LED COB bar project

bmdiyh

Well-Known Member
Hi guys, I have been reading lately all the LED threads, and have thought of a DIY project which I want to build. I really have a specific idea, you can see it in the picture below. If you can take a look at it and tell me is it doable? Because I am a really new to these I have now idea what I'll need to buy. Please take a look. Ask me if I am missing something.

I imagine a 1 meter (3.28 ft) long bar, having on it 4 chips each running at 50/60W. There are spaced so the bar can cover exactly a 1x4 ft space. I have a general idea and a simple design, but the thing is I have no idea about drivers, heatsinks, fans etc. Please tell me (if) how that idea can be made a reality, if its going to cost me a lot.

Thanks in advance! (:
COB.jpg
 

Rahz

Well-Known Member
One driver per 4 x 50w cobs, might want to check out 4 x CXB3070 and HLG-185H-C1400. Unfortunately, no perfect combo for 4 Vero 29s at similar wattage.

However... if you would consider 38 w/ft then you could pair 4 Vero 29s with an HLG-120H-C1050, saving about 10 on the driver, 45 on the COBs, and efficiency would rise a bit.

I think 50 w/ft is pretty bad ass, but 40ish would also give great results.

I like your thinking on COB spacing BTW.
 

bmdiyh

Well-Known Member
These seem very expensive to me for only one bar. FYI I need 8 of those, I was hoping to get by with $1000-$1500 for everything. $40 (for the CXB3070 and the kind) times 32 is $1280 only for the COBS..... It's just I read a tutorial in which a guy build something similar, but the LED he used were like super cheap see here http://www.ebay.com/itm/100W-Warm-White-High-Power-LED-Panel-9000LM-100-Watt-Lamp-Light-/380515680644?pt=LH_DefaultDomain_0&hash=item5898808184. There 100w max each, he ran them at 60W which is alright for me. Would you think that these kind of leds (the cheap ones) can do a good job. It's just I need a 1600W (200W per bar) LED setup and I don't want to pay as much as I would for a commercial fixtures. Don't tell me that good always means super expensive even if you build it yourself. ....
 

Positivity

Well-Known Member
A 200w 3' heatsink should cover a 4 x 2 easily...

So cut numbers in half. The newest chips should be capable with just 25w sq/ft

Plus you can always add when you are able to afford more

Could cut costs off a Cree build and use the Vero 18 or 29 and run them a little harder.

That's what's I'd consider a budget build...

Multi chips are in some other lighting category.... The hot category...:p
 
Last edited:

robincnn

Well-Known Member
These seem very expensive to me for only one bar. FYI I need 8 of those, I was hoping to get by with $1000-$1500 for everything. $40 (for the CXB3070 and the kind) times 32 is $1280 only for the COBS..... It's just I read a tutorial in which a guy build something similar, but the LED he used were like super cheap see here http://www.ebay.com/itm/100W-Warm-White-High-Power-LED-Panel-9000LM-100-Watt-Lamp-Light-/380515680644?pt=LH_DefaultDomain_0&hash=item5898808184. There 100w max each, he ran them at 60W which is alright for me. Would you think that these kind of leds (the cheap ones) can do a good job. It's just I need a 1600W (200W per bar) LED setup and I don't want to pay as much as I would for a commercial fixtures. Don't tell me that good always means super expensive even if you build it yourself. ....
Consider vero 18 @ $12.61 at digikey instead of ebay cob crap for $6.49
Just found out 18's are much cheaper than 29's
 

bmdiyh

Well-Known Member
Sorry, Positivity, I'm not sure I got the joke. I'm that much of a newbie;) Anyway the vero 18s are acceptable for a start I suppose, and for their cost of course. What else would I need? A driver for how many watts with the V18s, 100W? Then heatsink(s) and fans I haven't considered in my design, because I'm not really sure on what to place the chips. I was thinking a whole aluminium peace, but that's going to be pricy. How can I make this for less? Having 4 heatsinks with fans for each chip and somehow connecting them with a frame? will that work? I am really worried about heat. I don't want to burn the house down.... I think I am really making progress with the idea. Thanks guys for the suggestions!
 

