power? If my lamp is 10W how much should I expect to have as output?
Right now I use a lens to collect some of the light, thus I lose most of the source power. To calculate how much I can just compare angle of illumination with size of the lens, but what about the integrating sphere?

Thanks again,

There is zero chance that I will have a larger bathroom, it implies moving house inside Amsterdam which is virtually impossible with the current crazy housing market. And even if it came true, it will be my girlfriend’s domain, not mine, my hobby is doomed to be restricted to the dark corner behind the cupboard. :cowboy_hat_face:

Painting with latex may make a styrofoam sphere a bit more heat-resistent. I experimented with a PVA-bariumsulphate mixture too (PVA=polyvinylalcohol), that handles a bit better even and should give a better (more constant reflectivity over all visible wavelengths) coating too.

But there is no other reason to start with a styrofoam ball than easy availability, if you can find a sphere from any other material and give it a flat white coating on the inside, you are good. I hope to find an affordable nice metal sphere at some point, that should handle any output power.

It is a misconception that input-output efficiency plays a direct role in integrating spheres. It does not matter if the sensor picks up only a tiny fraction of the input power (which indeed it only does), it can be any fraction as long as that fraction is the same for all colours and output directions of the light source. Your multiplication factor then takes care of the correct calibration. The only reason that you want high reflectance on the inside of your sphere is indirect: high reflectance causes more reflections before the light is finally lost or reached the sensor, and more reflections equals better integration.

(As a side note: I suspect that most people on BLF have no idea what light integration is, and even if they have a clue, why it is an important feature of your device in order to measure light output. I hoped that my threads on integrating spheres would help a bit but it does not seem so.)

I see your point about integration. Maybe I should have pointed out I am most interested about power output than most people. I am considering using a sphere to “save” energy from a halogen lamp that I need illuminating an optical setup. The setup only takes few mm rays, thus now most of the light is lost shining elsewhere. An integrating sphere with a homemade output slit will give me the ability to change lamp without realigning the setup while using somehow all the light from the lamp, even tho I have no clue what would be the efficiency I get out.
Maybe a new thread is due for me =)

Ah, I see. My spheres are for measuring, yours is to make what they call a lambertian light source :slight_smile: Then my story does not apply to your situation. And yes, you can paint any sphere flat white on the inside, styrofoam spheres are simply the cheapest and most available spheres, that is why I use them.

I found that any white coating of the sphere warms up the light of the source, typically by 500K to 1000K (some blue is lost) It will depend on the coating. If that is not a problem, any coating will do.

I was just discussing spheres with member Moderator007 the other day and he found that molds for “Atlas stones” may be a good material for an integrating sphere. Unfortunately that is a very US product so on top of them being more pricy than styrofoam, shipping will be high to Denmark.

Thanks Djozz! I will look for Lambertian light to see what’s up =)

So I’m learning that my DIY sphere (shown back at post# 64) is apparently wildly inaccurate at higher lumen levels (/- 400 lumens and above) and it doesn’t seem to matter what size flashlight is in the hole (so I don’t believe it’s related to light reflecting off the bezel/glass lens). I’m wondering if it’s related to light loss through the sphere wall, where lower lumen levels wouldn’t see much light loss, but higher outputs would. Thoughts? A few years back I posted these lux readings where I experimented with covering my sphere with a thick black towel, but I only used my Convoy S2 EDC (high mode on that light is only/- 500 lumens). FYI - my sphere calibration/multiplier is derived from measuring three different Fenix HL55 headlamps and averaging the results. (Note that I’ve never had a light that shows steady well regulated output at levels about /- 400 lumens, so my multiplier(s) are computed with “low” and “medium” output levels). I’m not trying to nail down exact lumen #’s from lights (I understand I’ll be off/-10% of “true”), but looking to have a multiplier that works consistenly from 1 lumen to +/- 2,000 lumens.

Here are 3 sphere results that have led me to question my calulated lumens at higher levels (all lights measured using fully charged hi-drain cells):

Sofirn SP40 (XP-L 4,000k):
Low = 6.6 Lumens
Med = 62.8 Lumens
High = 306 Lumens
Turbo = 798 Lumens

Sofirn SD05 (on a brand new Molicel P42A 21700)
Low = 277 Lumens
Med = 792 Lumens
High = 2,223 Lumens

Astrolux HL01 (XP-L 4,000k)
Stepped Ramp Levels:
Step 1 – 13.0 lumens
Step 2 – 43.9 lumens
Step 3 – 124 lumens
Step 4 – 162 lumens
Step 5 – 253 lumens
Step 6 – 404 lumens
Step 7 – 690 lumens
Turbo – 992 lumens

I immediately questioned my results on all three lights above 300 lumens. In this post, user YogibearAl gave me his calculations for the SP40 (which sound good to me) and so I used those to re-calculate new multipliers for each mode. My previous multiplier was 0.4275, and now trying to calibrate to Yogi’s #’s I am getting varying multipliers, with the #’s for the high outputs being significantly higher (my expectation – that my sphere is “good” at low outputs, but gets exponentially worse at higher levels). Multiplier’s I get:

