Can someone explain 80 CRI vs 90 CRI- LED

_eNVy_

Member
I'm very curious as to why you chose 3000 CRI 90 and I chose 2700 CRI 97.

Did I make a wrong choice?

What was your determining criteria behind your choice?
No I don't think you made a wrong choice! I think that 2700k are a better flower option than 3000k but I chose the 3000k 90cri mostly due to the fact that It will be both a veg light (supplemented with T5HO hortilux UV) and my flower light. Also through my research It seems that 3000k with my setup would be more efficient vs 2700k. I basically wanted more red for flowering but still be able to veg with limited supplemental T5HO. Someone correct my thoughts on this if I am mistaken
 

_eNVy_

Member
If you refer to post #10 on page 1 here you can see the graph of 4000k - 2700k @ 90cri. 3000k wins out over the 2700k up to 630nm where they both top out at 100%, from there it's pretty much neck and neck between the two.
 

NoFucks2Give

Well-Known Member
pretty much neck and neck between the two.
That may not be true. If you look at the CIE xy chromaticity for citi 3000K and 2700K I posted a few post ago, it shows very little difference in Deep Blue. The center points of the x and y for both 3000 and 2700 are on the same lateral. If the 3000 had more deep blue than the 2700 the center point would be lower then the 2700. Deep Blue being in the bottom left hand corner.

The CRI 90 chart post on page 1 may be miss leading. The chart is not that great for comparison form one CCT to another. It's intended purpose is the show the relative wavelengths of the single CCT. Each CCT chart is scaled so the highest peak is at 100%. This is most obvious for the 4000K where the blue is the highest peak.

When the PPFD is measured all the wavelengths are summed. This means the width of the peaks is weighs very heavily when calculating PPFD or Lux.

Notice the widths of the higher CCT charts get wider. Yet the peaks at 630nm are all equal. Where is that unaccounted for radiant energy coming from? All these LEDs have the same type and number of dies. But the various CCT variations appear to have significantly different flux energy. That fact along with the CIE xy chromaticity plot indicating the deep blue values are about the same, says to me the charts for each CCT are scaled at different ratios.

This would mean the blue peaks should be about the same height if the curves were scaled at the same ratio. The curves for the 3000K show it has 25% more blue than the 2700K. That 25% is not reflected in the chromaticity plot. It is more likely due to the difference in scaling of the curves.

If you scale down the 3000K by 25%, then the red peak will come came down to about 75% of the 2700K. Then the variation in peak widths will make more sense.

Full disclosure. All of the above is based on my own personal observation and opinion. I believe it to be true because I am really good at this shit. That said, I will also admit I am not normal. Not even close.

A couple of years ago I tasked my self with designing the most efficient LED grow fixture. I have been working with the University of Florida having been hired as a contractor by the professor that leads their LED horticulture research. I have been working on this LED stuff 18 hours a day 7 days a week for the past 2 years.

My decision making process for selecting the 2700K LED is quite different. I bought them to be used as heat sources for thermal experiments. I figured I'd get the best possible PAR CoB so I can use the CLU028-1204 to grow stuff if they survive the thermal experiments. 3 of the 4 that arrived on Monday are still functional.

This week I may have discovered an exceptional method of keeping CoBs running cool. I just bought a milling machine to make some custom parts for the idea.

This is the test setup:

The CoB is mounted on a 0.062" x 1" x 24" copper bar. The copper bar is clamped, using C-Clamps, to a 1/2" copper pipe with water flowing through it.

In this photo, you can see on the multimeter, the thermal pad temperate is 45° C. But this is Florida and this is in my garage and the water temp is rising. Last night I had the thermal pad temperature down to 23° C where ambient was 28° C. Ice cubes in the water.


email3.jpg



heatsinkCobClamps.jpg
 

Rahz

Well-Known Member
So you are saying a 3000K 70 CRI has higher PAR value than a 2700K 90CRI?
If so, that is incorrect.
Readings were taken with an Apogee SQ-520. I am confident the readings were accurate. Each reading was taken centered under 4 cobs. All tests used the same driver. All cobs at the time of reading were at an exact height, and the x and z axis were within a millimeter or so. There's nothing I could have done to get a more accurate reading.

Additionally I did a spectral analysis for all the samples along with the Apogee 520 response curve for normalization and the Mcree curve. The Apogee response curve was factored in which hardly changed the results, at max just over 1% variance among the samples tested. The Mcree curve was also factored in and didn't change the results much due to all samples having an abundance of orange.

