James87
Active Member
My first post here, and I thought someone might appreciate what I've learned from building my own LED panels. I've only built three so far, and the first two were identical. The third basically came from what I learned about the first two, and I'm preparing to build a 4th (which would be my 3rd generation), and would correct some of the issues I discovered from the 2nd gen. I'm also hoping to learn from others while I'm sharing info. I'm also going to attempt to limit how much geekspeak I employ as I describe the electrical stuff. It would help the reader immensely if its already well-understood what V=IR and P=VI means...
I decided to use LEDs for many reasons. Right now I'm using a hybrid light system, of CFLs and LEDs, and I'm in my 2nd week of growing. I'm not new to growing, but this is my first indoor operation.
The Box
I built a nice cabinet, 2x4x7, with a hinged door and lock that creates an airtight seal. I'm using carbon filters intake and outtake, to eliminate scent and light leakage, and help prevent dust from getting inside. Intake and outtake are on the top of the box, and the intake is directly under a furnace vent. Furnace is natural gas, and I figure the CO2 emission would be a good thing. Also I figured that keeping all vents on the top helps heat escape, and between CO2 and O2, the CO2 will fall and O2 will rise, so its also acting as a CO2 scrubber, in a way. I have two 80mm fans on the intake, blowing warm air down. This creates a natural air circuit. I have another fan directly on my plant (yes only one.. the rest of these seeds didn't make it). I plan to get better seeds, but for now this is a test. Learn what I can, and the 2nd time around I'll have all the kinks worked out. The entire box, inside and out, I've primed with acrylic latex paint. I'll paint the outside another day, and the inside I've tacked up panels of the white/black plastic (visqueen, but this is some other brand, as it has several names). The box itself is mostly 2x2's and high-density fiberboard (1/8") walls. Floor and ceiling are 3/8" plywood. All the corner joints have those steel L brackets, and its very rigid. Before I put the ceiling up, I could do pullups inside it. I fed a 50' extension cord in through the top, connected to a timer, and the timer controls a power bar.... because right now I have two different power supplies and the CFL circuit.
Power
I'm a little concerned with what I've read in several posts about all the high-power HIDs and MH etc, multiple 1000 watt lights. A typical circuit from your breaker panel is only 15 amps for 110-120V AC. This limits power consumption to about 1650-1800 watts before the breaker pops on you. Having two 1000 watt bulbs will cause problems. Simply replacing the breaker with something bigger without replacing the house wiring is a big no-no, because typical wiring on a 15A circuit will use 14-gauge (American Wire Gauge) wire, and pushing 20A through this will make the wires in your house get very hot, similar to how a plug is warm/hot after using a hairdryer or whatever. Get an electrician to install a 30A circuit and you can use three 1000W bulbs no problem.. but not four.
Anyway, I want to discuss power for LEDs. There are various circuits you can use, and various supplies you can use... I thought I would use 12V from an old PC power supply (aka switching supply) and designed my LED circuits based on that. The problem here is that power supplies have a power curve, so voltage tends to drop as you draw more current, and now I'm faced with designing my own power supply just for the LEDs. I'm formally trained in electronics, so this is no problem! Its just a pain in the ass. I decided for this I'm also going to use a Constant Current circuit for each LED panel, because the first two are so different even though they are somewhat identical. I'll explain why shortly. I'm also going to provide photos and my schematics for the lights and the supply.
Why LEDs?
This seemed interesting for me, and that was enough to get me started. What I like about LEDs are that you can choose specific wavelengths of light, and have a selection of how wide of a beam you get from them. Regular bulbs produce light in 360 degrees and require a hood of some kind to redirect lost light. LEDs can focus that light and don't require a hood at all. They're also relatively cold, and with the right lights you start approaching 100% efficiency, of wattage spent to photons absorbed by the plant. The power losses are in the power supply and any resistors used, and a very little amount of heat. You can select an LED that has 20,000 mCd but a very tight beam, and its blinding. Create an array of them, and you have a very blinding spotlight. The downside is you have to articulate the panel so its facing where you need it; looking at the panel from any angle other than straight-on won't seem very bright at all, so, many of the photos of LED panels I've seen on the web look like crap. When I come out of that box, I feel like I stepped inside from noon sunshine, as regular house-lighting seems so dark.
