• Home
  • /DIY LED Grow Light

DIY LED Grow Light

This article will be in two parts: my history of the search for an optimal grow light for a small 2’ x 2’ space, and the construction of such a light to achieve over a gram per Watt.

How it came to be

The first part of my journey was trying to understand why growers (in general) were using two separate lamps, metal halide (MH) for vegetation and high pressure sodium (HPS) for flowering, while there was only a single outdoor light source: the sun, for the entire life cycle.

Blue for veg and red for flower is an oversimplification that underscores common misperceptions of plant physiology but it is still a good concept to understand.

In the HID world HPS is the champ in converting electricity to light (110-150 lumens per Watt (LPW)), why use the less efficient (90 to 105 lumens per Watt) MH? There is insufficient blue in HPS bulbs (though this is improving) to prevent extreme stretch in the early stages of growth, thus the much higher blue content of MH was chosen. Plants grow towards blue light. This is called phototropism. When there is not enough light in the blue spectrum, plants will keep searching or stretching trying to find it.

So why not continue all the way through the entire growth cycle with MH? Red chlorophyll A & B receptors have a higher response than the blue chlorophyll A & B receptors; therefore let’s give the plants the spectrum they respond best to. As upward growth slows or stops in early to mid-flowering, blue light is not as critical to marijuana growth at that point in time.

But here is the kicker, HPS puts out the majority of its energy in the yellow spectrum; read: the ‘wrong’ theoretical spectrum, so how come indoor growers get such great results? The simple answer is that plants use almost the entire visible light spectrum, though at different efficiencies, and can adapt to imperfect or out of balance light conditions. Through a series of highly complex chemical processes at the molecular level, cannabis can up or down-regulate the light received into a more useful form.

Enter bi-color LEDs that just focus on the red and blue chlorophyll A receptors. This was based off of work done at NASA using very early and inefficient emitters. Most commercial lamps mimicked this research and copied the 6 (or 8) to 1 red to blue ratio. I believe this to be less than optimal as can easily be observed by looking at the photosynthetic response curve. The red response is not multiples greater than the blue response.

Response Photosynthesis

What NASA did prove is that plants can be grown from seed to fruition on merely two narrow spectral bands. This is good information, but does it point to optimal growth?

My first foray into commercial LEDs was the purchase of three (one red, one blue and one mixed) 45 Watt Chinese LED panels using 0.5 Watt emitters. These actually worked OK, but for a few caveats.

  • The power supplies were not current regulated and all three burned up. This was just poor engineering and could have been redesigned. Good commercially made LED lights have self protecting power supplies.
  • One other problem was the laser-like focus of these narrow-angle LEDs. This could cause bleaching (super saturating the chlorophyll receptors turning them white) if too close and resulted in uneven lighting or spotting. Again if you look at how the pros make their products they use 90° angle beam lenses;
  • Then there was the issue of color mixing, even on panels using both color LEDs. Each leaf would not receive the same balance. Color mixing means you need the LEDs close to each other, with lenses and good thermal management to keep the light source strong and direct.

My next experiment was to use higher power red and blue (3, 5 and 10 Watt) emitters on a single chip. These were much less likely to cause spotting, but color mixing was still an issue.

Also a major consideration is that while lumens per Watt is not the metric we need to best evaluate a horticultural lamp it was what was available on spec sheets.

LEDs have four distinct advantages over HID

  • Directionality, LEDs have viewing angles of 15 to 150 degrees depending on the lens usually while most used in horticulture are in the 60 to 120 degree range. HIDs radiate in a 360 degree range meaning that reflectors are required along with accompanying losses as heat. Some HID growers have overcome this limitation by doing vertical grows where no reflector is required;
  • Long life, 50,000 hours vs. 10,000 – 20,000 hours for HID with ever decreasing light;
  • Spectrum can be custom designed;
  • Lumen decay is slower than with HID. Translated, this means that LEDs will hold closer to their initial brightness throughout their life than HIDs. The reasons are simple. HIDs have metal filaments that get extreme heat stress while the gasses contained in the bulb may also age and leak. Compare that to LEDs, which are solid state semi-conductors with few parts and subject to a much narrower range of heat-stress.