I also extracted the relative RGB values for those samples from the datasheets. 2700/90 emits about 10% more red (600-700mn) than 3000/90 as a percentage of total output.

BGR % values for 2700/90
8.7983044058
34.4864366489
56.7152589453

BGR % values for 3000/90
12.517040412
36.1719812191
51.3109783688

However, when factoring in the par readings it comes out to about 2% more absolute output in the red region. You need to do a spectral analysis and factor in par readings to get an accurate comparison. It's not the same as scaling the SPDs.

What wasn't backed up by the math was the 3000/90 and 2700/90 samples both providing better yield than the lower CRI samples. What this suggests is that while the YPF factor is necessarily accurate for the vegetative results in the Mcree study and quite possibly cannabis as well (3000/70 did really well in veg) since variation from one crop to another was within 5%, the actual benefit of using more red during the flower cycle for cannabis is greater than what a correction factor based on the Mcree study suggests.

I hope this give you a better idea of what I used to make the claims I have. Because of the grow test and the 90 CRI results compared on paper I decided to switch to 3000/90. It's possible that the new Veros among others will have less performance gap between 2700/3000 but it's difficult to guess without having LER/QER to look at. Does anyone have that data?
 

NoFucks2Give

Well-Known Member
I hope this give you a better idea of what I used to make the claims I have.
I am fuckin' impressed. I thought you were quoting someone else's data. I am not accustomed to anyone knowing what they are talking about.

I believe you now. But I am still going to buy a 3000K and test it for my self. I kinda want to be wrong. What do you think I missed? I still believe my assessment is correct, but I find it difficult to argue the point with you.

It's very difficult to argue with measurements backed by actual real life results. 1% deviance, I thought I was the only one crazy enough to to shit like that.

I just looked at your website. WTF? Someone selling fixtures actually knows shit? That's too fuckin' cool. What the hell planet are you from? The Coup de grâce, ""We also list the PAR watts of each lamp and the PPFD in the recommended usage space." God damn alien.

Fuck, I got to go, just ran out of beer.





 

_eNVy_

Member
That may not be true. If you look at the CIE xy chromaticity for citi 3000K and 2700K I posted a few post ago, it shows very little difference in Deep Blue. The center points of the x and y for both 3000 and 2700 are on the same lateral. If the 3000 had more deep blue than the 2700 the center point would be lower then the 2700. Deep Blue being in the bottom left hand corner.

The CRI 90 chart post on page 1 may be miss leading. The chart is not that great for comparison form one CCT to another. It's intended purpose is the show the relative wavelengths of the single CCT. Each CCT chart is scaled so the highest peak is at 100%. This is most obvious for the 4000K where the blue is the highest peak.

When the PPFD is measured all the wavelengths are summed. This means the width of the peaks is weighs very heavily when calculating PPFD or Lux.

Notice the widths of the higher CCT charts get wider. Yet the peaks at 630nm are all equal. Where is that unaccounted for radiant energy coming from? All these LEDs have the same type and number of dies. But the various CCT variations appear to have significantly different flux energy. That fact along with the CIE xy chromaticity plot indicating the deep blue values are about the same, says to me the charts for each CCT are scaled at different ratios.

This would mean the blue peaks should be about the same height if the curves were scaled at the same ratio. The curves for the 3000K show it has 25% more blue than the 2700K. That 25% is not reflected in the chromaticity plot. It is more likely due to the difference in scaling of the curves.

If you scale down the 3000K by 25%, then the red peak will come came down to about 75% of the 2700K. Then the variation in peak widths will make more sense.

Full disclosure. All of the above is based on my own personal observation and opinion. I believe it to be true because I am really good at this shit. That said, I will also admit I am not normal. Not even close.

A couple of years ago I tasked my self with designing the most efficient LED grow fixture. I have been working with the University of Florida having been hired as a contractor by the professor that leads their LED horticulture research. I have been working on this LED stuff 18 hours a day 7 days a week for the past 2 years.

My decision making process for selecting the 2700K LED is quite different. I bought them to be used as heat sources for thermal experiments. I figured I'd get the best possible PAR CoB so I can use the CLU028-1204 to grow stuff if they survive the thermal experiments. 3 of the 4 that arrived on Monday are still functional.

This week I may have discovered an exceptional method of keeping CoBs running cool. I just bought a milling machine to make some custom parts for the idea.

This is the test setup:

The CoB is mounted on a 0.062" x 1" x 24" copper bar. The copper bar is clamped, using C-Clamps, to a 1/2" copper pipe with water flowing through it.