First Generation LED Panels
The reason they are different is because when my LEDs arrived in the mail, I sorted them first by how much voltage they required, individually. I built a simple test circuit using a 12V DC supply, 5 LEDs and a resistor. Using a multimeter to measure current, I selected a resistor that gave me about the right current I needed, and then used the multimeter to measure voltage across two nodes where I place an LED to test. I ordered 440 LEDs, the 30mA kind with a 25 degree beam, 10mm diameter in blue, pink, red, white and yellow. It came to 300 red, 40 blue, 20 white, 40 pink, and 40 yellow. The reds were around 625nm wavelength. Pink is actually sort of blue/violet and tighter wavelength than blue, so I thought it would compliment well. Yellows are more of an amber or almost orange, something like 585 to 595nm. Obviously I wanted to avoid the green band, but they're looking kinda green as the blue and yellow are right next to eachother. Future builds will use a better selection. Anyway, so I sorted the LEDs and put them in piles based on color and voltage. Typically they have a tolerance rating around 5%, so some were 2.1V and others were 2.3V of the same color. The blue/pink/white all were around 3.2V. I figured when I assembled the panels, I'd use a combination of low's and high's until all that were left were the medians, and try to keep the combined sum of voltage-drops the same. The big problem here was I planned to use the same resistor value for every series circuit. I had five LEDs in each series circuit with a 39 ohm resistor. As I designed the panel, some of the 3.2V LEDs couldn't have any 2.1V LEDs, so I used a 120 ohm resistor with three 3.2V bulbs. Only three of these circuits are in each panel because I couldn't neatly arrange them without using wire jumpers, so I just used a different resistor. This produced about the same current. You see, LEDs are diodes which are current-operated devices. Voltage changes slightly +/- based on temperature, and whatever voltage is left is across the resistor, which determines how much current that circuit will use. I can show the math but its futile because you'll never get it right this way, and I tried really hard getting it right. So, I'm going to be using a constant current source, and everything else can stay the same. My recommendation is make sure every series-circuit you have on your panel is identical, use a very low resistor value (1 ohm metal oxide (1% tolerance)), and use a constant current source. Each series circuit is then connected in parallel together to the power source. I noticed while sorting that the LEDs seemed to be in groups of identical voltage, so I would have a pile of exactly 2.150, two smaller piles of 2.130 and 2.170, etc. I also ordered them at different times so that the shipper (ebay) kept my orders under $25 so I wouldn't have to pay customs fees or taxes. The problem with that was I also had a nice pile of exactly 1.950V, two smaller piles of 1.930 and 1.970, etc.. When they are produced, the machine making them will make them with 5% tolerance, but a second machine will have a slightly different rating when producing them. I ended up noticing that all the red LEDs must have been produced on exactly three machines, and I didn't account for that when designing the panels. Huge caveat, because when you lock-down the current for each 5 LED circuit with a resistor, some circuits will be overpowered and some underpowered, by as much as 400% both ways. What I should have done is test every single 5 LED circuit and replace one bulb at a time until I have exactly what I need for that circuit. Or maybe I should have written software that tells me exactly which bulb to use after measuring them all, so that all 88 series-circuits end up with 5% tolerance. 88x5 is 440. Right now, some of the bulbs have died on me because those circuits were overpowered, and I have to get the multimeter out again to figure out which ones popped, and replace them. When I installed these two panels, they drew about 1.75 amps together. The 1st panel (made of highs and lows) is only using 0.75 amps, and the 2nd panel (much tighter tolerances because I was left with the middle voltages) is using 1.0 amps, and that 2nd one has a much more consistent brightness across it. I also noticed that the pinks and yellows appeared half brightness, so I'm not sure if thats because of human sensitivity to those bands, or what. I used just these two for the seedling, and it seemed to love them after I moved them a little closer. It stopped trying to stretch/reach the panels and went right into veg.
