Robotic Farming

I’ve been looking at new LEDs and available components.  There are many more available now than there were last year.

We could save a lot of money on the power supply by wiring the strings in parallel, which I was anxious to do because of the risk of thermal runaway.  However, I saw an article about parallel wiring that uses current-mirroring to allow all the strings to use the same current … just what we need.

That would allow us to use a low-voltage, high current source to drive several strings. For example, LEDSupply lists a 24VDC, 6.5A, 150-Watt Power supply for $29.99.  Nominally, that could drive 50 3-watt LEDs.  We could avoid building the circuit by buying a BuckPuck for each string that runs off the 24VDC supply for about $10-15. 

Also, I saw several products in LEDSupply that provide optics and cases for LEDs mounted on the Luxeon star configuration, which would simplify the rest of the design.  I’m thinking we can make our own cases with the lathe pretty easily.

National Semiconductor has a nice series of articles about options to drive LEDs.

• A Matter of Light, Part 1–The ABC’s of LEDs
• A Matter of Light, Part 2–Buck Whenever Possible
• A Matter of Light, Part 3–When to Boost and Buck-Boost

From the article:

Meeuws and three other Dutch bioengineers have taken the concept of a greenhouse a step further, growing vegetables, herbs and house plants in enclosed and regulated environments where even natural light is excluded.

In their research station, strawberries, yellow peppers, basil and banana plants take on an eerie pink glow under red and blue bulbs of Light-Emitting , or LEDs. Water trickles into the pans when needed and all excess is recycled, and the temperature is kept constant. Lights go on and off, simulating day and night, but according to the rhythm of the plant – which may be better at shorter cycles than 24 hours – rather than the rotation of the Earth.

http://www.forbes.com/feeds/ap/2011/04/11/technology-eu-netherlands-sunless-farming_8401593.html

The plants under the metal halide light are fully mature.  They are clearly larger than the ones under the LED lights.  The 12-hour plants are clearly smaller than the 18-hour plants.  I’m going to reduce the light to 8-hours/day for the ones that were 18-hours and see how long we can keep them healthy, in a sort of “paused” mode.

I went back and looked at Eric’s old notes.  We know a few more things now about how the LEDs operate.  I include the range for all the bins, since I don’t know what wavelength bins we’ve been getting.  I also assume that we supply 1A of current and they have a typical voltage drop and radiant power.

LEDEngin Deep Red LZ1-00R205,    655-670nm,    2.48-3.44V,    1A and 2.8V typ, radiant power: .525W, eff. 18.75%, $8.55, $16.29 $/radiant W. 2 eV/photon.  272 microMoles of photons/sec.  $.03 / microMole
LEDEngin Far Red LZ1-00R305,    725-745nm,    2.24-3.44V    1A and 2.4V typ, radiant power: .310W, eff. 12.92%, $7.11, $22.94 $/radiant W. 1.687 eV/photon.  190 microMoles of photons/sec. $.04 / microMole
LEDEngin Dental Blue LZ1-00DB05,    457-463nm,    3.20-4.40V    1A and 3.6V typ, radiant power: .900W, eff 25.00%, $20.80, $23.11 $/radiant W. 2.696 eV/photon.  346 microMOles of photons/sec. $.06 / microMole

I’ve been setting up the water by adjusting the pH, then adding the food.  I’ve just discovered that the food is highly acidic.  Consequently the pH has been much lower than it should have been.  It was 4.0 when I measured it just now.  Has been like that since I last adjusted it.  After an hour of screwing around the front and back now have a pH of about 5.5 and conductivity of 2.8 (front) and 2.5 (back).

I can’t believe that I’ve only just discovered how acidic (~3.0) the food is. 

Reading the LEDEngin book on Thermal Management Practical Application for the High Power LED emitters.
It says that the fundamental equation is this:

RΘJunction-Ambient = (ΔT junction – Ambient) / Pd
Where
ΔT = T junction – T Ambient
Pd = Forward current (If) * Forward voltage (Vf)

That seems wrong to me.  I think it should be
RΘJunction-Ambient = ΔT junction / Pd

RΘJunction-Ambient = RΘJunction-Slug (J-S) + RΘSlug-Board(S-B) + RΘthermal interface + RΘHsk-
Ambient(B-A) (2)
Where
RΘJunction-Slug (J-S can be found in the specific data sheet
RΘSlug-Board(S-B) includes the thermal resistance from the slug in the die
package to the board material
RΘthermal interface is the thermal resistance of the material interface between the
MCPCB and the heat sink
RΘHSK-Ambient(HSK-A) is the thermal resistance from the heat sink to ambient air

So to calculate all of this, we need to know the thermal resistance of:
RTheta(Junction-Slug):=
RTheta(Slug-Board):=
RTheta(Thermal-Interface):=
RTheta(HeatSink-Ambient):=

I’d like to set up an array of LEDs on an metal core circuit board, so we’ll need the array formula as well.

