Possibility of Integrating Sphere Purchase

Hi guys,
I dont know if this has been done or tackled already, but considering the fact that this forum community could really use an integrating sphere to put all assumptions and manufacturers reted outputs to a rest, and considering the fact that there are almost 9,500 registered members, I think that it would a good idea to setup an Integrating Sphere Contribution Account where people could donate some money for the purchase of an integrating sphere.

Now i dont know how much this apparatus costs but maybe 2 - 3 USD per member ( or more for the generous ones ) would cover the cost ?

The IS could be delivered to a respected member/flashlight reviewer to be put to good use for all of us.

What do you think ?

Sounds cool, and we are crazy enough to do it. But who would have it and take on the task of millions of flashlights being sent to them to test? Plus aren't these things huge? I think you'd need a garage to house it alone!

Don't know how big a one is needed. Ebay has a 6 inch one (whatever that means) here:

http://cgi.ebay.com/Optronic-OL-IS-...ryZ73165QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

The problem isn't so much buying the sphere but rather keeping it calibrated long term. Also, the very narrow beam of most flashlights would require an integrating sphere with a lot of sensors. The method I've used to estimate lumen output is likely not more than 10% off although I don't know how well it will work with flashlights having a much narrower beam angle than bare LEDs.

Interesting concept. We could then even charge a small fee for every test to earn back the cash and avoid a bazillion lights sent in ...
The big question remains ... who will be the poor soul doing all the testing and maintaining the sphere?
bernie

Well i guess the size of the IS depends on what youll be testing, for a flashlight i dont think that we would need a huge IS. Hey according to the one on ebay it dont look like it is as expensive as i thought.

Im not sure about d many sensors as i think that an IS is supposed to diffuse the light into an even flood within the sphere since it is supposed to measure the total amount of light emitted by the object in question. then i guess it would convert the amount of light that falls on the sensor ( which should be the same anywhere inside the sphere )into Lumens. therefore it would not matter in which direction or how wide or narrow the beam is. dunno about calibration.
Im no expert but that is what i think an IS should do. maybe the experts could lend us a hand and explain what an IS actually does.

Maybe someone who already has a large collection of lights could have it so that he/she could test the already available lights thus drastically reducing the load of sending and recieving lights. the small fee per test would pay for the calibration ( if any required ) rather than to earn back the IS cost.

This is just vapourware but it would be really great and would answer many questions and make/break many myths if it could be turned into reality.

jtr1962 said:

The problem isn't so much buying the sphere but rather keeping it calibrated long term.

Click to expand...

Absolute calibration would require a calibrated source. Relative calibration should be fairly straightforward. Have a reference light whose output you trust will stay about the same, e.g. regulated, underdriven Luxeon running on a standard battery setup. Measure the output when you first get the IS, and check it periodically thereafter. Assume the light's output is constant and adjust the IS readings accordingly.

there you go, one problem solved ??

only 2,000 to 10,000 $$$ problems remaining ??

Plus one - the IS user

My understanding is that using an IS is rather complicated and time-consuming - the setup for each test involves a number of custom-made parts and may be damage to the flashlights being tested.

If the IS is not large enough to house the whole flashlight (which has to be masked white at the very least) and house the custom-made baffle(s) too, then it has to be larger enough for an opening (port) and custom-made adapter into which the flashlight can be shone. The IS will still need baffle(s) and you'll also need apparatus for supporting the flashlight) (the parts of the bezel visible inside the sphere would have to be masked white).
The larger the IS is the larger the port or disruption can be (otherwise the calulations required to compensate are horrific and make everything more complicated. It would be better to use like a 6ft diameter IS if you plan to put things inside it for example.

Then there are issues of sensor types (incandescent vs LED light need sensors that can measure the output accurately) and all the considerations of temperature inside the sphere from the flashlight and/or beam. It would be useful to be able to measure the colour temperature and such too.

The software to record the output over time is important. You'd need a dedicated PC as well as a supply of batteries and lamps to gain repeat measurements.