Second Generation LED Panel
I decided to use so-called "1 watt" LEDs this time, 40 degree beam, of the red and near-infrared variety. 625nm and 660nm, respectively. So-called because they're not 1W LEDs at all. They operate at 300mA, the 625nm's were 2.4V and the 660nm's were 2.7V. So, 2.7V x 0.3A is only 0.81W. The plan was to put 50 of them together on a panel, and I was expecting around 36W power consumption at 3 amps. Well, for some reason the old computer power supply I'm using just can't maintain the voltage to drive 3 amps, and I'm wondering if thats why it was just laying around... It drops to something like 10.5V and so, the current draw is just pitiful. So I installed some CFLs today so that later I can muck with a custom power supply without confusing the hell out of the plant. Its all on a 21/3 cycle and that 3-hour window to work on the panels is pretty tight, so I needed extra lights in there so I can remove the panels to work on them during the daytime. I used 1 ohm resistors, and discovered the PC supply I'm using has such a high internal resistance that I didn't even need resistors.... The circuit for this was much simpler, as its just a 10x5 to fit all 50 bulbs. All the series circuits go up/down and I ran power and ground laterally across the top and bottom. This configuration is much healthier for me, mentally! You'll see what I mean when I show the circuits used on the first two panels. 3 of the 50 are now dead, and I had to short the dead ones with a resistor to limit current.... I'm very unhappy about that. The supply I'm using is going in the trash when I upgrade; you can see some of the lights flicker and stuff, and it probably killed those 3 bulbs because they were weak to begin with. I could tell those were weak because they slowly required more voltage to stay on, and eventually 3 columns (15 bulbs) stayed OFF because there wasn't enough voltage to drive them. 2 still worked until I forced them to use 300mA and they went dead. But anyway, 47 are still good.
3rd Generation
I'm going to design the next panel after I build my power supply. The plan here is to avoid ebay, and order from reputable parts suppliers instead. I only used ebay because of the attractive prices of those chinese-made LEDs, and where I live any kind of sourcing is damn-near impossible without paying out the nose and ass! I mean, I know what parts should cost, but a store that specializes in selling individual components (like the one near me) wants $8 on a $0.20 part. Ridiculous! Parts add up fast, and this would have turned into a $1000 order instead of the $50 I spent. Digikey seems to be a good place to order from, and has a wide variety of LEDs, much wider than ebay, and better prices. I plan to use 3W LEDs, and target 8 wavelengths. A 300W panel, with 40 degree beam at 85% plant power efficiency would be equivalent to something like a 2700 watt bulb if it could also have 85% plant power efficiency. What I mean by plant power efficiency is the wattage spent vs what the plant can use. For this I am assuming 100% of the wavelengths are used by the plant, compared to 25% (r+b) for CFL and all the rest. I'm assuming 85% because some of the power is lost converting to DC, and lost as heat radiating from the semiconductors used for the constant current supply. It should be noted that ballasts aren't 100% efficient, either. If a bulb were at 50% (r+b) then it would have to be 5400W to match this 300W LED panel. There's a reason why NASA is planning to use LEDs for deep-space food production; LEDs really are the future. And I thought HIDs were cool, friggen arc-welding brightness doesn't even come close to the efficiency of LEDs.
The Constant Current Supply
I was looking at various circuits I can use for this, when I stumbled across one that is very simple and only uses a few parts. Basically, one NPN transistor and one N-Channel MOSFET. I have a spare 750W transformer that will give me 25VDC, which should be enough to account for most loads. The plan is to have several of these Constant Current circuits in a box so each one drives a single panel at a time. Each is rated to provide up to 15 amps or so. 15 amps at 12 volts is 180W, and that would be a pretty damn bright panel! Something else worth noting, is that when bulbs die, the whole series-circuit stops conducting where the bulb died. You lose all 5 when 1 goes, when you have 5 per circuit. You can actually place a regular diode in parallel with each LED so that if the LED dies, that circuit will still conduct current. What I'd like to do is make it modular, so I can easily replace each LED when they go, like some kind of minimal insertion-force plug, or like speaker connections on stereos.
The AC-only Supply
There is a way to connect these LEDs directly to a 120VAC outlet. Definitely do not try this unless you know exactly what you are doing! The idea is that the diodes only function when "forward biased", that is, when power is only going in one direction. But, you can have 50 or so in series going one way, and 50 or so going the other way, both in parallel. Then, 60 times a second one row lights up, then the other. In Europe obv this would be 50Hz and not 60Hz, but whatever. The current problem still exists, where trying to pin-down the exact current you want to drive the diodes will fluctuate wildly based on operating temperature and the tolerance of the mains.. 115V is common but sometimes it drops to 108V or spikes to 122V. So, its effectively unreliable. The only clear winner is a constant current source. Also, having 120VAC exposed near your work area will always be a hazard, forcing you to build an enclosure for them. DC is nice like that, you can handle the exposed backside of these circuit boards (led panels) barehanded while they're on.
