A Place Where No Dreams Come True...

Current Topic: High power LEDs can be a bit trickey to drive. In order for them to work reliably for extremely long periods they need be driven by a current source. NOT a voltage source. Or they may create an unsafe operating environment and burn out way before their expected lifetime of possibly 100,000 hours.

Wiring The Individual LEDs.

The LEDs we chose are effectively 1 watt Osram Oslon ssl80 series. I'm a real fan of Osram since they are originally an early scientific optical company whos origin predates the LED. I have great confidence in their engineering and manufacturing capabilities. And I have not been disappointed. So... The LEDs I chose have a clear lens with 80 degree transmission spread. I have chosen not to use reflectors to additionally steer the light downwards however I may decide to change that depending on the test results.

These LEDs, although only ~1 Watt, are still considered 'High Power' and as such require effecient heat dissipation or else they will burn up. To accomplish this I used a device called a star. The stars, named because of their shape, are a Metal Core Printed Circuit Board (MCPCB). Basically, the normal fiberglass core is substituted with an aluminum substrate. The actual circuit is printed on an extremely thin circuit which is thermally bonded to the aluminum. This allows the aluminum stars, in turn, to be mounted to a heatsink to allow proper transfer of the heat generated by the LEDs.

The heatsink I chose for this is a standard product called MakersHEATSINK SLIM manufactured and distributed by a company called LED Supply. This product is actually a kit. It comes with T-slotted channels for mounting the LED stars using any pattern desired. In addition to the heatsink the kit includes various screws, nuts and nylon washers to mount the stars to the heatsink. It also includes ABS side peices and an acrylic cover to seal in the LEDs. All in all... This is a marvelous product and not only worked extremely well for me but it also yielded a proper looking prototype. In my opinion a prototype should look great and work fantastic and this product certainly satisfied that goal. In the end I actually used more LEDs then the manufacturer recommends and as such I ran out of screws. I contacted them and they were happy to immediately send out some additionall screw/nut/washer kits. Without charge. Wow! I'm really impressed now. That's the best customer support I have had in quite a while. Thanks LED Supply!

Bare LED heatsink LED mounting screws

Now came the real tricky part. The LEDs are surface mount with three rectuagular pads on the bottom. One for the Anode (plus) one for the cathode (minus) and a thermal pad to help with the heat transfer. No way to solder these by hand. So I bought a tube of low temperature solder paste and with a little guidance from YouTube set out to bake these in a toaster oven. Based on what others were doing I spread a thin coat of solder paste over the pads on the star, placed the LED as best as I could and procedded to bake it in the oven. The most interesting part here is that when the solder begins to melt it tends to wick towards the pads and the surface tension pulls the LED into place over the pads. If you are not perfect on the LED placement you can actually see it jump, slightly, to the proper position. Exactly like the videos I watched demonstrated.

However... When it came time to attach the leads to the star it all went horribly wrong. I used a regular soldering iron and normal 60/40 led solder. At about 650°F. The first lead soldered properly with no problem. Or so I thought. But when I went to solder the second lead and I was done the lead accidentally hit the LED and knocked it off the star. What I hadn't realized is that the soldering iron was so hot, compared to the solder paste melting point, that it was actually reflowing the LED. Hmm! Well the good thing that came out of this discovery was that I didn't need the oven to reflow the LEDs. Instead the method I ultimately adopted was to solder one lead to the star. Then after it cooled down I applied the solder paste to the star and set the LED in place. Finally, the process of soldering the second lead generated enough heat, due to the thermal transfer of the aluminum substrate that is was enough to reflow the LED to the star. This method worked well for me for the over 150 LEDs that I soldered for this project so far.

Bare LED stars LED solder paste LEDs on heatsink

For this first attempt I am using five different wavelength LEDs. There are four Everlight Ultra Violet A 395nm running at 550mA. Fourteen Osram ssl80 Deep Blue 455nm at 380mA. Six Osram ssl80 True Green 528nm at 380mA. Thirty Osram ssl80 Hyper Red 657nm at 380mA. And Six Osram ssl80 Near Infra Red 721nm at 380mA. I was hoping to use shorter wavelength UVC LEDs but was unable to find affordable High Power models. I'm estimting that the total power consumed by the LEDs for a single light fixture will be about 58.52 watts.

