DiY LED - Cree CXA3070

The part number is CXA3070-0000-000N00Z230F. Here is the spec sheet. I would have preferred the CXA3590 but I have not found them available anywhere yet. They are both awesome though there was no sense waiting. Planning on running these at 720mA-740mA so they will dissipate about 25 watts each. Efficiency should be a minimum of 42%. Color temp is 2950K-3150K. 3500 lumens each or 139 lumens/watt. LER is ~328 (Thank you for that Mr Flux)



I will test the vF while they are cool and after the heatsink has warmed up and stabilized to try and estimate the Tj. I will also check ambient and heatsink temps with infrared to try and get some idea of the total thermal resistance.

These will be passively cooled just like the R/W/B modules with 105-110 sq cm /watt. After some testing I realized that the circulation fan was significantly reducing heatsink temp so that is a nice bonus. The hottest parts of the R/W/B heatsinks measure 25C with ambient at 21C so that is probably overkill. The sinks are 5453 sq cm each so I could probably run 55W and stay under Tj 50C even in the summer.

I might make some reflectors for the COBs. Nothing fancy just flat white cones better than nothing.

Sick!! Can't wait to see these in action

Heatsinks were already polished just had to drill them for the chains and solder the leads on. I used prolimatech PK2 thermal paste and 18 gauge 50V wire. The connections are bare crimp on sliders with heat shrink tubing. It makes it easy to swap drivers, modules and link strings without having to worry about polarity.

The lights are insanely bright and you can feel radiant heat if you put your hand in front of the beam. I tested them at 648mA 726mA and 1040mA. Each COB will get its own driver.

I got an infrared thermometer to play with today from the baby section of Walmart $20. It is designed to test the temperature of food so it has a decent range up to 140F and very accurate. The vF and mA were measured with a Fluke 115 which has very good resolution and accuracy.

I measured the vF of cold COBs and cold heatsink at 20C Tj/Tc it measured 34.6vF. Once the heatsink temp stabilized (which took 45 minutes) at average of 25C, the Vf measured 34.36. From the Cree graph I estimate the coefficient of voltage to be .01vF/C so that indicates a Tj of 44C. The COBs were dissipating a total of 47W and the heatsink is 5453 sq cm so that is 116 sq cm/W.

The heatsink was just sitting on the floor with its fins vertical so once they are hanging under the circulation fan it will be able to get rid of heat faster. In this test ambient was 20C but once they are in the flowering room ambient will be 24C.

Looks awesome Supra. Was wondering why two wires per terminal but that is just the mirror reflection... The photo's don't do justice to the massive scale that it must be.

I was surprised the Vf was that low. Based on the Cree graphs I was expecting 35.25. Hopefully that indicates that they are performing above the minimum specs.

You are right the photos do make them seem much smaller than in person and good call on the reflection I didn't notice that. The heatsinks are 10"X6"X3" each and weigh 6LBs.

Very nice supra........will be following with interest. Your right that these COBs make DIY easy for the masses with very good efficiency rates; surprising #'s TBH.

be safe

PetFlora said:

Along for the ride.

Yeah, why the double wires?

Click to expand...

at this LOL(lot of leds) in a cob it is a good way to use a current-mirror. http://en.wikipedia.org/wiki/Current_mirror
.
.
.
.
.
.
.









reefer madness

A confusing image, I polished the heatsink and the reflection makes it look like double wires.

That is bizarre reflections!

Good work on the CXA 3070!
I built a 2 - cxa 3050 array for a 1.8 sf filing flower cabinet.

I will be watching this one.

Can't wait to see how this pans out. I'm a huge fan of these.

Are cobs the future now SupraSPL? Might finally have some money to put towards a project I was thinking about for a long time. But if your using cobs now perhaps I should just go for them

So a newbie question. If you are using cobs does that mean you don't need a driver?

@rememberingnever That is a very good question. The RWB modules have about the same dissipation wattage and radiometric efficiency as the COB modules, 52W vs 47W, 39% vs 42% so I should be able to get some idea which is the better option after the next cycle. Even if the bud results are close, the COBs are cheaper, simpler, easier, more available, slightly more efficient, and have better "reach".

I know many DiYers would use multiple color temps because 3000K might only have 10% blue and I expect mixing them will work well. I will probably stagger the COB modules with the RWBs which are 19% blue. Also if I need more blue in the COBs I will add Luxeon ES deep blues because they are 55% efficient at 700mA and it would be very simple to add them to the string. I might add the blues anyway just to get more photons.

@GU420 yep COBs operate the same way as any other LED except they have a high vF. Drivers don't have to break the bank though. You can run a single CXA3070 on a $6 driver. I tested one and it worked great, ran at 648mA and 88% efficient with a perfect power factor. There is also a 1A $11 and 1.6A $11 both should work with the CXA3070. I don't have any to test but they are similar construction as the $6 driver and they all claim 87-88% efficient which turned out to be accurate.

I also tested the CXA3070 on this $6 12V DC driver hooked to a car battery, it worked great, ran at 646mA.

These should penetrate a 3' canopy just fine


Not that these pics can be directly compared but just for fun

hello,

I was wondering if you can help me out Supra...

when would one need to purchase a power supply? bbspills has a 3050 setup with 1.5a constant current driver hooked to each one wired straight to a power cord. I plan on doing 6 3050s with around 50 watts each for 300 total. Other than cost, is a power supply necessary if each led has a constant current driver on it?

thanks.

The most electrically efficient method is to use an AC-DC constant current driver like the ones I linked above, in which case you would not need a DC power supply. You could get the job done for $66, although I would recommend keeping one or two extras on hand. The only downside with those drivers is that you have to solder on the AC wires.

Sounds like an awesome project what color temps are you going with? If you haven't ordered yet, the 3070 is 12% more efficient than the 3050 @ 1.4A and I think they are only $4 more (32.3% vs 36.2%). May not sound like a lot but it is 12% more bud for free

thanks for clearing things up. I can solder no prob.

SupraSPL said:

The most electrically efficient method is to use an AC-DC constant current driver like the ones I linked above, in which case you would not need a DC power supply. You could get the job done for $66, although I would recommend keeping one or two extras on hand. The only downside with those drivers is that you have to solder on the AC wires.

Sounds like an awesome project what color temps are you going with? If you haven't ordered yet, the 3070 is 12% more efficient than the 3050 @ 1.4A and I think they are only $4 more (32.3% vs 36.2%). May not sound like a lot but it is 12% more bud for free

Click to expand...

i was going to go with 2700k for veg and flower. I was thinking of going 5k for a separate veg tent but I dont know yet.

Yea the 3070 isn't that much more at all.

I was planning on mounting each led to its own cpu heatsink and using a wall wart to power the fans. I haven't purchased anything yet.