Why are they any better? Who has an Ulbricht sphere at home? Who on earth uses a PPFD meter for PPF readings? What if the COBs have differences in beam angle? How about tempeature variations (ambient and COB)? What if the sensor that was used is inaccurate? What if the COB was slightly damaged already? What if it the sensor was slightly tilted for one (or more) of the measurements?
Probably worst yet, the efficiency binning of these COBs can be in 10% wide bins, so you need to test dozens of COBs from different batches to actually get a good average or your measurements might be 10% off.
So yeah, I'll take those measurements with a 15% to 30% sized grain of salt. Besides, you would still need the datasheet for actually finding out which is the best/cheapest solution.
How does that chart tell us how many COBs you need to fill a room with 800umol/s/m2 on average? The main thing that chart does is to sort COBs in size, with the biggest at the top and smallest COBs at the bottom. Mixed in with some efficiency differences.
I have no comment on the price/performance chart, or
@CobKits methodologies, but like
@robincnn and
@Malocan's tests, testing between COBs with a PPFD meter depends on trusting a few things:
All COBs have the same 120 degree FWHM angle. So a PAR reading from identical distances, or especially taking the average PPFD over a 7x7 grid like Malocan did, should be a way to compare -relative- performances between COBs.
Temperature differences...@malocan let them stabilize in the same setup,
@CobKits depends on overkill with fan and heatsink,
@robincnn was pulsing to the best of his ability, keeping Tc as close to Ta as possible. This was how
@SupraSPL worked as well. So in the first two cases I would argue they are testing thermal resistance as well. In robin's case, matching Tc and going from there.
Something you didn't touch on, but isn't being corrected for in these tests...quantum sensors are not linear in their response, so testing between two different brands or CCTs or both does skew the results, but not too badly. Additionally, once you push output past 700nm, it's not counted at all, so if far red is something you want, it's not being counted in the photon flux of any PAR meter.
You are correct about binning, robin has already shown spreads between samples. Maybe not 10%, but it's there between Bridgelux and Citizen COBs for sure.
But here's the motherfuck: I've voiced my concerns with methodologies here over and over again, but when every homebrewed half-assed test with PAR meter shows the same results, and points in the same direction, eventually you have to say, OK this is probably reality.
@Malocan's test is what really did it for me. He hung two COBs at the same height in a reflective tent at multiple wattages, took an average reading for each over 49 points, and the 2700K 90CRI Citizen out-punched the 3500K 80CRI CXB3590. The gap was small, it might be an over-performing Citizen, but that's impressive.
All we have is relative results between COBs anyway. Where's the integrating sphere results for any fixture anybody is selling in this place? A couple vendors have promised them, but so far there is nothing. Which makes me sad. It's all based on Supra's spreadsheets and some solid testing with PAR meters and DMMs.
Apologies if I mangled any details. This was all from memory.
