DIY LED 220w Cree XTE + Philips ES

SupraSPL

Well-Known Member
DIY LED has proven itself for vegging even with the older LEDs ~25% efficiency. I cannot imagine vegging with anything else. 100w DIY LED is more than enough to veg bushes for a 2400w HPS grow.

DIY LED works great for flowering too but it is a tall order to beat the 600 HPS because it is so cheap quick and easy. On the other hand larger HPS grows get hot quickly and you end up cranking fans and AC in the summer so there is lots of room for improvement.


LED has many advantages over HPS. It spreads light more evenly. It has a better spectral distribution that is also adjustable. It is more directional (less reflector losses). It does not require glass covering. It does not require active cooling.

IMHO canopy temps in mid to late flower should be below 80f. This is critical for quick finishing frosty dense buds. These LEDs create just as much heat as HPS watt/watt but the heat in the LED heatsink is not beamed into the canopy like it is when using HPS. This makes it dramatically easier to reduce canopy temps. Due to the advantages listed above we will be using less dissipation wattage to get the same job done so this also results in a huge decrease in canopy temp. Less air to move also means less fan noise and more contact time with the carbon filter.

Less heat in the canopy also means less humidity. Humidity can be a real pita because it reduces the effectiveness of carbon filters, promotes moldly buds and some say it reduces trichome density. Humidity can limit the size of your garden even more than heat!

LED disadvantages: More complex system for hanging and adjusting the modules. More difficult to access the canopy. Initial cost is higher. A very basic 600 HPS setup only costs $200 while this LED will cost $700-$800. Time consuming designing and assembling the modules. The drivers can be unreliable. If your soldering was less than perfect you can end up with blown drivers and LED strings!

So with all that said the goal here is to replace one of the 600 HPS with 220w of top bin DIY LED. That should be tough because I LOVE the 600 HPS except for the heat issue. HPS gives huge, fast finishing, icy, dense, easy to trim nugs
DSC06482a 16 50d.jpg

This will be KNNA style with constant current drivers, passive cooling, huge heatsinks and no lenses/reflectors. This is a 200w build from a few years ago. It still works great but we will be reusing the heatsinks for this new project.
dsc04975a.jpg
 

SupraSPL

Well-Known Member
LED efficiency

The Cree XT-E is a game changer for LED growing. The R3 warm white is actually equivalent to an R5 in terms of output because they are binned at 85*c. The R5 cool white is the equivalent of an S3 bin. XT-E has a wide angle but as long you keep the LEDs in tight on the canopy the wide angle actually helps spectral distribution. XT-E is also relatively cheap! The aquarium crowd also has available top bin Luxeon ES that are very imressive. The 442nm royal blue and 660nm deep red are top notch. Luxeon ES is binned at 700mA. All of these are sold already mounted except the OSRAM which are easily hand mountable and available by the reel.

Cree XT-E warm white R3 - Cutter $3.25
Cree XT-E cool white R5 - Cutter $3.25
Cree XT-E royal blue Q04 - LEDGroupBuy $3.25
Cree XP-E red P3 - LEDGroupBuy $4
Luxeon ES royal blue MR3 - $3 Steve's LEDs
Luxeon ES deep red EX6 - $3.40 Steves LEDs
OSRAM GD+ deep reds 1T-3T bin (assume 1T) - Mouser $2.00


If you are not an LED geek this might save you some time. When comparing LEDs for growing it all comes down to radiometric efficiency. Here are the calculated numbers of the very best from this list. The numbers are @ actual operating temperatures (Tj 50c) and represent the minimum output of these bins. Actual output will be higher.
@350mA (ES royal blue 63%) (XT-E warm white 41.5%) (ES deep red 42%) (XP-E red 48.2%)
@500mA (ES royal blue 59%) (XT-E warm white 38.3%) (ES deep red 40%) (XP-E red 45.9%)
@700mA (ES royal blue 55%) (XT-E warm white 34.8%) (ES deep red 38%) (XP-E red 42.4%)

As you can see it can be a tough choice deciding what mA is the best compromise. Running at 700mA makes driver selection easier, installation easier and keeps costs down but running softer requires dimming drivers and costs go up.

Here is the bottom line. Under actual usage conditions the lamp will average at least ~42% radiometric efficiency. A slight improvement over 600w HPS which is ~40%. A huge improvement from the old style DIY LED which was ~26%.


LED ratio
KNNA recommended a flowering ratio of 9 red 2 white 1 blue. The perfect ratio is not kn
own exactly but KNNA has a good understanding of it. It does not need to be exact as the plant will adapt. However the less the plant needs to adapt to the SPD the better. The ratio he gave back then depends on the efficiency of each LED so to be more accurate we need to know the ratio in terms of radiometric output. He suggests 15-20% blue for flowering and 20-30% for vegging. The way you divide up the rest is up to you.