I decided to use LEDs for many reasons. Right now I'm using a hybrid light system, of CFLs and LEDs, and I'm in my 2nd week of growing. I'm not new to growing, but this is my first indoor operation.
The Box
I built a nice cabinet, 2x4x7, with a hinged door and lock that creates an airtight seal. I'm using carbon filters intake and outtake, to eliminate scent and light leakage, and help prevent dust from getting inside. Intake and outtake are on the top of the box, and the intake is directly under a furnace vent. Furnace is natural gas, and I figure the CO2 emission would be a good thing. Also I figured that keeping all vents on the top helps heat escape, and between CO2 and O2, the CO2 will fall and O2 will rise, so its also acting as a CO2 scrubber, in a way. I have two 80mm fans on the intake, blowing warm air down. This creates a natural air circuit. I have another fan directly on my plant (yes only one.. the rest of these seeds didn't make it). I plan to get better seeds, but for now this is a test. Learn what I can, and the 2nd time around I'll have all the kinks worked out. The entire box, inside and out, I've primed with acrylic latex paint. I'll paint the outside another day, and the inside I've tacked up panels of the white/black plastic (visqueen, but this is some other brand, as it has several names). The box itself is mostly 2x2's and high-density fiberboard (1/8") walls. Floor and ceiling are 3/8" plywood. All the corner joints have those steel L brackets, and its very rigid. Before I put the ceiling up, I could do pullups inside it. I fed a 50' extension cord in through the top, connected to a timer, and the timer controls a power bar.... because right now I have two different power supplies and the CFL circuit.
Power
I'm a little concerned with what I've read in several posts about all the high-power HIDs and MH etc, multiple 1000 watt lights. A typical circuit from your breaker panel is only 15 amps for 110-120V AC. This limits power consumption to about 1650-1800 watts before the breaker pops on you. Having two 1000 watt bulbs will cause problems. Simply replacing the breaker with something bigger without replacing the house wiring is a big no-no, because typical wiring on a 15A circuit will use 14-gauge (American Wire Gauge) wire, and pushing 20A through this will make the wires in your house get very hot, similar to how a plug is warm/hot after using a hairdryer or whatever. Get an electrician to install a 30A circuit and you can use three 1000W bulbs no problem.. but not four.
Anyway, I want to discuss power for LEDs. There are various circuits you can use, and various supplies you can use... I thought I would use 12V from an old PC power supply (aka switching supply) and designed my LED circuits based on that. The problem here is that power supplies have a power curve, so voltage tends to drop as you draw more current, and now I'm faced with designing my own power supply just for the LEDs. I'm formally trained in electronics, so this is no problem! Its just a pain in the ass. I decided for this I'm also going to use a Constant Current circuit for each LED panel, because the first two are so different even though they are somewhat identical. I'll explain why shortly. I'm also going to provide photos and my schematics for the lights and the supply.
Why LEDs?
This seemed interesting for me, and that was enough to get me started. What I like about LEDs are that you can choose specific wavelengths of light, and have a selection of how wide of a beam you get from them. Regular bulbs produce light in 360 degrees and require a hood of some kind to redirect lost light. LEDs can focus that light and don't require a hood at all. They're also relatively cold, and with the right lights you start approaching 100% efficiency, of wattage spent to photons absorbed by the plant. The power losses are in the power supply and any resistors used, and a very little amount of heat. You can select an LED that has 20,000 mCd but a very tight beam, and its blinding. Create an array of them, and you have a very blinding spotlight. The downside is you have to articulate the panel so its facing where you need it; looking at the panel from any angle other than straight-on won't seem very bright at all, so, many of the photos of LED panels I've seen on the web look like crap. When I come out of that box, I feel like I stepped inside from noon sunshine, as regular house-lighting seems so dark.
