DiY LED - Cree CXA3070

E

Elibrium

Member
SupraSPL said:
Congrats on your COBs :) If they are AB bin, at 1.4A that is 43.3% efficient "typical" so 416W @ 43.3% = 180PAR W or 22.5 PAR W/ft². I use about 13.5 PAR W/ft² so you should have some serious intensity in that 2X4 :)
Click to expand...
Indeed! But I got the z4 from Ali.. However I'll upgrade to AB get the 20 pack after a few runs for sure. I think they are 39.3 ish % putting me at 20 PARw a sqft. Still very bright. I can't wait even payed the extra shipping to get them here in a week lol. I got 2 5ks as well prob run them each on a dimable for my veg set up and ditch my T5hos.
 
Greengenes707

Greengenes707

Well-Known Member
SupraSPL said:
Hey GG, that would be cool to see what it would take to get 4 running on the C1400. The "typical" figures are just a guide as you know, the actual Vf varies quite a bit from batch to batch and from COB to COB. So if you could sort through a batch of CXAs and pick out the lowest, and dim the driver a bit, it should be possible. But if we have to dim below 1050mA as the PDF data suggests, I guess that would be a bust. All in the name of science bongsmilie
Click to expand...
Exactly...typical. SDS's hlp's run over rated current...many drivers do(and lower too). So I would think that voltage is somewhat similar. And if like you say...the factors align...it could be doable without any compensation.
If you go by looking at the pdf then it comes out that the hlg1050 is better after efficiency is factored in.
But if it were to just over perform in the voltage part that would be great.
If it were to counter the voltage increase with a current decrease...then it would need to go down to about ~1.34a to stay at the 200w cap of the driver...freaking perfect imo.

We will see. In the name of science and curiosity.
 
S

stock

Well-Known Member
Greengenes707 said:
Exactly...typical. SDS's hlp's run over rated current...many drivers do(and lower too). So I would think that voltage is somewhat similar. And if like you say...the factors align...it could be doable without any compensation.
If you go by looking at the pdf then it comes out that the hlg1050 is better after efficiency is factored in.
But if it were to just over perform in the voltage part that would be great.
If it were to counter the voltage increase with a current decrease...then it would need to go down to about ~1.34a to stay at the 200w cap of the driver...freaking perfect imo.

We will see. In the name of science and curiosity.
Click to expand...
GG ,can you please let me know how you diy led preform compare to at600 ,yield wise ? if you don't want to post here ,please just pm me ,i promise i will keep it to my self .
Thanks
 
K

Kuifje76

Well-Known Member
stock said:
GG ,can you please let me know how you diy led preform compare to at600 ,yield wise ? if you don't want to post here ,please just pm me ,i promise i will keep it to my self .
Thanks
Click to expand...
I'm quiet curious myself, before i stumbled on to this forum and cxa3070's, i was also looking at Apache Tech
 
Last edited:
Greengenes707

Greengenes707

Well-Known Member
stock said:
GG ,can you please let me know how you diy led preform compare to at600 ,yield wise ? if you don't want to post here ,please just pm me ,i promise i will keep it to my self .
Thanks
Click to expand...
Kuifje76 said:
I'm quiet curious myself, before i stumbled on to this forum and cxa3070's, i was also looking at Apache Tech
Click to expand...
It's hard to say right now. But I will know shortly. This run has enough variables constant that I can say more confidently what I think, but it's not finished yet. Last run before this I hit ~1.2g/w with my at600, and just ~.8/w with the drone. They were different strains and pot sizes feeding regiments. But both were in the 40w/sqft range, so lighting was fair at least. But I don't call it fair or even a comparison at all. Too many different things going on for each.
This run I have the same strains and pots size and soil. while everything else more equal and enough to see.

There is defiantly development differences between spectrums. 3K is the closest thing to hps I have ever seen. Biggest buds I have ever grown indoors, even hps. Surprising penetration...as good as anything so far(I run 1400ma). Still amazing trichome production.
R2 brings out the colors in anything. Faster flowering. Produces a shorter stature. And a touch more resin/trichomes possibly...it just reaches out to the leaf tips more...but testing will tell the part that matters, flowers not leaves.

I do't need to hide anything for AT. I want the drone and my other cxa projects to beat them. And I think they can/will.
 
