Building LED lights from facts, no theories

[knna]

I was going to post this at the Perfect LED Grow Light thread, but as some of what im going to post was posted 2 years ago on the stickied thread about LEDs and people still continue developing lights from wrong ideas, i think a thread about this topic is largelly needed.

The main problem is related to efficacy of spectrums.

When the firsts LED experiments at Overgrow, we work on the hypothesis that blue and red light are more effective. It was an appealing hypothesis that promises large electric savings. But unfortunatelly, it has proven wrong

Now we know both by theoretical research and practice that only red and blue spectrums arnt more effective than complete spectrums, but less.

LED grow light sellers obviously not want to notice it, so they may claim efficacies of their lights 8 or 10X higher than HPS. But thats simply false, and any grower that have checked it got dissapointed.

Does it mean that we cant develop LED lights with higher perfomance than HID? No, we can still develop LED lights with better spectrums, by reducing (not elimitating) cyan, green and yellow light and adding it on the red side of the spectrum. This result on a enhanced photosynthetic efficacy, but its small, from 5 to 15% of enhanced efficacy of same amount of photons.

So any further electric saving from using LEDs must come from other factors: better light distribution and increased energy efficiency.

Better light distribution

HID lights, specially HPSs, are very energy efficients (emit a lot of photons per watt burned) and have good spectrums for growing plants, but have a mayor drawback: they run very hot, so they must work at some distance of plants. This mean all the light of the grow comes for a single point of light, and it creates a very uneven light distribution along the grow. Both horizontally (plants below the bulb have way higher irradiances than the edges) and vertically (bottom of plants gets insufficient light while top of plants are exposed to excess light).

This uneven distribution strongly drops photosynthetic efficacy, because as higher is the irradiance, the lower the photosyntesis rate (per photon). But you must expose the top of the plants to such high irradiances if you want the light reaches some into the canopy.

LEDs dont have this problem. They are solid (no risk of breaking them) and run cool enough to be in contact with plants without problems. So they may be distributed along the grow, both horizontally and vertically (sides and between plants) in a way plants gets the optimun irradiances (those at photosynthetic efficacy is higher) along all their height.

This increases photosynthetic efficacy strongly. When NASA scientistist thought on this, and changed half of the light from top to side and intracanopy (IC) lighting, the yield enhancement was of 35%.

Using LEDs on IC lighting not only enhances productivity, but allows to grow taller plants indoor than any previous type of lighting: the equivalent strategy with HIDs has been vertical growing, wich avoid optical losses at reflectors and achieves more even lighting. We all know how vertical setups increases productivity compared to top reflectorized HIDs. But LEDs allows to do it at any scale, still on very small grows, and control of light delivered at each part of the plant may be controlled accurately, in order to get the max.

Increased energy efficiency

Any enhancement on efficacy may be related to an amount of photons. A better spectrum or a better distribution enhancement is applied to the baseline of PAR photons.

So the first and by far most important factor to determine the growing efficacy is the energy efficiency of the light: how much PAR watts it deliver per each watt burned.

But this factor is almost ignored on most LED grow lights development threads ive read, while is the most critical parameter.

For example, the difference on photosynthetic efficacy from a 635nm peaked led to a 660nm peaked led is around 5%. But most 660nm leds are 50% less energy efficients. Independent of the wavelenght efficacy, using 660nm leds is a bad choice (except if they have same energy efficiency than 635nm, but find me such one and i invite you to dinner ).

In order to compete with large HIDs with energy efficiencies about 36-38% is to use leds at least 30% efficients (an that if the HID is reflectorized). A bit less if the HID has a glass barrier.

But if you use LEDs with energy efficiencies below 25%, and most LEDs on sale are so, there is no way they may offer electric savings (same yield with less watts).

Forgot to analyze this critical factor had led to many dissapointing LED grows. So please, do it yourself a favour and think on the efficiency of the LEDs before buying them.

