Thank you very much naoki. It certainly is a lot of information to take in for someone as new to LED lighting as I am but this gets me one step closer to taking the plunge.
LED stuff is a bit complicated, and it did take quite a bit to learn about it for me, too. Sorry if I don't explain things in an easy way... I'm trying to leave out too much details, but I frequently end up with difficult explanation...
Two more quick questions:
I've noticed that people seem to stick with <5000K lighting for LEDs, yet bulbs as high as 6500k are common with fluorescent lighting. Why is that?
My theory is that people took "day light" spectrum of fluorescent light literally. Here is my measurement of T5HO:
3000K 82 533 1-2mo old, Philips F54T5/830
4100K 83 590 new F54TH/841/HO/ALTO
5000K 80 620 1-2 mo old, F54T5/850
5000K 85 625 new F54T5/850/HO/ALTO
The measurement is taken at 12" from the tube in a typical fixture/reflector with Li-COR LI-190 (for PPFD) and LX1330B (fc).
2nd column is PPFD in micromol/m^2/s
3rd column is fc.
There is a bigger difference in terms of foot-candle (which isn't as relevant as PPFD), and higher K seems to have higher fc. But in terms of the PPFD, the difference is pretty small. I didn't have 6500K, but the trend should be similar. I used to use whatever available between 3000-6500K for T5HO. Those purple plant bulbs do indeed have higher efficiency than white ones in my measurement (
link to my measurement), but it is questionable if the extra cost is worth spending (since the output decays in 1-2 years).
As a related note, I got some data comparing PAR efficiency of T5HO (T8 and LED, too). Here is a calculation done by some planted aquarium people from South Africa to find best T5HO bulbs. It is sorted from the most efficient to less efficient (the top is the best). Basically they digitized the emission spectra, and calculated how much plant relevant light is emitted for a given amount of electric power usage. This is the "PAR efficiency" column (micomol/J is the unit). K (kelvin) column is the calculated (not the advertised) corrected color temp. This is a bit old data, so I'm not sure if all of these are still available. To be honest, I'm not sure if the PAR efficiency of the top one is really correct. It seems to be too high for fluorescent light. Top rated ones seem to have higher K values, so there may be some reasons why people use 6500K. But this is from calculation (not measurement). Also, some of the specialty bulbs may have quite different emission spectra from the typical T5HO.
The other factor is the emitted spectrum of fluorescent bulbs (jagged shape with a few major peaks corresponding to different phosphors) is very different from LED (smooth, 1 blue peak + yellow/red peak from phosphor). Once you start to go above 5000K with LED, the light is dominated by the blue peak, which doesn't look too good for phytochrome related response (my guess here). Many MJ people goes with 3000-4000K COB LEDs for flowering because their flowering is photoperiodic (phytochrome is relevant for this process) and requires red light.
Note that typical spectral power distribution (SPD) curve which is in data sheet has power (watt) on y-axis. But the photosynthesis is more related to the number of photons than power, so you need to correct the y-axis. 1 blue photon has more energy than 1 red photon. So if you cover y-axis to number of photons, the blue peak becomes a bit smaller.
What does Cree's two-step, three-step, and five-step binning mean? They have multiple "steps" for some color temperature/bin combinations but don't understand the chart at the end of the spec sheet about it.
Those are the color accuracy. For some applications, they requires that color of one light to match with another. I forgot, but I think 2-step is more accurate than 5-step. But for plants, accurate color match isn't relevant. So I ignore that part of binning. So I go with whatever the cheaper ones with lowest CRI and highest luminous flux bin. So for example, you can look at p.3-4 of the
cxb3070 datasheet. For 4000K/CRI70, you have BB or BD bins (they call it minimum flux "group", but this is the bin number which most of us are talking about). BD bin is the best, and the product number is CXB3070-0000-000N0BBD40E. This is the top bin for 4000K (the link to Cutter above is this one), and I haven't seen it available from other vendors. Now if you go to 4000K/CRI80, then the top bin is BB (which is lower than BD). There are 2-step and 4-step version, but I would go with either one (probably the cheaper one) if there is a choice. You can think that a higher CRI requires more phosphor, and more phosphor could result in a bit of light loss during the conversion of blue light to yellow/red light. Generally, if you go down in K, the top bin class becomes lower (e.g. for 3000K, AD is the top bin). So you have to look at the data sheet to decide what is the top bin of each category.