Barron's Ultimate Orchid Case

Current Status in the House:

(time in lower right of image)
Current status in the house

Pictures of the case are here.
Pictures of the plants are here

Background:

In order to set the stage for this project you need to understand some of the background. I grew up on an apple orchard (see URL above) and developed a love of plants. My degree is in Systems Engineering from the University of Virginia, with a special interest in software integration. My professional experience has been in supply chain computer systems. Put it together, and you have the genesis of the Ultimate Orchid Case.

Orchids are wonderful plants, and they can grow in most home environments; however, they grow much better when they have

warmth, between 65-85 degrees F
humidity, between 60-80% RH, and
light, 9-12 hours.

But it is a little more complicated that this. Most orchids like a little diurnal temperature variation (like they'd get in nature). And though they like it humid, they also like air circulation to prevent fungus and mold. And of course, they need some dark time, just like you do. Orchids are epiphytic so water goes right through their potting media, but they don't like saucers under them because they don't like to sit in any water. The challenge for the home grower is how to provide all this, AND enjoy the plants in your living space.

The Project

The idea for this project developed in Texas in 2003. I was reading about Peltier devices (aka TEC / Thermo Electric Heating & Cooling) and considered how they may be use to maintain a climate for plants. As is often the case, a simple idea snowballed. There was a lot of snowballing with this project.

Here are the highpoints of the effort:

homemade serial temperature sensors of air and water temperature
Heating and Cooling via TEC and a DPDT relay
computer control of 8 different circuits through parallel port relay, including grow lights, night lights, humidity, fountain, circulation fans, exhaust fans, and heating/cooling.
custom software to control all circuits based on sensor inputs, time of day, interval timing, and conditional logic
On-Off-On switches for all circuits, to provide manual override
Nice quality woodwork with teak racks, stained curly maple, poplar, and oak.
Completely waterproof case to allow easy watering in situ.
wireless internet connectivity providing near-live status to everyone in the world

Details

The House Frame

The Case is built from aluminum, glass, and wood. Various sizes of alumnimum "angle iron" form the frame. Two 2'x4' sheet particle boards form the floor for the inside components, and the floor of the case. The floor of the case is then lined with a heavy desk chair mat painted green. This is sloped by using 2x4's around the perimeter, with a shower drain mounted in the middle, and draining to a bar sink below. The bar sink is plumbed so that it drains to a valve mounted on the bottom of the case.

The case is topped with a removable lid that includes doors that open in either direction. Stainless sailing hasps form the closures. The doors are framed in poplar, and a 1" thick piece of curly maple forms the top.

The Power

An old PC power supply runs constantly providing 12v power to all the DC fixtures. Both AC and DC power runs through a fuse box where 13 fuses guard against dangerous spikes.

The Computer

Like much of the House, the computer was someone else's junk. It is an old 1994 vintage Tecra laptop that is built like a tank. No USB, but it came with a lan jack, serial and parallel ports which are enough for me. The best thing about a laptop is that the display folds up and doesn't require any space. It is also nice to have a short-term power backup in case the power goes out briefly.

Temperature Sensing

Temperature is sensed through a serial temperature sensor I built from a kit. The sensor is powered by a DS1820 chip. The kit allows 4 sensors, but I only use 2: air and water. The water sensor is dipped in epoxy to make it waterproof. Epoxy is amazingly conductive, and seems to provide accurate readings. I used telephone connectors to connect the sensors, which allows me to plug and unplug the sensors, allowing easy replacement if something needs to be fixed.

The software is the other important part of the temperature sensor. The kit provided a few clues, but obviously, I needed to integrate directly into my own software. A couple of friends at work provided insight on how to monitor the serial stream, and I was able to develop some VB code to intercept the temperature. This was really one of the hardest parts of the project, and I'm grateful for their help.

Parallel Relay

Just knowing the temperature doesn't do any good unless we can do something about it. In order to turn devices on and off using the computer, I use a 8 circuit relay board that is controlled through the parallel port. This too can be procured on the internet. It comes with some software, but I only needed the relay.c software that controls the port. The VB software calls the DOS relay driver for the port. I'm not sure it works with more recent versions of Windows, much less Linux. I'd be happy to know if anyone has tried this. I don't know how much longer I can keep running Win95. At some point, the PC will die.

