DiY LED - Cree CXA3070

My question was related to a 20% reduction in output over such a short period of operation (only 60 days). This is hard for me to believe if the array was cooled properly and not driven too hard. I figured they may have damaged their LEDs by misuse. What were these studies? Was it a sales figure for someone selling panels with quality american arrays?

I thought it was implied that they were less efficient by the much lower price. I have no idea what the actual efficiency or spectrum 'quality' is. I'm sure you're right that it's less efficient than cree by far. I'm just scrutinizing the claims. I wouldn't expect CXA like performance from a "chinese" no brand array. but I also fail to see how they'd fall apart that easily if driven at 50% or so rated max current or less, like DIYers tend to.

FranJan said:

They're multi-chips and by design they aren't as efficient as COBs. And remember they need a big heatsink because heat just kills their output. They're bright but aren't all that for flowering from what I've seen. Plus those are probably blue LED with just a yellow phosphor which doesn't produce a very good spectrum but does produce a pretty intense "white". Do you have any data on that chip? You may have to supplement with some red. I could be wrong, but without some kind of data it's a crapshoot whether it will work or not for you. But don't expect CREE like performance from that. At least not this month. Seems CREE and Lextar made a deal that according to some analysts was made so CREE can better compete with Chinese companies and their prices.
http://finance.yahoo.com/news/lextar-cree-inc-announce-led-034500204.html

Click to expand...

"but i noticed a lumen drop since day one... now id say theres atleast a 10% to 20% lumen loss due maybe to heat or low quality not sure but i doubt its the heat... most likely the bin"

that is a quote from a Chinese build and they used a lux meter

My 1st grow I ran 6x 30W chinese Arrays (4x ww, 2x R) and 40x red Epistar 3W near max.
2nd grow, 160W, i reused them all @ half current and added 4x 100W Arrays @ 15W,
6 from seed, soil, 3/6 unknown strains, 2 Plants yielding far less (one of them other pheno of same strain, sensi Skunk#1) than the others -guess genetics
gpw went from 0,7 to 1,12 -learning curve and things (even with most of the lights having lost 20% output prior to 2nd grow?)

So if these cheapos deliver less 30% PAR/Lumens per se compared to CXA and i resued them and 75% of my leds already lost 20% output.
Does that mean on 3070 i d reach like 2 gpw (leaving those 2 super low yielding ones out of the equation)?
Sounds too good to be true

So china's looking to have a bright future too?

Id get 3x 3070 now for the same Price. Regarding heatsinks, fans and other small things those 3 would be cheaper.
My space is 54x 74cm. Leaving the 40x 3W leds out, thats 10 Arrays covering that area.
With them nicely spread, I can put the lights really close to the plants -less light gets absorbed/reflected by/away from the walls.
I think thats an advantage over 3 cobs, that need to be further away to cover that area equally. Reflector could help on that.
On the other hand those 3070 would put out more light.
I think its light Distribution vs total photon Output in my case, idk.
If i wanted something that simply works Id just get 3070, but for me it was about building something and having fun while doing it.

I just remembered that there are larger luminus COBs than the CXM18, the biggest is CXM27 so that would be a more fair comparison with the CXA3070.

http://www.digikey.com/product-detail/en/CXM-27-30-80-36-AA00-F2-3/1214-1208-ND/4729071

So if you run it medium to hard, the CXM27 3000K 80 CRI should be on par with the CXA3070 3000K Z4 bin in terms of cost and efficiency.

So if we follow that logic, the CXM27 2700K would be equivalent to a decent bin of cxa3070 2700K (which never seems to become available). But the same problem with the CXM27, the 2700K is not currently available.
http://www.digikey.com/product-detail/en/CXM-27-27-80-36-AA00-F1-3/CXM-27-27-80-36-AA00-F1-3-ND/4755691

AVnet claims to have them if you buy 20
http://www.digikey.com/product-detail/en/CXM-18-27-80-36-AA00-F2-3/1214-1204-ND/4729067

So the next best option currently available for 2700K is the CXM18 2700K 80cri
http://www.digikey.com/product-detail/en/CXM-18-27-80-36-AA00-F2-3/1214-1204-ND/4729067

Does anyone have any data breakdowns on the Luxeon COB's? They are readily available and look to be pretty close to the Vero and the CXA's in performance

Here's an Amnesia Haze (hybrid) cola that still survives on the plant that went timber. There are a few others too. Even the immature buds from the broken branches were surprisingly decent smoke.

