DiY LED - Cree CXA3070

When you say 85c and 50c are you referring to their normal operating temperatures or max temperature? And if so what difference does that make?[/QUOTE]

At 2.8A I suspect it would be impossible to achieve a Tj of 50C using ambient air which will likely be about 25C. That is why I used Tj of 85C for the 2.8A example. Tj makes a very significant difference in efficiency.

tightpockt said:

This is what I'm after also. I want to switch to 3x3 one plant scrog. I'm tired of battling the heat and the noisy ventilation. if I happen to save on electricity that would be a nice bonus too...in short I don't want a 600 watt HPS if I can avoid it.

so you're saying I would need nine of these to accomplish that goal? where can I save some cost? Are CPU coolers cheaper? How about those liquid cooled ones they use for high-end gaming machines? Is that feasible?

Click to expand...

If you run 3000K CXA3070 Z4 at 1.4A, Id estimate you could get the job done with 300W or less (6) COBs. Here is the cost breakdown
(6) CXA3070 Z4 $275
(6) 1.4A drivers $75
(6) 92mm fan/heatsink combo $78
$428 + misc for connectors wiring solder thermal paste

Four 3070's at 2.8A puts out the same amount of light as six running at 1.4A, about 40k lm according to SupraSPLs numbers above. Using those same coolers and some $28 100W drivers I found in a quick search, 4 3070's costs only $345 to put together, but the cost per year with 90% driver eff. (hah) at 12 hours per day and $0.15/kWh is $332 vs $228 for the six pack at half current! That means you'll pay off the difference in like 10 months.

...of course, that assumes the 3070 at 2.8A is 85C while the 1.4A ones are at 50C on the same cooler. Not sure if that's reasonable.

We have to remember that it is not only the internal thermal resistance at play, it is also the interface between the COB and the heatsink as well as the thermal resistance of the heatsink itself. If ambient is 25C and the COB is dissipating 115W, I expect a Tj of 85C or above.

Edit: I think I misunderstood what you meant O&R. You are suggesting that we would end up with much higher Tj at 2.8A? The more I think about it, Tj might be quite a lot higher than 85C and the fan would have to be really cranking.

Do you have a link to those drivers?

bicit said:

Well I'm considering running the R/B emitters as supplement to some white cobs. Figured the cree whites wouldn't really be needed with the cobs providing ample white light.

Click to expand...

SupraSPL said:

Here it is all put together

32W module:
(4) Cree XM-L2 7a3 3150K T4 (aliexpress)
(2) Luxeon ES royal blue M4R (Steves LEDs)
(8 ) Luxeon ES deep red EX6 (Steves LEDs)
(4) Cree XPE 630nm red P3 (Cutter or LEDGroupbuy)

650mA driver

To answer your original question bicit, if you went with a COB for the whites you would be doubling the wattage of the whites and increasing the efficiency. Based on that you could put something like this on the red/blue string:
(3) Luxeon ES deep blue M4R (Steves LEDs)
(12 ) Luxeon ES deep red EX6 (Steves LEDs)
(4) Cree XPE 630nm red P3 (Cutter or LEDGroupbuy)

That would give you a vF of about 46.

Click to expand...

Hey supra I was looking at an older thread about RWB panels and was thinking was this 3 deep blue, 12 deep red, 4 630nm per cob and at what amperage was the cob assumed to be running at for this R&B supplementation?

Im sure you know by now I am going to be running my 2 3070 3000k at 1400ma each and I plan to run just the 3000k cobs the first run or two and probably add the blue and red to see how much of a difference I notice.

Would you recommend the 3 blue, 12 red, 4 red on the 650ma fasttech driver or would I be looking at adding even more red and blue than that? Also wondering if adding a small amount (2-4pcs.) of the xm-l2 u2 4300-4500ks would be useful or if the 3000k cob will do well enough alone? If they arent really going to make as much of a difference as the reds and blue, Id rather spend the money and space on the string on them.

Depending on how many leds I needed to string together to run on one driver for the r&b supplementation for the cobs, I was originally thinking something like 1 deep blue and 4 deep reds both from steves per cob but seeing this post quoted above seems like that wouldnt be enough r&b supplementation to make that much of a difference.

