Trying to understand light measurements

[CurrDogg420]

@AxisCat, I keep coming back to this problem. I hope you don’t mind if I play the role of foil here, I apologize if I am just slow and you have already considered this in your calculations. I might also be stoned, but it was a Sativa.

I think there is a flaw with using LUX brightness as any sort of benchmark for your calculations. LUX is a measure of temperature brightness, not quantity of photons directly. Really color temperatures with respect to quantity of photons would represent a wave function integral across the color spectrum.

For your problem you’re looking specifically for the quantity of say blue photons which cause a your Lux meter to have a certain reading. So you’re fitting a point measurement to a wave function which varies. This is where the neural network may be required in order to “reverse engineer” the output of that particular meter.

From Wikipedia:

• There is no single conversion factor between lux and W/m2; there is a different conversion factor for every wavelength, and it is not possible to make a conversion unless one knows the spectral composition of the light.*

So you have this light sensor, which can give a “power” reading. This was all my nerdy way of asking, wouldn’t it be better to use a PAR meter to derive your spreadsheet. Or at least use it as a double check.

[Storm]

I cant remember last time i used my meters…lol.

[dbrn32]

I would say per your context ppf is absolute and ppfd is not. Lumens=ppf and lux=ppfd. Just difference in photometric energy and radiometric energy. Truthfully even the ppf of a fixture will degrade a small amount over enough time. But this is something that currently all fixtures of all lighting tech share to some extent. Lux/ppfd will change with simple movements of light, unless you are expressly limited ppfd average over a given area.

Everyone here seems to gravitate or using ppfd, but is relatively useless if not ppfd average.

[AxisCat]

No worries about questioning my methods, I could still be wrong with my thinking. The next few weeks should sort that out for me.

There is no single conversion factor between lux and W/m2; there is a different conversion factor for every wavelength, and it is not possible to make a conversion unless one knows the spectral composition of the light.

So the key thing is if you have to know the spectral composition otherwise known as the spectral power distribution (SPD) in order to make the calculations. And they are made on a per wavelength basis which is what my spreadsheet does.

This all started for me when I ran across this brilliant article where the author details the math behind making this conversion. Do a search of:

Photometry and Photosynthesis by Ian Ashdown

His equations makes this conversion for a known spectral power distribution. It occurred to me that I could maybe use an inexpensive multichannel color sensor to measure the SPD of any light source. With only 6 channels it will be somewhat limited in resolution but it could do a fair job. The 11 channel sensor I am getting this week should do even better.

This is one area where my limited math skills breaks down. Right now I am just interpolating between the 6 wavelengths it reads to give a full spectrum readout at 1 wavelength intervals across the 400-700nm range. I am researching Gaussian functions which the channel outputs of the sensor resemble, but my poor old brain can only handle so much though.

My thinking at this point is to calibrate the sensor using light that has a nice smooth curve through the PAR spectrum. Which is why I started with tungsten bulbs. It has a nice, well known SPD curve.

I am not a scholar so look over me if I confuse some terminology, it is tricky stuff. I don’t pretend to know how the sensors on Lux meters or even Par meters work. My thoughts are a single sensor under the white dome outputs a current in proportion to the total intensity of light that strikes the sensor. I don’t know about a Par meter but a Lux meter filters the light in such a way that it mimics peoples perception of light intensity using the Luminous efficiency function.

This is from the data sheet of my Lux meter:

It gets a little weird below 500 but otherwise looks like a good response. My particular meter, about a $100 has an accuracy to within 5%. Don’t quote me but I think that is close to the accuracy of some of the Apogee Par meters. Point being the best I can hope for is 5% and will be happy with 10% error.

So getting back to calibrating using a Lux meter and Luminous efficiency function, here is my sensor measuring incandescent light with the Luminous efficiency function (red dashed line) on the same graph:

So my thinking is if I am going to use a Lux meter to calibrate my sensor, then firstly I have to calibrate it to measure measure Lux. Once it is power calibrated it should be a matter of not running the output through the Luminous efficiency function. Then it would be measuring the total number of photons under any SPD function.

I have to admit, I just barely understand all of this and the more I work with it the more it makes sense. But it is too early for me to say yes, this works. Tomorrow I could notice something that throws a curve ball at me and I have to scrap this whole method.

Which is fine if that happens. The journey has been fun!

[AxisCat]

yes sir, you are correct. I miss typed PPFD when I meant to type PPF… thanks for the correction.

[AxisCat]

I completely agree about Par meters reading a single point at the canopy and assume it is evenly distributed throughout that one meter squared.

Mine is a hobby grow, three plants in an area of about 3 square feet. So very small. It is pretty easy to move my sensors around above the canopy to get a better idea of the average PPFD across the entire area.

I have a few thoughts on how to better handle this particular problem, just not high on my priority list yet.

[Syndrix]

I’m just curious to see where this goes, and how accurate that 11 channel light sensor is you have coming.

Really interesting research you have done so far.

[beardless]

In your research I imagine you have come across Dr Bruce Bugbee. Just in case you haven’t, this a basics primer.

Cannabis Grow Lighting Myths and FAQs with Dr. Bruce Bugbee

He has lots of videos. All worth watching

[AxisCat]

Thanks… I have watched a couple of his videos… That fellow is a genius. My profession is in the construction contracting trades. Not extremely technical and I barely use anything more advanced than say 5th grade math. My handful of semesters at university was about 30 years ago. And I failed calculus. Just never could ‘see’ it in my head to really understand it.

[beardless]

@AxisCat

Not extremely technical

I will join you on that. I am also with you on wanting to have a better understanding of the subject without getting too far into the technical side.
In making my recent light purchases I mainly considered: spectrum, PPFD and in turn DLI, balanced against cost and efficacy.

[AxisCat]

Hi all
@beardless , @CurrDogg420 , @Fiz , @dbrn32 , @StonedCold13 , @spankyjr1

I thought I would give an update on my progress. I am setting up to do some data acquisition using an array of sensors.

My lux meter doesn’t have bluetooth so I had build a system to capture the readout with a webcam and translate the segments of each digit to numerical values. That proved to be harder than I thought:

The next step is capturing the data from all of my sensors so I can work some math on them:

This should be interesting.

Axis

[beardless]

Good luck. I could not even find the door let alone open it to enter the realm. Certainly not something in my tool chest.

[dbrn32]

Cool! Are you capable of laying out plotting axis 400-700 and placing bands properly? That should have your graph show up roughly the same as typical spectral distribution graph?

[AxisCat]

That’s on my list after I acquire all the data points… Probably be a few days till I attempt my first measurements… I will let you know

[StonedCold13]

Very nice work, I think you made my brain shift a touch

[AxisCat]

Here is the spectral power distribution of my LED lights. The Y-axis are arbitrary units but it shows the shape which matches the manufacturer (if I can trust their marketing)

I am still basing all my calculations on a measured LUX value as my reference. Hopefully I will be spending a day soon acquiring data so I can nail down my multipliers, I don’t know how accurate the one in this graph is but you get the idea.

[AxisCat]

Another update… the diffuser I put in front of the new sensor (8 channels between 400-700) didn’t have a large impact on the way this new sensor read out power levels.

the one on the left has one of those sphere shaped domes on top, the right is the raw sensor exposed to an incandescent bulb. Still chewing on that one a bit.