Happy growing
knna

knna Reviewed by knna on 10-25-2008. Building LED lights from facts, no theories I was going to post this at the Perfect LED Grow Light thread, but as some of what im going to post was posted 2 years ago on the stickied thread about LEDs and people still continue developing lights from wrong ideas, i think a thread about this topic is largelly needed. The main problem is related to efficacy of spectrums. When the firsts LED experiments at Overgrow, we work on the hypothesis that blue and red light are more effective. It was an appealing hypothesis that promises large Rating: 5

[knna]

Other esential parameter is often forgot when designing a LED array: what current we are going to use to run them.

Light emision of a led is directly linked to the current flowing across it. But any led emits at the higher efficiency as lower the current, and it drops as current increases.

Depending to the thermal design, there is a point when increasing further the current dont achive any increase of light. Increasing current from it raises strongly the electrical consuption without any increase of light emited. This is extreme, but what happen very often is using currents so high that efficiency is very low. As i explained on the last post, enough energy efficiency is key in order to achive electric savings.

So current used (or range of current used) may be carefully considered.

Some examples (thermal resistance solder point to ambient temp supposed to be 10 K/W, for a very good mounting with heatsink):

Cree XR-E coolwhite Q4 (luminous bin, 100-107lm@350mA): 28.5%@350mA, 22%@700mA

Cree XR-C red (R3:625-630nm dom wavelenght), bin M (39.8-51.7 lm@350): 29%@350mA, 25%@500mA, 20%@700mA

Cree XR-E Royal Blue, bin group 14 (350-425mW@350mA): 33%@350mA, 29%@500mA, 26%@700mA

Luxeon K2 red, bin S (typ 60lm@350mA): 26%@350mA, 23%@700mA

Luxeon K2 coolwhite, 200lm min group: 26%@350mA, 21%@700mA, 19%@1A, 13%@1.5A.

(Note: there is more unaccuracy on Luxeon products, as i have not digitalized their SPDs at operating conditions).

A LED cost the same independent of at what current you use it. So if you run 2 leds at 350mA instead of one at 1A, you get almost same light but using just a little more than half watts, but it initially cost you double.

But check that except for the Royal Blue, at 700mA or over all those leds that are state of the art currently gives efficiencies too low to be able to save watts against a large HID.

The Procyon, for example, probably use those red leds and lower bins of the whites (i believe P4). If they run them at 350mA (0.8 and 1.2W for red and whites/blues, respectively), and use double of leds, it may offer a good perfomance. Far for sustituting a 600W HID, but maybe able to get close to a 250w. But by driving them at 700mA to reduce costs, the perfomance is too low.

At the end, buyers only think on: "120watt for xx dollars, wow, its nice". If instead of watts, they say how many micromols of photons it deliver, we could compare perfomance. But they dont say anything about it, or how many PAR watts it emit. The only honest LED grow lights manufacturer who says the emission of the lamp is what makes the TI Smartlamp. And unfortunatelly its output (265uE if i remember well, similar to a reflectorized 250w HPS) is too low to do it cost effective.

Always there is a trade off between initial costs and efficiency. Always. So choose carefully the current or current range for your setup, in the way its the best compromise between both ends for your situation.

[justdirt]

procyon doesn't use any white leds dude

[knna]

procyon doesn't use any white leds dude

Yep, you are right. I wanted to mean blues (white leds are mostly a blue led covered with phosphors: bins of both are tightly linked, as the white always has about 20-25 lower efficiency than the correspondent blue due the Stokes effect).

And BTW, it should carry white leds. It probably enhances its perfomance. Although most people using it ends adding CFLs to give the white light it doesnt emit.

[redline]

Wow, finally someone on the forums that has a grasp of the situation with LEDs.

Here's my some of the first of many questions?

1. While HPS is more efficient the LEDS, how much of that disadvantage can LEDS make up by being closer? I have been building arrays that use no lenses, 140 to 160 degree view angle and trying to get them placed about an inch above the top of the plant. I think the closest you can get with HPS is about 6 inch for a 400 watter and 8 inch for a 600 watter. And that is with air cooled hoods and a light mover.
My personal thinking is 100 watts LED is equiv to about 60 HPS at this time.
I have yet to see a documented LED grow of more then .3 gram/watt/month and the grower has so far been unable to repeat that performance.

