90W homemade LED grow

[Chronikool]

Going down to 1000mA form 1500mA makes you loose about 100 lumen per LED. 12 modules with 10 LED each means I would loose 12000 lumen - what that relates to in PAR I don't know, but I'm sure it's significant.

But I have ideas for that

Fair enough

Yep, lower efficiency, but you will need fewer LEDs - so you gotta choose what you want.
Heat shouldn't be a problem, and the heatsinks are cheap.

Still trying to figure out how I want it to look. My first simple draft (500x480mm of heatsink area):

View attachment 2795259

So what are the measurementz there...? lookz like a mighty beast...is that hollow or solid...?

 

[JMD]

looks really nice, which programm did u use?

Regarding the pictures in the first post: did u start from seed with full power 72W?

Also, why did it work so great, so many ppl here r like screaming u need "white" for all sorts of things going on in the plants and yellow and green and so on...saying with just red and blue the plant is missing something...

Why r those heatsinks cheap, you have a good source? Or what do you consider cheap, a 500x500 heatsink would be here like 120$, thats like 180W of led modules

edit, oh just saw that "green" etc was already mentioned

I used Google SketchUp - it's simple and works very well.
Yes, from seed with just 72W. I'm sure the plants would have yielded more with a broader and more powerful spectrum.

The heatsinks are they cheap because they are bought directly from the manufacturer in China. Downside is that they only ship to China, so you gotta have some connections to help you.
I did a small scale test with 120W low quality and low efficiency LEDs on a 400x120x20 heatsink, and the temperature rise was low. Which tells me that I can use 200W of good quality LEDs on a heatsink that's slightly larger, without any issues. I might even be able to use 250W.

Fair enough

So what are the measurementz there...? lookz like a mighty beast...is that hollow or solid...?

Your point of driving them at lower current is very good, and my thoughts so far: I want to make it possible to run the LEDs a lower current, by running them in parallel with fewer drivers.
So for example 4x 50W modules in parallel driven by two drivers resulting in 750mA. It is very unlikely that any of the modules will draw a larger current than any of the others - else this wouldn't be possible.

The Beast is roughly 48x50x7 cm, so it's not that big. The inside is going to house all the drivers

 

[lax123]

but adding some Transistor circuitry to those parallel strings right, or what about thermal runaway? there is always a tolerance for everything manufactured.

Yes, from seed with just 72W. I'm sure the plants would have yielded more with a broader and more powerful spectrum.

would be interesting if adding a white or equally just more r b modules would increase yield more

 

[JMD]

but adding some Transistor circuitry to those parallel strings right, or what about thermal runaway? there is always a tolerance for everything manufactured.

I'm not sure yet how I'm going to control it all, but it's either gonna be standard switches or semiconductors.

Thermal runaway is mostly only an issue when you supply with constant voltage - I'm going to be using constant current drivers.
Of course there's a risk of one module drawing higher current than the rest, but as long as it's below the maximum rating it's fine. I plan to protect all of the modules with slow burn fuses rated at, or slight above, the maximum current rating of the modules.

In addition I will use some bimetal switches to insure that the temperature doesn't exceed the maximum allowed.

would be interesting if adding a white or equally just more r b modules would increase yield more

I read this: http://www.heliospectra.com/sites/www.heliospectra.com/files/field_page_attachments/what_light_do_plants_need_2012-10-05.pdf

So if that's true, the warm whites should be much better. Look at the spectrum:




I might add some 700+ nm LEDs, but I'm not yet sure if there's any benefit from using those.

 

[lax123]

Thermal runaway is mostly only an issue when you supply with constant voltage - I'm going to be using constant current drivers.

im a bit tired now so maybe i mix things up, but wont each "string" have a different combined forward voltage? this will result in different "current splits" so one string gets warmer then the other, so that forward voltage will shift even more, and one string gets even warmer, the other one gets cooler, also changing its Vf...and so on

I know that pdf, but im not sure if i should agree. I think r b is used most efficient right? So as long as the plant is not like really fully "satisfied" in this regard those other spectrums r a waste of energy...but thats just my theory. I see the issue in how to determine when that point of saturation is reached. So imo it would be interesting to see if rather an added r b source or an added white would have a larger impact in these "low wattage" regions. What do you think about that?