S

stock

Well-Known Member
Greengenes707 said:
It's hard to say right now. But I will know shortly. This run has enough variables constant that I can say more confidently what I think, but it's not finished yet. Last run before this I hit ~1.2g/w with my at600, and just ~.8/w with the drone. They were different strains and pot sizes feeding regiments. But both were in the 40w/sqft range, so lighting was fair at least. But I don't call it fair or even a comparison at all. Too many different things going on for each.
This run I have the same strains and pots size and soil. while everything else more equal and enough to see.

There is defiantly development differences between spectrums. 3K is the closest thing to hps I have ever seen. Biggest buds I have ever grown indoors, even hps. Surprising penetration...as good as anything so far(I run 1400ma). Still amazing trichome production.
R2 brings out the colors in anything. Faster flowering. Produces a shorter stature. And a touch more resin/trichomes possibly...it just reaches out to the leaf tips more...but testing will tell the part that matters, flowers not leaves.

I do't need to hide anything for AT. I want the drone and my other cxa projects to beat them. And I think they can/will.
Click to expand...
Thank you !
What is R2 ?
Do you have journal for latest run ?
 
DonovanWorbot

DonovanWorbot

Member
SupraSPL said:
Congrats on your purchase! Here is how they perform at each current using the "typical" figures at Tj 50C
View attachment 3326349

Regarding the water cooling, if for some reason the pump failed, is there a way to make a backup and would the water stay in the pipe?

That driver you linked is 36V but those CXA3590s are ~72V. Also it is only 83% efficient so it would create quite a bit of heat and the COBs as well. Even if it were 72V, each parallel string would have to be balanced somehow (circuitry that is over my head). So to give fair warning, if run hard, we could be flirting with a design where the (cutting edge 1000W DE) HPS could be a threat and certainly the commercial LED lamps. It is some tough competition I will admit.

You could use something like these $7 drivers to really keep driver costs down and they run 88-89% efficient. You could put (2) CXA3590s on each driver and it will put out about 280mA
http://www.fasttech.com/products/0/10008079/1714900-18-36-1w-85-277v-36w-high-power-constant-current

At that high of an efficiency (47%) your setup would be sure to 1000W HPS and it would dissipate 608W. With driver cost included = $6.88/PAR W

Another option would be the Mean Well HLG-185H-C700. They cost $70ea and can run (4) CXA3590s at 600mA. Your system would dissipate 1308W @ 40.7% efficient. Driver cost included = $4.54/PAR W
Click to expand...


Hey Supra, Thanks for your interest, and well given advice!! I am new to the forum as I said, but have read a couple of your other threads your in, and have to say, I like your style. I had not done any previous conversing with anyone about the switch to LED's, or other peoples ideas on the matter, but it is pretty awesome how far this wonderful technology has came in the industry. I have grown many successful crops in dirt under 600 and 1000w digital HPS bulbs. I have a 20x20' room to play with, and just got done building a 5x16 foot NFT table with (6) 4" PVC tubes with 3 3/4" net pots spaced every 16 inches for 78 sites total, and the Cree's are going over them.

The poor mans way is to take the led and drive it at say, 32v, and
measure the current it draws. Then again at 34v, whatever, to get a feel
for the sweet spot where the led will draw the 85 or 90% of the 3600mA
limit. This is WHY we drive them under the max, so we'll have a safety
margin.

So say you find out you can draw 3200mA (90%) at 33.5vdc. You set up
your 36v pwr supply to put out 33.5 (if it allows adjustment), or add a
dropping resistor inline with each led to carve a few volts off the top
and just run them that way.

It's not as fool proof, but the risk is low. And the price/ circuit is
much cheaper.

They'll tell you that every Led is different, and has a slightly
different voltage/amperage curve. But modern electronic devices,
especially from a reputable house like Cree are gonna be within a couple
of percent of eachother, not all over the map like maybe some chinese
varieties.
 
DonovanWorbot

DonovanWorbot

Member
Oh, and also, Yes I will have a pump running the water through the cooling tube heat sinks, I am planning on just hoping all goes well, and maybe incorporating some-type of thermal switch that the driver power would have to run through, in the event of a possible overheat scenario......
 