First time trying a couple plants in pure coco coir with synthetic nutes and I have to say I am impressed so far. We will see how the smoke quality is before we reach a verdict but I have already grown this cut in organic soil so I know how that turned out to compare.

Anyway 3L pot of cocotek coir being watered by hand drain to waste 2x a day now with just GH Floranova Bloom @ ~1.7ec (not the whole time but thats what I have slowly worked up to as of right now). It is a Sugar Kush x Chemmy Jones clone at day 35 12/12. These COBs plus coco and synthetics can really do some damage in a small space. This is growing in about a 12"x15" space in the corner of my tent, it just looks bigger since its flopping outward without the other plants and tent walls holding it up.

Wow how is it surviving CW? did that branch not fully snap or did you put a fresh cut on it? It looks like it is going to firm up nice and lots of sticky

BT, dang 3L that looks awesome especially for 35 days. Nice work cant wait to see how she finishes!

I am about to freeze the design for powering my 1 sqft mother & vegging area.

Driver, thanks to Supra:

SupraSPL said:

$3 driver ... 88-89% efficiency...it runs at about 290mA
....
This driver would be perfect for a string of Vero 10s. The 5000K Vero 10s should make good vegging lights and they run at 41.5% efficient at 290mA and only cost about $2.19/PAR W.

Click to expand...

For Cree CXA, I would have choosen the next larger COB size to increase the efficiency. But for Vero series, I am not sure if that pays out.

SupraSPL said:

Bicit, the Vero behaves differently than the CXA. The Vero 10 is just as efficient as the Vero 29 if both are driven at say 50% of max current as SDS pointed out....

Click to expand...

Fact is: for a given Vero COB, efficiency is not increasing much as a function of decreasing current, like that is the case for CXA. Seemed at first strange and confusing to me. However, that does n o t imply that efficiency is not increasing when choosing a larger COB while keeping the current fixed. I thought on that matter for a while:

For Vero, the "Thermal Resistance of Package" values are
1.49°C/W JC (Vero 10)
0.87°C/W JC (Vero 13)
0.55°C/W JC (Vero 18 )
0.26°C/W JC (Vero 29)
Note that Vero 13 and 18 have the same module diameter but thermal resistance differs significantly. Seems to correlate with the size of the a l u m i n u m substrate area within the COB. (see AN30-Thermal-Management-of-Vero-LED-Modules.pdf, Figure 2). BTW for the grinders and polishers, ever thought on grinding not until flat but until aluminum becomes visible? Reduces the number of thermal interfaces by 1 and might be worth the effort

For CXA (3070), I remind someone mentioned values around 0.3 °C/W JC. (Is there a source? I would like to compare these values for different CXA sizes). Furthermore, I noticed that Cree refers to a "c e r a m i c substrate", ?silicon-carbide?, ?aluminum-nitride?. For metals ( Vero´s aluminum), thermal conductivity decreases strongly with increasing temperature (Wiedemann-Franz law: linear with electrical resistance), heat transfer via electrons gets more and more restricted. Temperature dependance of heat conductivity for covalent materials like ceramics (CXA) or diamond is significantly lower compared to metals. That would explain why a chosen CXA´s efficiency gain is more pronounced than for a Vero when both are powered more softly.


So back to the original topic:
Supra, can you supply numbers for Vero 10, 13 (and 18 ) for 290 or 350 mA driving current? From what I think I understood, I would expect that for a chosen current, efficiency should increase when using the next larger COB (due to the significant difference of the thermal resistance of package).

I'm not LED guru but my guess would be Cree emitters are a bit smaller and/or doped differently than Vero thus getting better efficiency but at a cost of more current droop.

SupraSPL said:

Wow how is it surviving CW? did that branch not fully snap or did you put a fresh cut on it? It looks like it is going to firm up nice and lots of sticky

Click to expand...

All but one of the surviving branches is from below the break pretty much. I'm flushing her now, should still be some top shelf on there. Definitely one worth cloning.

^^ Cloning... at this stage?

