Building an Integrating Sphere ...

[wbp]

I've been pondering the accuracy of our polystyrene spheres. I believe that lights that have a very small intense spot are not measured as accurately (or the same) as those with wider beams. I observe a "hot spot" on the outside of the sphere where the beam hits, and surmise that some light is getting through, and more so with more smaller beam spots.

It occurs to me that we might be able to improve this with the application of something directly opposite the input port. Perhaps a small disc of aluminum, for example, or aluminum coated by BaSo4. Seems like this would be something interesting to experiment with.

Also, if anyone is interested, I have been able to obtain service literature for the AEMC CA813 meters, and can now calibrate them. This meter claims to have "CIE Photopic" sensitivity, but like all such meters it is calibrated using an Illuminant A light source, which is a long way from our LED flashlights. Also, with any meter using a cosine receptor, the placement of the sensor is critical - this can introduce significant error during calibration. I've developed a new technique for calibrating this type of meter that yields significantly better results.

William

 

[MrGman]

Another option would be to use a slight diffusion filter of a known percentage drop of total light efficiency to spread out the light. I and precisionworks have been experimenting with just such a set of diffusers from Gene Malkoff, very nice material indeed. But if you knew the diffuser dropped the total light transmission by a solid 6%, then you could reevaluate all the tight spot beamed lights and see how they fair in that manner. I just got all my supplies back from Precisionworks (Thanks) I could send you a couple filter samples.

 

[wbp]

I have a pretty good collection of diffusion filters here already - which one(s) did you find helpful?

I found another problem with the polystyrene sphere - color shift. I had been recording CCT and coordinate readings taken with the sphere. A couple of days ago I needed to add color data for a particular light, so I set it on the stand and aimed it directly at the SP-100. Then I measured one of my reference lights the same way, and got *very* different numbers than what I get using the sphere. So all my CCT data done with the sphere gets tossed...

William

 

[Ryanrpm]

Gentlepeople, I have installed a baffle, and the results are, I think, very interesting. First of all, a couple of pictures, showing the baffle and the sensor port:

http://www.meier-phelps.com/temp/IMG00025.jpg
http://www.meier-phelps.com/temp/IMG00026.jpg

I admit, the baffle is a bit of a hack, not exactly round, but I was in a hurry. The black ring in the sensor port is a bit of rubber fuel line, it hold the tip of the fiber optic probe very nicely.

Now the measurements (cd/m2, CCT in K, calc'd Lumens):

[FONT=Fixedsys]Fenix L0D Q5         Hi   2051  5455K    82
Muyshondt Aeon       Hi   3219  5077k   129
[/FONT][FONT=Fixedsys]Bitz Pocket          Hi   3252  6080K   130
[/FONT][FONT=Fixedsys]Nitecore Defender    Hi   3387  6213K   135
[/FONT][FONT=Fixedsys]Nitecore EX10        Hi   4220  5739K   169
[/FONT][FONT=Fixedsys]Olight T10           Hi   4455  6147K   178
[/FONT][FONT=Fixedsys]Dereelight DBS V2     -   5380  5633K   215
[/FONT][FONT=Fixedsys]Jetbeam II IBS       Hi   5522  6048K   221
[/FONT][FONT=Fixedsys]Lummi Raw NS         Hi   6688  6409K   258
[/FONT][FONT=Fixedsys]Milkyspit Creemator  Hi   7674  6577K   307
MTE SSC P7           Hi  11490  5674K   460
[/FONT]

(The Fenix has freshly charged batteries)

I can repeat measurements and get within three percent. For example:

[FONT=Fixedsys]Bitz Pocket          Hi   3332  -----   133    2.2%
Muyshondt Aeon       Hi   3291  -----   132    2.2%
[/FONT][FONT=Fixedsys]Olight T10           Hi   4580  -----   183    2.8%
[/FONT][FONT=Fixedsys]Jetbeam II IBS       Hi   5675  -----   227    2.8%
[/FONT]

Some of this error is due to the Lightspex recalculating dark frames every time it changes ranges. I'm being lazy and just covering the light being tested, and not actually pulling the probe and covering it...

These numbers look really good to me. I've got a Fenix T1 coming so I can check that.

At some point I want to do a run with a more affordable meter like one of the Extechs (I have a 403125 here) and see if I can calibrate that to get the same readings. If I can, then we could actually build these for a very reasonable price.

Other than that I think I'm done unless someone can point out something I'm missing. What fun!

William

Hi William, do you find that the Extech 403125 gives you pretty reliable lux readings?

 

[wbp]

Hi William, do you find that the Extech 403125 gives you pretty reliable lux readings?

Depends on what you want to do.

I prefer the AEMC CA813. The Extech is not very linear. It is more difficult to calibrate than the CA813. Neither of them will give accurate lux readings for LED lights as received, regardless of factory calibration.

William

 

[JWW]

Hello,

I am new hear and read this thread a few months ago. Any chance of showing a few updated images of your integrating sphere in use?