Where LEDs were most lacking was in efficiency and color mixing. Color mixing is giving each chloroplast an equal amount of the spectrums emitted. My red and blue LEDs were averaging about 30 lumens per Watt vs. 3 to 5 times that much for HID lamps. Even though optimally targeting the prime receptors, the lower luminous efficacy could not compare to HID.

During my research, commercial manufacturers were experimenting with, and offering, 2, 3, 5, 8 and 10 spectral band LEDs in an attempt to hit all of the various photoreceptors and not just chlorophyll A & B a couple offered true full spectrum with no gaps in the spectrum offered. There seemed to be no general consensus on what was best. People were groping in the dark (no pun intended). Were the ancillary pigments stimulated by orange, green, yellow, cyan and deep red; etc. necessary? I looked to the sun and the plant photosynthetic response curve as templates for the best possible grow lamp.

Getting the curve right AND getting a much higher efficacy are both necessary if we are trying to match and/or outperform HID lamps. I believe this became possible with the introduction of the CREE XLamp CXA25xx and CXA3050 series, check out the Cree website, it is pretty interesting.

Bridgelux, Seoul Semiconductor and others have somewhat similar products. These are COB (chip-on-board) high power white, full spectrum emitters designed for industrial and residential lighting. They have a very high CRI (color rendering index) which means they show all colors equally well to the human eye.

Cree White LED graph

Note that the peaks happen to coincidentally correspond to the main plant response peaks.

Chlorophyll Ab Spectra

These CREE emitters are putting out 105 to 118 LPW, depending on spectrum chosen, at a 120 degree viewing angle. This puts them on equal footing with MH as far as efficacy and surpasses MH in regards to color spectrum and longevity, but still trails as far as cost.

My grow area is very restricted as to space and ventilation so I calculated the maximum amount of heat dissipation I could have in my space at about 165 Watts, thus I chose to use 5 emitters running at about 28 Watts each. I went with four 2700K (warm or more red shifted) and one 5000K (more blue/green or cool) emitter(s) with the 5000K in the center, and the 2700s on the four corners.


I decided to go with passive cooling to keep it simpler and to not add the extra few Watts required for the fans. Using fans (active cooling) could greatly reduce the size of the heat sinks and probably the cost as well.

  • I went with five monster heat sinks rated at 50 Watts of heat dissipation each. Attaching these to an open air aluminum frame gave even greater cooling capacity;
  • The recirculating air exchange fan utilized to bring fresh air to the plants also helps to cool the emitters.

50W Heat Sink Foe LED Grow Light

I attached the emitters to the heat sinks with permanent thermal adhesive (Arctic Silver Alumina AATA-5G Two Part Thermal Adhesive Compound Epoxy).They are now non-removable without destroying them. Since my build, Molex has released a LED holder specifically for the CREE CXA 25xx series. These would screw into the heat sink, requiring tapped holes. Power leads can then be pinch-gripped instead of soldering as I did. If you want to change the emitters to a higher Wattage or different Kelvin rating or replace in the case of failure, it is a simple matter of loosening two screws and removing the leads with a pin.

I also chose the more expensive route of using 5 separate constant current LED drivers. Constant current means that as components age or heat and cool and input voltage varies slightly, the LEDs receive the same amount of current. This keeps the brightness constant as well and prevents premature aging or possible failure from current spikes as LED emitters are extremely sensitive to current fluctuations. I have noticed that only some of the better commercial LED grow lights use this kind of technology.

In this case, I went with the 30 Watt/700mA model # RACD30-700 units from RECOM. These feature overload, short circuit and over-voltage protection. Efficiency is good, but not great, at 85%. I also considered Meanwell as an established manufacturer which offers some more efficient drivers at about 92% efficiency, but at a higher cost.

