water cooled COBs

hey all, i'm planning a 3 compartment grow cab right now and I've been inspired by all of you. Please let me know your thoughts so I can take them under consideration!

My space is limited vertically so I'm going with water cooled COBs. 40mm square aluminum water blocks are $5 on eBay and copper ones are $10. They only add 10mm to the height of the COB and you can attach them directly to the top of the cab, so they can easily save you 3-4" of vertical space over a commercial fixture.

My flower compartment is 35L x 17.5W x 31.5H. I'm going to scrog 2-4 plants in perlite hempys. I'm planning on three 3000K Vero 29's run at 1 amp (or less) on dimmable Mean Well LDH-45 drivers. Thats about 110 watts of half power Veros, which should be enough, especially with reflectors (Angelinas and '29 clamps are available from Mouser with six week lead time.)

For the veg and bonsai mom compartments I'm not so sure. They are each about 17" square and 21" high. For veg I'm thinking a Vero 13 on a 350 mA driver right in the middle, not sure what temperature is optimal for vegging. On a water block there will be 19.5" of clear space below it.

For bonsai moms I'm thinking another '13 but dimmed down? Some XTE's or XML2's? I can get 5" long water blocks for longer arrays without having too many more blocks in the loop. There are 12V "10W" strips like this too http://www.miniinthebox.com/diy-10w-950-1050lm-3000k-warm-white-light-rectangle-cob-led-emitter-12-14v_p457287.html Any advice?


thanks!

sounds good, waiting for pics

For a cabinet it's easy to use the walls for reflection, no need for reflectors on the Vero. This gives a much better spread of light. Ideally the upper half of the walls with a mirror surface (to diminish the edge effect) and white everywhere else.

The 3000K provides just the absolute bare minimum of 40 umol/m2/s of blue light. In your situation with a height of only 80cm consider something cooler like 4000K to avoid stretch.

Have much does that new step-up convertor cost?

The LDH-45 is $12. Mouser has 40 on order with a 4/1 estimated ship date but Jameco says six weeks lead. I almost ordered LDDs which are $7 and shipping but only have PWM dimming at 100-1kHz. The controller I'm looking at for running all the lights, fans, pumps, monitoring, etc... (Aquaero 6) only does 15kHz PWM, and the rest of the Mean Well's only do like 10kHz PWM, so I'm stuck with analog dimming (0.2-1.8V on LDH-45). The LDH-45's are the only analog dimming drivers I could find that will shut off when dimmed to zero. Any other driver is going to require a relay which means more $$ and more ways to go wrong. I could, of course, forget the controller and use regular timers and manually dim the lights when I'm working in the cab like a normal person. I wish I was normal. I figure once I get two timers, dimmer knobs, temp and fan controllers I would be getting pretty close to the Aquaero ($250) which does safety monitoring, logging, has a touch screen, remote control, etc... If the Aquaero had a humidity sensor it would be a perfect grow controller but they still haven't shipped it despite announcing it years ago.

I want to dim the lights way down when I'm working in the cab, and also for general adjustment if it's too much light. Since I can't figure out how much PAR I'm going to get at a certain distance from the emitter. I have an Apogee quantum meter but getting drivers just to test and figure out how much current I need seems like a waste and I'm already planning to spend too much money. I will take some measurements and post them once I do get the gear and assemble it, however.

You mention a 40 umol/m2/s of blue light, how do you reckon? Shouldn't the quantity of photons vary by the amount of current and model of COB?

Thank you for the suggestion on the color temperature. I thought more blue meant more stretch, looks like I things backwards.

I agree with Flux, he won't steer you wrong...he has been doling out credible advice for a few moons so listen up and go hunt out his efficiency program, probably where he found the 40 umol figure.

I am currently flowering with different LED's [3k and 2700k]. I would press for the 4000k. I am leery of the UV-B produced by some of the "more" blue aka 5000k and 65000k, but some are blending with these ranges too.

Don't worry, I've read the "CXA Analysis" thread several times, I'm very grateful that Mr. Flux is paying attention to me!

Abiqua said:

I am leery of the UV-B produced by some of the "more" blue aka 5000k and 65000k, but some are blending with these ranges too.

Click to expand...

There aint no UV-B produced in these...

I think I see, from your absolute power graph in the CXA analysis thread the 3000K Vero puts out 1 mw/nm/W in the blue peak and I'm proposing running the '29 at almost 40W. However, 1W through the '29 puts out more than 1W through the '10, I guess not that much more. Also, I'm confused on how to take the photon flux numbers you've calculated and figure out how bright it will be directly underneath the emitter at N cm distance.

This is the makings of an informative thread.
Personally, I would water cool CXA 3590 (bd).
Maybe position some led cob bars on 4 sides
properly heatsinked with a little forced air.

CXA3590 looks great but I don't need 100 of them...

Look around and you might find a couple or 4.

.................

just a couple of things...

