Budget Lumen Measuring Rig

How would I calculate lumens from lux with this setup?

This is a 68 quart Coleman cooler that is stood up on its' short side and the lid is closed when collecting data. Data collector is a smartphone using the Androsensor app.

I've looked up lux to lumen calculators and formulas but I'm still not sure how to convert to lumens with this.

Thank you for your assistance!

I wrote an app for that… but you’ll still need a source for calibration. The readme talks more about calibration and reducing error with makeshift integrating boxes.

Zak, Thank you for developing the app! I'll look into it. I think I'll use an S2+ or my D4 v.1 at the highest regulated mode for calibration.

This is my budget solution :
5$ foam sphere
15$ luxmetre

Calibrate with some known value (at med/low levels) to find your multiplier and you’re good

1 Thank

Some smartphone sensors are highly directional (and all of them are to some degree…) so you need some sort of diffusor directly on top of the sensor.

Look up the brand and type of the ambient light sensor in your phone (which is what the app uses as luxmeter). It can be looked up on your phone, there is a list somewhere with phone hardware parameters. They differ hugely in their spectrum response, a bad one can easily cost you 30% in accuracy compared to a good luxmeter, but you may be lucky, some are really good (Texas instruments makes some very good ones) and are within 5%.

Yes! Much better setup! My wife is giving me funny looks with that cooler setup! Lol

X, K & D, Thank you for you recommendations and knowledge share. I spend a little money on nice lights but am too frugal to buy a light meter! I use what's on hand and sometimes, it's the biggest thrill to use that 'thing' I've had for years! It figures well with being on the BLF. And quality is held in high regard!

I looked up the approximate value for a D4 v.1 w/Luminus SST20s, at full regulation and at turbo and if I divide the Alcatel Idol 4 sensors' lux reading by 3, ie. 465/3, I get 155 lumens. Not bad. With a 3.9 V battery, I get right under 3000 lumens on turbo with the same calculation. I'll test a few more lights that I have and have been tested here on BLF. Pretty cool! Thanks!

:+1: that it works as planned!

Could you look up the brand and type of the ambient light sensor in your phone, for my curiosity?

is it linear ?, double the lux means double lumen ?

It is lineair, but the lineairity is not device-related, for any set-up good or bad, if nothing else is changed but the light output, your luxmeter will measure twice the value if the output is increased twice, if it does not do that you just have a very bad luxmeter.

I see, does the reflector affect the reading?

I did some searches and found various assemblies/modules but nothing about the ambient light sensor maker & specs. I'll need to attempt various related queries.

Nope, it’s the purpose of the integrating sphere.
Of course a really floody (mule) light could have an effect (inflated numbers are to be expected).

I'm back with some results! I tested a few of my project lights and a number of stock flashlights. These are peak output tests with some lower mode measurements. Or, in the case of the Wurkkos multi-emitter, a test for each LED. There are multiple tests for some of the same lights with the difference being the battery.

Here's a link to download the file: https://1drv.ms/u/s!AqeZmZFhBiGK4g_YxrNiKMhMhqG1?e=CKzpj9

Just select the file and download it. OpenOffice and Excel should open it.

The divisor of the lux measurements is between 2.5 and 3.0. to come up with lumens. I guess I'm still calibrating...

It was surprising to find that my Emisar D4v.1 XP-L Hi didn't perform as well as my D4v.1.x SST20 4000K running Anduril. This was bought just before the v.2 was released.

djozz, I bought a Sunche light meter from TA and have found there is approximately a 1% correlation between the Idol 4 sensor and it. The Sunche is ~1% less in readings.

I'll run them side by side to further verify this correlation. If I can find repeatability, I think I should arrive at a decent calibration.

I would recommend a much more detailed explanation on each Excel tab, explaining what you are showing. I couldn't understand it. Why do some of the graphs look like the light turned on/off repeatedly?

"time since start in ms" ? milli-seconds? minutes-seconds ?

1% difference over all the flashlights that are measured is very good, but maybe your flashlights all have comparable tint and CRI, then different luxmeters, bad or good, will correlate pretty well. The question is will the two light meters still be within 1% of each other if also warm light is measured, or other less normal tints?
But if these light sources are the type you will measure anyway, you are completely fine using the phone light meter of course.

The tabs are named based on the stock light and battery model. The models are typically known to belong to a particular manufacturer. In the cases where I modified a box store light, I put the LED also. And in the case of the D4s, I have different LEDs in the three I have.

Some of the graphs are displaying where I kicked the light back to turbo and some are runs from different times. You can see it in the data columns. The units are milliseconds. I just haven't done the conversion to minutes. The numbers on the X axis are the record number although I'm trying to figure out how to display time, instead, with the OpenOffice software.

I did this primarily to see the max OTF lumens, especially for my project lights. I recall a comment from ToyKeeper that tail current really isn't the best way to determine the actual output so it was bugging me to find something better with what I have on hand. Essentially, I miniaturized the bathroom ceiling bounce technique. Lux will drop when I open the lid so I could remove those readings from the graphs. I also tinkered with the idea to buy a DC capable clamp DMM but for me, the one I have is acceptable.

I agree the .ods file can be a little more polished. I appreciate you looking at it and providing the critique.