DiY LED - Cree CXA3070

SupraSPL said:

Very nice heatsinks!

Click to expand...

i tried to give a generic search instead of a specific listing so it would work in the future - the one I grabbed was one of the copper slabs. I'm thinking that these will likely pop up cheap as recycled/harvest parts for awhile. CPU cooling tech is constantly changing so you gotta find the stuff used for the last gen that is heading out the door to find the cheap buys.

Supra, sorry if this is too off topic, but I like you guys and feel like sharing one of my secrets for getting good deals.

Here's an ebay buyer master tip: Let's say you want one of these heatsinks, the search I gave you for this is:
(771,lga771,dempsy,woodcrest) (heat sink,heatsink)

if I look at the results now, i realize I really should change it to:
(771,lga771,dempsy,woodcrest) (heat sink,heatsink) -fan

which would cut out the fan only results. Most of the good copper chunks are server CPUs which didn't have fans attached anyway.

Ok, next you use the search refining tools to change to Buy-It-Now only. Then you put in your price params (low/high). I usually do all of this by searching the recently complete auctions to see what things are going for. If you zero in on the price you'd love to have, you'll find that many times, in terms of current auctions, you'll have zero results. That's fine for this exercise.

I refined it a bit more here:
http://www.ebay.com/sch/i.html?_odkw=(771,lga771,dempsy,woodcrest) (heat sink,heatsink)&LH_BIN=1&_sop=15&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313.TR0.TRC0.H0.X(771,lga771,dempsy,woodcrest) (heat sink,heatsink) -fan&_nkw=(771,lga771,dempsy,woodcrest) (heat sink,heatsink) -fan&_sacat=0&_udlo=4&_udhi=11

when you get it down to what you like, you append the following to the URL: "&_mPrRngCbx=1&_rdc=1&_rss=1" (without quotes)

then you get:
http://www.ebay.com/sch/i.html?_odkw=(771,lga771,dempsy,woodcrest) (heat sink,heatsink)&LH_BIN=1&_sop=15&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313.TR0.TRC0.H0.X(771,lga771,dempsy,woodcrest) (heat sink,heatsink) -fan&_nkw=(771,lga771,dempsy,woodcrest) (heat sink,heatsink) -fan&_sacat=0&_udlo=4&_udhi=11&_mPrRngCbx=1&_rdc=1&_rss=1

and now, you have a stripped down RSS feed of exactly what you want.....shove that into an RSS monitor service/script, have it email/txt you when something comes up and you can snag the buyer-priced-it-too-low-with-buy-it-now-price item.

How are they doing?

Thanks for the RSS tip caretaker. I use the same technique to try and find good stuff that hasnt been snapped up yet. Scored a nice box of CPU heatsinks (15 pounds) for $35 shipped. Normally I pay $150 for that much heatsink.

Hey Canna how is your 2700K order looking?

Ladies are doing good, 4 weeks in. Unfortunately I got early fading on the bigger ones. They depleted too much nutrient during Veg. I have increased the fertilizer in the upcoming batches hopefully that will increase yield. The smaller ladies that were not depleted are looking really nice and green though.

Kandy Kush on the left, will fatten up later. Ace of Spades in the center, Blue Widow on the right.


Close up of the Blue Widow, ignore the nug on the bottom that is a tiny Kosher Kush that I am letting go, ran it for the last time in the shade and it still made nice nugs. Flavor is delicious but stems are brittle and I have better OGs so OK about letting this one go.

This batch is almost to 9 weeks. Ready to go into the dark room for a few days. The Dark Star, Blue Dream, OG18 and Stacked Kush look great. The CC Original Sour D, LA Chocolat and Green Poison came out decent enough but I doubt I will run them again.

Dark Star on the left, Blue Dream center rear, Stacked on the right.


LA Chocolat center left, CC OSD in the center, Green Poison front right, stacked kush rear right, Blue Dream rear center, Darkstar rear left

Veg room needs an upgrade. I plan on cutting the big heatsinks in half and updating them with XTEs I have lying around, XPE reds and Luxeon ES deep reds/royal blues. Currently 80W soon to be 60W.


Upgraded seedling box from a 2W XML2 to a 4W RWB.

SupraSPL said:

Scored a nice box of CPU heatsinks (15 pounds) for $35 shipped. Normally I pay $150 for that much heatsink.

View attachment 3027963

Click to expand...

lol at DIY tragic buying heatsinks by the pound. Nah, good on you - I'm just envious at what you're going to end up growing with these . Cobs are such a perfect match for CPU coolers its almost obscene.

Have you considered getting a thermocouple to improve your raw data? It must be tempting to know what the temp your emitter resin is seeing in real time...


Beautiful weight of plant in this photo - clearly you're doing more than alright.


Final question Supra, because I'm interested in your thoughts. Assume we have a flower panel in 3000k giving a flowering plant sufficient illumination - what supplementary lighting should we consider?
Just trying to get a basic handle on the question of "where to next?"