Regards,
-JWW:

 

[sygyzy]

I want to build an IS or a lightbox this weekend. Can someone explain where the baffle needs to be placed and how large it has to be? I know you want to block direct light but size and shape, how far it drops down from the wall of the sphere and how much it wraps around the censor (180 degrees at flashlightreviews.com), are all important. It seems like the sizes an shapes of baffles being used here are just "random".

 

[milkyspit]

I want to build an IS or a lightbox this weekend. Can someone explain where the baffle needs to be placed and how large it has to be? I know you want to block direct light but size and shape, how far it drops down from the wall of the sphere and how much it wraps around the censor (180 degrees at flashlightreviews.com), are all important. It seems like the sizes an shapes of baffles being used here are just "random".

Sygyzy, I don't have exact answers to your questions, but can vouch for both Quickbeam and Leef as having particularly well-designed lightboxes... don't recall if they posted detailed threads about their construction on CPF, but I'd try a search for their threads... either one ought to provide an excellent model for your own construction.

(To all: granted these aren't true integrating spheres, but for relative testing of one light to another where all the lights are LED, or all the lights are incandescent, my own past experience has been these boxes can do surprisingly well! And it puts fairly decent measurement capability within reach of most folks, from both cost and complexity standpoints.)

 

[wbp]

I want to build an IS or a lightbox this weekend. Can someone explain where the baffle needs to be placed and how large it has to be? I know you want to block direct light but size and shape, how far it drops down from the wall of the sphere and how much it wraps around the censor (180 degrees at flashlightreviews.com), are all important. It seems like the sizes an shapes of baffles being used here are just "random".

The baffle needs to block the sensor such that no light from the source directly hits the sensor. Preferably it's made from the same material as the sphere, or coated with the same coating. It should only be big enough to cover the source opening from the perspective of the sensor opening.

Are you using a polystyrene ball? What size? What meter?

I've been "inspired" to try building a new sphere with a different type of ball. I've also been working on a way for anyone who does this to calibrate their sphere so we can all get accurate measurements. Stay tuned...

William

 

[sygyzy]

I want the baffle to be near the source and not the sensor?

I am still confused about the shape, size and position of the baffle. Needless to say, I am baffled <--- get it?!?!

I was thinking about using the ball like in this thread but then I thought about the other lightboxes made in, well, boxes and they seem to work fine. Does it matter if the container is a sphere, house shape, pyramid, cube, or rectangle if the user is using it consistently across their lights?

 

[wbp]

The baffle goes between the sensor and the source. Typically they are half way in between. I posted some pictures of mine somewhere around here, can post again if need be.

If you want accurate results, it has to be a sphere. Otherwise you'll end up with different reflections for different beam shapes, and that will throw off the comparisons.

I am currently testing a 12" diameter (ID) sphere made from something a little different. Like I said, stay tuned...

William

 

[wbp]

has anyone thought about using one of these:

http://www.walmart.com/catalog/product.do?product_id=7757565

12 inch ID, white plastic. I've got one I'm playing with now. Drilled a 3" hole for the source, 7/8" for the meter. Using an AEMC CA813.

William

 

[sygyzy]

wbp - Well I built a rectangular light box. I think it'll be good enough to compare my lights against each other.

I'll build a sphere later. I saw your picture of the baffle but due to the way it was shot, it was a little confusing as to what I was looking at.

Let us know how that new sphere turns out.

 

[run4jc]

Wondering if this GREAT thread can be rejuvenated...I purchased a 16 inch smooth foam sphere and the same Extech Easyview 31 light meter that was mentioned in the beginning of this thread.

I have no illusions of being able to create an accurate IS - but what I would like to do is build something that gives consistent RELATIVE readings. After reading through all 4 pages of posts in this thread, it seems that the most important thing to ensure such (readings given the various shapes of beams) is to avoid any direct light hitting the sensor, accomplished by the installation of some type of baffle.

And it also appears that the installation of some type of tube to properly position the sensor is necessary, but how long is the right length? Just 'through' the styrofoam, or actually extending into the sphere for an inch - two inches - whatever?

I've seen multiple photos of how baffles are positioned - debate over size and # of said baffles - any consensus on the best way?

Mine will be simple (read "cheap") and I will not glue the sphere - rather tape it like wpb's original so it can be modified/improved.

One more question - what makes us do this? I spent $130 on a meter (cheap, I know), $40 getting a 16 inch smooth foam ball shipped to me - about to purchase a couple of lengths of PVC pipe and who knows what else - just to get MORE readings (so many are readily available)....and I can't WAIT to do it!! Heck, that money would have purchased a decent Surefire.

BTW, that was a rhetorical question - I know why we do it - we are FLASHAHOLICS!! :twothumbs:twothumbs

And again, I spent an hour reading this entire thread, so many of my questions are already answered, but if anyone has any new insight(s) since the last post almost a year ago, please weigh in!

Thanks!

 

[jashhash]

This is such an awesome idea and I'm so glad I found this post. After reading this post I have decided to undertake an Integrating sphere project too. I have already made a bobber cooler integrating sphere and it works great but im still not satisfied with the 4% error margin. I'm going to make a new sphere by plaster casting a 16 inch ball kind of like this one:

Then I'm going to make a ultra reflective paint using 90% barium sulfate powder and 10% elmers glue.