Low – 0.4516
Med – 0.4422
High – 0.5000
Turbo – 0.5158

If I then run with the multiplier of 0.5000 from high mode, run that back against my SP40 measurements and also on my SD05 & Astrolux HL01 (XP-L 4000k) measurements, I get:

Low – 7.8 lumens
Med – 73.5 lumens
High – 358 lumens
Turbo – 934 lumens

Low – 324 lumens
Med- 926 lumens
High- 2600 lumens

Astrolux HL01
Stepped Ramp Levels:
Step 1 – 15.2 lumens
Step 2 – 51.3 lumens
Step 3 – 145 lumens
Step 4 – 190 lumens
Step 5 – 296 lumens
Step 6 – 472 lumens
Step 7 – 807 lumens
Turbo – 1160 lumens

So now my higher numbers seem much more in line with expectations, but I suspect I’m showing lumen #’s that are too high for the lower levels. So what am I supposed to do, use different multipliers for different lumen output ranges? And how would I know what I am doing is reasonably accurate Is anyone else using these round styrofoam spheres seeing the same issue I am? Do I need to paint the outside of my sphere to stop light loss? Should I test lux readings with the towel again with higher output lights? Is a towel really causing light not to get “lost”.


I will have a closer look at your numbers above, but there are a few things that I am fairly sure of:

*luxmeters are as far as I know very lineair, so if you start with a reading at a low setting, then 10 times as much light should get you a 10 times as high reading, and 1000 times as much light should get you a 1000 times as high reading.
*same goes for light loss through the sphere wall (light loss is no problem at all as long it does does get over a certain percentage), the same percentage is lost from very low to very high light levels.
*and thus: for the same spectrum (tint+CRI) the same multiplier should be used for all output levels.

So either your numbers are “more right” than YogibearAl’s, or something is going on with your measurements that I do not understand yet.

Thanks for checking into things djozz! The statements you’ve made above I’ve read elsewhere, but puzzle me since they don’t seem to be holding true for my sphere (i.e. that one single multiplier computed from lower lumen levels should work for higher outputs). I don’t trust my higher lumen numbers at all. I’m sure that the SD05 on medium should be at least 900 lumens, if not more. I’ve not scoured BLF looking for user’s results who have the same lights measured in light measuring devices which seem verified as trustworthy - I only posted asking in the SP40 thread and Yogi responded. It also seems it would be difficult to try to “calibrate” to a light with a higher output as most high outputs are not steady (as I’m sure you know). The SD05 is one that is fairly steady on medium, although another user would probably have to test it with an identical battery for me to compare to. I have one bike light (highly modded though) which seems very stable at higher output levels, but I would need someone else to measure my own light as no other light exists modded like mine.

Here is a recent video showing measurement & analysis on my SD05 (on a different battery) for what it is worth.


I’m also wondering if I should be adding diffusion material (like trying to put a sheet of DC Fix across the bottom of my sphere opening - though I could see it peeling away and being a constant problem). Any change with my sphere means trying to get a hold of those Fenix headlamps I originally calibrated with which is now much more difficult to do!

Also, what about the fact that my luxmeter is simply pressed into the side of the foam without the foam actually being fully cut out? Should light in the sphere reach the luxmeter sensor itself withut going through foam?


Okay, I’m planning to build sphere #2. I’ve picked up another (2) hollow 12” styrofoam balls identical to my first sphere. I did catch that I believe my styrofoam balls seem to have thinner walls than djozz’s (could this be part of my problem? - Perhaps why so much light seems to escape through?). Mine looks like this one.

Anyway, with this one I am looking to still have a +/- 90mm entrance hole, but this time fit a piece of glass with DC Fix attached to the underside of it into the entrance hole. Sourcing a 4” round piece of glass is more difficult than I thought it would be! I have looked into obtaining diffusion lenses used in LED recessed lighting, but that seem difficult and expensive if buying the entire light for the lens. It also seems that the 6” recessed lights frequently have a lens that is only about 3 1/2”. I did manage to pickup a clearanced recessed light which I was able to remove a 3 1/2” x 3 1/2” square diffusion lens. It is plastic, does a wonderful job of diffusing a flashlight beam into a smooth wide wall of light without any throw, but seems it would be difficult to mount and would be too small to still have a 90mm round hole. I think on this sphere I will paint the inside with white acrylic matte finish paint (after sanding the foam). I then think I will cover the outside of the sphere with brown paper and glue (got the idea reading this technical article on a cheap I.S.).

I’m unsure if I should change anything with the luxmeter. I might just start with the same setup I have now - dig out some of the wall (perhaps a bit more this time) and mount the luxmeter against it. I was also thinking to fully remove the styrofoam creating an opening for the luxmeter, but first attach a PVC pipe 90º elbow so that the light can’t directly reach the luxmeter sensor. Also thinking to maybe add DC Fix at this opening as well. I’m considering adding a baffle in front of the lux meter opening too, but not sure if it’s needed or not. Maybe I’m already changing too many things to see what the problem on the first sphere actually was!