Blue =/= Ultraviolet-B

UVB leds cost many thousands of dollars each.

Back to topic, while I am always excited to see watercooled LED, i have _never_ seen one reach completion here - either because they dont bother or because they realise that air cooling is effective enough.

One of the problems with watercooling is that efficiency loss from pumping the water around, plus you need to think about exactly what will happen if theres a catastrophic failure.

Observe & Report said:

Since I can't figure out how much PAR I'm going to get at a certain distance from the emitter.

Click to expand...

The PAR you can expect can be calculated. Your room is 0.4m2, you have planned 110W of LED power, and a Vero at half test current (3000K or 4000K) gives about 2 umol/s per Watt. That gives 110*2/0.4 = 550 umol/s/m2 on average. That is an adequate amount but it would be nicer to have double that amount since you can always dim it. Having a lot of light at your disposal is much more useful than all the controller gadgets as it will actually add to the result....

You mention a 40 umol/m2/s of blue light, how do you reckon? Shouldn't the quantity of photons vary by the amount of current and model of COB?

Click to expand...

The vero 3000K has 7.1% of quanta in the 420-490nm blue region, so 550*0.071 = 39 umol/m2/s.

Thank you for the suggestion on the color temperature. I thought more blue meant more stretch, looks like I things backwards.

Click to expand...

OK this is crucial.

^^^Good stuff!

Chronikool said:

There aint no UV-B produced in these...

Click to expand...

Sorry Freudian fuck up....I just meant UV more around 420-375nm, definitely not below 300, I doubt they produce that low. I don't know why I even put B, maybe I was reminded from the weird new BlackDog dongle

my bad

Hi, I got here via Googling "water cooled COB LED". This is a very interesting thread as I think you are experimenting with the future of these COB units -some form of integrated cooling seems almost inevitable given their power / efficiency.
I have made 'flat' computers and liquid cooling is the way to go. VGA coolers with heat pipes tend to be quite thin too, you might want to try this...
Here is a video showing some experiments with cooling I found, there is also a chap using his aquarium water as the cooling agent which is 'cool' as it doubles as a heater!
Have you considered the "Full Spectrum Grow LED" modules rather than standard white COB (which will waste energy in the green spectrum rather than mostly red and blue for the full spectrum ones...);

http://www.aliexpress.com/item/2014-New-Arrival-high-lumen-super-intensity-full-spectrum-380-840nm-120W-DIY-led-grow-chip/1606479453.html

http://www.aliexpress.com/item/2014-New-Arrival-high-lumen-super-intensity-full-spectrum-380-840nm-120W-DIY-led-grow-chip/1606479453.html

-I've just sent off for;

1 x 30W White COB
1 x 100W White COB
1 x 30W Full Spectrum
1 x 120W Full Spectrum (bit more than 100W...)
2 x 17W COB Strips White
2 x 17W COB Strips Ice Blue (these "strips" are 12V used in the automotive industry for fog and internal trim, etc. They are dirt cheap and usally don't need any sort of active cooling but low power)

-I plan to use the drivers that come with White COBs to test the full spectrum and do a comparison. Also try different cooling solutions like bare heatsink, small fans, VGA coolers with heatpipes and waterblocks. I use a Corsair H70 with two high pressure fans in a push/pull arrangement for my CPU and it keeps it <40 degrees...

-I don't think the lower power units are good for anything other than "winter maintenance" of smaller houseplants/herbs (e.g. Aerogarden type growing) as they don't really put out the intensity required...

Abiqua said:

Sorry Freudian fuck up....I just meant UV more around 420-375nm, definitely not below 300, I doubt they produce that low. I don't know why I even put B, maybe I was reminded from the weird new BlackDog dongle

my bad

Click to expand...

Im not sure you understand how color temp works with LEDS wether its 3000K or 6000K 99% of cobs are based on 440-460nm blue chips, they do not put out UVA or Violet light. and the blue spectrum starts at 400nm visible Violet light IS BLUE LIGHT and infact the 400-440nm is the most beneficial blue region for photosynthesis once you get past around 460-480nm photosynthesis starts to take a bit of a dive. If cobs put out Violet light (not Ultraviolet) that would be awesome in my books, violet is the often missing spectrum perhaps its not as efficient in generating photons for photosynthesis as blue, but for quality a full spectrum at both the red and blue side is important. however some cobs, like the luxeon crispwhite use both violet and blue chips even though they barely get 100 lm/w this is mostly due do the fact that its a high cri with lots of deep reds and half the chips are 410nm violet wich is barely visible thus barely contributes to any lumens. its actual PAR efficiency is probably pretty good and they would definitely make your grow room POP with colors as these cobs were designed for art galleries that brings out some really vivid colors and true whites (perfect for showing off the trichome and purps action)

hey @Abiqua tell us more about your 2700k

are they solo or mixed with other colors?
80 or 90 cri?
hows the stretch (and what strains?)

thanks!