I had some extra room on the frame so I added two E27 (common screw-in household sockets) ports in case I wanted to add UV, far-red (incandescent) or CFLs. Due to the limited space, I use the newer and shorter micro-mini CFLs.

Micro Mini CFL

Note the older, larger style mini on the left and the same Wattage micro-mini on the right.

For the frame design, I wandered the isles of Lowes hardware store. Any large chain hardware store should have most everything you need.

As this was just for my personal usage and to keep cost and complexity down, I wired the power directly to the terminal block and did not use a power switch. With a bit extra work, one could put each driver on a separate power switch for ultimate control. Either way, you be sure to ground the frame to prevent electric shock.

The heat-sinks were then pinch gripped between the U-rails using the long screws and oversized washers.

DIY LED Grow Light Front

DIY LED Grow Light Rear

  • Overall size is approximately 18” X 20” and only a few inches deep. This is not meant to be an engineering drawing as your needs will differ from mine, but the same principles can be adapted to any grow space.
  • Note that CREE also makes much higher power emitters such as the newly released XLamp CXA3590 rated at up to 18,000 lumens for those needing even more light or greater coverage. Of course, different heat sinks and drivers would be required.
  • For those wanting to get their feet wet, but not commit to a full build, using just one 5000K emitter, heat sink and driver and a simple mounting is a good way to go and is perfect for a small tray of seedlings up to 2-3 weeks of vegging.

So theory is great, but how does it work?

Having five light sources spaced 5 – 6” apart gives much better lighting than a single source with minimal shadows, even down low on the plants. The color is very natural – no harsh blue or murky yellow.

The plants absolutely love it. I have had less stretch than with any other lamp and lush, full growth all the way through to flower. This lamp provides the perfect amount of light for my 2’ X 2’ space and getting over a gram per Watt is easily doable for the experienced marijuana grower.


Note that this is the unadjusted natural color put out by the lamp without color correction.

I have spent untold amounts of money and hours of research to come to this design and I am very happy with it. I am convinced this is the way to go for personal use for those with minimal space and heat restrictions and a penchant for mechanical and electrical tinkering. As to commercial applications, the price/performance ratio is not quite there yet.

CREE and other cutting-edge semiconductor manufacturers have already accomplished much higher LPW figures in the laboratory and eventually these will make it to full production. I would estimate there are 3 more generations before they reach their maximum potential and easily surpass the mighty HPS. In the meantime they will focus even more on ramping up production and bringing costs down.

Many people ask me the price. Where do I start? From engineering school where I learned to read and evaluate specifications? My years of cultivation experience? Do I count the thousands of hours of research and the costly failed experiments and the many redesigns on drafting paper until I got it almost right? Then there is the cost of the tools and labor and so forth, but in the end – this lamp is a part of me and this reconstruction is a gift to you guys for your support at HowtoGrowMarijuana.com. Parts costs can be tallied from the links and your local supplier on the accompanying list. Figure on around $500.

If there is enough interest, I will help others build a similar lamp adapted to their particular situation either as a consultant or as part of a future column started here.

If you feel this build is beyond your current capability, or just can’t be bothered, fair enough, you might want to check out some fine, full spectrum manufactured LED grow lights that come with a 5 year warranty from the guys at SuperGrowLED.com.

VN:F [1.9.22_1171]
Rate this content...
Rating: 9.1/10 (61 votes cast)
DIY LED Grow Light, 9.1 out of 10 based on 61 ratings

From the forum

How to Grow Marijuana Forum - Growing Marijuana Indoors & Hydroponics

15 thoughts on “DIY LED Grow Light

  1. Vinnie says:

    Hello all,
    I have a serious question, please advise/comment.

    I went to Home Depot and found Philips LED 14.5 Watt, 1500 lumens, 103 lm/watt.

