Wednesday, January 18, 2012

The Arduino Learner's Project - Part I

I had originally considered this project many years ago while working for a hardware manufacturing company. And so I went about asking about details required for my project as hardware engineers were plentiful. I had heard rumblings of Arduino but really didn't know anything about it and I certainly hadn't explored any details. And so I trusted the answers I heard. After all, this is what he does for a living. The answers meant the project was cost prohibitive so I simply let the project go there - but not forgotten. I was disappointed to say the least.

Years later I learned the information provided by my learned professional was more or less, completely wrong. I don't know what he had based his information on but it turns out it can be done for a fraction of the cost I had been told and certainly by using more frugal methods than had been suggested or what I had originally considered. I realized all this after curiosity got the better of me and I started exploring the Arduino platform.

Long story short, creating your project is likely easier and more cost effective than you think.

The [Advanced] Wireless Sprinkler Controller
The project at inception was originally going to be a cheap Buffalo wireless router with some relays and custom Linux code. Once I started seriously looking at an Arduino solution, I realized it could be much cheaper yet. It was at this point I started learning about sprinkler system controllers, their defacto solenoids, and the voltage/current requirements. And so I started looking for various components to piecemeal my project.

I realized early on I wanted more than just another sprinkler timer. If you check, you'll find many such Arduino based projects. Some even complete with code and schematics. None of these really scratch my itch as they are typically a generic scheduler and timer and/or the former plus a web interface. All more or less variations of a theme. None of which are terribly exciting for me.

Well, I wanted none of that. I already have a nice multi-zone sprinkler controller and timer. Adding a web interface is slick but doesn't really do anything aside from save me a trip to the garage a couple times per year, plus timer adjustments after I realize the grass is getting burned. Not really sexy enough. That's when I realized I wanted a smart controller. Some investigation reveals the level of smarts I want is retail priced in the $600-$1200 dollar range depending on the specifics and brand of controller. Now that sounds like a project. While I don't expect exact feature parity with these costly commercial offerings, I do believe I can get mostly there for a fraction of the cost.

You might be wondering what this does that the others don't. Well, for starters, they automatically adjust water based on an array of environmental factors. Generally speaking, the person behind the controller provides operational criteria for each zone and the controller does the rest. This means, for example, no watering while its raining and extra watering during those triple digit heat waves, all without any input from me. What this really means is the opportunity to play with lots of different sensors and embedded development while creating something practical at frugal expense.

The Arduino Platform
There is a lot of confusion about what Arduino actually is. It doesn't help that the group behind Arduino seems to take pride in obfuscating what Arduino actually is.

Arduino is a Microcontroller (uC), a USB interface, (typically) power regulators (3.3v and 5v), any number of microcontroller pins pulled out to easily accessible headers, a bootloader, a set of high level libraries (which come at a performance cost), and a primitive IDE.

The USB interface provides a means of powering the board and small, attached devices, as well as a means for basic input and output (I/O). This means of I/O is also the basis for flashing (sending your program to) the microcontroller. This interface also alleviates the need to have a separate device known as an In Circuit Serial Programmer (ICSP); though headers are typically broken out to allow for easy attachment of an ICSP to the board. Its important to understand, this interface does not actually extend USB access to the Arduino platform as the available USB interface is completely abstracted away (and frequently controlled by a second on-board microcontroller or USB TTL chip) so as to look and act like a serial port.

Next is the language. Its extremely annoying they call programs, "scripts", when in fact, they are absolutely are not. "Scripts" are typically associated with a scripting language and none of the languages supported by Arduino qualify as a scripting language. You program Arduino in C, C++, and Atmel AVR assembler. The standard Arduino library does, however, do a good job of simplifying some of the more involved tasks associated with embedded development. So don't let the newly revealed truth scare you away.

In a nutshell, Arduino is a rapid prototyping and development platform. And the best thing is, you need not actually deploy your project with an Arduino. You can build a system using the minimal components required, thusly helping to reduce deployment costs. Meaning you can use Arduino strictly for development and deploy with a much lower cost solution; frequently for a couple of bucks plus any external devices and sensors. Of course, to do this requires some extra hardware expertise. But if you, like me, are lacking this expertise, visit the Arduino Forums. You'll find you're not alone and that lots of help from a very active and friendly community is ready to assist so long as you're willing to put in the effort.