First Generation LED Panels
The reason they are different is because when my LEDs arrived in the mail, I sorted them first by how much voltage they required, individually. I built a simple test circuit using a 12V DC supply, 5 LEDs and a resistor. Using a multimeter to measure current, I selected a resistor that gave me about the right current I needed, and then used the multimeter to measure voltage across two nodes where I place an LED to test. I ordered 440 LEDs, the 30mA kind with a 25 degree beam, 10mm diameter in blue, pink, red, white and yellow. It came to 300 red, 40 blue, 20 white, 40 pink, and 40 yellow. The reds were around 625nm wavelength. Pink is actually sort of blue/violet and tighter wavelength than blue, so I thought it would compliment well. Yellows are more of an amber or almost orange, something like 585 to 595nm. Obviously I wanted to avoid the green band, but they're looking kinda green as the blue and yellow are right next to eachother. Future builds will use a better selection. Anyway, so I sorted the LEDs and put them in piles based on color and voltage. Typically they have a tolerance rating around 5%, so some were 2.1V and others were 2.3V of the same color. The blue/pink/white all were around 3.2V. I figured when I assembled the panels, I'd use a combination of low's and high's until all that were left were the medians, and try to keep the combined sum of voltage-drops the same. The big problem here was I planned to use the same resistor value for every series circuit. I had five LEDs in each series circuit with a 39 ohm resistor. As I designed the panel, some of the 3.2V LEDs couldn't have any 2.1V LEDs, so I used a 120 ohm resistor with three 3.2V bulbs. Only three of these circuits are in each panel because I couldn't neatly arrange them without using wire jumpers, so I just used a different resistor. This produced about the same current. You see, LEDs are diodes which are current-operated devices. Voltage changes slightly +/- based on temperature, and whatever voltage is left is across the resistor, which determines how much current that circuit will use. I can show the math but its futile because you'll never get it right this way, and I tried really hard getting it right. So, I'm going to be using a constant current source, and everything else can stay the same. My recommendation is make sure every series-circuit you have on your panel is identical, use a very low resistor value (1 ohm metal oxide (1% tolerance)), and use a constant current source. Each series circuit is then connected in parallel together to the power source. I noticed while sorting that the LEDs seemed to be in groups of identical voltage, so I would have a pile of exactly 2.150, two smaller piles of 2.130 and 2.170, etc. I also ordered them at different times so that the shipper (ebay) kept my orders under $25 so I wouldn't have to pay customs fees or taxes. The problem with that was I also had a nice pile of exactly 1.950V, two smaller piles of 1.930 and 1.970, etc.. When they are produced, the machine making them will make them with 5% tolerance, but a second machine will have a slightly different rating when producing them. I ended up noticing that all the red LEDs must have been produced on exactly three machines, and I didn't account for that when designing the panels. Huge caveat, because when you lock-down the current for each 5 LED circuit with a resistor, some circuits will be overpowered and some underpowered, by as much as 400% both ways. What I should have done is test every single 5 LED circuit and replace one bulb at a time until I have exactly what I need for that circuit. Or maybe I should have written software that tells me exactly which bulb to use after measuring them all, so that all 88 series-circuits end up with 5% tolerance. 88x5 is 440. Right now, some of the bulbs have died on me because those circuits were overpowered, and I have to get the multimeter out again to figure out which ones popped, and replace them. When I installed these two panels, they drew about 1.75 amps together. The 1st panel (made of highs and lows) is only using 0.75 amps, and the 2nd panel (much tighter tolerances because I was left with the middle voltages) is using 1.0 amps, and that 2nd one has a much more consistent brightness across it. I also noticed that the pinks and yellows appeared half brightness, so I'm not sure if thats because of human sensitivity to those bands, or what. I used just these two for the seedling, and it seemed to love them after I moved them a little closer. It stopped trying to stretch/reach the panels and went right into veg.
Second Generation LED Panel
I decided to use so-called "1 watt" LEDs this time, 40 degree beam, of the red and near-infrared variety. 625nm and 660nm, respectively. So-called because they're not 1W LEDs at all. They operate at 300mA, the 625nm's were 2.4V and the 660nm's were 2.7V. So, 2.7V x 0.3A is only 0.81W. The plan was to put 50 of them together on a panel, and I was expecting around 36W power consumption at 3 amps. Well, for some reason the old computer power supply I'm using just can't maintain the voltage to drive 3 amps, and I'm wondering if thats why it was just laying around... It drops to something like 10.5V and so, the current draw is just pitiful. So I installed some CFLs today so that later I can muck with a custom power supply without confusing the hell out of the plant. Its all on a 21/3 cycle and that 3-hour window to work on the panels is pretty tight, so I needed extra lights in there so I can remove the panels to work on them during the daytime. I used 1 ohm resistors, and discovered the PC supply I'm using has such a high internal resistance that I didn't even need resistors.... The circuit for this was much simpler, as its just a 10x5 to fit all 50 bulbs. All the series circuits go up/down and I ran power and ground laterally across the top and bottom. This configuration is much healthier for me, mentally! You'll see what I mean when I show the circuits used on the first two panels. 3 of the 50 are now dead, and I had to short the dead ones with a resistor to limit current.... I'm very unhappy about that. The supply I'm using is going in the trash when I upgrade; you can see some of the lights flicker and stuff, and it probably killed those 3 bulbs because they were weak to begin with. I could tell those were weak because they slowly required more voltage to stay on, and eventually 3 columns (15 bulbs) stayed OFF because there wasn't enough voltage to drive them. 2 still worked until I forced them to use 300mA and they went dead. But anyway, 47 are still good.
