Beam Profiles was created to give people some powerful tools: an accurate way of measuring lumens without a sphere and quantitative alternative to white wall shots.
Version 1 was designed for businesses that wanted high quality beam data. Version 2 reduced costs and simplified construction for reviewers. Version 3 focuses on bringing this tool into reach of anyone with a luxmeter.
What is needed?
A luxmeter ($30) and an angle finder ($15) and a tape measure ($10). Plus the bits to mount everything. No giant sphere/pipes. No sanding and painting. No calibration. You can have it assembled and accurately measuring lumens in minutes. The guide has links to all the parts I used.
How does it do this?
Through mathematical integration instead of physical integration. “Integrating spheres” work by smoothing out the light and taking 1 measurement. But it is also possible to take many measurements and mathematically integrate all of them into lumens. The app does all of this behind the scenes for you.
V1 and V2 used motorized platforms to automatically collect 1000s of data points. V3 brings human judgement into play to collect the minimum number of data points. You simply add data points until what you see looks the same as the actual flashlight beam.
Limitations
Spheres are instant. This takes a few minutes. So it can only measure modes with flat regulation.
It can only measure clean radially symmetric beams. Weird beams (like bike lights with cutoffs) need multiple-axis scanning and that is just too cumbersome to do by hand.
Future
Those were some of the results of V1. Similar charts might come to V3. (Right now you export the data to Excel and make your own charts.)
Possibly a database of beams. This mostly depends on how many different lights are contributed with the share results button….
Interested in giving this a try?
Take a look at the guide and I’ll answer any questions you have.
Forgot to mention that you can use a smartphone with a bubble level app as a digital angle finder too. I’m not the biggest fan of that though. At steep angles they tend to start giving wrong results.
Someone had jokingly suggested that I dropship the commercial equivalents this project from Aliexpress. After a little digging we found a shop with them. Priced between $7500 to $98000. Youch.
Very interesting!
I have done 100’s of measurements, and even with an integrating (self made) sphere, it’s a P.I.T.A.
Results will vary due to many factors, insertion depth, light loss due to fitment into the opening, heat dissipation of the light, variability of the luxmeter itself, etc, etc.
This looks like a more easy method, especially when you say : V3 brings human judgement into play to collect the minimum number of data points. You simply add data points until what you see looks the same as the actual flashlight beam
That's a really cool project. I sent one of my Maukka lights for recalibration a few months ago, and that company measured it with a Goniophotometer. And this reminds me of it.
Goniophotometers take 30+ minutes to take all measurements (usually 1+ hour), and therefore could be a bit more work to measure flashlights, because of the drop in output.
Your project is really cool. Do you have any comments on how to get about the output drops?
Thanks! 1st is to do the measurement in minutes instead of an hour. 2nd is to only measure modes where the output doesn’t drop.
Its pretty trivial to measure a dim mode and then scale the result up. The lumens will scale with the candela after all. And nearly any light worth measuring will have a mode that is fairly flat. The 350mA of a single AMC7135 would be about 150 lumens and that is plenty to for the Beam Profiler to get a reading from. Even with a meter that only has 1 lux resolution.
“But what about lights that have no flat modes?” This project simply can’t measure all lights. It makes some heavy tradeoffs that are much worse than requiring flat output.
Those professional goniometers are fully automated and multiple axis. They are also designed to be universal and work for any illumination pattern. So they need to take a 2D grid of readings and that takes time. Beam Profiler V3 uses a 1D line of readings and makes some major assumptions about symmetry. A professional goniometer running like this could be finished in a few seconds.
(V1 and some versions of V2 were basically a standard goniometer but people were scared off from building or buying a good 2-axis platform.)
I do plan to add options for more types of beams with different types of symmetry. Next would probably be lanterns that have a 360° “donut shaped” pattern.
- – - – -
I had an idea for a way to improve the interpolation algorithm. I think it worked much better than I hoped. Now it produces good results with half the data points. Even removing the 0 degree hotspot reading only nudges the final lumen number by 1%. (Not that I would recommend purposely skipping that reading!)
It turns out this is also good for testing non-flashlights. I was looking for an LED floodlight and there are a lot of cheap dodgy things out there. Eventually I went with the Sansi 70W floodlight because it seemed to be an original design from a decent brand instead of a generic clone rebadge. (I also got it for $25 instead of full price.)
But does it actually meet its specs? Plugging it into the Kill-a-watt had me worried. They claim “70 watts” but I measure just 45 watts which suggests that the “7000 lumens” might also be overstated by 40% too.
However I do measure 70 volt amps. The power factor on this light is not good! But that does explain where the “70” number might have come from.
A light this big would be impossible to measure on a DIY sphere or tube. Its just too big to fit. 18cm across the diagonal of the LED array means you’d want a sphere at least a meter across.
But that is no problem for the Beam Profiler.
A few minutes and 7 readings later I have an answer: 5700 lumens. So it is overstated by 23%. That is on par with typical LED lumen claims.
It is probably a safe assumption that all these floodlights will be using LED lumens and should be derated somewhat.
Nice charts! Originally I imagined horizontal (left-right) movement to measure the beam but now I’ve read the guide, I see it’s measuring in the vertical (moving light up-down).
I can really see this as a benefit for comparing ‘before/after’ LED swaps.
Might have to get myself some of this gear this year and have a go myself, looks easier (and smaller to store) than a lump of PVC piping or polystyrene box!
Very Helpful.
I made something similar. I used a 5 foot tube with baffles and flat black paper inside. Lux meter at the end.
This I pointed at the light axis. Cuts down on room and reflective light. And I could make measurements in moderate room light as opposed to waiting till dark.
I used a protractor and rotated the light horizontally to get the beam profile. Using the emitter as the rotation point.
I was working on trying to automate the thing with a photo diode - data logger - motorized table - when health reasons brought all that to a stop.
A graph type of info allows comparison of different beam profiles. Also good for showing up dark donuts in the beam.
On some lights it necessary to get far enough away to let the reflector/TIR get things to “come together” as it were.
Great info,
All the Best,
Jeff
Tonight I tested a bunch of AAA keychain lights. Some of them were far lower than official specs but that is probably because they were tested with 1.5V alkalines and I was using 1.2V NiMH.