Do tell

I am revegging a SIlver Skunk plant (Blueberry Cheesecake) that I had harvested, and basically put outside to die. Daily rains kept it alive, so I put it back under the light (Aug 02) and damn she is now a mini-monster.

Will get a few nice clones from her and flower the rest

PetFlora said:

^^ Cloning... at this stage?

Do tell

I am revegging a SIlver Skunk plant (Blueberry Cheesecake) that I had harvested, and basically put outside to die. Daily rains kept it alive, so I put it back under the light (Aug 02) and damn she is now a mini-monster.

Will get a few nice clones from her and flower the rest

View attachment 3243827

Click to expand...

Lower branches that are shaded and not fully developed can still be cloned, even towards the end of flowering. I carefully pick the budlets off after cutting. I try to leave something on the plant to regrow too, but that doesn't always work.

The Haze plant in the last picture has plenty of undergrowth to regrow on it, but I want to take clones so I can get rid of the plants and clear the flower room.

Great thread! Just found it the other day and am 30 pages in or so. Looking to build a diy setup to reduce heat and get rid of my ventilation from my 1k and 600 watt hps setup.
I would like to build the equivalent of 2 1k hps lights with led, all though not all at once. How many 3070 3k would i need to accomplish this? Also i was thinking of adding maybe a few 5k cobs, is their any benefit? i see you guys discussing the amount of blue ( 19% i think it was for 5k ) vs the 3k cobs at 11 or so percent. keep up the good work!

@Growerx2014 Good insights and good questions.
For the 3000K
Vero29 350mA 39.5% (from vF curve rather than chart)
Vero18 350mA 39.8%
Vero13 350mA 35.6%
Vero10 350mA 33.3%

It is interesting that for the first "current category" in their charts, the efficiency is about the same:

Vero29 700mA 39.3%
Vero18 350mA 39.8%
Vero13 165mA 39.3%
Vero10 115mA 39%

@jay5coat welcome to the thread. to replace a 2000HPS I would suggest about 1000W of CXA3070 (20@ 52W). You could use the CXA3590 (8@ 122W) but I think they are only available in a 50 pack at the moment.

The 3000K ended up having about 14% blue based on SDS work and I have had no problem running them on their own so for simplicity sake 3000K works very well on its own and that avoids the issues of spread and color mixing. That said, you could add some 3500K 4000K or 5000K to experiment with. Or you could simply add royal blue luxeons, especially easy if you went with the CXA3590s because of their lower current.

Most recent batch came in over 1.36gr/W, will update when the last lady finishes drying. Averaged dissipation was 611W of CXA3070 3000K AB passive cooled, running at 700-900mA with a few red/blue strings of luxeons mixed in. The catch is, it took 9 weeks to finish this time. Maybe because of the slightly higher summer temps or maybe because of the 3000K spectrum (less deep red).

Reserva Privada - Kandy Kush - 75.97
Cali Connection - Purple Chem - 44.2
TGA - Qleaner - 85.47
Sin City - Alien Nightmare - 47.06
Sin City - Nightmare OG - 82.33
Sin City - Rappers Delight - 80.71
Sin City - SourJefe - 92.42
Reservoir - SSSDH (Shire cut) - 56 (estimated)
TH Seeds - Sage n Sour - 68.4
TH Seeds - Darkstar - 38.7
DNA - La Chocolat - 80.12
DNA - Stacked Kush - 80.62

For the ladies that came in with low numbers it was not necessarily their fault. Squeezing 12 under 600W was pushing my luck, so some of them received a lot more light than others. Ideally I would use 10 ladies for that wattage, or for 12 I should add another pair of 50W modules. Also, some were vegged up bigger than others. What I am really interested in is the overall yield capabilities of the lamp and searching the varieties for the best quality. They alll came out awesome but the very best were Kandy Kush, Qleaner, Darkstar, LA Chocolat and Sage N Sour (stuff that was already selected for the most part).

As I have notoriously admitted, my average yield from HPS was about .4gr/W over a period of several years. So 1.36 is quadrupling efficiency and the COBS can produce a much higher overall yield from the same space. I would have been happy with 1g/W so I am blown away by the efficiency.

This was my first all LED summer and the heat was no problem. When I was running 1200 HPS flip flop I had to crank my fans and crank the AC. During heat waves I would shut down half of the lights to try and keep canopy temps decent. I still use the AC but it is mostly for dehumidification.