So any thoughts on what I’m planning? Does any of this seem wrong? Djozz?


I like those styrofoam balls from your link, they are nicer than mine because they are completely spherical on the inside while the balls that I buy have some holes that I have to fill up. The wall thickness looks fine to me, similar to the 2cm of my spheres.

About light passing through the wall, don’t worry about that, reflectivity will not be 100, almost 90 if you are lucky, that 10% will go somewhere and part of it makes it through the wall and can be seen. What is important is the 90% that is reflected, not what happens to the 10% that is lost with each reflection.

Coating the ball with brown paper in the article is to prevent ambient light to enter the sphere and influence the reading. My integrating sphere is built into a plywood box that works as such (and is convenient because you can put stuff on top of the flat top).

You can try all sorts of diffusion but mind that every diffusion step selectively removes some blue light from your lightsource, so lightsources with a high blue peak in the spectrum will more and more undermeasure compared to souces with little blue. (but I don’t know by how much). So keep the diffusing steps limited.

In case you make a hole for the sensor: a baffle before the sensor is not good, the sensor should “see” as much of the sphere as possible, only the hole where the light comes in must not be in sight for the sensor, it must be blocked by a baffle, about halfway the entrance hole and sensor.

Btw, I still do not see what is wrong with your first sphere, when you posted pictures it looked pretty ok to me.

Umm . . . I started working on my new sphere and ran into an issue. While sanding the inside with 400grit sandpaper, the sandpaper would pull small pieces of styrofoam out causing what look like gouges. Is this going to be a problem for me? I’ve already applied two coats of matte white acrylic paint to the inside, plan to apply two more coats, but the paint isn’t really filling in the gouges.

Some pics:

And Painted (one coat in this pic):


Should be no problem at all for integration of the light.

Ok, good. I’m still trying to decide how to proceed with the rest of the construction - paint outside black? (Read in a forum post not to because black absorbs light. Also, will it matter since I am painting the inside with paint? I’m not trying to keep ambient light out, I’m trying to keep the source light in.) Cut mount hole for lux meter sensor completely through to inside of sphere? Set lux meter sensor back off sphere surface with PVC? Add neutral density filter in front of lux meter sensor? Add baffle(s)?

One thing I will definitely do is add diffusion film/lens at the light entrance hole. I can’t think of a good solid way to mount that to the inside edge of the foam, so I’m planning to cut slits into the sides of the entrance hole and insert the glass/lens into those slits. I will have to do this where the two foam halves meet so I can assemble the halves together with the glass/lens in the slits. I can only find a 4 inch glass at 3/32 inch thickness and even that is a little costly. (It happens to ship from just 30 miles away from me, so I’m tempted to try to pick it up when I’m nearby.)

I know what will happen - I will go through all of this and still have a sphere that gives erroneous readings at certain levels.

I wish more foam sphere builders would post up their findings on what works, what doesn’t, etc. I did find a post over at CPF where the person used their sphere for awhile and discovered it to be inaccurate above +/- 400 lumens just like me.


I wonder how you are sure that the reading of your sphere at higher lumens is much off? Nothing what I know of the construction of spheres or the working of luxmeters, supports that. I’m not saying that is not possible (I do not know everything) but please also consider that your readings are correct and your outside information about the lights you are measuring is wrong or incomplete (like they fail to mention that it was measured at 0 seconds instead or 30 seconds, things like that).

Sorry I haven’t read this whole thread.

I know for higher output lights you really want a bigger sphere.

Have people tried mixing barium sulfate into a paint like mixture to coat the insides?

Directly for the measurement itself, the size of the sphere is unimportant, a sphere integrates the light and does that regardless of the amount of light.

But there are three other reasons for a bigger sphere at higher output:

1)A bigger sphere is needed because high output lights usually have bigger heads and thus the entrance hole must be bigger and to keep the surface percentage of the hole within the required maximum the sphere needs to be bigger.

2)the amount of heat may melt your styrofoam, going bigger spreads the heat over more material.

3)your measurement may get your luxmeter out of range, a bigger sphere increases the range. But a ND filter before the sensor does that too.

I have done BaSO4 coatings on test spheres, tried mixed with latex paint, and mixed with PVD. I did all sorts of checks but I found it difficult to find clear advantages over bare sanded styrofoam.

Yeah, I didn’t want to write all that out. I’m lazy. :stuck_out_tongue:

Was the coating thick enough to keep the light from going through?

I was thinking of smearing Elmer’s glue on the interior surface and then pouring the powder on it and shift it around. That way you have a thick, jagged surface to help integrate the light.

I cannot see images 2 & 3 in post #100.

This seems to always happen with images hosted with Google.