DiY LED - Cree CXA3070

[Gaius]

Man I think you need to build a few Battlestars! The thing was custom-designed for 4x3 area.

 

[Gaius]

3 Battlestars will run you about $2200 in parts, a couple days of work, and will run at 1440 watts in the configuration I made. I suspect you may be able to skip the 5000K COBs to save some money if all you are doing with them is flowering, but Supra could probably tell you if they'll have enough blue or not.

 

[tallen]

I'm already thinking along those lines but I want to do a little research first before I commit to going that way. I'm starting out knowing very little and I want to do some learning before I start ordering parts. I'm sure I'll be hitting you up for help though, so thanks in advance

 

[SupraSPL]

Glad you decided to try diy! If you replace 2000HPS with 1000W of CXA3070 you might actually increase your yields and get better coverage in your canopy. Here is the formula I would recommend:

(8 ) Heatsink USA 10.08" X 6" = $350
(32) 3000K CXA3070 = $1375 (might save some $ by ordering from arrow instead of digikey)
(32) 1A drivers = $320
fans, wiring + misc $150

That would give you a 39% efficient lamp dissipating about 1200W for $2200

--------------------

This option is slightly less efficient and not spread as well but very economical and easy to build.

(4) Heatsink USA 10.08" X 6" = $175
(16) 3000K CXA3070 = $690 (might save some $ by ordering from arrow instead of digikey)
(16) 1.5A drivers = $160
fans, wiring + misc $100

That would give you a 36% efficient lamp dissipating about 880W for $1125. Maybe replace one of your HPS with these 4 modules for starters?

 

[tallen]

Great info Supra!! I'd be more inclined to go with the more efficient option, but wouldn't I be better off putting each COB on it's own (smaller) heatsink and spreading them out more instead of putting 4 on each one? I don't mind spending a little extra on heatsinks as long as it will be worthwhile to do so.

I'd also like to be able to light each 3x4 table (3 of them) independently so that I'm not lighting empty space during harvest and cleanup. I was thinking 8 (or maybe 12?) over each table. If my thinking is off let me know, like I said I don't really know what I'm doing but want to get it right from the start. Thanks Supra-you da man!!

Oh, and that arrow link to the COBs is in increments of 20, so I'd get 40 of them for about the same price (just slightly more) as 32 (your suggestion) from digikey, so I'd just get the 40 from arrow and have extras to play with

 

[Gaius]

Great info Supra!! I'd be more inclined to go with the more efficient option, but wouldn't I be better off putting each COB on it's own (smaller) heatsink and spreading them out more instead of putting 4 on each one? I don't mind spending a little extra on heatsinks as long as it will be worthwhile to do so.

I'd also like to be able to light each 3x4 table (3 of them) independently so that I'm not lighting empty space during harvest and cleanup. I was thinking 8 (or maybe 12?) over each table. If my thinking is off let me know, like I said I don't really know what I'm doing but want to get it right from the start. Thanks Supra-you da man!!

Oh, and that arrow link to the COBs is in increments of 20, so I'd get 40 of them for about the same price (just slightly more) as 32 (your suggestion) from digikey, so I'd just get the 40 from arrow and have extras to play with

Sounds like a good plan. Based on the brief experience with my light so far, I believe a single 3000k CXA3070 per plant is about perfect. So 6-8 over each 4x3 would probably do really well. 3 Panels with 8 3070s might bury you in bud. 10 over each tray may even surpass your 1000w HID yields.

Only downside I see to going with individual heatsinks (like I used) is more wiring to deal with.

 

[SupraSPL]

Yep you could certainly go with individual sinks and it wouldnt even affect the cost. You are correct 4 COBs on each sink will be some serious light density. Either way it would be easy for you turn turn a few modules off if you arent currently filling that space.