TNX Smokey I have a new Multimeter on order that has a temperature setting. It will be interesting to see how close the infrared reading of the heatsink next to the COB compares to the thermocouple test point.

Funny you ask that question, I just finished this today. There are 3 XPE reds P3 bin, 5 Luxeon ES deep reds EX6 bin, 5 Oslon SSL80 deep reds 3T bin, 3 Luxeon ES royal blue M4R bin all on one string, will run it at 650mA. The COBs will be running at 900mA (32W each) so total dissipation for this module will be 89 Watts and efficiency will be 41.4%. Total blue output will be 15% (PAR Watts). There is a 140mm fan on the back of the heatsink it will be running at 8.4V.


Of course the spacing could have been better. I was modifying a module that already had 4 COBs on it so I just worked with what space I had.

Nice new light!

All the multiple sources work great..beautiful garden. I'm limited to a small tent so one fixture will do it for me.

I'm slowly getting my parts in for my 3070 build. Hope all goes well..cross my fingers and be careful. Decided on 2 inventronics drivers at 200w each. Enough volts to run up to 5 3070 at 1a or 2 3590 when they become available. Tried to plan around that upgrade in the future.

That sounds like a badass build cant wait to see it!

Positivity said:

Nice new light!

All the multiple sources work great..beautiful garden. I'm limited to a small tent so one fixture will do it for me.

I'm slowly getting my parts in for my 3070 build. Hope all goes well..cross my fingers and be careful. Decided on 2 inventronics drivers at 200w each. Enough volts to run up to 5 3070 at 1a or 2 3590 when they become available. Tried to plan around that upgrade in the future.

Click to expand...

Hey positivity, SDS posted a link, but now I can't find it, but it discussed the newest Cree CXA's and the specs are out on some....this is for Supra too, as it would be interesting to see hey they rate out....a few are for sale, but the only one I found in single quantity was the 2590 @$61. !!!!

Big #'s from smaller dies with "supposedly" more efficiency....Moore's law is hitting this form of diode quickly!

http://www.cree.com/LED-Components-and-Modules/Products/XLamp/Arrays-NonDirectional/XLamp-CXA1850
http://www.cree.com/LED-Components-and-Modules/Products/XLamp/Arrays-NonDirectional/XLamp-CXA2590
[there were a few more, but the 2 most interesting...]

The 2590 puts out similar or greater numbers than the 3070 @ 1a less power, smaller die size too @ 17mm to 24-ish mm, while being nearly as powerful [intensity wise?] but hopefully more efficient than the 3590's.

Abiqua said:

Hey positivity, SDS posted a link, but now I can't find it, but it discussed the newest Cree CXA's and the specs are out on some....this is for Supra too, as it would be interesting to see hey they rate out....a few are for sale, but the only one I found in single quantity was the 2590 @$61. !!!!

Click to expand...

What Cree is doing with the CXA-HD line is cramming the dies closes together, this makes it easier to create tight beams since the COB becomes more point-source like. It used to be Philips that had the highest lumen density in COBs but on the Philips website this claim has now been dropped.

The performance of the HD differs from the older CXA but not really for the better. Here are the output per Watt graphs that you all have come to expect, plus comparisons with the older CXA



The only HD part worth noting is the 3000K high CRI one. It is actually producing a bit more (usable) photons than the normal 3000K.

Here are the numbers:

Flux, what about the 2700 models? Are they not worth it?!

RedCarpetMatches said:

Flux, what about the 2700 models? Are they not worth it?!

Click to expand...

Cree does not give spectral curves for the 2700K unfortunately. There is a large lumen drop when going from 3000K to 2700K which tells us that the 2700K is not very efficient. Bbspills is doing very well with the 2700K though, he just pulled 100gr from what I believe is 100W. Judging from his grow there seems to be enough blue in it too. Although personally I would prefer a bit more blue than even the 3000K has, but this is where things stop being scientific and start to become more of a intuition/belief/religion.

Positivity said:

I'm slowly getting my parts in for my 3070 build. Hope all goes well..cross my fingers and be careful. Decided on 2 inventronics drivers at 200w each. Enough volts to run up to 5 3070 at 1a or 2 3590 when they become available. Tried to plan around that upgrade in the future.

Click to expand...

Do you have a link for the driver. I have decided on 4 x 3070's, but trying to decide to get either 4 x small drivers, 2x drivers to power a pair, or one driver to power the lot.

Has anyone researched this, which option is more cost friendly.

MrFlux said:

Cree does not give spectral curves for the 2700K unfortunately. There is a large lumen drop when going from 3000K to 2700K which tells us that the 2700K is not very efficient. Bbspills is doing very well with the 2700K though, he just pulled 100gr from what I believe is 100W. Judging from his grow there seems to be enough blue in it too. Although personally I would prefer a bit more blue than even the 3000K has, but this is where things stop being scientific and start to become more of a intuition/belief/religion.

