Top bin COB comparison

Abiqua said:

I used to to, but not any more umol/j seems to be a better baseline, because not all umol created equal

Click to expand...

You saw all the dust that got kicked up a few posts back just trying measure PAR with a brand name meter...and now you want to talk about how some photons are better than others in that PAR number? Please let me know how youll measure that. Im interested.

BuddyColas said:

You saw all the dust that got kicked up a few posts back just trying measure PAR with a brand name meter...and now you want to talk about how some photons are better than others in that PAR number? Please let me know how youll measure that. Im interested.

Click to expand...

Search is your friend, talked to death around here with PPFD calculations, what more is there to say?

Abiqua said:

Search is your friend, talked to death around here with PPFD calculations, what more is there to say?

Click to expand...

BuddyColas said:

You saw all the dust that got kicked up a few posts back just trying measure PAR with a brand name meter...and now you want to talk about how some photons are better than others in that PAR number? Please let me know how youll measure that. Im interested.

Click to expand...

combined average input of every NM from the RQE log, digitized and actualized as well into graph form.

Maybe @alesh can elaborate further and correct any fuck of my fuck ups of past said tense

if you want real par measurement you need a li-cor quantum sensor or better. not 100% flat response but way better than the apogee/sun systems

http://www.apogeeinstruments.com/content/Quantum Sensors-LEDs.pdf
http://www.advancedaquarist.com/2013/2/equipment

i just picked up a digital version of the licor quantum meter and can probably be talked out of my old analog meter (same sensor but uses basically an old school multimeter calibrated to the correct ranges). whether its a par meter or a guesstimate from a lux meter, *everyone* needs a basic meter at minimum to find the dark spots hiding in your garden

Abiqua said:

combined average input of every NM from the RQE log, digitized and actualized as well into graph form.

Maybe @alesh can elaborate further and correct any fuck of my fuck ups of past said tense

Click to expand...

Ok...so you are talking a spectral graph of a cob. So when you are tweeking your lights and adjusting drive currents, light heights, light spacing etc. What do you use to measure photons at the canopy?

How about a Li-cor Li-190r. Looks pretty flat from 400 to 700. Probably big bucks.

that sensor itself is $400-500. the meter to read it is upwards of 1500

BuddyColas said:

Ok...so you are talking a spectral graph of a cob. So when you are tweeking your lights and adjusting drive currents, light heights, light spacing etc. What do you use to measure photons at the canopy?

Click to expand...

No i am talking about the amount of "power" for each NM across the entire PAR spectrum, that is optimum for plant growth for each wave length, [although it is hardly static] and GG even showed the "cannabis rqe" in the chat as well, linked from SDS thread.
This is how you differentiate the diff between blue and red for instance...if something[aka a white source] is leaning more red, chance are the umol/ joule are probably higher than a cooler Kelvin temp and hence with same wattage and space as a "competing" 4k spectrum, would have a higher PPFD in theory. Then a field measurement comes into to test the PPFD.......

im trying to interpolate the data from the chart on page 1 for CXB3590 3500K

0-50 W is damn near linear and PPFD can be modeled as P= 9.25-0.03W

so.. i wanted to calc an ROI for 4@50W vs 8@25W vs 16@12.5W etc.

the above equation can be normalized to represent efficiency = 1 @ 50W:
E=1.1935-0.00387W

thus if 50W has a relative efficiency of 1
25W has a relative efficiency of 1.097
12.5W has a relative efficiency of 1.145

but if i look at Supra's efficiency tables we have a different relationship for PAR/W (relative to 2.62@48.846W) / umol/W)

A /W / formula /table
1.4 / 48.846 / 2.62 / 2.62
1.05 / 35.54 / 2.756 / 2.846
0.77 / 25.37 / 2.859 / 2.950
0.7 / 22.90 / 2.883 / 2.976

or to put it a different way, on a umol/W basis, according to the tables, 8 cobs at 25.37W should be 12.6% more efficient. according to the formula it is 10.0% more efficient.