    I’m so excited and thinking to buy 20 of 2700K bulbs for my flower room.

    But I’m not so sure these LED could be used for flower room.
    I need help, please give me some advice asap so I can make a decision to go out and purchase these LED or not.
    Thank you very much in advance.

    VA:F [1.9.22_1171]
    Rating: +1 (from 1 vote)
    • georgie1 says:

      I would like your help in a led design for a flower room 7x7x7 and a flower room 12x13x6.5ft
      These are 2 total different rooms. Thank you

      VA:F [1.9.22_1171]
      Rating: 0 (from 0 votes)
  2. jeffdagrowdude says:

    man thats a great disgin but it cost alot
    i did this same idea to start of but ended up finding this amazing led company from sothenr california called Dragon Grow Lights


    my freiend ended up getting one and took the risk and all i can say is thank you this thig has been amazing and only$599
    my yeild and trichcome prodution is off the hook, check it out viking

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
  3. viking says:

    so out of the game a while and thought led were a gimmick until i saw the recent spydr1200 vs 1000w de HPS dry yield results video and a few other similar videos. according to those videos and articles led has actually surpassed HPS but it costs $2400 so i set about figuring out how to copy but three days later getting mixed signals. you seem to disagree yet seem also to understand it very well. so i guess im curious if you would comment on the above. I can build anything id like to build a customized possibly even improved spydr1200 if it can really beat a 1000 w dehps in a 4×4 room hell i might even buy one of theirs if its really that good. but even diy its expensive trash if its not really as good. one video tested several leds with expensive equip and found spydr far superior i noted it red light seemed unique to it again youre skeptical about red the question seems to be are these tests a scam. i would add they seem to be testing flowering only and while i might consider being able to switch on more blue for instance i am also most concerned with the flowering being a root and go type of guy for the most part. now it wasnt lost on me your observation about the early nasa work just seems somethings not adding up like several people including yourself keep the lights high but spydr is saying 9-12′ inches now thats a big descrepancy they used 27 top bin osram claim thm to be 67% deep red, 13% red, 13% white and 7% blue, another guy claims to like the whites anyway im getting ready to move on this would like your help id be willing to document my efforts for you if you can help the research beforehand. guess ill just check back here for a reply.

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
    • johnny says:

      did you ever figure out how to replicate the spyder1200? I was interested in that light too but also found that the price was ridiculous. I’m sure maybe in the longrun itll make sense but without more tests and results done besides a simple youtube video, thats a huge investment especially if you are growing commercially and need 50-100 of those. I have an engineer who I told about my interest in LED lights and he said he could build one for cheaper but I dont know if hes quite skilled enough to tackle recreating something like the spyder1200. it’s a really sexy light with amazing output but the price is just too high to invest in multiple lights.

      VA:F [1.9.22_1171]
      Rating: 0 (from 0 votes)
    • bruce says:

      ok found a bit more specific info on spectrum heavy on the red blue and white for the balance reds are more efficient. still id like it more specific particularly for flowering. Ans I havnt been able to answer the big question which is can a spydr1200 like led really beat a dehps1000 on energy and yield. also i think part of the answer lies in my question about can spydr really be 12″ off canopy. It seems if theres the potential spydr claimed in that video then it might be worth building a really big light with far too many leds but separate controls to experiment with spectrum specific to cannabis during different times. it might cost more but hey its science

      VA:F [1.9.22_1171]
      Rating: 0 (from 0 votes)
  4. dy1ng_br33dz says:

    don’t give up, keep your head up and keep blowing that smoke up.