Good questions bueno time. I suspect that all of those scenarios would would great. So when it comes down to it, I say do whatever is the most convenient for your setup. In my case I have 2 COBs and a 650mA R/B string on a large heatsink with active cooling. I could run the COBs at 350mA 650mA 700mA 900mA 1050mA 1200mA or 1400mA depending on which driver I swap in. I have been running the COBs at 900mA, so overall I end up with 15% blue (a nice boost up from 10%) and add a nice 660nm boost.

The R/B string is 3 Luxeon ES royal blue, 3 Cree XPE 630nm reds, 5 Luxeon ES deep red, 5 Oslon SSL80 hyper red. I am using them in that ratio because that is what I have on hand but you could skip the 630nm reds, use more blues, etc.

You could use neutral XML2 if you have them on hand, although I tend to think of neutral and cool whites as better used for vegging stage because of their strong output in the blue/green/yellow range and the COBs are more efficient and cheaper, so they are more suited to the "white" task.

SupraSPL said:

Good questions bueno time. I suspect that all of those scenarios would would great. So when it comes down to it, I say do whatever is the most convenient for your setup. In my case I have 2 COBs and a 650mA R/B string on a large heatsink with active cooling. I could run the COBs at 350mA 650mA 700mA 900mA 1050mA 1200mA or 1400mA depending on which driver I swap in. I have been running the COBs at 900mA, so overall I end up with 15% blue (a nice boost up from 10%) and add a nice 660nm boost.

The R/B string is 3 Luxeon ES royal blue, 3 Cree XPE 630nm reds, 5 Luxeon ES deep red, 5 Oslon SSL80 hyper red. I am using them in that ratio because that is what I have on hand but you could skip the 630nm reds, use more blues, etc.

You could use neutral XML2 if you have them on hand, although I tend to think of neutral and cool whites as better used for vegging stage because of their strong output in the blue/green/yellow range and the COBs are more efficient and cheaper, so they are more suited to the "white" task.

Click to expand...

Thanks for the quick reply. Yeah, I hear you on all points. I will skip the NW XML2s for flower. Im thinking about using a few NW XML2 and run them soft on passive heatsink bars for a veg light when I get to that point. Going to focus on the flowering setup first for sure though.

I think I read you or someone else knowledgeable in LEDs say that the CXA 3000k has a pretty decent enough amount of 630nm in it that doesnt really need much if any supplementation but it lacks some in 660nm. Also could use a small boost in blue. So when I get around to it, Im probably going to go with passive heatsinks for the R/B string on a 650ma fasttech driver and build a R/B string similar to yours.

The cree xp-e p3 are sold out on ledgroupbuy so I might skip those unless by the time I am sourcing the parts if I can find some elsewhere and keep shipping to a minimum, since I would only be using a few of them shipping might kill the cost effectiveness of adding them.

edit: ledgroupbuy has solderless ones in stock

I think I might have found a cheap source for the xp-e p3, just sent them a message. I will post a link up if it turns out good. If so I will be throwing in some of those on the string.

Might try 4 Luxeon ES royal blue, 4 Cree XPE 630nm red, and 8 Luxeon ES deep red. Sounds decent. I like the idea of a 4/4/8 combo I think, so I can split them evenly on 2-4 separate passive heatsinks and be able to spread the red and blue around better. Im thinking 4 passive heatsink bars with 1 Deep Red on each end, a 630 and a royal blue in the middle of each bar. Or two longer bars with 2/2/4 evenly spaced along the length. I am picturing this all in my head right now and its lovely lol.

Time to fire up sketchup...

I agree with that description of the CXA 3000K. They work great on their own, but you can tweak the spectrum if you like. Here is another source for cheap XPE P3s. $2.60 on stars + shipping. They have a bunch of cool DIY LED stuff like fast blow fuses and dimmable drivers.

Got the final numbers on the last batch which was HPS/LED and was the very last of the HPS and the ladies were all brand new, completely unselected. Turned out .67 gr/W, a huge improvement for me and nug quality is all I could ask for. Next batch will be 100% LED and all selected.