The Software

(see image at top of page)
I really didn't know VB until I started working on this project, but found VB to be very easy to pick up. Once I knew the temperature of the air, I needed to determine a "target" temperature so that we could decide to heat or cool. But the target temperature really varies by time of day. So I wrote an algorithm that determined the optimal temperature by using a Sine wave with a mean temperature and diurnal range. The sine wave reaches its peak at 2:00PM and of course its minimum at 2:00AM. The user specifies the mean temperature, the diurnal range, and the allowable variation before heating/cooling (for example, it may be OK to be within 2 degrees of target temperature).

Temperature is one thing, but many of the other circuits don't have a sensor. So all the other circuits can be controlled by any of 3 methods:

on/off based on time of day
run for x minutes every y minutes
run if something else is running (for example, the pump runs if the mister is running)

...And as you probably guessed, the color indicates whether the circuit is on or off.

Because I knew the temperature, I also built alarm capability into it. The system plays Kool and the Gang's "Too Hot Baby" if it gets too hot (user variable); or "Your as Cold as Ice" (Foreigner) if it gets too cold. OK, this is kind of wierd, but I needed an alarm .wav anyway.

One of the best things about the software is that it also graphs the temperature based on a user-defined interval throughout the day. A red bar indicates it was heating, and a blue bar indicates it was cooling. Yellow means it was doing neither (meaning it was within the variance of target temperature). It also keeps track of the maximum and minimum temperatures observed during the past 24 observations.

Every event is logged and written out. The log is truncated when it reaches something really long. The "silent mode" ensures that the slightly noisy exhaust fans don't run between 10:00PM and 6:00AM.

Heating and Cooling

The House was originally designed to heat and cool via the TEC (Peltier) which is mounted on a used bar sink (proving that you can even buy the proverbial kitchen sink on Ebay). The insulated sink serves as a 2.5 gallon reservoir for water that flows up a fountain, and down through a drain in the case.

A TEC is a really neat device that can heat or cool, depending on the direction of current flowing into it. I found a site that showed how to wire a DPDT switch so that you can reverse the flow of current, and then I was in business.

Based on bench tests, I thought that the TEC would be powerful enough to heat and cool the water sufficiently to transfer the heat to the case. But I found that the TEC really didn't get the water hot enough. Perhaps my TEC is too small, or my power supply too wimpy. In any case, I added a secondary relay that trips an AC circuit which powers a 300 watt immersion heater that you can buy to keep your soup hot. This gets the water really hot, so I added the water temperature sensor and programmed the software to cutoff at 105 F, which is as hot as you want water around plant tissue. I also modified the software so that it would continue to circulate the water if the water temperature varied from the air temperature by a user-defined amount.

Cooling is underpowered, but works to some extent. When the case is cooling, the water temperature typically falls into the low 60's, but this isn't really cool enough to drive the House temperature down very far.

The picture at right shows the TEC on top of a solid aluminum block that conducts heat/cool through the bottom of the case down to the heat sink and fan (standard CPU cooling set). The fan is wired to the same circuit as the TEC so that it always runs when the TEC is running.

Lighting

Three "100 watt" compact flourescent bulbs (which of course only draw 20-something) provide nice growing lights for the plants. These bulbs never get really hot, but I lined the "ceiling" with reflective aluminum flashing so act both as a mirror, and a conductor of heat away from the bulbs. The aluminum fans also provide some ventilation up there, so it never seems too hot.

I also threw in a pair of green cold cathode night lights mounted in the top that come on in the evening hours, providing a nice transition between day and night.

Air circulation

Two 120mm computer fans are mounted in a "Mexican" clock tower constructed from plywood, and faced with drywall in a stucco pattern. Patios on the clock tower are paved with "copper slate" mosaic tile. The clock tower was a fun project by itself. The fans pull air in from below the floor of the case. I am training a ficus bonsai on top of the hosue. The humidity makes for lots of aerial roots. In the picture, you can also see the teak racks I built to help organize the plants.

Two 70mm aluminum fans provide exhaust ventilation. One fan points out, the other points in, providing good flow. These fans really seem to move more air than the 120's.