You'd need to buy and maintain several calibrated light sources, and a stock of high quality IS white 'paint' and the materials for the baffles and masking of the flashlights or creation of the port fixtures and support apparatus.

It would also be a good idea to send a tested sample off to at least one real lab or other testing facility willing to assist for "round robin" testing to check that everything was working correctly (you want to ensure that your test equipment and methods are up to scratch so not to do a load of testing only to find out it was invalid).

I work for an testing and research lab, and several years ago I toured SureFire's facilities and saw their IS rigs.

I'm sorry to say this but I don't think it is realistic for CPF to obtain an IS. It would be easier and cheaper to build a relationship with a lab that already uses IS's.

Al

Integration Sphere Information

I had a long talk with Chris at www.sphereoptics.com about our "hobby" needs for an integration sphere. He is going to send me a manual to check out and is willing to help us set a system up.

The first question was about how large the lights we are wanting to measure are. I told him the range ran from ¼ inch to 8 or 10 inches. He said that for an 8 or 10 inch aperture, we would need about a 40 inch diameter sphere. Of course the cost goes up with an increase in diameter and the logistics of a "pass around" of a large sphere can be a bit difficult as well.

He suggested that the 12 inch unit would be a good compromise for our use. It has a 4 inch aperture, which would cover a lot of the lights we use, but you could not test a Thor in it.

There are some maintenance issues as well as some set up issues, but they don't seem to be that difficult to do. Fixtures need to be made to hold the lights in place, and the unit needs to be calibrated prior to each use.

The standards call for 3 independent calibration source checks. This is done with a NIST traceable set of 3 lamps that are checked prior to the test run. They sell a "3 pack" of these calibration lamps for $2000.00, or you can purchase them individually for $900.00 each. The lamps have a "calibrated" life of 50 hours, then need to be replaced.

This is not as bad as it sounds. Once you have the system set up, you can pick up some automobile tail lamps and do a cross calibration. Some people have used their calibrated lamps less than 50 hours in over 10 years… The calibration lamps need to be free from fingerprints and need to be protected against corrosion and breakage. Also, a log needs to be kept to track lamp use.

The suggested unit is the SLM12. It consists of a 12 inch sphere, a calibrated lab quality power supply, a spectrum analyzer that runs in the 300-1050 nm range, and the Windows based software to run the unit and gather data. This will give us Luminous Flux (lumens) of a lamp (or flashlight) operating at a user specified current or voltage. It can also accommodate a battery operated light as well. In addition, the spectral data on the light is collected, including the chromaticity values of x, y, z, u, and v, correlated color temperature, peak wavelength, and dominate wavelength. This data can be exported to Excel (or your favorite graphing program) for further analysis.

The cost of the SLM12 is $13000.00.

He is willing to consider a small discount (maybe 5%) in exchange for a banner ad, or other form of advertising.

This unit is somewhat portable. The sphere, power supply, and spectrometer can be packaged up and shipped without too much difficulty. To run the unit all that is needed is a Windows based computer with a USB connection. There will be some minor additional costs to come up with fixtures to hold the lights and lamps, but I think we have some clever and creative people here with access to fabrication equipment that may be convinced to help.

Let's see now… $15000 divided amongst 500 people works out to about $30 per person…

Chris did also mention that they offer testing. The set up cost is $450 and then it is around $30-$50 per test. He suggested that we may be able to earn some of our cost back by offering testing at a "competitive" price.

Tom

Hello Al,

The information you have is very true for older systems. It seems they have simplified things a large amount with the newer systems, or at least that is the impression I am getting.

Do you have anyone you can contact to check into this? I believe the current unit calibration involves installing the source lamp and pushing a button to collect the data into the software. The software then gives the test data in relation to the calibrated source data.

The fixtures to hold the lamps do need to be fabricated and as with all testing, some fiddling around will be needed to get the set up correct.

Tom

Silverfox, thank you very much for all d info. i dont think it is as bad as it sounds, maybe we could shop around for some other quotes, or a possible self calibrating IS.
Baffle/light holder making would be the only requred work to b done. What happened to the other 9000 CPF members ? I think 500 ppl is a very conservative number dont u ?