This is the early bird Ace of Spades, 3000K COB nugs

SupraSPL said:

I agree with that description of the CXA 3000K. They work great on their own, but you can tweak the spectrum if you like. Here is another source for cheap XPE P3s. $2.60 on stars + shipping. They have a bunch of cool DIY LED stuff like fast blow fuses and dimmable drivers.

Click to expand...

Nice, thats the link I was talking about. Sent them a message just a while ago asking if the optional 20mm pcb boards pre-solder option meant that the led chips would be soldered on the stars. Wanted to make sure nothing was lost in translation from Chinese to English since I wouldnt want to have to deal with soldering tiny chips onto boards if I can avoid it.

Good job on the .67gpw, the Ace of Spades looks nice. Your GPW should go up even more now that it will be all LED and no HPS. Should be good.

Thanks for all the help and knowledge everyone especially Supra, this isnt my main mj forum but you guys have the best knowledge base and documentation for LEDs and DIY LEDs by far from what I have seen.

Thanks BT. I am expecting close to 1 g/W which is about double what I was getting from HPS when bulbs were new. Some of them fade too early on me I could probably increase yields with a top dressing.

This is a sample I ordered from AC-RC

I had a little problem with stretching when I tried just using 3000ks now I am running 1 50 watt 3000k with 1 50 watt 6500k and 3 10watt blues plus 2 10watt 660 nm reds and seeing much better results in veg.the test plant was started under T5s and both the T5s and LED will go into flowering in about three weeks. same strain grown in soil. everything except lights identical. it will be interesting to see how these leds compare with my T5 aquarium lights( 3 650nm 3 6500k). (same power and output).

Hi supra was wondering if anybody seen this 1900k sunset led ???
http://ac-rc.net/catalog/product_info.php?products_id=62

★★★kushed_out★★★

Maurice I should qualify what I said, I meant 3000K was OK on its own for flowering. For veg I would recommend a higher blue percentage. I use 25-30% blue.

Interesting chip (the 1900K) but I would be wary of it because there are very few LEDs on the market that can approach the efficiency of Cree. That chip is likely only half as efficient or less.

SupraSPL said:

Maurice I should qualify what I said, I meant 3000K was OK on its own for flowering. For veg I would recommend a higher blue percentage. I use 25-30% blue.

Click to expand...

I have not tried flowering with them yet so can not comment on that but they definitely need more blue for sprouting and vegging. when I start flowering I will switch to 2 3000k 50 watt and 1 6500 k 50watt.I am a member of the "prof" club and grow with T5 aquarium bulbs and am just starting to experiment with leds and they may flower with just 3000ks but I am use to running a little more blue and will reduce it gradually. the combination I am using for vegging provides equal amounts of red and blue and slightly less green. a balance full spectrum. but this is your area of expertise.I am shifting spectrums slowly to try and stay out of trouble.I do not pretend to be an expert. I am very new to leds and am proceeding carefully. in your post I did not see any comments about increasing the amount of blue for vegging until now. but you are the expert on this and I am a newbie to leds.

DiyKindaGuy said:

Hi supra was wondering if anybody seen this 1900k sunset led ???
http://ac-rc.net/catalog/product_info.php?products_id=62

★★★kushed_out★★★

Click to expand...

that 1900k is only 68 lumens per watt. I have not seen anything hotter than 3000 k that would produce 100 LM/watt.or better to bad that looks like it would produce a lot of red.

Bueno Time said:

Thanks for the quick reply. Yeah, I hear you on all points. I will skip the NW XML2s for flower. Im thinking about using a few NW XML2 and run them soft on passive heatsink bars for a veg light when I get to that point. Going to focus on the flowering setup first for sure though.

Click to expand...

Hey Bueno I am currently using the 4200k XM-L2's as part of my veg cabinet. Running at 1500mA. Amazing growth i must say. (Been under them about a week)

I have also got what you mentioned in my flowering cabinet. Running the 4200k at 700mA on passive heatsink bars as a supplemental lighting to my main COB's and multi-chips.







Just been sprayed....



Just re-veging a couple of my outdoor buds