Help with my D.I.Y. Sphere Calibration

I read that story last year and it was a really fun read, and indeed a pity that he did not measure at 30 seconds, not only because 30 seconds is according the ANSI method, but also because measuring after 5 seconds does not give an as repeatable indication of light output than measuring after 30 seconds.

Btw, last year I spent a really enjoyable weekend with folks from TLF and had the opportunity to re-measure 4 of those 68 actual flashlights in one of my integrating spheres that I brought with me. It appeared that my readings across those 4 lights (measured at 5 seconds to repeat the method) were almost consistently 9% higher. A TLF member present with his own lumen measuring device who had done his own calibration as well, also measured those 4 flashlights and found his values also 9% higher than what was measured earlier at the Zweibrüder sphere. So our calibrations agreed, but were higher than Zweibrüder.

djozz, did you see my post #13 above?

-Garry

Oh, sorry, yes I did, and then something else happened en then forgot to respond. (and now I need to sleep again, so a short answer for now):

I went along your lines with using a mildly 7135 driven power led for a constant light source, then for my sphere#3 (link via my sigline) I found that a ordinary white 5mm led driven at a constant 6mA by a simple circuit with a LM317 regulator is about the ideal constant light source for building into a sphere, but the low light level requires a luxmeter that can measure quite low levels.

Did the testing suggested and it appears my sphere has pretty good integration. However, I do see quite an affect of different size lights placed at the opening (and the Fenix headlamp I calibrated with fit the full width of my hole). I shot video of the tests and will try to upload them tomorrow.

-Garry

Videos of my tests posted to YouTube. During the 3rd video (testing my 3 inch PVC Light Tube) I ran out of battery in the camera. Basically my Thrunite Ti’s readings would vary greatly depending on where I placed the light, HOWEVER doing the 2nd test (testing for increased reflective surface area due to size of light placed at the opening) I saw extremely little affect even when placing my largest light (Courui) at the opening - which made sense to me since it seams the PVC Tube system is less likely to be reflecting light back toward the entry hole. I am however concerned how accurate measurements can be in the PVC since they vary so much as you move the light around. Perhaps me using only 3 inch PVC amplified that issue. Other’s see this issue in their PVC Light Tubes?

By the way, I had issues trying to do my test with dropping the light too far down into the entry hole and skewing the readings. (Was trying to watch the meter and not paying enough attention to where the light was positioned.) I tried to redo readings if I saw myself doing that, so be sure to watch the videos through to the end.

Videos:

Testing My Sphere for Proper Light Integration:

Testing Affect of Various Sized Lights Placed at the Opening:

Testing my PVC Light Tube:

Thoughts? Sphere seems to integrate light very well, wouldn’t you agree? I didn’t actually compute up how much change the size of the lighthead is causing. Seems to me if I calibrate with a medium sized light I should be within reason as I test smaller and larger lights. Problem is if I calibrate to a small light and then try to test a very large light. As I said in the previous post, the Fenix headlamp I calibrated to took up the full width of my opening, but it’s somewhat narrow leaving open space in the hole. Therefore I would consider it a medium to medium-large sized calibration light.

-Garry

So I ran a few lumen calcs on the Thrunite Ti by itself and then with another light in the opening (sphere, not PVC) and get the following increases:

With Home Depot 3AAA XB-D light: 2.57%

With 2D Mag LED: 8.2%

With Courui: 12.57%

I don’t know what affect the fact that I’m holding the Ti way off-center plays in though.

-Garry

The exact output of the reflectivity measuring light does not need to be measured, it just has to be very constant, and you can use your sphere to test that. Narrowing accuracy from 15% to 8% (I totally made up these numbers, a calculated wild guess you may call it, it higly depends on which other known and unknown variations are build in the measuring method) sounds reasonable for this measure, if a wide variety of flashlight sizes and textures are measured.

It looks like you use a standard chinese luxmeter, if I’m correct it measures down to 1 lux. Then the 6mA 5mm led lightsource for measuring reflectivity that I used in sphere#3 has not enough output to measure reflectivity in a significant number of digits. So for your sphere you better use a brighter light source. Yes, for a fairly constant output you could go for an S2+, on a 1x7135 chip driver. I think a simple AK-47 driver with two chips removed should do the job. Best is to use a high voltage led, like the newer batches XP-G2, so that the 7135 does not have burn off too much voltage and avoid temperature effects (these cheap 7135 clones suffer from temperature variation, but I can not find a source for original ADDtek 7135 chips). To avoid having to make a baffle inside the sphere between measuring light and detector, you can mount the flashlight not flush with the inner surface of the sphere but a bit recessed, with the sides of the hole made black (just like your measuring entrance hole). When inserting this light from the outside it needs a ‘hard stop’ (same for the detector, but you managed that by mounting it flat against the polystyrene so it can not go anywhere) because the measured light will be extremely sensitive to how deep in the hole the flashlight is inserted. Perhaps you can make a collar around the flashlight head for that that and hope that it does not deform the polystyrene when it is pushed against it. The position of this reflectivity measuring light I think would best be right next to the detector near the entrance hole.