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
    • Jo says:

      Great article. Thinking about giving it a try, do you have a basic wiring diagram or schematic diagram? Did I understand correctly is there a part II coming out soon? Keep up the good work…

      VA:F [1.9.22_1171]
      Rating: 0 (from 0 votes)
    • Jonathan says:

      Thursday, September 17, 2015

      Book Idea? = GREAT IDEA? We can START IT HERE?
       1st,, BSEE, and while companies use the new 2530’s, and so on, the average Joe does not know where, what, or how much of, what is needed to purchase.
       The Diamond uses X-ML?I believe, but things are changing fast and newer products are soon arriving, noticeably, I’ve noted many companies that use cree, sometimes use the other diodes, some CREE, others, Epi’s, again, Joe is not going to know that.
       Make a list of parts.
       Coverage? Per lamp, set-up? Amount needed, 4×4, 5×5,4×8,25×25, et-cetera.
       Parts and how many?
       Tools, cost and savings vs buying something you can build quite easy!
       F.W.I.W. I enjoyed technical writing in college(Manuals, and so forth). back in Boston, MA.

      I can help write that book if you want. People need to know how easy, even solder less is possible, if done correctly, but DO NOT not RECOMMENDED.

      People need to know where to buy the lights cheap, at a good deal, not cheap knock offs, just fair and real cree’s, epi’s, et-cetera.

      I believe having spoken with many in the industry, (still in its infancy in the U.S.), your possible builder, may not know where to get all the parts or parts list.

      Perhaps its already done, but am just leaving it just in case.


      VA:F [1.9.22_1171]
      Rating: 0 (from 0 votes)
  5. Tim says:

    Nice work!

    1) I’d thermally attach the heat sinks center to the underside of the aluminium frame and gain additional thermal transfer through the frame.

    2) I’d move the drivers off of the heat sinks – they are blocking airflow – maybe attach to the frame.

    VA:F [1.9.22_1171]
    Rating: +1 (from 3 votes)
  6. david says:

    hi can you tell me the best driver for 6 of this LEDS Cree: CXA2520: 2700K, thanks

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
  7. Yuri says:

    It is my first growing tent and I beside to do it based on your recommendation. Can you be more specific for a LED names. I tried to research CREE XLamp CXA3050 5000K and I have at least 10 option same for CXA25xx. I’m not sure if I need cool white or worm white. And they different lm too.

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
  8. Will says:

    Hello. Excellent post. Question.. how are you running each emmiter at 28W? are they not 10w LEDs and why one driver per LED. Doesn’t the driver deliver the same current and voltage (hence watts) to each LED?

    I’m ready to build this except for the 28w per light confusion. Thanks for the large amount of time this must have taken. Because of all your info on this site about LED’s I am excited to get rid of the fans and heat and begin learning as much as I can about this new way to grow indoors.

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
  9. Jonathan says:

    Hey, just wanted to say thanks for such a great, detailed write up! I too have been pondering building my own light from scratch. Your guide is one of the most advanced and detailed pages I’ve come across, so I really appreciate the large amount of time and money that went into it.

    I’ve mainly been going back and forth about whether to build or buy. I’ve definitely decided not to go with any foreign lamps of dubious design and warranty, no matter how enticing their low cost is. There are just too many reports online (including yours) of early failures, usually just after what little warranty they have expires.

    Have you by chance looked at the Diamond series from Advanced LED Lights? If so, I’d love to hear your thoughts. They claim to be US made and the warranty is pretty decent. The XML series seems to be most similar to your design as it uses larger COB emitters.

    If the cost wasn’t so high I would be more inclined to DIY, but the element of risk of failure without any kind of warranty or financial recourse makes me a bit nervous. If I can get the same amount and quality of light from a reliable pre-built unit with a warranty for the same cost as what I can build I would choose the former every time.

    On the flip side, warranty work usually means sending your light in for weeks or even months in some cases. That can mean a mad scramble to save your plants. The DIY option with totally independent drivers would make repairs possible in situ.

    And I thought researching LED flashlights was maddening – researching grow light technology brings it to a whole new level! Thanks again.

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
  10. dennis says:

    Hi which part is the most effective and best, cob or cree

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)

Leave a Reply

Your email address will not be published. Required fields are marked *


one × 8 =

Email This Page



Your Message

Navigation Menu