Gaius has a point regarding wiring. Individual sinks means more fans and fan wiring but also more adjustability. So far in practice I have found the big heatsinks are a good size for spreading light in my canopy but I am running at 700-740mA. You can see it on the bottom of page 20, the modules in the center have 4 COBs and active cooling. That is 500W in a 5'X4' canopy

 

[CannaBare]

I am excited to see what the buds will be like under 3000k! How are the plants doing thus far? Any extra leaf growth lol? Do you think the 3000k is actually better than a redder 2700k? My buds under my 3050's are killing at 1 week and 2 weeks flowering

One week Critical Jack Herer (Since pistles)


2 weeks Blueberry Gum(since Pistles)


Thanks I love the testing and math you do for this community!

 

[hyroot]

So in your guys opinions. How many cobs for a 4x4 - 5x5 for veg?

 

[Gaius]

4 CXA3070s would give good coverage for vegging.

 

[hyroot]

4 CXA3070s would give good coverage for vegging.

oh that's it. Cool Its going to be cheaper than I thought.

 

[SupraSPL]

Yep I veg in a 4X4 using 4 RWB modules (60W) to get good coverage. If I were to switch to COB I'd go with 4 modules and run them soft 650mA (80W). For the Cree CXAs 4000K has 19% blue and 5000K has 24% blue so Id probably go with the 5000K. That would emit 150lm/W at 47% efficiency.

If you want a more efficient spectrum you could mix Cree XM-L2 4500K with some deep blues and deep reds. Here is the thread for that one.

 

[SupraSPL]

Cannabare your buds look like they are coming along nice!

I am excited to see what the buds will be like under 3000k! How are the plants doing thus far? Any extra leaf growth lol? Do you think the 3000k is actually better than a redder 2700k?

I have been adding more and more 3000K so we will definitely get a good feel how they are doing on the next batch. The newest batch is 100% LED and they are varieties I am familiar with so that will be a good test. The 5.5 week batch (bottom of page 20) is now 100% LED but it had a few weeks of HPS exposure, has new varieties I have never run before, and some of them faded a bit too early so it won't be a great test. Despite all that, the Darkstar and HSO Blue Dream are impressing me and the Blue Dream was nowhere near that HPS it was under 50W of RWB LED.

One thing I can say for sure, the 3000K LED buds are frosting up much more and much earlier than the HPS buds.

Honestly I would prefer the 2700K because it has a redder SPD but the difference in efficiency is too great for me to accept. I figure the 2700K 3050 may need deep blue and deep red supplementation anyway so I went for the 3000K 3070. The 3000Ks will get some benefit from the RWB modules and vice versa so I am trying to stagger them somewhat.

 

[PetFlora]

2700K led A19 bulbs are available from a company in CA. I saw their adv in Electronic Products magazine. I contacted them, can be purchased direct, but I didn't buy any

Since we may only need < 10%, this might be a good interim option

Cannabare your buds look like they are coming along nice!



I have been adding more and more 3000K so we will definitely get a good feel how they are doing on the next batch. The newest batch is 100% LED and they are varieties I am familiar with so that will be a good test. The 5.5 week batch (bottom of page 20) is now 100% LED but it had a few weeks of HPS exposure, has new varieties I have never run before, and some of them faded a bit too early so it won't be a great test. Despite all that, the Darkstar and HSO Blue Dream are impressing me and the Blue Dream was nowhere near that HPS it was under 50W of RWB LED.

One thing I can say for sure, the 3000K LED buds are frosting up much more and much earlier than the HPS buds.

Honestly I would prefer the 2700K because it has a redder SPD but the difference in efficiency is too great for me to accept. I figure the 2700K 3050 may need deep blue and deep red supplementation anyway so I went for the 3000K 3070. The 3000Ks will get some benefit from the RWB modules and vice versa so I am trying to stagger them somewhat.

 

[SupraSPL]

Surprisingly all the 2700K screw in led bulbs I have been seeing have been only 80lm/W which is similar to fluorescent versus the COB which can do 130lm/W. Did you find some better ones?