Abiqua

Well-Known Member
You can build 1 bar of CXB3070's [x4]....for around $250-$300 US
Thats with a Meanwell HLG 185 and a nice long heatsink from HeatsinkUSA....

Going for Vero's is nice for price but lag in efficiency compared to an under driven CXB....

If you were a smart, thrifty shopper, you could probably build all 8 for around 2k total.....
 

SupraSPL

Well-Known Member
Those generic COBs are near the efficiency of a CXA3070 IF you drive the 100W COB at 5W. That means a lot of labor to build though and without the reliability of CREE COBs, you may get many partially damaged generics.

When talking about generic COBs and photon cost, it has to be said that the cheapest photons you could get with decent efficiency would be 1000W DE HPS
40.8% efficient
408 PAR W
$550
$1.35/PAR W (must be corrected for reflector losses/PPFD, SPD and spread. These numbers are not directly comparable with the LED cost below)

As far as LED goes, I realize there are no perfect-match drivers for running Vero29s hard. The closest thing is probably Mean Well HLN-80H-36, puts out 2.55A but ~$40 ea. So when you take driver cost into account it changes the picture. HLG-185H drivers $64 each. From lowest cost/PAR W to highest:

(5) Vero13 4000K @ 1.05A = 182W
33.2% efficient
60.4 PAR W
$36 + $64
$1.66/PAR W

(10) Vero10 4000K @ 700mA = 197W
32.4% efficient
63.8 PAR W
$44 + $64
$1.69/PAR W

(1) Vero29 4000K @ 2.55A = 98W
39% efficient
38.2 PAR W
$28 + $40
$1.78/PAR W

(4) Vero18 4000K @ 1.4A = 168W
38% efficient
64 PAR W
$50 + $64
$1.78/PAR W

(6) Vero18 4000K @ 1.05A = 184W
40.6% efficient
75 PAR W
$75 + $64
$1.85/PAR W

(4) Vero29 4000K @ 1.4A = 200W
44% efficient
88 PAR W
$113 + $64
$2.01/PAR W

(4) CXB3070 3000K AD @ 1.4A = 200W
48% efficient
96 PAR W
$150 + $64
$2.24/PAR W

(4) CXB3590 3500K CD @ 1.4A = 196W
56.2% efficient
110 PAR W
$190 + $64
$2.31/PAR W

So it looks like the Vero10/13 don't save much $ and sacrifice a lot of efficiency. The Vero 18 and 29 setups seem like good compromise for value and efficiency. If you want to prioritize efficiency for heat reasons or if you have very high electrical costs where you live, might be worth paying the extra for CXB3070 or waiting for the CXB3590 3500K CDs to land.
 
Last edited:

SupraSPL

Well-Known Member
One driver per 4 x 50w cobs, might want to check out 4 x CXB3070 and HLG-185H-C1400. Unfortunately, no perfect combo for 4 Vero 29s at similar wattage.
I have heard someone say that 4 Vero29 @ 1.4A works on HLG-185H-C1400 so I have been doing some experimenting. If you test the open voltage of a batch of HLG drivers there is a slight variation to the maximum voltage. So you might get 143-144V max. The interesting part, as long as the voltage of your string is not way over, as the driver and COBs warm up the current will gradually increase.

For example I found an HLG-185H-C700 that will go up to 287.4V instead of 286 and I hooked up (4) CXA3590s. On paper that should not work (292V) and when I fired it up I got only 630mA which is OK, but as it warmed up it rose to 690mA which is close enough for me. Based on that I think 4 Vero29s on an HLG-185H-C1400 should get 1.4A in most cases.
 