I put a spreadsheet together for radiometric output and played with the ratio until I ended up with blue radiometric output at 17%.
7 Lux ES 660nm red
3 XP-E 630nm red
5 XT-E warm white
2 Lux ES 442nm blue
 

SupraSPL

Well-Known Member
Why warm whites? This is debatable. Either way will work and work well but I am planning on skipping the XT-E R5 6500K cool whites and using XT-E R3 3000K warm whites. The KNNA theory is to use about half 660nm deep red, half 630nm red, cool whites to provide some 500nm-600nm wavelengths and 450nm for the blue. Cool whites are more efficient because they use less phosphor to convert blue to white. On the other hand we don't need much blue (15-20%) and the Luxeon ES royal blue is 59% efficient so it does that job very efficiently already. The warm white only has about 10% of its output in the blue spike and the other 90% peaking at about 610nm providing a wide spectrum of light with a high PAR value including the 500-600nm range. Based on all that and the fact that the XT-E R3 7A bin is incredible I suspect that now warm white is as good or better than the cool white R5 1C. The cree graph supports that but I am not sure exactly how accurate it is.

The warm white 3000K R3 7A bin (the maximum produced in XT-E) is 41.5% efficient at 350mA@50c while the cool white R5 bin (maximum produced in XT-E) is 52.4% efficient at 350mA@50c. The warm white has 90% of its output outside the blue spike while the cool white has 64% of its output outside of the blue spike. Which one will translate into more bud?

XT-E spectral distribution:
 

SupraSPL

Well-Known Member
Heatsink

The increase in efficiency from the old design affects how much heatsinking we will need to achieve the target of 50c junction temperatures. KNNAs rule of thumb for passive heatsinking is 21 sq in/watt. That figure was based on the 26% efficient lamps so now that we are up to 42% we should be able to use 16 sq in/watt and get the same job done. The heatsinks will also benefit somewhat from the circulation fan and exhaust fan so that will slightly improve efficiency.

You can estimate your junction temperature (Tj) based on your heatsink temperature. You have to figure out the thermal path resistance and multiply that by each watt disspated. Using the XT-E for example it has a thermal resistance of 5c/W. According to Cutter there is a resistance of about 2c/W from the star itself (professionally reflowed) and its junction with the heatsink should be minimal since the stars will be mounted to the heatsink with Prolimatech thermal compound. Estimated total 8c/W with a dissipation of 1.5W @ 500mA so I would add 12c to the heatsink temp to estimate the Tj. In the current lamp the heatsink temp reaches ~36c so we know we are in the correct range for a Tj of 48c.

Before the GD+ deep reds came out we were using Dominant SPNova 660nm deep reds which have a high resistance of 20c/W. KNNA determined that our method of copper over kapton was resulting in a disappointing thermal resistance so we used larger copper pads to help transfer the heat into the heatsink. So there is room for improvement which is great. Since then I was able to find 1mil thick kapton on eBay and copper tape with a thinner adhesive layer so if you are planning on hand mounting GD+ you can go that route.


Drivers
Lots of constant current drivers have shown up on eBay. Unfortunately not many of them are 500mA or 600mA drivers but there are plenty of 700mA available. Ideally they should be high efficiency and high power factor. Meanwell and Inventronics drivers are high quality, efficienct and some are dimmable with a simple potentiometer. The Inventronics is expensive but there is a 40w 700mA dimmable driver on LEDGroupBuy for $33 that looks perfect for a 500mA-600mA string. RapidLED has 33w 700mA Meanwell for $16.
 

jubiare

Active Member
Amazing project and intro +rep!

Looking forward to this, I'll chime in with more time. Subbed!
 

guod

Well-Known Member
You can estimate your junction temperature (Tj) based on your heatsink temperature. You have to figure out the thermal path resistance and multiply that by each watt disspated. Using the XT-E for example it has a thermal resistance of 5c/W. According to Cutter there is a resistance of about 2c/W from the star itself (professionally reflowed) and its junction with the heatsink should be minimal since the stars will be mounted to the heatsink with Prolimatech thermal compound. Estimated total 8c/W with a dissipation of 1.5W @ 500mA so I would add 12c to the heatsink temp to estimate the Tj. In the current lamp the heatsink temp reaches ~36c so we know we are in the correct range for a Tj of 48c.
hello......
 

PSUAGRO.

Well-Known Member
Hey Supra , nice to see you back.............subbed to this awesome project/ amazing how much better the xt-e is vs the xp-e series===good luck with the build

I'm expecting an awesome grow here......
 

jubiare

Active Member
SupraSPL

I like a lot of things you said and the way you put it. I also think that you are very practical and immediate which I like

could elaborate more on:

- needing less dissipating area given the diodes are more efficient

- your plan of wanting to run the blue at 500mA and the reds at 700mA

Thank you very much!

