Texas_Ace integrating PVC sphere with no math involved

So people have been requesting to see this sphere even though I warned them it was really nothing that amazing when you broke it down but here it is anyways.

To start out with it needs to be noted that I got the idea for the centering rings from JoshK’s spheres, they are truly a genuine idea and vastly improves the spheres performance.

Cliff notes version:

- Tired of screwing with a “normal PVC sphere”, so Dale sent me his JoshK sphere

- I learned a ton from messing with the Josh sphere and then decided to apply that to the PVC sphere in order to measure larger lights

- I added an “integrating disc” that I made out of styrofoam that diffuses the beam and ensures stable lux readings, other options exist as well.

- I 3D printed some flashlight “centering rings” and coated them in reflective tap to make the opening always look like a giant reflector and thus vastly lower any effects of head size on the readings.

- I tweaked the thickness of the integrating disc and added a small amount of electrical tape in order to get the lux meter to read lumens directly on the meter with no multiplier needed.

- I enjoy being able to take a quick reading without having to do math and know that it is consistent.

  • Pictures at the end show the stability of the readings even with a silly angle on the light, in real world usage you have to try to get over a ~2% variation.

So you have all seen the many threads on building a PVC pipe based sphere, many of you have done this already. I had done it as well but quickly abandoned it as it was just too unreliable for me, slight changes in the way the flashlight was put in the sphere would drastically change the readings. Plus having to use a calculator to get the lumen readings was a step I didn’t feel was needed.

I gave up on it for awhile and then Dale was nice enough to send me his JoshK sphere which I then set about improving to allow it to handle the kind of output my lights put out. During this I figured out some simple things that I could apply to the PVC sphere and now I use the PVC almost exclusively, simply because it fits on my desk and allows for much larger flashlights to be tested.

So, now to the sphere itself, Some of these steps are not totally in order as there were a lot of trial and error steps not mentioned here in between.

The biggest issue with normal PVC spheres is that it doesn’t integrate the light, it simply reflects it until it reaches the lux meter and because of this the readings vary wildly due to the beam pattern and position.

I also wanted the lumens to read out on my lux meter directly instead of having to use a calculator, why make things harder then they need to be?

I started out with a basic 3” “P-Trap” setup, I used a 3” to 1.5” reducer on the meter end and with a little dremmel work the HS1010A sensor pops right into the hole and mounts snugly. The HS1010A works great for me BTW, I have 2 of them and they both read within a few lux of each other.

So after the tests with the Josh sphere I knew that an NR lens for a camera works well at reducing the lux reading without effecting the accuracy. Although finding one large enough and with the right transparency proved hard. So I reverted back to an earlier idea I had which was to use a styrofoam disc.

It reduced the lux reading nicely AND functioned as a great integrating layer. The light passing through it was perfectly smooth and “soft” thus there was no beam that would reflect around the pipe and cause odd readings.

I cut it to size and then started sanding it down slowly checking as I went. My goal was to sand it down until the lux meter read the lumens out and thus didn’t need a multiplier. I got it just right and then fully pressed it into the pipe and blew out all the dust, turns out the dust had been effecting the readings and it was now a bit too high.

Not to worry, I simply added a small amount of black electrical tape to reduce the reading back down and bam, the lux meter now reads lumens!

So the sphere now read within ~3% no matter where I pointed the light around the entrance point but I wanted it better and I also didn’t want to be blinded while testing lights.

So I printed up some “centering rings” for the end of the PVC. I put reflective tap on the inside so that to the sphere the entrance hole always looks like a giant reflector no matter what size the flashlight itself is. It also centers the light and gives me a consistent depth to insert it.

And that is basically it. I made a normal PVC sphere, added an “integrating disc” and a “reflective centering ring” and now instead of 40+ change in readings based on flashlight position it is around 3. How far I insert the light past the ring easily has a larger effect then that.

If I was to do it again I think that a 3D printer integrating disc could be used if it was printed thin enough. Or one of a bunch of other options as well. Besides that I would not change anything at this point, I love it!

Now for the pictures!


Here you can see the integrating disc and the reflective tape on the centering ring

The dremmel work to fit the light sensor:

Centering ring:

Electrical tape to adjust the meter reading, it takes a surpringly small amount to adjust the reading:

Example of the lack of effect that angle has, a lot of those losses are actually from having to stick the light further into the sphere to keep the beam “inside” the opening. In pratical terms the most variation you will get without an extreme like this is sub ~2% vs the 4% seen here. Just showing the worst case as an example.


Thanks for the write-up, and nice results!

Before making true integrating spheres, I have been thinking before of light measuring devices, and in my thought process I discarded diffusers in the light path because of the fear of changing the light spectrum by that (some wavelengths getting through better than others), but I’m glad that you are more practical than me and just did it, and it works fine. Spectrum effects probably play a too minor role to be measurable anyway.