3rd Generation
I'm going to design the next panel after I build my power supply. The plan here is to avoid ebay, and order from reputable parts suppliers instead. I only used ebay because of the attractive prices of those chinese-made LEDs, and where I live any kind of sourcing is damn-near impossible without paying out the nose and ass! I mean, I know what parts should cost, but a store that specializes in selling individual components (like the one near me) wants $8 on a $0.20 part. Ridiculous! Parts add up fast, and this would have turned into a $1000 order instead of the $50 I spent. Digikey seems to be a good place to order from, and has a wide variety of LEDs, much wider than ebay, and better prices. I plan to use 3W LEDs, and target 8 wavelengths. A 300W panel, with 40 degree beam at 85% plant power efficiency would be equivalent to something like a 2700 watt bulb if it could also have 85% plant power efficiency. What I mean by plant power efficiency is the wattage spent vs what the plant can use. For this I am assuming 100% of the wavelengths are used by the plant, compared to 25% (r+b) for CFL and all the rest. I'm assuming 85% because some of the power is lost converting to DC, and lost as heat radiating from the semiconductors used for the constant current supply. It should be noted that ballasts aren't 100% efficient, either. If a bulb were at 50% (r+b) then it would have to be 5400W to match this 300W LED panel. There's a reason why NASA is planning to use LEDs for deep-space food production; LEDs really are the future. And I thought HIDs were cool, friggen arc-welding brightness doesn't even come close to the efficiency of LEDs.
The Constant Current Supply
I was looking at various circuits I can use for this, when I stumbled across one that is very simple and only uses a few parts. Basically, one NPN transistor and one N-Channel MOSFET. I have a spare 750W transformer that will give me 25VDC, which should be enough to account for most loads. The plan is to have several of these Constant Current circuits in a box so each one drives a single panel at a time. Each is rated to provide up to 15 amps or so. 15 amps at 12 volts is 180W, and that would be a pretty damn bright panel! Something else worth noting, is that when bulbs die, the whole series-circuit stops conducting where the bulb died. You lose all 5 when 1 goes, when you have 5 per circuit. You can actually place a regular diode in parallel with each LED so that if the LED dies, that circuit will still conduct current. What I'd like to do is make it modular, so I can easily replace each LED when they go, like some kind of minimal insertion-force plug, or like speaker connections on stereos.
The AC-only Supply
There is a way to connect these LEDs directly to a 120VAC outlet. Definitely do not try this unless you know exactly what you are doing! The idea is that the diodes only function when "forward biased", that is, when power is only going in one direction. But, you can have 50 or so in series going one way, and 50 or so going the other way, both in parallel. Then, 60 times a second one row lights up, then the other. In Europe obv this would be 50Hz and not 60Hz, but whatever. The current problem still exists, where trying to pin-down the exact current you want to drive the diodes will fluctuate wildly based on operating temperature and the tolerance of the mains.. 115V is common but sometimes it drops to 108V or spikes to 122V. So, its effectively unreliable. The only clear winner is a constant current source. Also, having 120VAC exposed near your work area will always be a hazard, forcing you to build an enclosure for them. DC is nice like that, you can handle the exposed backside of these circuit boards (led panels) barehanded while they're on.
LEDs being an equiv. of 36W draw, what would you say would be equivalent to this in terms of how many 25W CFL you would need for the same light absorption to the plant? I guess assume optimum spectrum of LED and CFL. I am looking to start a small cabinet grow with maybe 4 to 6 plants in flower at a time. Can you give some more information on maybe how many LEDs per panel you would need for a grow like that.