Ocean optics also makes optical test equipment like this. Based on what I have seen, your rep is correct that at least a 1 meter dia would be required.

For me personally, I think the total Lumen value from a sphere can be approximated well enough with the cat litter box (within 10 % or so). The basic lux readings can be done with a hand held light meter. What we really lack is a spectro photometer which can give us better spectral information about the beam.

Of course, if someone did set this up and offer the service, I would certainly use it. Perhaps we can find a lighting test lab and have some kind of agreement with them to place test equipment there and we share the service fees.

The numbers I crunched a couple of years ago were to see whether the test lab I work for could set up an IS and actually get a return on the investment (of money, technician time, floor space etc etc).

My concern about getting a 'cut price' set-up is that it will not be able to rate the HID lights if it is too small. More importantly however is whether we can perform these tests to the accuracy we desire (to make meaningful statements and compare different flashlights).

For example, I would expect Lumen ratings to be accurate to 1 Lumen.
Otherwise we would not be able to spot variations and trends.

One concept I'd be interested in is to run a 5W Luxeon flashlight constant-on from a bench supply (recording the data) whenever the IS is not in use to see how it's output changes over time. This test unit would travel with the IS as an on-going experiment.

The magnitude of what 'hard-core' flashaholics could achieve with an IS is quite staggering really. We have the ability to rate any(all) flashlights and report the findings. Something manufacturers can't do.
But unless it is done well it will be waste of time and money. It's a huge investment and requires an even bigger commitment.

Al

I have to agree qith Harry. The commercial builders need measurements that are tracable back to the NIST calibrations. The average user would just want to know if the light is close to the claims.


The niche for a CPF sphere would be fairly simple;

1) Disprove the more outrageous manufacturer claims
2) Provide approximate readings for the custom lights and mods.

The kitty litter bucket works as a very crude integrating sphere. It's an 18 inch deep white bucket. The LUT (light under test) is pointed into the box 18 inches from the bottom. The light meter (set to measure LUX) is set up pointing into the bucket from 1 meter. It's calibrated by moving the meter back and forth while using several lights of known output.

It's not accurate, but it is good enough to tell when a light advertising 200 lumens is blowing smoke. It's also neat to compare a lionheart with a VIP, for instance. I bring my homemade IS to the CPF get-togethers and we all have fun comparing our lights.


OK, Having said all that....

It should be possible to get a used IS cheap. It should be possible to use a simpler light meter instead of a spectrometer. It should be possible to have a lab measure a sample light source (like a small desk lamp with a regulated power supply) and use that for the unofficial "calibration" source.

It would not be legal for commerce, but it would help confirm our feelings of confidence in our favorite light makers.

Daniel

I'd be in for a contribution for sure
And imagine ... there are so many CPFers who'd want their units tested ... if they all pay a little fee then we can easily recover the investment over time.
Biggest problem is the poor sould who has to do all this IMHO ...
bernie

Hello Daniel,

I have been involved in Non Destructive Testing for nearly 30 years. I constantly get asked if there is a cheaper way to get the information needed. My answer always is that it depends on what you are going to do with the information.

If you are simply curious about a generality, a simple inexpensive test on a limited number of samples will do just fine. If you have to make a decision based on the test results, you will probably need better information.

I have done tests that started out with 3 samples, and as more and more questions came up, ended up with testing over 100 samples.

The key is having testing equipment that is accurate, and following established testing procedures. The testing procedures can be altered for special circumstances as long as there is a valid argument for doing so. I own and maintain over $200K of specialized testing equipment, and am constantly questing the accuracy of my test results. It has been my philosophy that the post test discussions should be centered on the interpretation of the test results with comments on the procedures used, and not on the accuracy of the test equipment used to gather the data.

Al touched on an idea that I feel is very important. If we as a community purchase a piece of test equipment like this, we should be able to offer services to the modders of our group to help them come up with better ideas.