2. I am undecided on getting better efficiency by running about mid-range on current and allowing better area coverage with more LEDs. Or running them rascals as close to max ma as possible, spending more $$$ on cooling in order to get better penetration. Any thoughts?

3. What is your thinking in light ratios? Mine has been approx 3 to 1 red/blue with a little white light for insurance. However, I really want to experiment with a heavier blue ratio. I am close to the point where I think I can use LEDS in a vegging set-up where it will be able to outperform both CFL and HPS.

4. For some reason, LED in combo with CFL seems to produce the best results. Any thoughts?


I am currently building a test fixture where I can start experimenting with different set-ups. I will post a pic tomorrow of current progress.

[redline]

So I might be better off running discrete royal, blue, cyan instead a cool whites of equal wattage ....because I won't have to take an efficiency hit from the phosphor coating ?

[knna]

Wow, finally someone on the forums that has a grasp of the situation with LEDs.

After OG's fall i registered here. I posted some about lighting, but at that time this board becames very unconfortable, and i get very little feedback. So i went to other boards. Now things seems to go smoother here, and ive read good threads about LED on this board, so im here again. :rasta:

1. While HPS is more efficient the LEDS, how much of that disadvantage can LEDS make up by being closer? I have been building arrays that use no lenses, 140 to 160 degree view angle and trying to get them placed about an inch above the top of the plant. I think the closest you can get with HPS is about 6 inch for a 400 watter and 8 inch for a 600 watter. And that is with air cooled hoods and a light mover.
My personal thinking is 100 watts LED is equiv to about 60 HPS at this time.

Mant people had a bad understanding of the "square law" and its meaning. It refers to how fast irradiance drops with distance. But its due to a given amount of photons are dispersed on a exponentially increasing volume of space as distance increases. When you distribute small points of light along the grow, you deliver the light exactly where its needed, so you can keep irradiances at leaves surfaces very even, as you dont need the light of each LED cover a large volume.

But light from a top HID dont dissapear magically with distance :wtf: irradiance goes dropping from very high figures on top to very low on bottom and edges, but its not due to less photons avalaible, but to almost same photons diluting on a larger volume as distance increases.

Of course there is optical losses as the light must travel larger distances. Dust in air absorb some photons (more as dirtier the environment), aswell as light bouncing on walls let part of the light at them. And it obviously happen more as more far is the light source (and the reflector footprint match less the plants area).

So distance affect light efficacy by two ways: less total photons avalaible and suboptimal photons densities at leaves.

How much less photons are avalaible depends of a lot of factors: how well the reflector footprint matches the grow area, how much reflective are walls, how clean is the air, if there is a glass barrier between the bulb and plants, how much reflective is the light's hood...so its impossible to generalize those losses. It depends of each setup. A well designed LED setup almost eliminate these losses, and it allows to use less total photons than a HID to get same results.

How many less photons? Little difference if you compare with a vertical HID setup surrounded by plants. For top reflectorized HIDs, optical losses at reflector are 20-25% of the total amount of photons emited by the bulb. With a glass barrier (cooltubes and such), its a minimun of 35% of the light emited. These are the minimun less photons we may use on a LED setup to get similar yield. As i noted before, there are other optical losses, but it varies strongly between setups. For a well designed HID setup (reflective walls, ventilated and filtered (clean) area, reflector footprint mtaching the shape of grow area) they exist but are small, so we may ignore it for doing estimations.

On the other hand, what have a large enhancement of light productivity (yield/uE of photons) is keeping irrandiance at leaves on the optimun range. For a whole grow room, the only data i have about this is the 35% enhancement from the NASA study. We must calculate it for cannabis, but until we do it, we may use this figure (with a grain of salt ).

So, answering directly your question the fact of using the LEDs closer reduces slighty the light losses, although keep irradiances at optimal levels. We may suppose initially an enhancement of 35% on photons efficacy. It means we can get same yield using about 75% of the total photons emited by a top HID.

