Building an Integrating Sphere ...

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
There are already articles on CPF about different light measurement designs, but none that I found on Integrating Sphere construction. For me, finding a hollow, white sphere was the first challenge. After quite a few searches, a 12" hollow styrene sphere was found at Barnard (phone 888-584-3637). Price with shipping was about $38 - the shipping seemed high until the sphere arrived in a carton almost twice as big as the item.

sphere1.jpg



A top and bottom plate were cut from 1/2" plywood. The sphere measured exactly 12" diameter, and the plywood was cut to 11 7/8" (square) to give the sphere a small amount of compression when assembled.

sphere2.jpg


Four posts (also 11 7/8") were cut to connect the upper & lower plates, and the posts and plates were assembled with yellow glue & brad nails.

sphere3.jpg


sphere4.jpg



The sphere was placed inside the assembly, and four side panels were cut from 1/8" plywood. These are attached with yellow glue & brad nails.

sphere5.jpg


The upper hole is cut with a holesaw to match the diameter of the light meter sensor. The light port is also cut with a holesaw to fit the light tube.

sphere6.jpg


A piece of 2 3/4" stainless tube is used here, but PVC pipe would work just as well. The light tube was coated with Satellite City Special T Super Glue, the inside of the hole was misted with accelerator, and the tube was inserted & twisted until the glue set (about 5 seconds)

sphere7.jpg



Except for applying a finish to the wood, the IS is complete. The light meter is an Extech EA31 (range to 20,000 Lux). It was found on Amazon for under $100 & gives repeatable readings. Since the IS is a comparative device, my concern is that the readings are the same every time a light is retested.

Here are the first Lux readings obtained:


Malkoff M60 in Surefire 6P host - 10,400

Surefire E1B on high setting - 6,540

Surefire U2 on high setting - 6,350

Novatac 120P on high setting - 5,790

Muyshondt Nautilus high setting - 4,900

Surefire L4 - 3,840

Malkoff M60LL in Surefire G2 - 3,330

Surefire E2L (single stage Cree) - 2,970

McGizmo Sundrop - 2,550

Surefire E1L (single stage Cree) - 2,180

Muyshondt CR2 Ion - 1,580

Again, these are comparative readings ... they are not absolute.
 
Last edited:

Blue72

Flashlight Enthusiast
Joined
Aug 24, 2007
Messages
1,138
That is a cool set up.

What difference in readings do you get with and without the sphere
 

half-watt

Flashlight Enthusiast
Joined
Jul 12, 2007
Messages
1,095
Location
Southern Connecticut
very nice construction job. kudos to you for your fine effort and excellent fabrication.

based upon some of the comparative readings, not sure if the setup is truly integrating. obviously, the meter is still reading in Lux. that said, in some respects it still seems to be reading mainly "hotspot" (Lux) readings (plus some, perhaps???) and may not be truly integrating. why would i say this? well,...

unless i have my head screwed on totally backwards, just compare two or three of the SF lights. SF rates the E1B at, IIRC, 80lm, the U2 and L4 at 100lm. now, the lux (hotspot) readings of these three lights are quite different, i would imagine, based upon hotspot (E1B > U2 > L4). your readings are ordered in accordance with what i would imagine it would be if only HOTSPOT was considered (again, E1B > U2 > L4).

your readings indicate, comparatively, for those three SF lights what one might expect from Lux/hotspot readings and not what SF claims for Lumen ratings.

i would like to know what the Lux readings are at a straight line linear distance equivalent to the diameter of the sphere. how different are they from the Lux readings using the sphere? how do they compare to each other?

all this is *NOT* said to denigrate in any way either your exemplary initiative, nor your fine efforts. i just have these questions based upon the readings obtained and published in your Post.
 
Last edited:

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
it still seems to be reading mainly "hotspot"
I understand your thoughts, and agree that some of the readings are not what I expected. Perhaps 12" is too small for full integration of the reflected light. However, the diffusion dome of the meter is looking straight down at the bottom of the sphere, where it sees no direct light. The hotspot from the light is projected against the sphere wall opposite the light port.

I may need to fabricate & install a baffle at the bottom of the sensor dome port. This would lower all readings by an equal amount & eliminate any overreading from a hotspot.