K

kkman

Member
SupraSPL said:
Whoops I forgot the smallest 185H was the 500, you are right it is the HLG-120H-C350 I am referring to. If you use the B version, you could dim as low as 35mA, although I am not sure how the efficiency would hold up driven that low (I will check). I have the HLG-185H-C1050B on hand and have tested the potentiometer method for full dimming to make sure. You could dim it even lower but I hear it is not advised to go below 10%.

Yes the AB is still a lot cheaper/PAR W, for example if you compare the $/PAR W at any given efficiency. The Z2 at 46.7% (350mA) cost $6.37/PAR W and the AB at 46.9% (900mA) cost $3.45/PAR W. I believe as you go even lower in drive current the AB would gain even more ground because at the lowest drive currents there is not as much gain from current droop.

I usuallly base the spacing between COBs on the drive current. If the drive current rises high enough, the cones misses the canopy and reflectors are required. I am not exactly sure at what current that happens but in my application (700mA) I find it tricky to implement reflectors effectively. You could use very short reflectors in that case, to try and shape the light to your canopy without reducing the even spread.

It is very hard (for me) to say how much reach into the canopy you will get or need. I havent sorted it out in my head whether the reach into the canopy would be reduced when we run softer, even if we maintain the same PAR W per area. But I can say this much, I run my CXA3070s at 700-800mA (24-28W ea) at about 30W/ft², with no LST or SCROG and my ladies sometimes finish up to 5 feet tall when they are from seed. As long as I cut out some of the underbrush, I get no larfy buds. In other words, as long as the bud site is directly illuminated it works out great, even from quite a distance from the COB (2.5 feet). But if the bud site is shaded out, I try to move it or prune it away. Since you are planning to use LST you should have no worries there :)
Click to expand...
Thanks for the answer.

I guess it shows how little I know about the subject of dimming as I thought the A version functioned like the B version. From what you said I need to buy the B version, which method of dimming is more efficient the 1-10V dimmer or PMW. Can you recommend a model I can use to run up to 430V through it. Lastly does that mean I could buy the 185H-C500 and dim that down as well as the C350 ( I prefer the 350 as I can run upto 13 cobs, I think ).

As to the cob to use, I looked at mouser and digi key and the AB is the best I can do for a 3070 cob and costs around £46 (maybe £35, thanks Super), or z2 for £29. An alternative which looks better is to use the 3590 36V version. I can buy from mouser the AD for £48 or the BD for £57. The idea is to make a unit per driver, so at 430V I should be able to run 12-13 cobs per driver, can you advise the type of wire I need to use, I would perfer thin wire rather the the thick ones but I dont know if that is safe at such voltage.

So the bit which is doing my head in is working out the grid and distance between cobs to be able to get around 15-20W per sqft. In my current setup I run 4x3070 Z2 bin covering 60cm x 60cm at a distance of 15-20 cm ( in flowering room ) @ 700ma, towards the end of the run I felt it could do with more and used 900ma to finish. Based on that I thought 700ma would do a good job covering a 25cm square, so to cover a 1x1 square I would expect to use 16 cobs.

So the question is, by how much should the grid increase to from 4x4 to 5x5, 6x6 or higher? or could I use the 4x4 array but with 3590 cob driven softer to get closer to 15-20w per sqft.

I think going as low as 35ma would be a little OTT what would be your suggestion?

Lastly, if I can run each cob under 350ma I will move to passive cooling. The idea is to use 4-9 230cm fans above the cobs blowing wind in a circle route around the room, each running around 1W. Since the cobs will be producing less heat I hope to recoup the watts by running the extractor from once every 2 mins, to once every 3-4 mins.
 
Last edited:
alesh

alesh

Well-Known Member
kkman said:
Thanks for the answer.

I guess it shows how little I know about the subject of dimming as I thought the A version functioned like the B version. From what you said I need to buy the B version, which method of dimming is more efficient the 1-10V dimmer or PMW. Can you recommend a model I can use to run up to 430V through it. Lastly does that mean I could buy the 185H-C500 and dim that down as well as the C350 ( I prefer the 350 as I can run upto 13 cobs, I think ).