Nice!
What was used to calibrate the device?
Did you check at multiple brightnesses to make sure it was 100% linear?

I thought about this as well but I dismissed it since we are not building a mathematically calibrated sphere where things like that would matter but instead a standard calibrated sphere.

Thus all the little things will be factored into the standard calibration.

I don’t think any of us truly believes that his $30 DIY PVC “sphere” is reading ANSI certifiable Lumens here. I know mine doesn’t. But it is consistent which makes the numbers comparable which is what matters.

Via trial and error I also think that the numbers themselves are quite reasonable for the things I have tested it on as well.

I calibrated it based on the Josh sphere to start with and then fine tuned it using the Cree data sheets for the most part, along with the few lights I have that actually have ansi ratings I half trust.

It lines up with the cree datasheets great now, well within the margin of error in most cases.

As far as being linear, this was hard to test but I finally tested it be measuring several lights individually and then I made up a disc to measure them all together. The reading of them all matched the individual readings added up so I am happy. Since there is no light lost even at 10,000 lumens this makes sense.

It actually takes the full 10,000 lumens to even start to see a faint glow through the PVC.

How do you judge how far the flashlight should go into the opening?

Just far enough for the light to not spill out over the top or whatever gives the highest reading. Even a small amount further into the opening can change the reading by quite a bit on a black flashlight due to it absorbing the light photons.

Great !

Now you can calibate it with a calibrated spectrometer or/and with pretested emitters and flashlights, its no matter that it costs 10$ or 50, as long as its calibrated it will work and serve your needs

Thanks a lot TA. :+1:

Very nice little device! It looks like you have done enough testing to ensure that it is a very useful device for relative comparisons and a pretty strong connection to and direct linear relations ship with official measurements.

Do you plan on assembling any for sale?

I am quite happy with the consistency, my fenix “test light” measures within 2-3 lumens every time. The overall calibration is as good as I can get it with the lights I own but I am quite happy with the results that I have seen. They line up with the Cree datasheets which is a good indication you are doing things right plus all the lights I have that do have ansi ratings are generally within reason as well. Plus it seems to line up pretty well with numbers posted around here.

I don’t plan to build them as the bulk of the cost is in the PVC and that is even more costly to ship.

I suppose I could sell the centering rings for others to build there own and I could share the CAD files for anyone that wants to print them out themselves. Although someone could also make them out of poster board.

The real issue is calibration, we should have a “standard light” that could be used to calibrate spheres but sadly nothing like that exists.

Brilliant. Hard to believe you only have 10 posts to this thread. Must be the holiday business because this thread is gold.

279 reads.

Incredible! Looks so simple to the naked eye, but its a very well thought out contrapment for those that know what really went on into creating this lumen tester.
How sturdy is it? If the area around the horse shoe was longer, if would be a very stable base, but I suppose you wanted it as small as possible, and if its sturdy enough this small, then even better.
I guess your not really pushing down on the pvc, so it shouldnt tilt forward.
So how big is the opening? Can the head to my btu shocker fit? my biggest headed lights are my trustfire j20, shocker, and the tn42.
I’m not quite sure in understanding the black electrical tape….so is the tape still left inside the pvc?
And lastly, so how does the readings work, just put the hs1010a on auto and whatever number you see, is the lumens? So in your photo, it shows 123, does that mean 123 lumens?

One ‘beef’: This is more a “Light Pipe” than a sphere :wink: I did a lot of reading-up about these creatures some time back and I’m a stickler for nomenclature accuracy :blush: I like both the electrical tape ‘tuning’ and the ‘baffle’; neither is conventional but both are brilliantly simple solutions. The only potential failure point I can see is the electrical tape which could lift, shift, or curl in time but that is easy to check.

We tend to over-think things sometimes and that can mislead by inducing ‘tunnel-vision’ which often leads to dead-end paths we get stuck on. The only thing here which really matters is the results- if they are accurate and consistent who can complain (and can the complainer do better themselves?). I’d like to see a few more details such as the highest and lowest lumens and the largest and smallest head diameters you’ve tested- that would complete this ‘review’ very well.

It’s a given that our home-made measuring devices will never be lab-grade in every aspect but that isn’t really necessary for our uses. To get within a few percent is totally adequate and a major achievement. I always say that if something is doing what it’s supposed to do there can’t be much wrong with it; and it seems you’ve got a winner here :+1:

Phil

Yeah, the idea and execution is quite simple indeed. It was the months of testing to figure out what worked and what didn’t that was hard. I overthought it a lot at first but turns out that the simplest option is the best one as well.

It is quite sturdy, PVC pipe is pretty tuff stuff. If you meant stable then it is actually surprisingly stable as well. I have never had it even attempt to tip forward. I lay my heatsink/LED test rig on top of the inlet port and it holds up just fine.

I used 3” PVC for this sphere because at the time I was too cheap to spend the extra $8 for 4” and the extra space would make it more of a nuisance on the desk. You could easily build this with 4” or even 5” PVC to fit any light you could want though.