What if we discovered that pulsing a 3 watt LED at 6.8 watts gives very similar aging results as driving it with constant current at 4 watts?

It would also be very useful for narrowing down the tint issues for custom lights.

I spent about an hour with Chris talking about our milk carton and kitty litter bucket spheres. It seems the problem with what we are doing has to do with the accuracy of the light meter. In side by side comparisons using an expensive light meter (i.e. costing more than $1000 and NIST certified) and a spectrometer, there has been as high as a 40% difference in lumens from LED's that have a bluer tint.

I also think an Integration Sphere would enable us to check the calibration of our light meters response to the various tints we see in LED's and other lamps. We all have witnessed the variation in light meter readings, and I am not entirely convinced it is solely due to the variations in LED's.

Harry brings up an important consideration. The 40 inch sphere would allow for testing all our current lights (I think) and although it would be less portable, it may be better suited for our use.

Tom

Silverfox, I would be very interested in any comments you might have on the lumen testing methodology I described in this thread since you seem to be better qualified than almost anyone here to give such a critique. I was surprised that my results for a Q-bin Luxeon actually were right where they should have been.

I own and maintain over $200K of specialized testing equipment, and am constantly questing the accuracy of my test results.

Click to expand...

Same here except I'm not even close to owning that much equipment. Then again comparing the temperature chamber I made to what's available commercially I might have $10K just for that alone.

If we as a community purchase a piece of test equipment like this, we should be able to offer services to the modders of our group to help them come up with better ideas.

Click to expand...

Agreed. I'll also add that there are only a very small number of members here who would have the background to properly run such tests, and an even smaller number who have the necessary time and space. I'd say your name is pretty close to the top of that list at least in the background department (don't know if you have the time, desire, or room for such a massive undertaking). I would also like whoever does this to test bare LEDs, both the small indicator type and power LEDs.

SilverFox said:

I have been involved in Non Destructive Testing for nearly 30 years.

Click to expand...

I've been involved in destructive testing for nearly 4 (four) years! :naughty:
(Technically most of the tests are non-destructive but things break!)

SilverFox said:

I constantly get asked if there is a cheaper way to get the information needed. My answer always is that it depends on what you are going to do with the information.

If you are simply curious about a generality, a simple inexpensive test on a limited number of samples will do just fine. If you have to make a decision based on the test results, you will probably need better information.

I have done tests that started out with 3 samples, and as more and more questions came up, ended up with testing over 100 samples.

Click to expand...

I normally only have to test one sample of a product. Although for testing toughened glass and such the more samples the better.
It is very true that you need to know what you want from the testing so you can do the testing that is required.
There is no point only doing a safety evaluation if you need to know whether a product is durable over the course of many years.

SilverFox said:

The key is having testing equipment that is accurate, and following established testing procedures. The testing procedures can be altered for special circumstances as long as there is a valid argument for doing so.

Click to expand...

Exactly! Most of the tests we perform are accredited by UKAS under ISO/IEC 17025 (2005) and we take calibration of our test equipment extremely seriously as well ensuring the test procedures are followed.

SilverFox said:

I own and maintain over $200K of specialized testing equipment, and am constantly questing the accuracy of my test results. It has been my philosophy that the post test discussions should be centered on the interpretation of the test results with comments on the procedures used, and not on the accuracy of the test equipment used to gather the data.
Tom

Click to expand...

My point with regards to CPF getting an IS is that test procedures must be in place and CPF Testing Standards created. Additionally, people using the sphere must be trained to perform the testing as agreed - otherwise the data is as meaningless for comparison as using various lightmeters in Heath Robinson contraptions...

Al

Hello JTR,

I believe your methodology is very sound. The problem is that you are basing your calculations on numbers obtained from a light meter that has a peak response at 550 nm. If your LED peaked at 600 nm, you could only be getting 60% of the response you would get with one that peaked at 550 nm.

I also have a question about some of your numbers, but will PM those to you.

If you substituted a spectrometer for your light meter, you may end up with an even closer approximation. I would be interesting to take the LED's you have tested and send them to a test lab for comparison...

Tom