As to the cob to use, I looked at mouser and digi key and the AB is the best I can do for a 3070 cob and costs around £46 (maybe £35, thanks Super), or z2 for £29. An alternative which looks better is to use the 3590 36V version. I can buy from mouser the AD for £48 or the BD for £57. The idea is to make a unit per driver, so at 430V I should be able to run 12-13 cobs per driver, can you advise the type of wire I need to use, I would perfer thin wire rather the the thick ones but I dont know if that is safe at such voltage.

So the bit which is doing my head in is working out the grid and distance between cobs to be able to get around 15-20W per sqft. In my current setup I run 4x3070 Z2 bin covering 60cm x 60cm at a distance of 15-20 cm ( in flowering room ) @ 700ma, towards the end of the run I felt it could do with more and used 900ma to finish. Based on that I thought 700ma would do a good job covering a 25cm square, so to cover a 1x1 square I would expect to use 16 cobs.

So the question is how by how much should the grid increase to from 4x4 to 5x5, 6x6 or higher? or could I use the 4x4 array but with 3590 cob driven softer to get closer to 15-20w per sqft.

I think going as low as 35ma would be a little OTT what would be your suggestion?

Lastly, if I can run each cob under 350ma I will move to passive cooling. The idea is to use 4-9 230cm fans above the cobs blowing wind in a circle route around the room, each running around 1W. Since the cobs will be producing less heat I hope to recoup the watts buy running the extractor from once every 2 mins, to once every 3-4 mins.
Click to expand...
• No real difference between 1-10V and PWM dimming. Go with whichever you can do easier.

• HLG-120H-C350B is good from 215 to 430V with current up to 350mA.

• At such low currents, 3070 AB > 3590 (36V) BD. 3590 (36V) BD is more efficient than 3070 AB from about 1A.

• For such low current, thin wire is OK. Just make sure that its insulation is rated >495V.

• 35mA is definitely OTT. Actually I think that efficiency goes back down at such current. It's like < 5mA/die. I personally wouldn't go lower than ~200mA. I also wouldn't use that driver dimmed to 35mA as its efficiency would be in lower 70's. What you gain by running soft is be burned in the driver.
 
SupraSPL

SupraSPL

Well-Known Member
alesh said:
• No real difference between 1-10V and PWM dimming. Go with whichever you can do easier.
Click to expand...
Even better (IMO) and simpler, is using a 100K potentiometer. The only reason I can see to use 1-10V or PWM is for automated dimming, but they both require a power source and the POT does not.

I agree running at 35mA would be extreme. I am only pointing out its potential to dim that low, but not a recommended current to run at regularly. It would come in handy during a power outage to keep the light cycle steady.

350mA is an aggressively low current to run the CXA3070s at as well, but the cost and build complexity is still within reason if you are off the grid or growing in a hot climate. Can you get the CXA3590 3K BB from Digikey UK? This chart compares the CXA3070 AB vs CXA3590 BB, priced in pounds

KK.png
 
DonovanWorbot

DonovanWorbot

Member
Thanks for all the likes, and trophies, I am finishing construction this weekend on my first NFT table, it is 16'x4' with 78 grow sites. I looked and don't know how to send a "like" or "trophy" back to people? Thanks
 
K

kkman

Member
alesh said:
• No real difference between 1-10V and PWM dimming. Go with whichever you can do easier.

• HLG-120H-C350B is good from 215 to 430V with current up to 350mA.

• At such low currents, 3070 AB > 3590 (36V) BD. 3590 (36V) BD is more efficient than 3070 AB from about 1A.

• For such low current, thin wire is OK. Just make sure that its insulation is rated >495V.

• 35mA is definitely OTT. Actually I think that efficiency goes back down at such current. It's like < 5mA/die. I personally wouldn't go lower than ~200mA. I also wouldn't use that driver dimmed to 35mA as its efficiency would be in lower 70's. What you gain by running soft is be burned in the driver.
Click to expand...
Thank you for the input.

When you say 30070 ab > 3590 bb is that only for the 36V 3590 or the 77V version as well. Also is that based on the cost of the units or the actual light they produce.

This confused me as the next post after yours from Supa was suggesting using an even lower bin 3590 bb.

Lastly if 200ma is too low, what would be a safe min to go for. This is the hardest thing for me to get my head around. what size array 4x4,5x5,etc. can I use to get to 15-20W per sqft and still have a light which works as well or better than my current 4x4 array running 700ma-900ma using 3070 z2 cobs. So currently running 25W-35W, so in total I would use between 400W and 560W for 1 meter square area.