The black electrical tape is my “multiple fine tuning adjustment”. Basically by sizing the tape just right and leaving it in the sphere it absorbs a fixed amount of light photos before they reach the light meter.

This in turn lowers the reading on the meter and allows me to have the meter output the lumens directly on the readout and not need an external multiplier. It makes taking readings much more pleasant since what you see is what you get.

Yes, the meter is on auto and in this case it was indeed reading 123 lumens. When the lumens go over 2000 it adds in a 10x multiplier on the meter but that is easy enough to figure out, if it has the 10x sign in the corner then you know that it would for example be 1230 lumens.

The name could be argued, IMHO sphere is anything that integrates the light to the point of beam profile/shape/angle not drastically effecting the readings. A light pipe classically does not do this so I feel it is closer to a sphere then a light pipe. Although the largest reason for calling it a sphere is that is what everyone else calls these devices and thus what they would search for.

Yeah, the first several prototypes of this setup were very overthought, didn’t work that well and were hard to use. When I finally said to heck with it and decided I didn’t care I went simple and to my surprised worked way better that way.

The real eye opening moment for me was when I realized the difference between a mathematically calibrated sphere and a standard calibrated sphere. DIY spheres are always standard calibrated. In this case there are only 2 things that matter, consistency and linearity. This checks both boxes so it is then just a matter of calibrating it to a good standard.

On a mathematically calibrated sphere (such as an ANSI certified sphere) things get way harder and that why they cost $10,000.

The electrical tape for tuning was another eye opening moment. I tested extensively to see if it would effect the constancy but it turned out that if the light was truly “integrated” it had no effect other then reducing the meter reading. As far as it peeling or currling, I have not had an issue yet but I also keep that fenix light nearby and check the readings from time to time. It always comes out within 2-3 lumens.

The largest issue I have is actually dust. I left the centering rings off of it at first when not is use and after awhile the dust started to effect the readings slightly. I blew it out and it was back to normal. So now I leave the smallest hole centering ring on it at all times to keep the dust out.

As far as the lumen measurement range, the min lumens it will detect is around ~5 or so, generally doesn’t detect moon modes and under 5 lumens I am not convinced of the accuracy.

The highest it has seen so far is a bit over 10k lumens from an 8X SRK.

For head sizes, it depends on the size of the PVC you use. 3” that I used Is cheap but the price goes up quickly as you go bigger so I decided to pass on that. You can get PVC upward of 6” though so max head size is not really an issue, just a matter of cost and how much space you want to take up.

I personally have measured from a BLF-348 up to a D01 myself, those are the smallest and largest lights I own at the moment, no issues with either at all.

I totally agree on being “good enough”, consistency is way more important then accuracy, as long as it is consistent then it is comparable.

Thanks for your posts, it helped me learn a lot. I especially like the way you modified the device so you can read values directly off the meter without a need for conversion factors.

Inspired by your work, I did a quick test, using a vitamin jar cap as a diffuser. I had to add 1 piece of white envelope paper, and one piece of kleenex, under the cap, to “calibrate” it.

I used my new Integrating Cap to test two Lumintop lights.
On the left a ReyLight Tool w Nichia, ReyLight modes are mfg spec to be 1-16-80
On the right a Lumintop Tool w Nichia head, on a Maratac body, Tool modes are mfg spec to be 18-3-80

you can see in the beamshot that the ReyLight obviously has a lower Low

it just helps to be able to quantify the differences, here is the ReyLight Low calibrated to read 1 lumen

and this is how that calibration reads the Tool head on Low

here are the beams on High side by side, and their readings (these are both N219b lights, you can see they use different batches of CCT)

ReyLight High

Tool head High

I then tested an L11c w Nichia, claimed high is 140 lumens… surprisingly it is only about 5 lumens brighter than the Tool on High.

Very interesting idea, I never considered going that small. That is thinking outside the box, good going!

I also tried many items like plastic and such for the integrating disc but nothing was just right. I didn’t think about 3D printing it until I already had the stryfoam done.

You also did remind me of one of the big reasons that I did this in the first place and realized it would work fine. The light sensors we use already have a diffuser/integrating dome over the sensor that it is calibrated to. So why would another one effect anything any differently?

A larger “integrating space” does help even out variations from beam pattern which is real important once you start dealing with different sized lights with vastly different beam patterns.

That is a tricky set-up, jon! To have a fixed calibration for all kinds of flashlight sizes and types, your set-up depends on a luxmeter that correctly reads light coming from any direction, hitting the detector at any postion (and as I understand, that is not an easy requirement to match for a luxmeter).

But the proof of the pudding is again in the eating, if in a series of tests, the calibration of this set-up proves correct enough in various types of flashlights, all is good. :slight_smile:

Edit: TA had a parallel answer to mine .