For an array of n LED emitters, the total RΘjunction-board-interface would follow
the equation below
1 / RΘjunction-board-interface = Σ(1/ RΘ(junction-board-interface)) i where i = 1…..n

For the 5-Watt LEDEngine:
Deep Red LZ1-00R205 RΘJ-C  5.5  °C/W (Voltage @ 1A, 2.8V) (Voltage @ 1.5A, 3.1V)
Far Red LZ1-00R305 RΘJ-C  5.5  °C/W (Voltage @ 1A, 2.4V) (Voltage @ 1.5A, 2.6V)
Dental Blue LZ1-00DB05 RΘJ-C 5.5 °C/W (Voltage @ 1A, 3.6V) (Voltage @ 1.5A, 3.8V)

Since its the same for all of these, we can just calculate the total resistance:
2 = 1/(2/5.5) = 2.75
4 = 1/(4/5.5) = 1.38
6 = 1/(6/5.5) = 0.92
8 = 1/(8/5.5) = 0.69
10 = 1/(10/5.5) = 0.55
12 = 1/(12/5.5) = 0.46
14 = 1/(14/5.5) = 0.40
16 = 1/(16/5.5) = 0.34
Interestingly, if you multiply the power (5W) each times the number of LEDs, you get 27.5 C.  That means that at best we will get a 27.5 degree difference just between the junction and the case, before any other calcs.

20-25C is ambient temperature.
The maximum current (and therefore the maximum power output) decreases linearly until the temperature gets to 125C.  The nominal ratings are at 25C.

The maximum junction temperature is 125C.
Typical MCPCB values < 3 C/W
If we use thermal grease between the board and the heat sink, such as Chomerics T660, it has a very low resistance 0.02 (C-in2/W), so it is practically no effect. 

So the total thermal resistance is dominated by the junction to slug and should be around 8 plus whatever the heat sink provides. At 5 watts, that means a difference of 40 C, giving us 125-25 = 100 C margin.  So we need a heat sink capable of better than 5 C/W.

It seems pretty easy to get large heat DC convertor heatsinks for about $3, that are on the order of 3-4″ finned aluminum blocks.  These have a heat resistance of 2-4 C/W, which should be fine.

The 2 plants in the front, 12-hour condition now have noticeable die-off of the older leaves and speckled brown spots.  The disease seems to occur on the old leaves as a general yellowing as well as spots.  There is also noticeable leaf burn on the newest leaves.  I think the yellowing may be caused by a nitrogen deficiency and the leaf burn by a calcium deficincy, but I have no way to tell.  I looked at one of the older leaves (which also had some leaf burn and rotting/browning at the edges) under a microscope and couldn’t see any bugs.  There is no evidence of bugs on the plants.  I didn’t completely change the formula, so it is reasonable to assume that some of the micronutrients have decreased.  It may be that the plants at the end of the trough are suffering more than those at the head.  The very last plant appears to have a root in the bucket.  Perhaps that is helping it somewhat.

All the plants seem to have something, but those in the front seem to be worse off than those in back.  The ones in front have more light and are growing slightly faster, possibly putting more demands on the resources.

I’ve completely changed the water and bumped the conductivity level to 2.3 to see if it helps the problem.

Front pH 6.02 cond 1.8

back pH 5.81 cond 1.8

The blue lights in the middle condition were off when I checked.  Loose wire.

C: 1.7 pH: 4.65 69.8F

C: 1.7 pH: 4.62 69.8F

Pictures from today.  Clearly the control is outpacing the others.  I’m pretty convinced that the issue is that the light from the control is very intense directly below it.  It can sustain 4 nearly developed plants, max.  The LEDs, however, are much more distributed and are currently sustaining 8 plants.

Here is the 18 hour LED condition.

And here is the 12-hour condition.