The fountain

I bought a solid lava rock fountain off the internet, and asked that they make the top resevouir extra deep to allow me to submerge an ultrasonic fogger in it. When the fogger is running, it looks like a volcano about to blow. It also adds nice humidity to the case.

The switchbox

The switchbox is aluminum, and provides the ability to switch all 8 circuits to either manual on, off, or automatic based on the computer relay. This is really handy. Green LED's next to each switch indicate whether the circuit is on or not. A blue LED indicates whether the TEC is cooling; and a red LED indicates whether the TEC is heating. I haven't figured out a nice way to label the switches on an aluminum box, so if anyone has an idea let me know. For now, labels are taped on. Yuck.

And internet too...

The Case's laptop runs Windows 95, and takes a screen snapshot every 3 minutes. A wireless Access Point in the case connects the laptop to the rest of my home network. The image is saved directly to my Linux Ubuntu computer which is always on. A CRON entry on the LINUX machine FTP's the jpg to this website every 5 minutes.

What didn't work

You learn a LOT in a project like this. I learned that design is way more important than materials. It is a lot easier to design stuff in from the beginning than to try to retrofit it later. Here's a list of stuff that didn't work, some of which I just abandoned because it was too hard to fix after the whole thing was assembled:

water level alarm: I had originally wired up a reed switch that was triggered by a ring magnet floating up and down a tube. When the circuit was triggered it would sound an alarm. This didn't work because the float did not move smoothly within its PVC pipe, AND because I didn't have a very good way to position the reed switch within its tube, so even when it did work, it wasn't properly calibrated to the high water mark of the case. I abandoned this part of the project because I didn't think it was all that necessary anyway.
watering station: In some of the pictures I eventually hope to get posted, you'll see a hand sprayer like you'd see from a kitchen sink. I planned to use it to water the plants using water from the reservoir. It is hooked up to a windshield washer pump. This worked OK for a while, but I found that this kind of pump really isn't designed to run for more than 15-30 seconds. It seems to die down after that. Perhaps it isn't getting enough power. The other problem is that all that hose is cumbersome to get in and out of the case. AND because I don't have a filter in the line, the pump probably needs to contend with a lot of dirty water.

What I'd do differently

Besides the stuff that didn't work (which is taking up half of the room in the component section), I would do a few things differently:

make the fuses more accessible: I did a good job of fusing all the components, because I wanted it to be safe, and to isolate any problems. I did not do a good job of making it easy to get at these fuses. The case is way to heavy to move without a lot of effort, and I had put the fuse box in an inaccessible place in the case. Moreover, I didn't make the fuse box easy to open without disconnecting a lot of wires. With 13 fuses, something goes wrong about every 3 months. Spring 09 Update: Actually, nothing fuse-related has gone wrong in over 18 mos. Knock on wood.
use a bigger TEC, and probably a bigger power supply: this was already mentioned
do a better job of laying out the wiring inside the case. It is too cluttered, which decreases airflow and can make the laptop hard to slide in and out.

Log of problems

clogged drain: I originally used mesh wrapped around the bottom of the drain to keep junk out of the sink and submersed fountain pump. But this eventually clogged and backed up inside the fountain. Fortunately the pump ran dry before it overflowed the case. I took that mesh out, and put it up on the top of the drain where I can see it better, and can clean it
one of the LED's has burned out in the switchbox. Would be a real pain to fix.
after one year, the fogger died. Should be easy to replace, as long as I can find one that is the same voltage and plug layout.
after one year, the epoxied temperature sensor died. I think the problem was that I didn't epoxy all the way up around the heatshrink tubing, and water finally worked its way down. My new sensor should do better.
May 08: the DS1820 sensor died again. This time, I've epoxied it into a copper tube. Maybe that will keep the water out.
Sept 08: the DS1820 died yet again because I let the water get over the top of the tube. This time I put the new sensor in a longer copper tube. Right idea, better execution.

It took me one year of planning, and one year of actual construction time. The case "went live" in January 2006, and has been running constantly ever since then.

I hope you enjoyed reading about my project. If you'd like, you can contact me here.

Or return to www.shaworchards.com.