Positivity

Well-Known Member
Sorry, Positivity, I'm not sure I got the joke. I'm that much of a newbie;) Anyway the vero 18s are acceptable for a start I suppose, and for their cost of course. What else would I need? A driver for how many watts with the V18s, 100W? Then heatsink(s) and fans I haven't considered in my design, because I'm not really sure on what to place the chips. I was thinking a whole aluminium peace, but that's going to be pricy. How can I make this for less? Having 4 heatsinks with fans for each chip and somehow connecting them with a frame? will that work? I am really worried about heat. I don't want to burn the house down.... I think I am really making progress with the idea. Thanks guys for the suggestions!

sorry bro, wasn't really a joke. Those cheaper cobs aren't very efficient and produce loads of heat. Lifespan, output, reliability, everything is below acceptable that I've seen.

Looking at the big picture, building them, places to mount things, fans, etc. It's a lot easier to build a light with one long heatsink. You'll gain a little protection from your cobs overheating if a fan gives out for any reason. 1 heatsink, 1 plug, 1 fan, 1 power supply for fan, 4 CXB 36v 3590s. Splitting it up is going to produce more wiring which is no joke, more time connecting things together. The price is rock bottom since your building it yourself, just build as many as you can afford and they will provide you years of trouble free growing.

I see more chance of buyers remorse skimping on efficiency than building it as well as possible.
 

Rahz

Well-Known Member
I have heard someone say that 4 Vero29 @ 1.4A works on HLG-185H-C1400 so I have been doing some experimenting. If you test the open voltage of a batch of HLG drivers there is a slight variation to the maximum voltage. So you might get 143-144V max. The interesting part, as long as the voltage of your string is not way over, as the driver and COBs warm up the current will gradually increase.

For example I found an HLG-185H-C700 that will go up to 287.4V instead of 286 and I hooked up (4) CXA3590s. On paper that should not work (292V) and when I fired it up I got only 630mA which is OK, but as it warmed up it rose to 690mA which is close enough for me. Based on that I think 4 Vero29s on an HLG-185H-C1400 should get 1.4A in most cases.
Interesting. You are saying the driver limits the current when it's set to output .7A? I was under the impression it would provide a constant current and over voltage protection would kick in if the voltage was too high.
 

bmdiyh

Well-Known Member
Those generic COBs are near the efficiency of a CXA3070 IF you drive the 100W COB at 5W. That means a lot of labor to build though and without the reliability of CREE COBs, you may get many partially damaged generics.

When talking about generic COBs and photon cost, it has to be said that the cheapest photons you could get with decent efficiency would be 1000W DE HPS
40.8% efficient
408 PAR W
$550
$1.35/PAR W (must be corrected for reflector losses/PPFD, SPD and spread. These numbers are not directly comparable with the LED cost below)

As far as LED goes, I realize there are no perfect-match drivers for running Vero29s hard. The closest thing is probably Mean Well HLN-80H-36, puts out 2.55A but ~$40 ea. So when you take driver cost into account it changes the picture. HLG-185H drivers $64 each. From lowest cost/PAR W to highest:

(5) Vero13 4000K @ 1.05A = 182W
33.2% efficient
60.4 PAR W
$36 + $64
$1.66/PAR W

(10) Vero10 4000K @ 700mA = 197W
32.4% efficient
63.8 PAR W
$44 + $64
$1.69/PAR W

(1) Vero29 4000K @ 2.55A = 98W
39% efficient
38.2 PAR W
$28 + $40
$1.78/PAR W

(4) Vero18 4000K @ 1.4A = 168W
38% efficient
64 PAR W
$50 + $64
$1.78/PAR W

(6) Vero18 4000K @ 1.05A = 184W
40.6% efficient
75 PAR W
$75 + $64
$1.85/PAR W

(4) Vero29 4000K @ 1.4A = 200W
44% efficient
88 PAR W
$113 + $64
$2.01/PAR W

(4) CXB3070 3000K AD @ 1.4A = 200W
48% efficient
96 PAR W
$150 + $64
$2.24/PAR W

(4) CXB3590 3500K CD @ 1.4A = 196W
56.2% efficient
110 PAR W
$190 + $64
$2.31/PAR W