I do not know a single BLF source for measured outputs. I know selfbuilt over at CPF incudes output numbers in his reviews (made with his corrected milk carton device :sunglasses: ), and so does UPz in his reviews, who btw made a very thorough write-down of how his sphere is made and tested (although in spanish).

I think it is great that you made these video’s, I will watch them tonight.

The PVC-pipe device does not suffer much from entrance hole effects indeed because not much light is not coming back, it does not integrate very well. In fact: how better the integrating properties of the device the bigger the entrance hole effects, so improving integration creates an extra problem to be solved :-/

Thanks djozz. I’m liking the idea to “keep it simple” and have a calibration factor calculated for a medium sized light which should be within reason on smaller or larger lights. I’m okay with readings staying within about 10% range. I’ll have to read up on your Sphere #3 soon.

I’m using the LX1330B which reads down to 0.1 lux (not saying it’s accurate that low though).

mhanlen has graciously agreed to swap some lights (I send some of mine to be tested and he sends me some of his that have been tested) with me so that I have measurements in common with another user to compare my calibration against. I know what will happen though, I’ll get these measurements and then will change my sphere thereby negating all the measurements!

BTW djozz, where did your signature go?

Thanks,
-Garry

No idea where the signature has gone, it is there in my account information. :frowning:

edit: I deleted my signature line and then created it again, and it seems to be back now :slight_smile:

Disappointing is a good word for it. It looks like they only measured the high modes, which are usually the most variable and the least accurate. Would be nice if they had measured low and medium modes too, or perhaps every mode on each light. It looks like a big missed opportunity, using the most-precise equipment only on the least-precise modes.

Just watched the video’s, nice work! I’d say the sphere is working convincingly well :slight_smile: , it integrates fine even with the fairly large entrance hole, and as expected it is somewhat sensitive to the physical type of flashlight that is measured. I agree that with a mid-size calibration light, the deviation of the multiplier is not that great of a worry.

When you say measured output what are the characteristics of a measures output. I include my measurements for all modes in all the lights I review. So you mean a real ansi measurement from a real integrating sphere?

No, I did not mean to suggest that they are the only ones with valuable lumen information. We are all amateurs with home-made contraptions. I mentioned selfbuilt and UPz because they have written down a thorough explanations of how their methods work and what their flaws are, and I like how Garry has done some characterisation of his sphere (sorry, I’m a bit OCD in testing stuff ;-)) , but there are quite a few others too: you provide your own measurements, Dale does, TomE, manxbuggy, pflexpro to name a few.

I actually plan on doing a guide on how I do it, for a future video. Thanks!

djozz was also answering my question about a “single source” - a thread or Google Spreadsheet or somewhere I could go to see where members have collectively posted their measurements. I wouldn’t even know where to go if say I wanted to see what other members measured for a 2D LED Maglite.

Wouldn’t this be a handy reference to put together? At least a thread, but then you might have to scroll through many posts to find various lights.

-Garry

I have a notebook where I collect my findings, and would be happy to share them if someone were to create it. I don’t have a lot of time to create the document at the time being though.

Yeah I don’t have a lot of time either, plus I’ll disappear from BLF for 6 months at a time. I am thinking to link to a document in my signature for my own though.

Still contemplating whether or not to mess with the calibration reference light. Part of me says it’s good enough and part of me says I can do better (though usually when I move ahead with trying to do better I screw something up).

-Garry

I call my lumen the ‘djozz-lumen’ because I’m not sure if my SWM D40A on high setting, which is my single calibration source, indeed exactly puts out the specified 550 lumen. But I do know that the D40A gives very repeatably the same output, at various drainage levels of the (Eneloop) batteries. So I take very good care of my D40A (I use it for nothing else than calibrations) and I stick to this as a reference, even though the true lumen may differ somewhat, because it keeps all my measurements repeatable and comparable.

I was thinking to call my calculated lumens “GBL’s”.

-Garry