I have a friend who used 700+ for his last grow, but only for a specific time period each day. He said it had a positive effect...but more like a good tweak rather then something with an epic impact.

 

[JMD]

im a bit tired now so maybe i mix things up, but wont each "string" have a different combined forward voltage? this will result in different "current splits" so one string gets warmer then the other, so that forward voltage will shift even more, and one string gets even warmer, the other one gets cooler, also changing its Vf...and so on

You're right, but there's one thing fighting the thermal runaway - the heatsink. Lets say that LED is not mounted on any heatsink, and that for the LED to rise 1ºC we would have to increase the current by 10mA. If the heatsink is good it will reduce the temperature increase when we increase the current by 10mA. This would mean that we at some point reach a maximum increase of the current, in relation to junction temperature.

From experience the problem isn't that big when using current sources. I will see if I can get my hands on a few dataloggers and show the voltage/current/temperature relationship.

I know that pdf, but im not sure if i should agree. I think r b is used most efficient right? So as long as the plant is not like really fully "satisfied" in this regard those other spectrums r a waste of energy...but thats just my theory. I see the issue in how to determine when that point of saturation is reached. So imo it would be interesting to see if rather an added r b source or an added white would have a larger impact in these "low wattage" regions. What do you think about that?

I have a friend who used 700+ for his last grow, but only for a specific time period each day. He said it had a positive effect...but more like a good tweak rather then something with an epic impact.

Yea, it's not that easy to choose the exact perfect ratio of wavelengths. I'll give those 700+ nm another look.




Small update on the design phase:

Almost everything will be riveted together. Top features 3x 120mm fans.



The four heatsinks will be screwed together and fastened to the case. The underside shows the 16 modules.



The 'cover plate' will allow the fans to suck air through the heatsinks. This should, hopefully, draw air in from both ends of the heatsinks, thereby getting a nice flow over as much surface as possible.

 

[CaliJoe]

Design is nice and no doubt will cool just about anything you can throw on it. Do they make special optics for those LEDs? If not, I would go a different route. You can easily increase the output 3 fold by adding optics.

 

[JMD]

Design is nice and no doubt will cool just about anything you can throw on it. Do they make special optics for those LEDs? If not, I would go a different route. You can easily increase the output 3 fold by adding optics.

They do not make optics for these. Optics doesn't change the total flux - just the opposite, because the optics doesn't pass through 100% of the light.

Optics will focus the light on a smaller area, thereby increasing the flux in that area. Of course if you measure the light output with a direction dependent PAR meter, then it will show an increase. But what it doesn't show is all the reflected light from the scenario without optics.


If you think I'm mistaken, then please elaborate

 

[CaliJoe]

Energy stored in photons decreases with distance. Personally I do not think reflected light is nearly as good as directed light and you will get a lot more usefulness out of an LED light with optics vs no optics. Typical spread on LEDs without optics is 120 degrees, that is a lot of light going out the sides and not directly to the plant. The light may reach the plant eventually after it bounces off a few reflective surfaces, but by that time it is 10% the energy it would have been if you went the shortest distance from light source to plant. The directed light also has much more penetration power to reach down into the plants.

 

[JMD]

Energy stored in photons decreases with distance. Personally I do not think reflected light is nearly as good as directed light and you will get a lot more usefulness out of an LED light with optics vs no optics. Typical spread on LEDs without optics is 120 degrees, that is a lot of light going out the sides and not directly to the plant. The light may reach the plant eventually after it bounces off a few reflective surfaces, but by that time it is 10% the energy it would have been if you went the shortest distance from light source to plant. The directed light also has much more penetration power to reach down into the plants.

But that's mighty far from 3 fold But what's the efficiency of the optics?

Feel free to suggest other LEDs - I haven't chosen yet which I want.

 

[anomuumi]

How did I miss this, congratz for the harvest! Do they list the bins used in those cree multichips from ebay?

 

[PetFlora]

But that's mighty far from 3 fold But what's the efficiency of the optics?