In this photo, you can see on the multimeter, the thermal pad temperate is 45° C. But this is Florida and this is in my garage and the water temp is rising. Last night I had the thermal pad temperature down to 23° C where ambient was 28° C. Ice cubes in the water.


View attachment 3943776



View attachment 3943784
I'm still a noob when it comes most of this led cob shit. That being said I'm a pretty smart guy and thoroughly enjoy learning things. I guess I take my hobbies a bit too far like most of you seem to also do. Probably my OCD kicking into overdrive lol. I have been soaking all this light science up like a dry sponge. Definitely not biased as to 2700/3000 etc. in fact I plan to fully research and experiment all these theories and ideas people here have. Like I stated in a different thread, this led project is a hobby to add to the already existing grow hobby. Anyways I hope we can all share whatever proof we come to find out through research and/or trial and error.

Btw copper would probably be by far one of the better heatsink options (besides gold/silver) especially when water cooled! That shit ain't cheap though. Would be fun to do what you're doing though and I might engineer my own heatsink from scratch in the future. Totally awesome! All of this:bigjoint:
 

_eNVy_

Member
So you are saying a 3000K 70 CRI has higher PAR value than a 2700K 90CRI?
If so, that is incorrect.

I plotted the CIE xy chromaticity for citi 3000K and 2700K with an app I wrote today.
Not too much difference between the two, but 3000 has more green and less red than 2700. Difference in deep blue is negligible.

The bigger difference is in the CRI 90 vs. CRI 70 values and in particular the R9 value.


View attachment 3943591



Notice below, the CRI 82 R9 is below 20.



View attachment 3943611

View attachment 3943622




What color is missing in the CRI 70?



View attachment 3943603


View attachment 3943602
Think you would be willing to plot another chart featuring citi CLU048 1216 3000/90 and 2700/90? I'd like to see if the outcome is the same
 

Rahz

Well-Known Member
Thanks and I look forward to your results. As far as what you missed, as I mentioned normalizing 2 curves at a particular point isn't going to give you accurate results at any other point in the curves.

You can use math to predict absolute PAR values (PPF) (and abslute RGB values) but you need a bit more information, which @alesh has provided along with instructions here: https://www.rollitup.org/t/math-behind.868988/

Even still, if PPF and even YPF doesn't tell the whole story we're still left with grow tests being the best way to compare lights.

To be accurate we need more grow data but if we were to assume my little grow test was accurate we could make a determination about the efficacy of the two spectrums.

2700/90 = 89.7% output -vs- 3000/90 (paper prediction)
2700/90 = 91.1% output -vs- 3000/90 (raw PAR readings)
** 2700/90 = 93.6% efficacy -vs- 3000/90 (raw PAR readings with Mcree factored in)

3000/90 - 367 grams.
2700/90 - 344 grams.
2700/90 = 94.5% yield -vs- 3000/90

Comparing the raw PAR readings to the yield indicates the 2700/90 spectrum is a better flowering spectrum, but we still choose 3000/90 because in the end it provided more grams per watt presumably due to output efficiency outweighing spectrum efficacy.

** I hadn't examined the data this way before. How it would look if I ran the test multiple times with better conditions (the grow was problematic) I'm not sure but it seems to indicate that the Mcree data does in fact get the final numbers close to the yield results at least for these two samples.
 
Last edited:

_eNVy_

Member
T
Thanks and I look forward to your results. As far as what you missed, as I mentioned normalizing 2 curves at a particular point isn't going to give you accurate results at any other point in the curves.

You can use math to predict absolute PAR values (PPF) (and abslute RGB values) but you need a bit more information, which @alesh has provided along with instructions here: https://www.rollitup.org/t/math-behind.868988/

Even still, if PPF and even YPF doesn't tell the whole story we're still left with grow tests being the best way to compare lights.

To be accurate we need more grow data but if we were to assume my little grow test was accurate we could make a determination about the efficacy of the two spectrums.

2700/90 = 89.7% output -vs- 3000/90 (paper prediction)
2700/90 = 91.1% output -vs- 3000/90 (raw PAR readings)
** 2700/90 = 93.6% efficacy -vs- 3000/90 (raw PAR readings with Mcree factored in)

3000/90 - 367 grams.
2700/90 - 344 grams.
2700/90 = 94.5% yield -vs- 3000/90

Comparing the raw PAR readings to the yield indicates the 2700/90 spectrum is a better flowering spectrum, but we still choose 3000/90 because in the end it provided more grams per watt presumably due to output efficiency outweighing spectrum efficacy.