(Important math's note: we are talking of enhancement of photons efficacy, so we may apply it to the baseline of amount of photons. If we use 75% of photons with 35% enhanced efficacy it result of a total efficacy of 75%+(35% of 75%=0.35*0.75=26%)=101%. )

If the enhancement on cannabis is of 25%, we must use 80% of photons (to get same results than a top HID).

Having this reference in mind, we need know to know how many photons emit a HID avalaible to plants. It depend of the setup, so the amount of uE to be delivered by the LEDs in order to get same yield depends of with what setup you compare it. For example, a generic 600w HPS emits 1000uE (per second) (good ones as Grolux, Hortilux, Plantastar, Green Power or Sunmaster may reach up to 1100uE):

-Verticaly unreflectorized . Very little enhancement, especially if plants are kept short (usually with this type of setup), so we must use near same amount of photons delivered (1000uE).

-Top lighting with reflector. Optical losses at reflector are 20% (very good ones), 25% on average (dirty or low quality reflectors may have way higher losses). It mean about 750-800uE. If we suppose a 35% of enhancement for optimum irradiances, we need 550-600uE of leds.

-Sealed (glass barrier) top lighting. Average losses of 35% on clean setups. So just 650uE avalaible. We will need about 475-500uE to get similar yield.

Take in mind HID are less efficients at smaller wattages. So its easier (less watts of leds for each watt of HID) to get similar yields as smaller is the grow.

I have yet to see a documented LED grow of more then .3 gram/watt/month and the grower has so far been unable to repeat that performance

Yep, current perfomance LED commercial units are producing about 0.2 g/W per month. But this parameter is very important to decide when comercial units are cost effective, but of little usefulness on the actual stage of development. g/mol of photons is a way more meangliful and useful parameter, due LEDs are improving very fast, and for a given amount of uE you need now 100W of leds, but one year later only 60W. At it depends too of how hard (efficiently) you are running those leds. As i noted on my last post, it affect strongly to their efficiency (uE/W).

When we determine production in terms of g/mol of photons we can extrapolate that data to watts required of any LED with known emission, so we could estimate the amount of watts to install to get a given yield target.

I dont know how many times ive claimed that we may use LEDs with accurately known emission on our experiments so we could get universal conclusions that can be extrapolated. BTW, Ive been little sucesfull. So im running now a group buy of components of known emission with about 20 growers that are going to do logs with them, using differents styles of growing (soil, hydro, scrog, LST, sog...) so we can get conclusions fast of the amount of light required to grow cannabis and get similar yields than with HID, fluos, etc.

2. I am undecided on getting better efficiency by running about mid-range on current and allowing better area coverage with more LEDs. Or running them rascals as close to max ma as possible, spending more $$$ on cooling in order to get better penetration. Any thoughts?

Thats a very personal decision. You must find the best compromise between initial cost and efficiency: more leds at lower current means higher initial cost but lower efficiency.

On the actual LED's state of art, they just may be cost effective on very small grow spaces. So i suggest to build an experimental LED setup very small and use them at very good efficiency figures. The size of such grow determined for your budget.

I try to avoid as much as possible active cooling. Of course its doable, but a failure of fans may result on permanent damage or strongly reduced life of LEDs. I prefer to rely on passive cooling and LED distributed sparingly along the grow so power concentration is low (thus easy to cool down): lower cost on fans, less electric consuption, increased efficiency.

At least, i suggest to follow this way for side and IC modules. If you want max power on top, modules on top may be actively cooled. Indeed, PC's socket heat dissipators are one of the cheapest options as heatsinks, and they comes with fans. You can conect the fans or not, depending on the density of leds you install and current used. Those bulky heatsinks arnt a problem on top, butthey are on IC lighting. As on IC we need lower power densities, often passive cooling are a better choice.

You dont need any penetration ability if you use the LEDs between plants, so you always may place LEDs very close to plants and using wide emission angles.

Relying on light penetration is a limitation of HIDs, not an advantage. We are so used to work that way that we not realize how bad its it: produce uneven light distributions (reduced productivity), and light passing though leaves increase strongly its wavelenght, resulting of a strongly far red lighting on bottom plants that produces fuffly buds.

