Thanks for the test.
If you already have all the data in Excel, you could multiply current and voltage to calculate power, and then add lumens divided by power as an extra graph.
Maybe seeing the point where efficiency drops sharply will help decide on a sensible drive current.
You probably meant efficacy (lm/W), not efficiency (radiometric W out divided by electrical W in). Such a point does not exist because output (lm) is a very smooth function of input power, youâll never get a non-smooth dropoff before the point of catastrophic failure. In fact, the efficacy drops fastest (in the sense of having most negative derivative) at the extreme low end of drive power.
A related useful rule of thumb is that 50% of the max current gets you 70% of the max output, and 80% of the max current gets you 90% of the max output. I generally stick with 50% for most lights, and 80% for squeezing the most performance out of large throwers.
You are right. sorry iâm not a native English speaker as Iâm sure you can tell. so sorry for messing up the semantics but I see how this might be relevant in this context.
You are referring specifically to Luminous efficacy - Wikipedia
if I get you right.
is a measure of how efficiently a light source produces visible light. It is the ratio of luminous flux to power, measured in lumens per watt in the
The per Watt refers to watts in optically output radiated âpowerâ I assumed and not electrically, but
Turns out this assumption was wrong. Confusingly (for me) Wikipedia calls âLuminous efficacyâ âa measure of how efficientlyâ
but what I thought of as Luminous efficacy actually would be called âLuminous efficacy of radiationâ
It is obtained by dividing the luminous flux by the radiant flux.[4
so a more green light has more Luminous efficacy ratio than a blue or red one, and lower CRI generally more than higher CRI.
(correct me if Iâm wrong I just want to make sure I get it right)
To measure this OP would need some sort of spectrometer or compare PAR or LUX readings or in some way compare a radiometric optical power to photometric lambda V corrected âluminousâ
This would be interesting to see a curve of this measurement for some white LEDs as it would indirectly show the âperformanceâ or efficiency? of the phosphorus layer in converting blue photons to green ones with higher luminous efficacy (ratio).
This curve might be relatively smooth but certainly it is not linear.
Anyways this metric was not what I meant, sorry for the confusion.
I was talking about how efficiently the led is in the sense as a ratio of electrical input power to luminous flux. As you rightly pointed out it would be called Luminous efficacy
But since we are already deep down in the rabbit hole in âwell actuallyâ theory I doubt OP actually measures luminous flux.
and assume he measures illuminance and calculated the total luminous flux.
And since I donât know what instrument was used to measure illuminance itâs possible a cheap âlux meterâ was used that actually doesnât have a V lambda filterâŚ
So this might actually be an irradiance measurement.
But to come back to topic:
What I previously referred to is that I would want to see an âoutputâ / electrical input power graph in lm/W(electrical) and this is, as you correctly pointed out, luminous efficacy.
but in the same way I would appreciate an illuminance or irradiance per watt electrical graph since many hobby measurements use them kind of interchangeably with just a fixed correction factor.
But honestly thank you for pointing out my mistake. this is how I learn and hopefully can be more precisely articulate in the future to avoid misunderstandings.
1 Thank
The measuring device I used for this test is my TKlamp integrating sphere.
Its a budget device. Not laboratory calibrated, but they do include a calibration light that was tested in a laboratory.
I 3D printed some holders that fit Maglite so that can get consistent (though perhaps not perfectly accurate) results.
1 Thank
Cool, thanks.
So you are one of the few users that actually has a device that does the full lumen integration measurement in one, and you donât have to calculate it yourself.
This is very convenient, and Iâm kind of envious of you for both having the device, and especially the calibration lamp.
With DIY measurements, you are never quite sure about your results if you donât have a trusted, laboratory-calibrated lamp to compare your results to.
would love to ponder this orb⌠i mean⌠sphere for some time 
They will sell you a calibration lamp separately for like $75. I used it to roughly calibrate other devices like my 6" lumen tube so I can measure lights that are too large in diameter for the sphere.
Thank you for clarifyingâyour understanding is fully accurate. I appreciate your caring about correct terminology usage. Most native English speakers (including writers of popular science articles) get it wrong, and some of them refuse to acknowledge the mistake even when it is pointed out.
Since the spectrum is largely the same for the same LED at different drive currents, it is fair to assume that measuring luminous efficacy (lm/W) lets you infer the efficiencyâthey differ only by a multiplicative constant known as spectral efficacy, which is lumens per watt of light having a given spectrum.
I see. I donât know of a publicly available graph with this exact pair of axes, but: many LED tests by koef3 feature a plot of output (lm) against input power (W), and pretty much all LEDs have the same curve shape, up to scaling.
Hereâs an example test; I canât give you the exact plot because none of the images are loading for me at the moment, but just scroll down to find such a plot. You can then get an efficacy-vs-power plot manually via dividing y-axis by x-axis. As you intuited, the plot is not a simple linear decay; the decay is fastest close to 0W, and slows down as drive power increases.
1 Thank
Hereâs a plot of output against power: https://lychee.lichtundstrom-blog.de/uploads/original/9d/73/157a2a0b0a15ca553afba2c55317.jpg
2 Thanks