DiY LED - Cree CXA3070

Yodaweed

Well-Known Member
Anyone know when the CXA3590 will be coming out? When I look it up its always out of stock or something...
 

mtnstream

Active Member
Org, get get a 6 channel pc fan controller and copper Tim pc Heatsinks placed on cheap aluminum threshold
rails with a through bolt/washer/wingnut for frame. You will need 3 pin pc extension cable to reach sink to
controller.

Good work Sup! Those 4-3050's cree cob chips are performing superbly in 1.8 sf...I am going to add 2-30 w
660 nm chips a sog AK landrace run.

Man this thread is moving; I hadn't looked in a few days and had to read like 6-7 pages - it was OK though :joint:... I was sad to hear about digikey and Canna's 2700k order but on the bright side at least I can save up some money for the next order. I just wanted to give a status update on my experiment with 4 3070s I've got everything working but I've had some issues with the fans - I think I blew them out with too many amps they are rated at .25mA and the wall wart I had them hooked up to were 12v 850mA; I think I just need to throw a resistor in there.
View attachment 3034272
Supra what kind of thermal tape are you using? the thermal paste and electrical tape just don't cut it.

Here is a picture of the drivers:
View attachment 3034275
I like these drivers because they have a volt adjustment so they are rated 48v/1.25A but I adjusted them all the way down and I'm getting 36v/860mA; this will give me a chance to play with the volts. I also like that it has connections for two leds and slides onto a din rail.

I still have tons to do before I will have a setup that I will feel comfortable flowering with but that is OK though :joint:

Thanks again to Supra and his dedication to this thread.
 

RedCarpetMatches

Well-Known Member
I want a 'clean' set up due to my OCD. Don't want bulky heat sinks and fans dangling to bump my head and concuss on :) Should I just go for spread with 3w, or go with the cobs? Does anyone know the best 2700k cobs atm? Sorry if I missed it, as I can't keep up the daily updates and upgrades.
 

SupraSPL

Well-Known Member
Do you even realize how inefficient that is...
88%-90% efficient depending on the driver. That includes the AC-DC conversion and the constant voltage to constant current conversion. Due to the laws of physics it is very hard to get more efficient than that, especially considering the give-away price for these drivers $10 each.

KNNA taught us way back that constant current drivers were the way to go, but in those days we were using multiple colors on one string with a variety of vFs so it was a necessity. Now with COBs we can get away with constant voltage, but I still prefer these drivers unless you know of a better option? If I recall you are using large benchtop power supplies, have you tested the overall efficiency of that system? I suspect they cost more than $10 each...

Of course I am not going to run them all in a pile like that picture, they will be mounted on a board and spread out as much as possible with plenty of air flow.
 

PICOGRAV

Well-Known Member
88%-90% efficient depending on the driver. That includes the AC-DC conversion and the constant voltage to constant current conversion. Due to the laws of physics it is very hard to get more efficient than that, especially considering the give-away price for these drivers $10 each.

KNNA taught us way back that constant current drivers were the way to go, but in those days we were using multiple colors on one string with a variety of vFs so it was a necessity. Now with COBs we can get away with constant voltage, but I still prefer these drivers unless you know of a better option? If I recall you are using large benchtop power supplies, have you tested the overall efficiency of that system? I suspect they cost more than $10 each...

Of course I am not going to run them all in a pile like that picture, they will be mounted on a board and spread out as much as possible with plenty of air flow.
By using multiple smaller cores you amplify the eddy losses developed inside of each of them. you can make a single CV source from a handful of components in a few hours that would drive all those LEDs...
 

SupraSPL

Well-Known Member
Deus, those lights are very convenient but it works out to only 64LM/W whereas we are getting up to 150lm/W in 3000K with the COBs so I highly encourage those who are willing to give DIY a shot.

If you are using those E27 bulbs, I highly recommend removing the plastic diffuser and point the LEDs at the canopy. It should boost you up to 80 or 90 lumens/W.
 

SupraSPL

Well-Known Member
By using multiple smaller cores you amplify the eddy losses developed inside of each of them. you can make a single CV source from a handful of components in a few hours that would drive all those LEDs...
Wouldn't that create a very high voltage in the grow room? I will have up to 20 COBs in each room. Currently my DC voltage is no higher than 36V.

Even if we build a custom power supply, would we be able to reach above 90% efficient?
 

PICOGRAV

Well-Known Member
Wouldn't that create a very high voltage in the grow room? I will have up to 20 COBs in each room. Currently my DC voltage is no higher than 36V.

Even if we build a custom power supply, would we be able to reach above 90% efficient?
If you know how to wind your own coils, yes. I thought this was a DIY thread....
 

SupraSPL

Well-Known Member
Pico, I am not an electrical engineer. If you have a solution that can beat this one, please by all means enlighten us :) How efficient are your power supplies? If I understand correctly, bench power supplies were not designed with efficiency in mind. Even if we made a custom supply, I am not convinced we could outperform these Chinese CC drivers.

My flower room #1 consumes 574W and 74 of those watts are driver losses. Driver cost was $200. If you were to offer me a 95% efficient driver, I would save 37W. Would that require more than $200? Would we need to create high voltage in the grow space in order to reach 95%?
 

Scotch089

Well-Known Member
...offer me a 95% efficient driver, I would save 37W...
Holy shit. Now that makes you step back see how much more each bit countss..

Sup-- you could have SO many stickys just breaking shit down. (how to test efficiencies, personally ha, I know you said it somewhere, when I get time I am gonna have to go back through..) You have done a hell of a job...... Thank you.

Ever think of keepping a list of alllll the most efficient stars/cobs for each color range/nms to current date? if you want, lol. Im sure it could be tedious. But if it were to be stickied, I am sure the other DIY Heads would chime in different perspectives on the details of the spectrum, driving em and the flux and.....yea.... that stuff. You guys blow my mind some times and I can never get enough.

Im strongly considering throwing down in the COB game, but NW or CWs to keep eff to the MAX. seriously. If I build a primary panel........ But I think mr flux had even mentioned it, or postivity? Fuck idk, but CW:R has proven ground with Commercial panels, possibly incorporate a FEW ww's, but after my custom spectrum of the A90 vs. my all 4250k A90 over an Apogee Meter----- (besides the specific color errors w/ the meter) that really opened my eyes to what the phosphor can do to what we get to the plants. Im pretty well convinced I want to use "the best of the best," of both worlds.. if I end up doin this.

Someone had mentioned sticking with 3000k instead of 2700k and I am all about that. Redcarpet has been running around asking bout 2700s and hope he knows bout all this...
 
Top