The reasoning behind the various wavelengths are as follows... The Red and Blue are specifically taylored to the majority wavelengths that most plants utilize for generating the different chlorophyll types. The Infra Red is supposed to help with early growth and can be used to taylor the vertical development for the plants. The green is generally not used by plants during the photosynthesis process and being not absorbed by the plant will penetrate deep into the canopy and reflect upwards to the underside of the foliage. Asside from a minor whitening effect on the radient light NASA suggests that it might help with developing a more natural color for certain plants like lettuce. The UV, particularly UVC which I was not able to provide exactly, has been rumered to enhance flavor and fragrance and might improve taste in fruits and vegtables as well as scent for flowers. Since this is just a first test I was willing to give the UV a try however without several isolated experimental setups I will be unable to tell if this is actually true.

Applying thermal compound

The final step in wiring the LEDs consists of applying thermal compound between the stars and the heatsink. The purpose of this is to basically fill in any air gaps caused by the uneven (not perfectly flat) surfaces of both the heatsink and the stars. For this I use a silicone based thermal compound (grease). The jar I have is over 40 years old and is still in perfectly good condition. Primarily it is a special type of grease that is electrically non-conductive while being thermally conductive and it basically never fully dries. The drying issue is especially important for maintenance reasons. If an LED fails I will need to remove the star from the heatsink so I can remove and replace the LED.

NOTE: I have seen many DIY prototypes that use a thermally conductive epoxy to attach the stars to the heatsink however I went out of my way to find a system where the stars can be fastened by screws so I can repair or change my design if it doesn't work out like planed. My goal is create a product for market so I need to get the most out of my R&D process so an adhesive was not a possibility.

A Power Supply Suitable Only For A Prototype.

The power supply was, well, a disaster. Asside from requiring quite an effort to breadboard, and a bit expensive, it was way underpowered for my requirements. I unfortunately underrated the main transformer (cost choice) which caused it to run extremely hot. Secondly I neglected to expect that the peak voltage from the RMS rating of the transformer, as well as its transfer regulation, would yield a rectified, filtered, voltage almost 50% above what I was expecting. Ugh! Further... I forgot to order one of the transformers and was forced to use one I had laying around which produced a voltage below what I required. Finally... The overwhelming AC ripple presented to the current drivers made them run extremely hot. Hence... Total failure.

Prototype power supply

Ultimately... I had to replace the power supply with a 24VDC 120W switching supply purchased from China on Ebay. And wow, what a difference that made. The current regulators smartened right up and the voltage is a lot closer to another target goal of being able to power the system from solar panels and batteries. I also purchased, on Ebay from China, small DC to DC USB charging units to supply 5VDC power to the controller boards that drive the current regulator boards which provide the PWM dimming for each LED color band. Asside from reducing the total cost of the power supply by about 70% it reduces the total weight for the complete lighting unit by several pounds.

Controlling The Individual LEDs.

Here's where it all comes together. With the current drivers and the network control processor it is possible to control each set of color LEDs individually. This feature allows for providing a circadian cycle over a 24 hour period as well as a longer cycle over several weeks. Basically control can be fed to each light indivdiually, or as a group, to provide proper light exposure to meet the needs of any plant. Be it flowers, fruit, vegetable, etc...

LED controller LED control methods LED network control

Additionally each light controller has several serial, digital and analog inputs for handling various sensors and to control external devices. Like, for instance, temperature, humidity, moisture sensors and maybe watering system valves. To name only a few. There are enough spare resources on each lights controller to meet the needs of any botanical system wheather it be a closed or open system.

Horticulture Light Ready For Testing.

Some people might think that I made a mistake with the wiring so close to the heatsink. What's not obvious are the materials I used. I did think about that and have used suitable wire and ties for this project. The wire I used is Military grade MIL-W-16878/4 which is Teflon coated and silver plated. The ties are also not normal zip ties but are instead Military cloth lasing. Both have a minimum operating temperature range from -55°C to +125°C and both these items will actually withstand much more. In reality, the solder will melt before either is damaged.