In case you might be interested you can get 500 grams Barium Sulfate powder from here: http://www.carolina.com/jump.do?keyword=barium+sulfate&Search.x=0&Search.y=0 for$13.50.

 

[Mick]

This paint system used for video projection screens might be good for the inside of the Styrofoam balls. The first coat is highly reflective and the top coat is a diffuser. I should block any light loss.

http://www.goosystems.com/

This place will sell the GOO in 120 mL size which should be plenty for a IS.

http://www.rosebrand.com/shop/results.aspx?keywords=screen+goo

I have a wall painted with the stuff for my home theater.

 

[precisionworks]

Didn't realize how long I'd been away from this thread :nana:

I've added a few lights since I last posted ... will measure those today & update the chart.

Here's a copy of the chart from post #21, and including the Catapult V2 SST-50, plus the McGizmo Haiku 6 volt (High CRI emitter with custom set driver for lower lumen output/greater runtime).

First reading is my lumen calculation, followed by factory stated lumens:

Mac's Custom P7 (AW C-cell LiIon) = 782 lumens (about 800 per Mac)

ThruNite Catapult V2 SST-50 (2xAW18650, 2900mAh) = 238/722 (250/1000 emitter lumens)

Pocket Rocket - 8th run XM-L (T6-3B bin, 1xAW18650, 2900 mAh, ) = 51/184/680

PR8 XM-L flux should be about 7% higher with the U2-1C bin, giving 55/197/728 Lumens (estimated)

Surefire 6PX-Pro LED = 16 / 235 lumens (15/200 per SF)

Malkoff M60 with 60Ω Sandwiche Shoppe mod = 15/192 lumens (235 per Malkoff website)

NiteCore Extreme XR-E Q5 = 5/186 lumens (190 on Tactical High)

Early Surefire U2 = 7/114 lumens (2/100 per SF) / Latest U2 1/161 lumens

E2L two stage KX2 head = 12/119 lumens (1x17500)

E2L two stage KX2 head = 11/118 lumens (2xCR123)

E2L two stage KX2 head = 11/113 lumens (1x18650)

Surefire E1B = 11/113 lumens (5/80 per SF)

Novatac 120P = .23/11/99 lumens (.23/10/120 per NovaTac)

Muyshondt Aeon = 10/92 lumens (?/114 per Muyshondt site)

Muyshondt Nautilus = 6/91 lumens (?/107 per Brightguy site)

McGizmo Haiku 6v HCRI = 5/23/89 (5/25/90 per Don IIRC)

McGizmo LunaSol 20 = 9/73 lumens (9/78 per Don)

Surefire L4 = 66 lumens (100 per SF)

M60LL in Surefire G2 = 61 lumens (80 per Malkoff site)

Surefire E2L (single stage Cree) = 57 lumens (45 per SF)

McGizmo SunDrop = 46 lumens (40-50 per Don)

Surefire E1L (single stage Cree) = 40 lumens (30 per SF)

Muyshondt CR2 Ion = 3/29 lumens

Surefire KL1 (Luxeon) head = 29 lumens

Gerber Trio (2 AA Lithium) = 22 lumens (24 per Gerber)

Does anyone reading this remember when ThruNite brought out the V2 Catapult (SST-50)? Quite a few people with calibrated spheres were reading about 700 OTF lumens, which caused David at ThruNite no end of distress. From everything I've read, emitter lumens X .66 = OTF lumens (more or less, plus or minus, approximately, etc.). If 66% is a valid number to use, the emitter lumens are closer to 1100. Whatever the case, it's my favorite thrower, possibly because it's my only thrower Car drivers cannot move over quickly enough when the V2 is brought to bear.

 

[HooNz]

Is one lit Candle one Lumen in a sphere? and is the size of a candle uniform eg: a small birthday cake candle compared to a 6 inch candle , then petroleum based wax to bee's wax , a thick wick and the opposite , different wick compositions and all that stuff .

A quick answer would do me , unless one has the time or experience .

 

[precisionworks]

Finally received the Surefire 6PX-Pro that was ordered a few weeks ago from OMB Police Supply. They said it was on backorder, but it was worth the wait:thumbsup:

One of the nicest beam profiles that I've ever seen from a Surefire - not super tight like the TIR lights, not extra wide like the L4, but a well defined hotspot that fades nicely into the spillbeam. This light is a keeper, and $72.99 delivered is a price I can live with.

Surefire claims 15 lumens on low, my sphere showed 16 lumens, which is 6% higher than they state.

Surefire claims 200 lumens on high, my sphere showed 235 lumens, which is 15% higher than they state.

Again, my sphere is not calibrated to ANSI standards, but rather shows relative brightness of the lights I've been able to test.

The chart in post #117 has been updated with this addition.

 

[precisionworks]

Added NiteCore Extreme XR-E Q5. That's a little pocket rocket NiteCore claims 190L, my reading is 186L, not bad for a single CR-123. Chart in post #117 updated.