If I try 350ma for example, using the same 4x4 grid I get 176W or if I use a 5x5 grid I get 275W which with my bad maths gets me 24.75W per sqft. If I try with the 5x5 using 240ma I get 180W so 16W per sqft. But I cant see how such a light would cover the canopy well (each covers 25cm squared). From my own testing I thought I would technicly need to run more cobs with lower amps.

Anyway please give any feedback as what would be a good working model. IE. array size to cover 1 square meter and what current. Any advice on cobs would also help.

Tanks and you have been liked.
:-)
 
Last edited:
K

kkman

Member
Thanks Supa.

First 100K potentiometer... based on what alesh was saying about not running below 200ma and the C350A can be dimmed to 175ma, Would I want to still buy the B model and use any current lower than that.

I checked for 3590 bb bin on digi key, costs 39.67 for either 77V or 36V the only difference is min order of 50 for the 36V version.

Is the only reason you mention using it rather than the bd bin cost as I can get it for £45 ( min 50 ).

If 200ma is too low, what would be a safe min to go for. This is the hardest thing for me to get my head around. what size array 4x4,5x5,etc. can I use to get to 15-20W per sqft and still have a light which works as well or better than my current 4x4 array running 700ma-900ma using 3070 z2 cobs. So currently running 25W-35W, so in total I would use between 400W and 560W for 1 meter square area.

If I try 350ma for example, using the same 4x4 grid I get 176W or if I use a 5x5 grid I get 275W which with my bad maths gets me 24.75W per sqft. If I try with the 5x5 using 240ma I get 180W so 16W per sqft. But I cant see how such a light would cover the canopy well (each covers 25cm squared). From my own testing I thought I would technicly need to run more cobs with lower amps.

Does the maths make sense?
 
Last edited:
alesh

alesh

Well-Known Member
kkman said:
Thank you for the input.

When you say 30070 ab > 3590 bb is that only for the 36V 3590 or the 77V version as well. Also is that based on the cost of the units or that actual light they produce.

This confused me as the next post after yours from Supa was suggesting using an even lower bin 3590 bb.

Lastly if 200ma is too low, what would be a safe min to go for. This is the hardest thing for me to get my head around. what size array 4x4,5x5,etc. can I use to get to 15-20W per sqft and still have a light which works as well or better than my current 4x4 array running 700ma-900ma using 3070 z2 cobs. So currently running 25W-35W, so in total I would use between 400W and 560W for 1 meter square area.

If I try 350ma for example, using the same 4x4 grid I get 176W or if I use a 5x5 grid I get 275W which with my bad maths gets me 24.75W per sqft. If I try with the 5x5 using 240ma I get 180W so 16W per sqft. But I cant see how such a light would cover the canopy will. From my own testing I thought I would technicly need to run more cobs with lower amps.

Anyway please give any feedback as what would be a good working model. IE. array size to cover 1 square meter and what current. Any advice on cobs would also help.

Tanks and you have been liked.
:-)
Click to expand...
Both version of 3590 are the same efficiency-wise. 77V has twice as much Vf and half the current.

I don't think Supra was suggesting 3590 BB, he just posted comparison with 3070 AB. As you can see in that chart, 3070 AB is significantly more efficient at low-mid power. At max power they're equal. BD bin would be equal at slightly lower current.

As Supra said, even 350mA is "aggressively low current". What about 5x5 grid without the central COB? This way you could use 2x MW HLG-120H-C350B (or even A) with 12 3070's each. Provided that you wouldn't dim under 60%, it'd give you 14~26W/sqft with driver efficiency over 92%. Overall efficiency, including drivers, would be ~50% - an amazing number no other light can compete against.

nestor said:
Been awhile since I've checked this thread. Nice to see its still going strong a year later.

I stumbled across some interesting drivers that might work well for fans. Thought I'd share:
http://www.ebay.com/itm/261653490178?_trksid=p2060778.m2749.l2649&ssPageName=STRK:MEBIDX:IT

At $1.59 each seems kind of hard to beat and gets away from using multiple wall sockets.
Click to expand...
Great find, thank you! Looks nice, mounting holes for 2 dimensions! Just ordered a few of them.
 
You must log in or register to reply here.
Top