Complete wired prototype Complete assembled prototype

The first light has been completed and is operational. Now I need to test it. But... Not only does it need to produce the right light for plant growth but it needs to operate for between fourteen and sixteen hours a day, continuously, without over heating. In order for the Return On Investment (ROI) goal to be acheived I need this light to perform, without degrading, for at least 50,000 hours and preferably 100,000 or more. Only then will I be able to call this a complete success.

First Test With Live Subjects.

The only way I was really going to know if the lightwas working was to put it into service. For this phase my sister has been helping me out. So... Off to the local gardening center we went to purchase some planting supplies and seeds. We chose to try tomatoes and mini eggplants. We started them in these expandable starter pouches. They come as flat rings dehydrated and in a small box. Just add a little water and they puff up about three inches into a perfect mini environment designed specifically for germination. We let them percolate for a week or so and a few days after they sprouted transplanted them into small pots. For the eggplants we just used the pots as they are supposed to just grow into small bushes which can support the eggplants themselves. For the tomatoes we bought some chicken wire and using some old coat hangers kludged up a mini trellis system to help support the tomatoes. If we get any. Neither of us are gardeners so there are no great expectations but we're still optimistic.

Seed packs Sprouted seedlings Potted seedlings

At this house there is an old, unused, photography darkroom. It has been unused, except for storage, for many many years. We did a little bit of cleanup, moved lot's of old unused stuff out of the way and set up the light over a couple of chairs and fired it up. Since it is no longer a dark room we are now refering to it as 'the lightroom'. It's still a little bit crowded but it is getting the job done for now.

First grow test First grow test

With the light in place I set off to build another one. When I purchased all the parts I bought enough to assemble two lights and as the plants are growing it is becomming obvious that we are going to have to expand our operation to scale up with the growth. This presented a little logistics problem. First off the lightroom needed to be cleared out to accomodate the additional fixture and I needed to complete the networking code so I could control the lights from a single point. Basically I needed to create a client/server setup for the system.

Lightroom cleaned up Adding the second light A growing success

A Little Timelapse Video To Prove It's Really Working.

There was a Raspberry pi 3 laying around which I had not used yet and this was a perfect time to check it out. It had no display or camera yet for it so I purchased some on Ebay. There was also a builtin program (raspistill) which is capable of taking pictures, at intervals, over long periods of time. I set that up to acquire JPEG images every 30 seconds for a twelve hour span. I quickly ran into a problem with storage. The default camera setting yielded 5+ mega-byte images, and, there was only 1.6 giga-bytes of storage available on the memory card. The Raspberry Pi org website said very little about the features of the program but luckily the program has builtin help (--help) which allowed me to reduce the image size, rotate the view and set the JPEG image quality. With those features I was able to reduce the image size (on disk) to about 500 kilo-bytes. That was small enough that I was able to cache twelve hours of images at the chosen interval.

Raspberry Pi timelapse

I had previously written a program, for another project, using libjpeg and libvpx (the google VPX codec) which allowed me to take a series of JPEG images, scale and combine them to output a video in VP_8 format. I had also written code to pack the video output in a Matroska (.MKV) container. This combination creates a "WEBM" format video which any HTML5 complient browser can display natively, from my own server, without having to load and pay FLASH server royalties.

If you are unable to view this video on your device you can view it on YouTube by Clicking Here.

If you are unable to view this video on your device you can view it on YouTube by Clicking Here.

You can notice in the first two videos some banding effects, especially at night when we reduce the lights to only green. Supposedly the green lights do not disturb the circadian cycle and were quite useful for filming at night time. After viewing the videos I have reservations about that assumption. Anyway... I tracked down the problem with the camera to the auto white-ballance and auto exposure settings. I disabled them and provided constant settings and the result looks mugh better. I'm still not completely satisfied with the settings but at least it no longer looks like the lights are flickering. Because they don't. They're quite stable as the next video demonstrates.

Raspberry Pi timelapse

Oh yeah! We also cleaned out and rearranged the lightroom to make the time lapse much easier to manage.

If you are unable to view this video on your device you can view it on YouTube by Clicking Here.