by the formula, 12.5W is only 14.5% more efficient than 50W. i was anticipating it being 20% more efficient

obviously for that sort of investment (16 cobs vs 4 cobs), that 5% is kind of a big deal

what am i missing? are my calcs wrong? are supras measurements wrong? is cree's reported data erroneous?

also is there a trick to getting data out of crees product characterization tool? i enter parameters, and it spins and 'thinks' after each change, but the table never populates. same in both firefox and chrome

regardless....

lets say that 25W is 'only' 10% more efficient than 50W
and that 12.5W is 'only' 15% more efficient than 50W

lets do a thumbnail ROI

if you go with 4@50W vs 8@25W vs 16@12.5W you *should* have the same driver load in watts but voltage is getting too high w/ 16 so we have to add a driver (too bad we are lacking the HLG-185H-C350!) lets us (2) HLG-120H-C350, +$50 over one HLG-185H-C500.....

heatsinks = same (in fact you need less but keep them the same for this thumbnail)

so if the ONLY cost difference is number of COBs
base price of 4 COBs = $180
price of 8 COBs = $360 (additional $180 over base)
price of 16 COBs = $720+$50 driver (additional $590 over base)

so lets just say your crop of choice delivers 1 g/w at the 'base' 4 COB configuration

if you run 8 at the same 200W total wattage and bump up your yield by 10% you yielded an addl 20g@$4.4 = $88 per cycle. you'd cover the addl $180 cost of the COBs in about 2 cycles! even with a relatively short cycle life of a year or two between tech that's a phenomenal return!

AND you enjoy:

• better coverage
• less heat!
• longer life due to less heat!

but lets get kooky

16 COBs @12.5W sounds crazy, right? must be some diminishing returns we will hit, right? if you run 16 at the same 200W total wattage and bump up your yield by 15% you yielded an addl 30g@$4.4 = $132 per cycle. youd cover the addl $590 cost of the COBs and driver in a mere 4.5 cycles! and continue to run with close to 70% efficiency from there on out!

AND you enjoy:

• EVEN BETTER coverage
• EVEN LESS heat!
• possibly even more longer life due to less heat!

if your efficiency at 350 mA is closer to +20% as i suspect it would be even shorter return (less than 3 1/2 cycles)

@BOBBY_G you are describing the "compression" that I referred to. I should be able to use the curve to predict the lumens/W at 50W by knowing the lumens/W at 75W, but in practice it does not work. The curves should have more slope to them, based on the data in the spec sheets. This does not cause me to doubt the lumen figures or efficiency we see in the CREE spec sheets. It may be a limitation of the constant-distance measuring method that I have to use. An integrating sphere would solve that problem and I have no doubt that is what CREE uses for lumen measurements.

The CREE PCT is a bit fiddly, but if you change the measurement range to coarse the chart will populate. It shows "minimum" lumen figures versus my spreadsheets which use "typical" figures. The PCT also uses Tc/Tsp even though it defaults to Tj so that can be misleading as well.

Regarding Apogee correction factors, in theory they can work perfectly to measure PPFD and make comparisons from one light source to another. BUT that would require:
-every Apogee sensor must respond with the exact same curve
-every COB must emit the exact SPD shown in the data sheet
-the correction factor would have to be generated correctly (I would trust it more if @alesh peer reviewed its accuracy) For example did Apogee use Vero V1.2 or V2.0 curves?

Since the Apogee response drops off in a critical area, any slight deviation of the COBs SPD or the meter's response would result in a significant distortion to the results. So unfortunately we are still guessing quite a bit when it comes down to it. What the meter does best is give you a general idea of the PPFD in different parts of your canopy. But for $100-$150 the SQ120 is a miracle worker compared to a lux meter I will give it that.

A caloric test can give us some idea what is going on in terms of efficiency as well. High powered COB on a small heatsink results in very high temps and that should exaggerate the difference in efficiency. In my first experiment this worked reasonably well, revealing the efficiency difference between the CXB3070 and CXB3590. It should help us compare the efficiency of the Vero29 with the CXB3590 as well.