FWIW, Chevrofreak used essentially the same locations for both light port & sensor port:

completed%20sphere%205.jpg
 
Last edited:

McGizmo

Flashaholic
Joined
May 1, 2002
Messages
17,290
Location
Maui
Very cool!! :thumbsup: You might want to install a PTFE or other white type baffle blocking the meter from direct viewing of your source port. I believe most IS have such a baffle.
EDIT: Nevermind! :eek:

BTW, I have a 6" sphere with 1.250" port access and it seems to do fine with anything I can introduce properly to the sphere.
 
Last edited:

half-watt

Flashlight Enthusiast
Joined
Jul 12, 2007
Messages
1,095
Location
Southern Connecticut
understood about the relative positions of the light source and sensor. just having the right angle relationship between source and sensor does NOT guarantee that proper integration is occurring. one could use one flat mirror to reflect light at a right angle from source to sensor. this obviously would not be integration, and the meter would only read the hotspot/highest/brightest portion of the reflected light. i'd like to try to determine how much (in a purely gross qualitative sense; not a quantitative sense) integration is really taking place in your sphere.

it still seems to be reading the reflected hotspot plus possibly some non-hotspot reflected light, and NOT integrating more/"all" of the light output. if it did the U2 and L4 ought to read higher than the E1B (again, unless i have my head screwed on backwards).

if you measure the Lux output straight on at 12" (the diameter of your sphere) without the sphere, there are only three relative possibilites. the readings of each will be higher, lower, or the same when compared to the "in sphere" readings.

if higher, then your sphere [readings] and Lux meter are only measuring SOME of the reflected light which appears to the meter to be reduced by the reflection in the sphere.

if lower [, but still the same relative relationship E1B > U2 > L4], then there is some integrating taking place, but probably (again, based upon U2 and L4 being SF rated at 100lm vs. E1B rated at 80lm) still to a degree only measuring the brightest portions of the beam output.

if lower, but a different relationship (e.g. E1B, NOT being the highest reader), then there may be something amiss in the straight line Lux readings as one would expect the E1B to be the brightest (i really don't know how they compare in Lux testing; i'm just guessing here based upon the brightspot of each light - i own all three of these SF lights).

if the same (i don't expect this to be the case), then no integration is taking place.

this is the values of performing a "sanity" check by taking normal straight line/linear Lux readings. i suggest using the very close 12" distance since that is the diameter of your sphere, or use 24" (or any other distance - easier to use a multiple of the sphere's diameter - makes purely mental calculation of the inverse square easier; i can't easily do inverse square with decimals in my head) and use the inverse square law of radiating energy to compare the results to the sphere's Lux reading.

again, excellent work on the concept and fabrication. obviously, a valuable original Thread and Post on your part.
 
Last edited:

Norm

Retired Administrator
Joined
Jun 13, 2006
Messages
9,512
Location
Australia
I'll start by saying I know nothing about this subject but I would have thought the best position for the light sensor would be close to the input port, if the tube at the input was set an inch or two into the sphere it would also act as a baffle.
Norm
 

MrGman

Flashlight Enthusiast
Joined
Feb 6, 2007
Messages
1,777
Very interesting work indeed, Can you see white light coming through the sphere to the outside world? Is it glowing? If so you are losing light because the material is not as reflective as it should be. What ever amount of light is coming out is not being internally reflected.

Half-watt's test sounds valid as a way to determine if its integrating or just getting most of the hot spot readings.

Also what does IIRC mean?
 

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
I looked again in the sensor port (which is 1/2" plywood followed by 1" of styrene foam) and saw that the styrene foam is capturing a some of the spillbeam, leading to an inflated reading on some lights.

Since the sensor port is 2" diameter x 1 1/2" thickness (to the inside of the sphere), a piece of 2" tubing was painted flat black inside, and pushed into the port. The new readings are as follows:

Malkoff M60 in Surefire 6P host - 2,880

Surefire E1B on high setting - 1,780

Surefire U2 on high setting - 1,774

Novatac 120P on high setting - 1,582

Muyshondt Nautilus high setting - 1,370

Surefire L4 - 1,203

Malkoff M60LL in Surefire G2 - 935

Surefire E2L (single stage Cree) - 864

McGizmo Sundrop - 727

Surefire E1L (single stage Cree) - 614

Muyshondt CR2 Ion - 454

If you look at these numbers as a percentage, they are nearly identical to the first test without the black tube baffle. The numbers below are in percent, with the Malkoff M60 used as a 100% standard (numbers are test 1/test 2)