Arduino Selection
Once anyone starts looking at the full spectrum of official and unofficial Arduino hardware, board selection can become somewhat daunting. Doubly so when you have no idea what your code and pin count requirements will be. Since I was new to Arduino I decided I would err on the side of caution and simply get a big Arduino with the understanding I can always step down in a variety of way. After all, this is in large part, Arduino's claim to fame as Atmel makes a large selection of arduino compatible 8-bit microcontrollers.

I quickly settled on a Mega 2560. The specifications show it to be a fairly robust 8-bit microcontroller (uC) and so it seems like a very sane board into which I can grow. Sadly, experience has taught me otherwise. In fact, for a general purpose development board, especially for a beginner, I can not recommend strongly enough you stay away from this board. The board is very poorly supported, making it the exception rather than the rule. Reportedly it has a minor and rarely observed hardware bug which can require physical reset to recover. Worse, the bootloader has a critical bug such that the WDT can not be used. And even more frustrating, simply flashing code which contains the character sequence of, '!!!', will cause the flash to fail. Made even more frustrating, that sequence can be generated by the compiler, so flashing your code can suddenly just start failing without any indication as to why. Also, there is an issue where the EEPROM can't be programmed when its flashed, meaning your program must always setup initial state. There is also an issue accessing all of the 2560's memory in many circumstances. In the Arduino world, its by far the odd man out.

It also turns out the physical form factor is at a disadvantage for two reasons. These boards are not compatible with all Arduino Shields. The second physical issue is that the uC is not socketed. This means if you burn out a pin on the uC, that's that. In contrast to most other Arduinos, they are much more forgiving because you can simply replace the socketed uC for a couple of bucks and lesson learned. Whereas, with the Mega 2560, you need to lay down another ~$55 bucks or so. Costly lesson.

Compiler support is another real hit or miss issue with the Mega2560 and various Linux distributions. Turns out my then Linux distribution, Mandriva, happily provides an extremely buggy compiler. It was so bad, many of the Arduino examples would simply fail to run. I wound up having to compile my own gcc tool suite and avr-libc, which proved to be another project in its own right. And I can assure you, that's the very, very terse version of it. Eventually with the help of others on the Arduino and AVR forums, I was able to create a working environment. Sadly, Mandriva is far, far from alone here. I guess the good news is, if you run Windows, the Arduino environment is very well maintained so Windows users need not fear. If you're a Linux user, do not trust the AVR compilers out of the box unless you know for a fact it works with your uC of choice.

All of these issues are known for well over a year or more at this point and there seems to be little incentive for the developers to support this hardware. So in a nutshell, unless you are absolutely positive you need the Arduino Mega 2560, avoid it at all costs. If you still want a high end Arduino, consider the Arduino Mega in its place. It doesn't address the socketed uC issue but it also doesn't have all the other bugs.

"The Controller"
"One Controller To Rule Them All"
To pull this project together, we need a number of different components and sensors. So the generic and very rough list looks something like this:
  • microcontroller (uC) with enough ports
  • LCD for human interface
  • keypad for human interface
  • Real Time Clock (RTC)
  • power supply
  • wireless interface of some type
  • method for controlling irrigation solenoids
  • humidity & temperature sensor(s)
  • moisture sensor(s)
  • and eventually a nice case of some type
The uC needs to have storage to hold a program large enough to meaningfully handle LCD output, keypad input of some type, a wireless interface and associated protocol, read various sensors, and enough computing power to perform some type of environmental simulation to keep my lazy self out of the garage. It also needs to have enough EEPROM storage to maintain its configuration when power is out.

Notice the wireless component of the project keeps popping up. The wireless component exists to optionally enhance environment information, allow for configuration of zones, and to obtain logged irrigation zone information (when, how long, sensor readings, etc). And of course, to not be left in the dark with more traditional Arduino sprinkler projects, the wireless component is to allow for manual operation of each zone as well as simple timers.

Stay tuned to see how this progresses.