I definitely intend to thoroughly (re) test the Vero 29. I had a pair of Vero29 3000K and tested them against CXa3070 3000K AB and the CREEs won. BUT that was awhile back and I was using a luxmeter which is horrible to make comparisons with. I hugely appreciate what CREE has done for us but I have no special loyalty of bias toward their products. My goal is to fill the jars as efficiently as possible. If Veros are consistently overachieving their specs that is very awesome news for growers but we should keep in mind the Bridgelux spec sheets indicate Vero 29 V2.0 to be similar to CXa3070 top bins.

At the end of the day we do have some promising empirical results coming in. Growmaus 2.03g/dissipation W in 63 days using CXB3590s at 25W ea and recently I got 1.48g/dissipation W in 56 days using a similar setup ~700PPFD. Of course it will be highly variable from grow to grow, but I think we could be seeing numbers like these and even higher popping up all over the place if the COBs really are 64% efficient or even close to that.

@SupraSPL what are your thoughts on the performance of the Vero 29 at low current? Is there an easy way to explain why the efficiency increase is keeping up with CXB as current drops?

yeah 2gpw will be the new 1gpw once people are in the 60% efficiency range

Supra you wanna borrow my li-cor?

Rahz said:

@SupraSPL what are your thoughts on the performance of the Vero 29 at low current? Is there an easy way to explain why the efficiency increase is keeping up with CXB as current drops?

Click to expand...

The change in efficiency is easily detectable but not quantifiable. I suspect this is the same type of problem I have been calling "compression" of the curves, they are probably not absolute like I had theorized (I expected a linear percentage change in PPFD/W should be able to be taken from any distance). I am not 100% sure what is causing it but I think it has something to do with constant-distance measuring of a light source that is not increasing or decreasing linearly, it is behaving based on a modified inverse square law. If so, the slopes would be increased, so if the CREE slope is slightly steeper than Vero, it would become even more steep once corrected. ^Honestly this is a half baked theory I really do not know.

In other words I think we need integrating sphere to get the accurate slope and comparison. I am glad to see the efficiency increasing at very low current in all cases though. It would be great if we could build lights in the future that are 70% efficient or more, for a similar cost or less than we are paying today.

BuddyColas said:

Ok...so you are talking a spectral graph of a cob. So when you are tweeking your lights and adjusting drive currents, light heights, light spacing etc. What do you use to measure photons at the canopy?

Click to expand...

Oh why, that's a perfect job for a PAR meter. That's exactly what it is made for.

alesh said:

Oh why, that's a perfect job for a PAR meter. That's exactly what it is made for.

Click to expand...

Yes. I know that. The question was what does Abiqua use in his garden to measure light at the canopy.


It is interesting to me how all the dust gets kicked up about PAR meters when even the vaunted Li-Cor PAR meter is calibrated to just +/- 5% from a NIST standard. And the slopes on their response curve are pretty much straight up and down and not the more gentle slopes of the true McRee curve response. So you lay down the bucks for a Li-Cor and you are still leaving some true PAR photons on the table…so to speak…and at the end of the day it’s just a “pretty close” meter. But a great tool for tweeking light heights, COB spacing, evaluating reflectors/lenses, the effect of various reflective walls surfaces etc.


And in the end…in the garden…pretty close it plenty good enough. That’s all I’m saying.

I may have missed this but how do you determine what is "top bin" for various lights and spectrums?

I'm particularly interested in CXB 3590 3500s, 3070 3500s, 3070 4000s and 3070 5000s. I intend to do a mixture of spectrums using top bins for each but I don't know how we figure out CXB 3070 3500 top bin is BB? I know this is extremely rudimentary, so apologies to those who see this as "old hat".

Mary's Confidant said:

I may have missed this but how do you determine what is "top bin" for various lights and spectrums?

I'm particularly interested in CXB 3590 3500s, 3070 3500s, 3070 4000s and 3070 5000s. I intend to do a mixture of spectrums using top bins for each but I don't know how we figure out CXB 3070 3500 top bin is BB? I know this is extremely rudimentary, so apologies to those who see this as "old hat".

Click to expand...

http://www.cree.com/LED-Components-and-Modules/Landing-pages/CXA2

Check out the spec sheet on the CXB of your choice and you will see the binning information. And check out most any thread with CX in the title and there is a gold mine of information on RIU.