Feel free to suggest other LEDs - I haven't chosen yet which I want.

While I agree with Cali to a point, the issue can be resolved by designing a parabolic mounting plate (the 2 outside mounting boards angled in @ ~ 20*s

Also stagger the rows of chips to better blend the light between them

 

[PSUAGRO.]

But that's mighty far from 3 fold But what's the efficiency of the optics?

Feel free to suggest other LEDs - I haven't chosen yet which I want.

Top bin Osram/Cree already mounted on a quality MCPCB which you prefer for your design............AND 80 degree optics, so no wasted light (credit to guod)
http://ideal-spectrum-led.com/epages/box21231.sf/en_US/?ObjectPath=/Shops/box21231/Categories/"Moteurs de lumière - MCPCB multiLED[2]" they have awesome stuff, and their DIY kits are one of the best around.

 

[PetFlora]

Great find! Did you notice they are arranged in Merkaba shape? The 7 engine with the center engine should amp that up even more.

Top bin Osram/Cree already mounted on a quality MCPCB which you prefer for your design............AND 80 degree optics, so no wasted light (credit to guod)
http://ideal-spectrum-led.com/epages/box21231.sf/en_US/?ObjectPath=/Shops/box21231/Categories/"Moteurs de lumière - MCPCB multiLED[2]" they have awesome stuff, and their DIY kits are one of the best around.

 

[lax123]

their led complete kit is about 560$ for 100W, isnt that a bit steep?

 

[PetFlora]

their led complete kit is about 560$ for 100W, isnt that a bit steep?

Yes and no. For someone who has the disposable income, it's no big deal.

Keep in mind it will last ~ 5 years of steady use before losing ~ 10% (saw an interesting life graph breaking it down over 10 years) and at ~ $420/ the payback is pretty fast

That said, anybody who is handy can DIY their own for ~ $250 using quality parts/engines

 

[JMD]

Top bin Osram/Cree already mounted on a quality MCPCB which you prefer for your design............AND 80 degree optics, so no wasted light (credit to guod)
http://ideal-spectrum-led.com/epages/box21231.sf/en_US/?ObjectPath=/Shops/box21231/Categories/"Moteurs de lumière - MCPCB multiLED[2]" they have awesome stuff, and their DIY kits are one of the best around.

They look great, they are also super expensive. The cost of the entire build will triple - or more.
If I were to choose something a bit more expensive, I would go for something like this: http://www.led-tech.de/en/High-Power-LEDs-Cree/CREE-Solutions/7-x-CREE-XP-G-R2-on-round-PCB-LT-1768_120_117.html

At just under $0.50 per watt for the 50W modules, it's really hard to find anything else that even comes close to that.


There's no doubt it would be nice with modules with optics, but I think it's a bit too expensive. An other option might be to use good reflectors instead.

 

[PetFlora]

9w for ~ $55 + driver No wonder guys are taking apart the 9.5w Cree globes

 

[JMD]

9w for ~ $55 + driver No wonder guys are taking apart the 9.5w Cree globes

I think the 50W modules will do great - and with a maximum of 800W, I don't think I will need more than that in my small 60x60cm cabinet

 

[PSUAGRO.]

They look great, they are also super expensive. The cost of the entire build will triple - or more.
If I were to choose something a bit more expensive, I would go for something like this: http://www.led-tech.de/en/High-Power-LEDs-Cree/CREE-Solutions/7-x-CREE-XP-G-R2-on-round-PCB-LT-1768_120_117.html

At just under $0.50 per watt for the 50W modules, it's really hard to find anything else that even comes close to that.


There's no doubt it would be nice with modules with optics, but I think it's a bit too expensive. An other option might be to use good reflectors instead.

Well they are the BEST of the BEST.....so the price is high, don't forget that it includes the vat tax in those prices(20%) which US customers don't pay, just our EU brothers get fucked...lol

Remember those ^^^ Cree xp-g/xp-e modules(boards) have been flooding the market(alibaba/etc.) and Bin # are questionable as are the ACTUAL wavelengths emitted.......you get what you pay for with leds.

Good luck on your build JMD