So it looks like the Vero10/13 don't save much $ and sacrifice a lot of efficiency. The Vero 18 and 29 setups seem like good compromise for value and efficiency. If you want to prioritize efficiency for heat reasons or if you have very high electrical costs where you live, might be worth paying the extra for CXB3070 or waiting for the CXB3590 3500K CDs to land.
Thanks a lot man! I needed something like this. It'll make it easier to make a choice for my needs and possibilities.

So if I get it right - lower efficiency equals more heat, and bigger electric bill? Isn't t about the watt they run on, I mean the electricity used? What exactly is the efficiency? Give me a formula or something. I get the correlation between volts and amps but how is the efficiency calculated and what does it matter how efficient is one fixture if it put's out as many watts as you need?
 

SupraSPL

Well-Known Member
@Rahz They do go limp mode as you increase the voltage above its range the current drops off fast, but it seems that it can automatically and gradually increase current as the voltage decreases. If you go way above its voltage range it may go into a protect mode but I have not tested that.

@bmdiyh So for example if you have 2 lamps that draw the exact same power but one has more efficient COBs, it will create more light and less heat, but same electric bill. You are correct, in a way efficiency does not matter as long as it has enough PAR W for your needs. The consequence of low efficiency would be a lot of heat. But if you have a very cheap electric rate and you don't mind running a lot of HVAC, HID might be more attractive than LED because of the cheaper up front cost and the fact that you don't have to build your own lamp, a lot of labor saved (many LED diyers enjoy building the lamps)
 

SupraSPL

Well-Known Member
The way we estimate efficiency is somewhat complex. We (@alesh / @stardustsailor ) analyze the spectral curve to get the LER (max theoretical lumen/W for that particular color curve). Then we estimate how many lumens the COB will output at a particular current and temperature, based on the the manufacturer specs. Then divide lumens/LER and you get conversion efficiency. Efficiency is a decent place to start when comparing various LEDs.

Once you have efficiency and dissipation W (amps * volts on the DC side) you can estimate the PAR W emitted.
Dissipation W * efficiency = PAR W.

Once you have PAR W, if you have the QER (max theoretical photons/s/W for that particular color curve) you can convert PAR W into PPF. Once you have PPF, you apply penalty for reflector/lens/cover losses/scatter/wall losses etc, divide that result by the canopy area in m², and then you have estimated PPFD.

PPFD is a decent way to compare the effectiveness of two different lights and it is a good reference point. There are a few more things to consider on top of PPFD, such as how evenly the light is spread in the canopy, how evenly the wavelengths are spread across the PAR spectrum, lumen/reflector depreciation etc. You can measure/verify PPFD with a PAR meter (approximately) or you can measure it accurately with a calibrated spectroradiometer.
 

bmdiyh

Well-Known Member
...

(4) Vero29 4000K @ 1.4A = 200W
44% efficient
88 PAR W
$113 + $64
$2.01/PAR W
...
OK, if I decide to go for example for the above ( 4 V29s @ 1.4A), which will add up to 200 W per bar. How hot are we talking? What cooling do I have to get for them? What temperature could they reach, and at what temperature should I run them? How many fans would I need, how powerful? Or could I run passive cooling for this kind of setup? The heatsink(s) - are there available kind of what i showed in the picture above or will I have to change the desing concept... I want them to be as harmless as possible to the plants at, let's say a foot of distance (comments on the distance? ).

One more thing! You mention color temperature of 4000K. I am going to use these light only for flowering do you think this is the best CT. What is 2700-3000-3500K range about? Isn't it better for flowering? Or a mix?

Thank you all again for the "lessons" ;))
 
Last edited:
Top