...anyway those xt-e are impressive!


I did also go for WW lately instead of NW/CW ...

I did that for the more FR actually! (xp-e)
 

SupraSPL

Well-Known Member
Finally have everything ordered. Longest wait will be from Cutter. Regarding the heatsink here is an example. If we dissipate 1 watt with a 26% efficient lamp we end up with .74w of heat to spread. 21 sq in / dissipation watt gets the job done. But if we use a 41% efficient lamp there is .59w of heat. That is 20% less heat so we can use 20% less surface area to get the same job done (17 sq in). KNNA confirmed this concept when we moved up from SPNova 660nm to Golden Dragon+ 2T although I think it was in a pm I will try to dig it up.

I entered the XT-E numbers into a spreadsheet to improve accuracy and changed my mind about running the blue/white string at 500mA. The loss in efficiency at 700mA was not enough to justify it. So now I am looking at a 41% minimum efficiency for the lamp overall. The XP-E 630nm does suffer from 10*/W thermal resistance so Tj will be a bit higher for those. But it will be used in a room that has a ceiling fan for circulation so that should help make up the difference. I won't know for sure until I can measure heatsink temps in use.

I am becoming more confident about the warm white. There are some specific spectral analysis charts on LEDGroupBuy website for the Cree warm white and a bunch others
Cree_XPG_warm_white.jpg
 

jubiare

Active Member
well I think you're right on that one ... watt that doesn't turn in flux turns into heat! So ... the more efficient the led, the less the heat!!!

Nice one .... yes exactly the cree 630 is ideally still driven not past 600mA! But it's doable .. maybe you want to consider some extra dissipating area if you can? like 18inches for every watt burned or something like that, given you can afford it.

Looking forward to this one
Thanks for sharing
j.
 

PetFlora

Well-Known Member
Do you think a few CWs for veg only (to jump-start) would be worth while... or, mostly NW + a few WW, perhaps some on a separate circuit to supercharge flower?

Thinking outside the box... I think of harmonic resonance v dissonance, and how much of a factor this is when combining dissimilar lighting tech
 

jubiare

Active Member
I think what the fella is doing is playing with some monochromatic blue ... than prob going into the WW's. One could skip the blue's and go CW's ... but than the peak on the blue is slightly elsewhere? (I might do that on my next mevo, but I already bought blue's hihihi)

Don't forget pet that, given that MJ isn't that difficult demanding in spectrum .... one would want to go for the most efficient chip rather than the best spectrum? I mean, in choosing between whites. So CW would be spot on .. but given the efficiency is better with WW, diyers are using them more as it looks more "ideal" spectrum. (it has some blue in it anyhow)

 

PetFlora

Well-Known Member
Jubiare I am very much in agreement with you- mj is highly adaptive. That said, sub optimal spectral balance would likely cause the plant to spend time (generations?) adjusting to something other than optimal FOR A PARTICULAR STRAIN. One could/should trace the plant back to its' origin and approximate light spectrums throughout the life cycle + amount of light during both cycles, or actually all three + soil conditions + PH +...

As far as I know, ~ 20% after veg, PERHAPS a bit more during veg say ~ 30%?

This, not having been scientifically resolved, is why I would go with NW as the primary and having a few CW and WW on separate o/o switches to supplement NWs
during veg and flower stages
 

SupraSPL

Well-Known Member
MJ isn't that difficult demanding in spectrum .... one would want to go for the most efficient chip rather than the best spectrum[/COLOR]
This is an important point. Royal blue is the efficiency king here but we only need 15-20% blue. The efficiency advantage of cool white comes from its large blue output and the fact that it does not use as much phosphor. That is why I skipped the cool whites and depend instead on the royal blue to provide that spectrum. The whites should be a small part of the total though so either way will work very well.
 

FranJan

Well-Known Member
Hey SupraSPL, looking forward to your project here. I was wondering what your opinion of 10,000K-20,000K whites? I've got a 20 watt 12,000K that has this massive blue output because of the use of additional blue phosphors in the blend. Ever try any of those? The one I have has great output from 420nm to 465nm or so with minimal green and red, but I only use it for clones and seedlings for now. Anyway Good Luck and Happy Growing.
 

Hosebomber

Active Member
Do you mind showing a link to the datasheet for the 10k or 20k. I don't recall ever seeing an LED in that Kelvin range.
 

FranJan

Well-Known Member
^^^Give me one diode doin lotsa blue and I'll just pay the electricity bill at the end of the month ;). Point taken.

20000K


Can't find mine. Hold on Hose, I'll see if I can dig it up.
 
Top