Click to expand...

Love the 5000 for veg! I would personally go with 2700s mostly...then to the 3000s for the last couple of weeks.

MrFlux said:

Cree does not give spectral curves for the 2700K unfortunately. There is a large lumen drop when going from 3000K to 2700K which tells us that the 2700K is not very efficient. Bbspills is doing very well with the 2700K though, he just pulled 100gr from what I believe is 100W. Judging from his grow there seems to be enough blue in it too. Although personally I would prefer a bit more blue than even the 3000K has, but this is where things stop being scientific and start to become more of a intuition/belief/religion.

Click to expand...

The large lumen drop ,while for sure it means Photometric efficiency decreasement ,does not necessarily means radiometric efficiency decreasement .
2700 K high cri leds usually combine one green and one red nitride phosphor.
But the most part of power of emitted light,is 'weighted ' towards the red part of the spectrum.
Comparing to all other white leds,it is absolutely normal for these ww 2700 leds to have lower lumen output .....
Anyway ,highest lumen output has the green light of ~555nm ,which has 683 lumens per watt of light ...
But that doesn't mean that is ideal for growing ,or even that we are referring to an efficient (radiometric-wise) led ...
Green leds of ~555 nm are just , the most efficient photometrically leds !
Also ww 2700 K leds do emit some power in IR,more than any other white pc led...
That decreases lumen ,as IR wls have not any luminosity to human vision ...

Still ,I do not understand why many-respectable- led growers ,(especially led growers ),are utilising photometric values ,
when talking about plants and light ...Plants do not have highest light sensitivity ( or absorption if you like ) in green region,like human eyes do ..
Lumens is a total alien and irrelevant measuring unit ,to plant growing ..It does not provide any useful information about a light source ,
regarding plant horticulture ..It tells us nothing at all,by on it's own ...
Only in one case and only then, when combined with a known spectra and some complex maths , it can provide some useful info ..
Other than that,null....
Lumens for Humans ,mWatts or umol/sec for plants ....
( Still that info ain't enough ,to judge the growing efficiency of a led ..
Light power distribution, diffusion, spectrum,total system output power,emission angle ,etc are some more parameters that should be considered,
maybe more than solely the x efficiency of a led ....

Cool white leds are much more efficient (radiometrically & photometrically ) than WW leds .
Still they can not flower mj ,in the way WW do...
Neither they do yield ,not even close , to the yields of WWs ...

Have the most efficient cool whites in a panel ....
Have the most un-efficient Warm whites in another ...
Do a test grow..
And you will see with your own eyes ,what I'm talking about ...

So...Designing a led grow light ...
A good grow light...
"Efficiency " stays between same type of leds ...
-This WW 2700 K or the other brand WW 2700 K ?
There is where efficiency ,should be accounted ...

There are way more important issues ,than led efficiency,that make a "powerful" led grow light ..
( This is what is going on with blue leds ...."More efficient than Whites are "
So what ?
Is this enough good reason to be part of a led growing light with mj underneath it ?
Orthis reason only,make blue leds appropriate for a grow -room / tent /cabinet ?
Or maybe they do belong really only to greenhouses and/or leafy veggies ? )

Blah-blah-blah....
A lonely voice sounding in the middle of desert ....
I know ...

Back to 2700 K leds ...

Truth is that they are indeed ,also less efficient radiometrically ,for quite a few reasons ,more or less known to us ...
( combined phosphors, re-excitation,long nanometer 'gap ' between excitation die and peak(s) of phosphor emission,,etc ... )
But not because of their spectra or Kelvin Temperature

And yes ,2700K WW led's blue emitted light ,is usually not enough for good vegging ..
.But is adequate -if not ideal,in some cases-for flowering ...

Personally I like getting the blue wls out of NW 5000 + WW 3000 K combo....

I like also more blue than the WW 3000K have...
But not that much and certainly not from blue diodes ...
Then there's also 'no room ' for violet-UVA-UVB ...
Blue leds if along with violet-UVA-UVB,will certainly impact dry flower mass ..
In a very bad manner ..
So providing the blue wls ?
That's the main role of the 'Sweet daylight' neutral white leds....

Edit : Mr Flux and Supra SPL ...

What do you think of the 'minimal' combo of NW 5000 K & WW 2700 K COBs ?

It is like providing blue 450 & red 660 ,but in a way 'softer' and 'fuller' manner ..
Ain't that so ?
(More or less ...)
It just hit me ...

Up till now the NW 5000K -WW 3000K -640/660 nm 'combo' has given me the best results,
by far from any other led combination ....

But WW 3000K +640/660 nm can not be replaced by WW 2700 K ?
(Ok, that spike in 640-660 nm will be missing ...But maybe for better ...)

Any ideas / thoughts /knowledge to share for that issue ?

Star,

Like your, perspective on thee---2700s

Agree (I do)

2700 + nothing else = ***better yield***