** I hadn't examined the data this way before. How it would look if I ran the test multiple times with better conditions (the grow was problematic) I'm not sure but it seems to indicate that the Mcree data does in fact get the final numbers close to the yield results.
Thank you Rahz. I've read a lot of your posts and appreciate what you contribute to this community! Sure helped me in deciphering all this madness.
 

nfhiggs

Well-Known Member
Well I'm sold on the 90cri. Correct me if I'm wrong, 90cri = more red with the disadvantages of slightly less efficiency and losing photons to green. 80cri = a bit more efficiency and a good balance with less loss to green? This would be for 3000k. I'm a noob with led cobs but I'm about to purchase (16) citizen CLU048 1212 3000k and was told to check this thread out since I originally wanted 80cri. Thanks
What do you mean by "loss to green"?
 

_eNVy_

Member
"green" being the 550 part of the spectrum that's useless to our plants. I was under the impression that 90CRI gave off more of this vs. 80CRI. Which is where I get the less efficient statement from. I've stated many times that I'm new to this and learning as much as possible so please correct me if I wrong.
 

nfhiggs

Well-Known Member
"green" being the 550 part of the spectrum that's useless to our plants. I was under the impression that 90CRI gave off more of this vs. 80CRI. Which is where I get the less efficient statement from. I've stated many times that I'm new to this and learning as much as possible so please correct me if I wrong.
Ahhh.. but green is not useless. That is a common myth. I also thought that way when I began growing as well, but I have learned that plants can and do use green light, and its actually a very efficient driver of photosynthesis at higher PPFD levels. At lower light levels, yes, the red and blue dominate, red particularly due to the Emerson effect. But the advantage of the Emerson Effect falls off rapidly as light intensity increases.

I do not discount any portion of the white light spectrum based solely on absorption curves.
 

_eNVy_

Member
Ahh well thanks for the info. I'm going to look into that "myth" right after this post. Man I absolutely fucking hate myths, wives tales, etc. just waters down the truth. Which is all that I seek :bigjoint:
 

_eNVy_

Member
So through a quick read from a Michigan State University study. And according to that study, green isn't exactly useless but is inefficient (as I stated previously) so you're absolutely right on that one. I guess the best part of this would be that it is also able to penitrate the canopy far better than any other light wavebands. Kinda cool little read and glad you steered me into mythbusting that one.

EDIT: when I say "inefficient" I mean to say that's it's the least efficient of the light wavebands. Not to say that it is useless for promoting photosynthesis, as it is the opposite.
 
Last edited:

nfhiggs

Well-Known Member
So through a quick read from a Michigan State University study. And according to that study, green isn't exactly useless but is inefficient (as I stated previously) so you're absolutely right on that one. I guess the best part of this would be that it is also able to penitrate the canopy far better than any other light wavebands. Kinda cool little read and glad you steered me into mythbusting that one.

EDIT: when I say "inefficient" I mean to say that's it's the least efficient of the light wavebands. Not to say that it is useless for promoting photosynthesis, as it is the opposite.
Here's another page you might find interesting:
https://www.reddit.com/r/HandsOnComplexity/comments/17nxhd/sags_plant_lighting_guide_linked_together/?st=j28zyko4&sh=d4ba3204

He busts a few myths... Guy knows his shit....
 

_eNVy_

Member
Dude that's a gold mine worth of information! Appreciate the link. I have a lot more reading to do.
 

NoFucks2Give

Well-Known Member
Btw copper would probably be by far one of the better heatsink options (besides gold/silver) especially when water cooled! That shit ain't cheap though.
Copper today is $2.53 / lb.
A 0.062" x 1.0" x 48" copper bar is less than $10 and weighs just under 1 lb.
1/2" Copper pipe a Home Depot today is less than $1.30/ft.

So one solid copper 4 foot fixture would cost $15.30 plus CoBs and LED drivers.

Assembly would be something like this:

cobMounting.jpg

I use two 60 gph water pumps @ $8.00 each and less than $10.00 in fittings to get the water flowing in the pipe. I use two pumps in case one fails.