Malkoff M60 in Surefire 6P host - 100/100

Surefire E1B on high setting - 63/62

Surefire U2 on high setting - 61/61

Novatac 120P on high setting - 56/55

Muyshondt Nautilus high setting - 47/48

Surefire L4 - 37/42

Malkoff M60LL in Surefire G2 - 32/32

Surefire E2L (single stage Cree) - 29/30

McGizmo Sundrop - 25/25

Surefire E1L (single stage Cree) - 21/21

Muyshondt CR2 Ion - 15/16

To me, the only reading that falls outside the expected lumen curve is the E1B. I bought a few of these and sold all but this one (which is on a McGizmo Ti McClicky Pak host, not the standard SF body). It has about the same low setting as any other E1B, but the high is the brightest of any that I've seen. Maybe SF put the wrong driver board in this head, maybe the emitter bin is at the top of the XR-E series, but without a doubt it is brighter than the U2 on a ten foot, white wall bounce.

A five foot ceiling bounce gives the U2 @ 42, the E1B @ 44.
 
Last edited:

half-watt

Flashlight Enthusiast
Joined
Jul 12, 2007
Messages
1,095
Location
Southern Connecticut
regarding your E1B observations: understood. seen this type of thing before (though mine was a Fenix P1D-CE which was brighter than four P1D-Q5's; the fifth P1D-Q5 was brighter though). it's probably the LED and *NOT* the E1B electronics responsible for what you are seeing. the disparity b/t U2 and L4 readings though still indicates more of a Lux type of reading rather than a Lumen type of reading, if i'm not mistaken (and feel free anyone to correct me if you feel that i am mistaken). however, this assumes that SF has correctly rated both the U2 and L4. SF appears to rate conservatively and perhaps they are both >100lm, but that the U2 is more conservatively rated??? don't know. also, only one specimen of each light is being used, so perhaps the U2, in this case, has more overall light o.p. than the particular L4 specimen being tested???

also, thinkin' a bit more, i think my early contention that you would get a higher Lux reading in the sphere vs. linear is just plain wrong. since the sphere diffuses the light, including the hotspot. this ought to even out the brightness. looks like i did have my head screwed on backwards - wouldn't be the first time!!!
 
Last edited:

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
I would have thought the best position for the light sensor would be close to the input port
The sensor can be located in a number of positions, so long as the sensor does not see either the light source or the hotspot. The easiest sensor position (for a closed sphere) is 90° to the source. Larger spheres are split in half and hinged - this allows the source to be located in the exact middle of the sphere, with a baffle blocking direct radiation to the sensor.

Can you see white light coming through the sphere
The foam is 1" thick, so none of my lights (even the 240 l Malkoff) penetrates.
 

MrGman

Flashlight Enthusiast
Joined
Feb 6, 2007
Messages
1,777
I just checked with one of the optical engineers here at work. The integration sphere must have a baffle so it cannot read the light source directly. In case no one has already made that certain.
 

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
I understand your concern about the baffle, and the need for no direct light to strike the sensor dome, or strike inside the sensor port where it would reflect into the sensor dome.

I did conduct a test for this. All room lights were turned off. The photosensor was removed from the top port and a flashlight was turned on & placed in the light port. I inspected the sensor port to make sure that no direct light was hitting the inside of the sensor port tube. In a dark room, the only light I saw was the light at the bottom of the sphere. For this reason, I believe that the meter is reading an integration of total luminous flux.

A calibration light source is something I'd like to find. The commercial ones plug into a wall outlet, and the output varies with the size of the sphere. That way, the relative readings could be converted, using a conversion factor, to lumens.
 

MrGman

Flashlight Enthusiast
Joined
Feb 6, 2007
Messages
1,777
I looked inside the IS we use at work today and also talked to our head PhD Optical Engineer about this. The Baffle is very important in not having false high readings. Its a small circular 2 dimensional object that is literally directly between the 2 ports. Its tilted up in the circular cavity so that any light directly from the input port would cast a circular shadow on the meter port. Since its on an curving slope the best way to describe is that it is parallel to the flat vertical opening surface for the meter port. It is the same flat white color/texture as the rest of the sphere.