3. What is your thinking in light ratios? Mine has been approx 3 to 1 red/blue with a little white light for insurance. However, I really want to experiment with a heavier blue ratio. I am close to the point where I think I can use LEDS in a vegging set-up where it will be able to outperform both CFL and HPS.

Check the Inada curve. It shows the amount of photosynthesis promoted for 1W of light of each wavelenght (2nm bands) (isolated, each waveband at a time; study performed on sunlight grown plants):

(i was to post here the attachment, but i dont know how to do it; anyway, its on the bottom of the post, open it).

The curve 1 is the average response of 27 herb plants (curve 2 are for trees). Note how the minimun photosynthetic response is about 465nm, the emission range of blue leds! Still lower than of green light.

Blue light is added because plants need it, not because increased photosynthetic efficacy (at least, at no CO2 limited irradiances). Cannabis is a little demanding plants in term of light quality: it needs very little (if any at all) blue to be able to grow (opposite to most plants). But blue light has some "colateral" effect that advice to use it: first, the lack of blue light strongly increases internodal distance, so you need to add some blue if you want the compact plants we seek for indoor. But just 30-40uE per sq meter are enough for this, while we rarely use less the 500uE/sq meter of average radiances. So we need less than 10% photons of blue, often just about 6% of total radiance in order to keep plants short and compact. As blue photons carry way more energy than red ones (in direct correspondence to its wavelenght: a 450nm photon carry 650/450 more energy than a 650nm photon), if we talk about PAR watts instead of number of photons (uE), then that 6% correspond to about 10% of total PAR watts. This are minimum requeriments.

But blue photons have another collateral effect that may advice to use higher amounts of it, and its especially important on LED grows due the high percentage of red photons: blue photons promote the opening of stomatas, while red photons promote the opposite (maximum effect peaked at 450 and 660nm, respectivelly). It affects the transition point between the light limited part of the photosynthetic response curve to the CO2 limited part of the curve.

So the higher the irradiances used, the higher the percentage of blue you need to add. Thats why blue enhanced HPS bulbs (or MHs) tend to work better as higher the w/sq ft used, while it offer little enhancement of gardens using low light densities

So if you keep irradiances low to medium (up to 250-300 uE/m2), you wont need to add more blue. But we often use higher irradiances in order to get fat colas, and then the internal (leaves) concentration of CO2 becomes very important to get good photosynthetic efficacies. So in that situation, if you are growing with LEDs and you want to keep light's productivity high, you need either supplement with CO2 or add more blue light so plants are able to keep CO2 internal concentration higher.

On a side note, the ratio of Blue, Green and Red (simplified ranges for B:400-500nm, G:500-600nmn, R:600-700nm) must be refered to the amount of uE or PAR watts, not to the number of leds or watts installed, as leds of different colors and running currents have different efficiencies.

4. For some reason, LED in combo with CFL seems to produce the best results. Any thoughts?

Thats due commercial units using just blue and red have spectral lackings that results on reduced perfomance. By adding white light, the perfomance of all photons are improved.

Apart of it, people are using to little photons per sq ft because they are wrong about LEDs perfomance. They strongly overstate it so they end installing too little watts. By adding CFLs, they complete the spectrum and add some more photons.

(Will you expect a good yield using a 100w HPS on 1 sq meter? LEDs arnt different, if you dont give enough photons you cant expect good yields)

[knna]

So I might be better off running discrete royal, blue, cyan instead a cool whites of equal wattage ....because I won't have to take an efficiency hit from the phosphor coating ?

The problem is yellow and green leds have low efficiencies. As we require very little of it, aswell of cyan, the most profitable way of adding them is with white leds. Coolwhite leds have a huge peak around 450nm and then a continous spectrum peaking on yellow but that go beyond 700nm. So they complement very well with red leds, as gives the rest of wavelenghts and still some far red, wich its needed too. Coolwhite leds often have about 21% less efficiency than comparable Royal blues due to the change to longer wavelenghts at phosphors. But although it lower in that figure the energy efficiency, most of that drop happen on wavelengh conversion of photons, so they emits a very similar amount of photons than the Royal Blue inside it.