 

[PetFlora]

No, I gave up the hunt to pursue growing with 2 ufo 90s (3500/5000K), then once buds set, I added 3 @ ~ 16w x 4 ft @ 3000K tubes. That worked really well

AYK, this grow replacing the tubes with 2 cob/reflectors, in effect doubling the tube wattage + a dab of 660

Surprisingly all the 2700K screw in led bulbs I have been seeing have been only 80lm/W which is similar to fluorescent versus the COB which can do 130lm/W. Did you find some better ones?

 

[tallen]

Does anybody know how many umoles these COBs put out and how well they spread the light? Trying to figure out how far apart I'd need to space them to get the best mix of intensity and efficiency, and therefore How many to put over each tray.

 

[bondoman]

Surprisingly all the 2700K screw in led bulbs I have been seeing have been only 80lm/W which is similar to fluorescent versus the COB which can do 130lm/W. Did you find some better ones?

well comparing CFL to LED bulbs lumen to lumen, I think it's obvious that lumens don't tell the whole story as the LED blows the CFL out of the water.

 

[SupraSPL]

True, normally you cant compare bulbs by lumens but if they have a similar spectrum you can. CFL 2700K spectrum is not that far off from LED 2700K. Another thing that makes this comparison valid is that we know LED maximum lumens/watt for 2700K is about 325-330. So if you are only getting 80 lm/W the light is about 24% efficient, which is similar to the very best fluoro tubes but compares poorly to what LED is now capable of (48%). The reason the LED blows a CFL away is because its output is directional (although the glass diffuser removes that advantage to some extent).

-----------------------

Tallen I am afraid we might need Mr Flux for that one. I only focus on the radiometric efficiency % and spectrum and I go from there, but he knows how that information can be converted into photons (umoles) and I think it requires software and calculus that he has experience with. If you are planning on running at 1A the radiometric efficiency will be about 39% and here is the CXA 3000K curve


Once you get a umol figure (ppf) then there is the issue of the 115 degree output, reflector (or none) and distance to canopy (ppfd). Quite a process lol.

 

[Observe & Report]

Tallen, look at the numbers in Flux's https://www.rollitup.org/led-other-lighting/743645-cree-cxa-analysis.html thread. Find the table for CXA or Vero in the color temperature you want. Divide the "photon flux" number (in umol/s) by the "power in" number (in Watts) to get umol/s/W for that chip. Multiply that number by how many watts your chip will run at, found by looking up the Vf in the spec sheet with the current you're driving at. Multiply the result by how many COBs you have. That result is total PAR in uMol/m/s. Divide this by how many square meters you're covering and you'll get the _average_ PAR for any point under your array, assuming there is a perfectly reflective wall right on the edge of your coverage. Right under a COB and/or right in the middle of the array the intensity will be higher than average and it will be lower on the edges. How much variance will mostly depend on how spread out the cobs are, distance below the cobs, and also reflectors, reflector loss (including walls) etc...

You can adjust that number, which is based on the test current, by how much harder/softer you are running using the graph in the spec sheet that tells you how much brighter/dimmer the output is vs. current.

Example: three 4kK Vero 29's run at 1.9 Amps in a 0.4 square meter cab.

[B]4000K, 80 CRI
Power in              : 9.35 W
Luminous flux         : 1180 lumen
Efficacy              : 126 lumen/W
LER                   : 341 lumen/W
Radiometric efficiency: 37.0%
PAR efficiency        : 85.9%
Combined efficiency   : 31.7%
Radiant flux          : 3.46 W
Photon flux           : 16.65 uMol/s[/B]

Vero 29 Vf @ 1.9A = about 39V
1 COB wattage = 39V * 1.9A = 74.1W
flux per watt = 16.65umols / 9.35W = 1.78 uMol/s/W
Total flux = 74.1W * 1.78uMol/s/W * 3 COBs = 396 uMol/s of output from all COBs
Average flux = 396uMol/s / 0.4m2 = 990 uMol/m^-2/s^-1


Thanks again to Mr. Flux for computing these numbers!