In this way lights with a wide angle head cannot have any direct light impingement in the meter port.

You may not think that the lights are going directly to the port but the results you posted may have a false high reading. There should be nothing that sticks into the inside surface from the outside world that goes beyond the wall thickness.

You have gone this far, might as well try it, then divide all your readings by 36 (radius squared) to see how close they get to the rated lumens of the lights you use. This may be an over simplification, but its quick and easy for now. May need a fudge factor. The sensor in the one at work is only about 1.8" diameter opening.

I notice on the one I keep testing with at work (its a 6 inch sphere by the way and its calibrated with the meter and our engineers trust the readings of it for all kinds of LED light sources), that if I move the light into the sphere at all the readings actually go down. It appears the best readings are with the light source directly at but not past the opening of the sphere. That is why I believe that the black tube at the inlet port is really going to affect the readings negatively. If you are going for all comparative readings, it still doesn't buy anything to have that black tube on the inlet. Real IS's have no such thing.

I would expect all the readings to go down some what with the baffle installed.

In theory a 235 lumen light source of white light should read 8460 lux if the baffle does its job right.

If I am off on these numbers it means I should have gone to bed already. Good luck.
 

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
it still doesn't buy anything to have that black tube on the inlet. Real IS's have no such thing.
Real IS's aren't made with hollow stryofoam balls & cost just a little more than $38 ... and the light/lamp/LED being tested is typically located centrally in the sphere, separated from the photo detector by a baffle that prevents direct light readings. I couldn't figure a way to easily construct such a sphere, which involves cutting the sphere in half to produce two hemispheres hinged together.

The baffle probably should be installed, as it is common to most IS designs available, including the simple design I used:

integrating_sphere.png


In the example above, there is a baffle both above & below the light port (entrance aperture). The dual baffle arrangement is most likely used to prevent a wide angle light source (like a McGizmo Mule or Sundrop) from directly striking either the photodector or the bottom of the sphere - both or either of which would give an inflated reading.

the results you posted may have a false high reading
I also believe they did. The photo detector was looking through the 1" styrofoam sphere wall. That's a little more than 6 square inches of reflectance being added to the reflectance from the "bottom" of the sphere. The black tube does have a high negative impact, taking the first reading of 10400 down to 2880, a reduction on almost 73%. But, the percentage values from the first test are awfully close to those of the second test - the SF L4 having the largest test to test spread. Not a great surprise, with the wide angle, flood type beam dispersion of the L4.

I'll see if I can figure a way to mount both the lower & upper baffles.
 

MrGman

Flashlight Enthusiast
Joined
Feb 6, 2007
Messages
1,777
I just talked to my PhD Engineer coworker at lunch about the dual baffles. The only reason for multiple baffles is for units that have multiple input ports. Only one baffle is needed and used for a specific input location to the photodetector port. Our integration spheres only have 1 baffle. FWIW.
 

precisionworks

Flashaholic
Joined
Apr 19, 2007
Messages
6,623
Location
Benton Illinois
divide all your readings by 36 (radius squared) to see how close they get to the rated lumens of the lights you use.

Even though the OD is 12", the ID is 10" (walls are 1" thick). My radius squared is 25 ... but the flat black tube to the light meter dome absorbs enough that my "corrected" radius squared becomes 12.5.

I've thought about this quite a bit, and plan to line the flat black tube with a bright white tube that has an angled cut end. As long as that angle is 45° or more, the long side of the tube will have the same effect as adding an internal baffle - direct light from the flashlight lens will strike the outside of the tube but not the inside where an inflated reading will result. At that point, the factor of 25 will probably be correct.
 

MrGman

Flashlight Enthusiast
Joined
Feb 6, 2007
Messages
1,777
I am confused??? Are you talking about the cut up maglight in the inlet port or is there something else on or around the actual light meter you use as your photodetector? I would have no built up walls to prevent light from coming in from all around the sphere to the detector except for the one baffle directly in line with the front end of a light at the inlet port. If you can't put it inside the sphere because the sphere is now sealed up, I would put it only at that limited angle along the opening of the inlet port that faces directly to the detector.

I also realized that your inside diameter was smaller than your outside one about 1 minute after I made that last post but I figured you would figure that part out real quick since you built it and would just see what you came up with anyway as you did, so I didn't bother to send a correction.
 
Top