But today currently there is a way of giving that white light with a similar spectrum by using fluorescent tubes, at least from top ligting, wich is way cheaper. Using a Reflex tube with electronic ballast you get still higher energy efficiencies than a reflectorized HPS, and more than with white leds.

White leds emiting 150 lm/w are expected to be released in about 2 years (they are already developed in labs). By that time LEDs wont have alternative. But currently is reasonable to cut initial cost by giving the white light with fluorescent tubes or still with CFLs (for experimental purposes it dont mind if its slighty less efficient).

As red leds are cheaper, using a conventional white light and red leds is a cheap way to have good productivities on small gardens without investing huge money. And splitting the investment: now red leds and in two years, white ones.

There are Royal Blue LEDs very efficients, so they may be used together with white leds or white light in order to obtain the desired spectrum.

A very interesting page about optimal light spectrums for plants is this article: OPTIMIZATION OF LAMP SPECTRUM FOR VEGETABLE GROWTH.

We need to perform similar experiments with cannabis, but im working with the optimum RGB distribution of tomatoes as starting point (10-20%:15-20%:60-75%; B:G:R).

As coolwhite leds (aswell as daylight floros, with similar CCT ratings (6400K)) emits a bit more on the green than in the blue, but balanced, just adding to them red leds result on very well balanced spectrums.

For example, this is the result of 3 reds and 1 coolwhite all running at 350mA (CREES XR, white color WF (5700-6400K) and red R3 (625-630nm dominant), luminous bins Q4 and M):

[attachment=o206256]

It have a B:G:R distribution of 11:14:75% (R/B=7) wich seems pretty well balanced for flowering at medium-low irradiances.

The nice thing about leds is they are easily current adjustable, so we can modify the RGB distribution to adapt a given setup to different stages (veg, early bloom, late bloom) by modifying the current feed to each color. For example, with the same mix, if we put the white at 700mA but keep reds at 350mA, the spectrum is 14:19:67%.

Running all at 700mA results on 9.5:13.5:77% (both efficiencies and spectrum varies with current and temperature).

Options including Royal Blue are possible, especially when thinking on vegging lights, or using high irradiances along flowering. A mix 4R3+CW+RB result on this:

[attachment=o206257]

With a distribution of 29.5:9.5:61%.

Or, for example, 8R3+3CW+RB:

[attachment=o206258]

With a distribution of 22/14/64%.

Just a few samples of the spectrum you may create using red and white or red, white and blue.

[knna]

Well, this thread isnt been sucessfull. My only purpose when i started it was to help poeple wanting to develop thrier own LED lights.

I noticed there was here some people who want to do it the right way, and thats why i posted some info. Im lighting designer (meaning i work on it every day), i talk often with LED companies distributors and still sometimes with developers. I have a first hand experience in this field and i ve been researching from years ago about LED grow lights.

So i thought (wrong) i could help guiding others by giving them the tools and basic knowledge aswell as informing them of ways that has been proven wrong.

I use same nick on all boards, anybody may do a search and check that. For me is very sad that some "smart" people took the early research a group of growers did at the Overgrow boards and use it to do LED grow lights that simply dont work because they are based on biased premises and wrong research.

But its still sadder than people trust on obvious biased and false info from LED sellers that are mostly marketing hype than on people who provides the scientific facts and the tools to check them by yourself.

But i realized its not my problem. If people want to waste their money is their problem, not mine. Ive already posted on the net the info needed to avoid it. I cant do more than that.

From now on, i give up giving free reliable info about LED products. Im going to concentrate on earn money. Ill develop my lamps based on what i learned on the MJ boards: people prefer a reasonable and appealing lie than a hard truth.

It has been a nice social experiment, from what i learned a lot of the right way of marketing. Product quality dont matter on the industry for stoners. I already knew it for nutes products, but i thought that the strong impact of light efficacy should be evident. It isnt. People prefer to believe what they want to believe, still when real world clearly clearly points up its wrong. Selective blindness is the strongest power on the MJ boards.

BTW, this way i dont waste my time trying to write understable enough on a lenguage that isnt mine.

[justdirt]

bitter much?