waterFittings.jpg



And my proof of concept:

cobTestSetup.jpg
 

Greengenes707

Well-Known Member
Not bullshit. Orange isn't as effective as red in the Mcree study but it's no slouch either. 3000/70 has lots of orange so the enhanced spectrum of a 90 CRI sample wasn't enough to offset the large difference in the initial par readings. All it means is that YPF/McCree isn't a perfect way of measuring spectrum efficacy for flowering.
You have it backwards. Orange(590-620) is most effective(Relative 100%) at photosynthesis according to McCree. The whole range has 99-100 RQE. As where RQE falls slightly in the mid red region(630-645) and more so into deeper red(645-700)...but is where the photomorphogenic is said to make up for the slight decrease in carbon fixation. That photomorphogenic is only important to us if it's in dried flower weight and is quantifiable. But that is not what I saw in your experiment.
Your experiment results followed no pattern. They were random at best. There is no explanation other than a test that parameter got out of control in addition to a small sample to begin with(which even you acknowledge). Speaking of sample sizes...brings up the other experiment going on, the dawgs. Also too small to make any judgments other than, "they grew great pot". All great grows...but the conclusions being drawn from them are not accurate. I did the spectral side by sides(3v4, 3v3.5, 4v3.5) and the yields were negligible and not devoted enough to call a difference. Structure was notably different, but total yield was the same. Grows with larger plants and numbers. All documented in GG's garden thread that is now shut down.

@dro-man80 Has a nice experiment run going on. Good numbers and canopy size.
https://www.rollitup.org/t/one-more-run-with-the-cxbs-3000k-80-90-cri-qbs-are-next-in-line.938932/


Copper today is $2.53 / lb.
A 0.062" x 1.0" x 48" copper bar is less than $10 and weighs just under 1 lb.
1/2" Copper pipe a Home Depot today is less than $1.30/ft.

So one solid copper 4 foot fixture would cost $15.30 plus CoBs and LED drivers.

Assembly would be something like this:

View attachment 3944109

I use two 60 gph water pumps @ $8.00 each and less than $10.00 in fittings to get the water flowing in the pipe. I use two pumps in case one fails.

View attachment 3944116



And my proof of concept:

View attachment 3944123
That wheel has already been invented and turning for a while by @wietefras ...

You'll do fine with it.
SteamPunk_Leaktest.jpg

Quote from builder below VVVVV
I created this "steam punk" fixture:
View attachment 3834744

I soldered 3mm thick copper plates onto copper pipes and attached the COBs to the copper plates.

I think I'm going to try again with a single pipe bent into an S-shape instead.
 
Last edited:

NoFucks2Give

Well-Known Member
He busts a few myths... Guy knows his shit....
I don't buy into the Green. What he says is true except some plant species do not like Green. There have been studies on green and green is no longer used in most University studies. I do not want to take the time right now to find the studies.

I was impressed he wrote about the view angles of LEDs. This is a topic I have been working on for the past year.

I have written an app that uses the View Angle Spacial Distribution from the datasheet, it models the distance between LEDs in the fixture and the spacing of the fixtures to get optimal uniformity across the canopy.

From a paper I am writing on the topic:


Radiation Distribution Pattern for High Density and High Intensity.jpg



One problem with the Radiation Distribution Pattern the measurements are done in a sphere.

measurmentAnglesDistribution.jpg




So I recalculate the angles for a flat earth.

measurmentAnglesDistributionSphere.jpg


I can visually compare View Angles

The small insets are distribution curves, I added color and did the flat earth translation in the other.

compareViewAngles.jpg



I loop through many different distances between LEDs and fixtures and measure evedry 0.100" along a 4 foot line, measuring the Photon Radiance at each point, summing the flux at that point from every LED. About 2 Million calculations per run.



Simplified visual of the process:



anglesDistribution.gif
 

OLD MOTHER SATIVA

Well-Known Member
You have it backwards. Orange(590-620) is most effective(Relative 100%) at photosynthesis according to McCree. The whole range has 99-100 RQE. As where RQE falls slightly in the mid red region(630-645) and more so into deeper red(645-700)...but is where the photomorphogenic is said to make up for the slight decrease in carbon fixation. That photomorphogenic is only important to us if it's in dried flower weight and is quantifiable. But that is not what I saw in your experiment.
Your experiment results followed no pattern. They were random at best. There is no explanation other than a test that parameter got out of control in addition to a small sample to begin with(which even you acknowledge). Speaking of sample sizes...brings up the other experiment going on, the dawgs. Also too small to make any judgments other than, "they grew great pot". All great grows...but the conclusions being drawn from them are not accurate. I did the spectral side by sides(3v4, 3v3.5, 4v3.5) and the yields were negligible and not devoted enough to call a difference. Structure was notably different, but total yield was the same. Grows with larger plants and numbers. All documented in GG's garden thread that is now shut down.

@dro-man80 Has a nice experiment run going on. Good numbers and canopy size.
https://www.rollitup.org/t/one-more-run-with-the-cxbs-3000k-80-90-cri-qbs-are-next-in-line.938932/

/QUOTE]

again thanks for dropping in with your valuable input ...you are an oasis sometimes
 
Top