- 1 AVR Microcontrollers
- 2 AVR Hardware at CCRMA
- 3 AVR Software at CCRMA
- 4 Setting up your Computer to do AVRLib / CCRMA-Style AVR Development
- 4.1 ALL PLATFORMS
- 4.2 Linux Machines at CCRMA
- 4.3 PlanetCCRMA Linux on your own PC
- 4.4 Mac OS X / Linux
- 4.5 Windows
- 5 Compiling and Downloading Programs
- 6 Communicating with Pd via OSC over USB
- 7 Communicating with Max/MSP via OSC over USB
AVR 8-bit RISC microcontrollers by Atmel are currently at the heart of the embedded technology platform we use in Physical Interaction Design research and courses at CCRMA. These chips have a modern Harvard Architecture, executing most instruction cycles in a single clock cycle. An extremely efficient open source C compiler is available, and these devices are used and supported by a large community of developers. The AVR architecture and instruction set is available in a large family of 8-bit devices that come in a variety of packages. AVR-series chips come with as little as 1 Kbyte of Flash program memory, up to 256 Kbytes; 32 bytes to 8 Kbytes of RAM; and up to 20 MHz clock speeds. Specialized AVR devices exist for automotive, LCD and lighting control, USB and radio applications. Code is easily portable between devices.
AVR Hardware at CCRMA
We currently use the ATMega 32 (Product Page). It features:
- 32 Kbyte self-programming Flash Program Memory
- 2 Kbyte SRAM
- 1 Kbyte EEPROM
- 8-Channel 10-bit A/D-converter
- Up to 16 Mhz clock
- I2C, SPI interfaces
- 1 USART, 2-channels PWM, 2 ext. interrupts
ATMega 32 Documentation:
AVRmini Development Board
The AVRmini development board, made by Pascal Stang, is a general purpose prototyping board that provides easy access to the AVR's I/O and programming interfaces, along with a number of useful peripheral devices. We are currently using the AVRmini v4.0. The AVRmini v4.0 provides:
- Support for most 40-pin AVR processors
- Ethernet, USB, SD/MMC, RS-232 I/O
- I2C header
- Character LCD header
- 10-pin I/O headers
- 4 buttons and 4 LEDs
- 5-V regulator
CCRMA's AVR Prototyping Kit
For our PID courses at CCRMA, we assemble a prototyping kit based on the AVRmini. We mount the AVRmini on a sheet of 1/8" acrylic, and add solderless breadboard strips, an LCD display and a small loudspeaker. Pins 9 and 10 of the AVRmini v4.0's "AUX" header (JP13) are connected to the ground and +5V power rails of the breadboard strips.
- kit Photo
Spyglass User Interface Board
The LCD display on CCRMA's prototyping board is part of the Spyglass User interface Board, which communicates with the AVR via I2C. We purchased these on eBay, and from a few other places, but they are hard to come by these days. Pascal Stang has written a brief Spyglass User Guide. He has also developed a replacement I2C button module/LCD driver (no documentation available) to which we can connect an LCD display and use just like the Spyglass.
- Spyglass photo
- buttonmodule photo
AVR Software at CCRMA
We use a set of open source tools for compiling C programs and downloading them to the AVR. Here is a brief description of what these tools are. See below for detailed instructions on setting up these tools on your own computer.
Programming the AVR
The avrlib is a library of easy-to-use C functions for a variety of tasks on the AVR microcontroller. It gives high-level functions for accomplishing tasks that might otherwise be tedious to program. These include serial communication, analog-to-digital conversion, displaying text on LCD displays and interfacing to a whole range of other devices such as hard drives, GPS units, Ethernet devices and sensors. The avrlib is distributed under a GNU Public License. At CCRMA, we make a few slight modifications to the original avrlib.
The avrlib directory contains mainly a collection of .h and .c files containing the library function. They are well-documented and compatible with the doxygen documentation generation tool. Here is a link to the avrlib on-line documentation.
The avrlib-demos are a set of example AVR development projects developed at CCRMA, some based on code from Pascal Stang's examples in the avrlib. Each project in the avrlib-demos contains at least 3 files:
This is the .c file containing the code for the project. Since microcontroller programs are typically very small, all the code we need to write is normally contained in one .c file. We normally use other functions from the .c files in the #avrlib, but these are linked in when we compile the code. By convention, the name of this file (minus the .c extension) is the same name as the directory in which it is contained, i.e. button.c is in the button directory - this is the "name" of the project. If you want to create your own project based on one of the avrlib-demos, you should copy the directory, change its name, and change the name of the corresponding .c file. You'll need to change the target name in the makefile as well (see below). See the tutorial on writing C programs for the AVR for more information.
The global.h file is required by the #avrlib. In most cases, it simply defines the clock speed of the AVR microcontroller that you are using. The general philosophy of the avrlib is that the library should remain general-purpose, and able to support a range of projects for different microcontrollers and hardware without modification. So, since you may have different projects on your computer for different AVR devices at different clock speeds, there is a global.h in every project directory (rather than keeping a common one in the #avrlib directory).
This file contains project-specific settings for compiling and downloading your code to the microcontroller. The makefile is read by a program called make that is a sort of scripting language commonly used for generating compilation instructions. The makefile in each project contains settings that are passed to a global makefile called avrccrma_make that resides in the avrlib/make directory. The 2 most common things you may need to modify in the makefile are the target name of the .c file that you are compiling and the list of avrlib .c files that you are using. See the tutorial on writing C programs for the AVR for more information.
There is one other type of file that is present in some of the avrlib-demos directories that you may need in your projects. These are configuration files that contain settings for specific avrlib modules. For example, in the lcdtest directory, you will see a file called lcdconf.h. lcdconf.h contains settings relevant to the lcd display you are using, such as the number of lines and columns. As described above, you may have different projects using different lcd displays, so rather than change a global configuration file each time you compile your code, the avrlib requires that you put a different configuration file in each project directory.
avr-gcc is actual compiler that generates AVR byte code from the c files that you compile. We don't invoke avr-gcc directly, rather it is called from the makefile. At the end of a successful compilation, the compiler will generate a .hex file that must be loaded onto the microcontroller. See below for instructions on installing avr-gcc on your own computer.
Downloading Compiled Programs to the AVR
After the compiler has generated a .hex file, it must be loaded onto the microcontroller. There are a variety of ways of doing this, using different programming hardware and/or protocols for transmitting the data. We use an option that requires no additional programming hardware and uses freely-available software. The avrmini v4 also supports programming by external programming hardware, such as the STK500 development board and the AVRISP programmer. Use of these programmers is not covered here.
At CCRMA, we install a bootloader on the ATMega32 microcontroller. The bootloader takes advantage of the AVR's self-programming feature. Residing in a special section of program memory, the bootloader is a small program that is executed as soon as the AVR is reset. For a certain amount of time, it looks for incoming programming instructions on the serial port, which it uses to load a new application program into memory, overwriting the rest of the Flash. If no programming instructions are received, the bootloader times out and branches to the beginning of the main program memory, running whatever application program was previously there. The serial data may come from the DB9 serial connector on the avrmini connected to a real serial port on a PC using the RS232 protocol, or via USB through the FT232R chip that can appear as a serial port both to applications on a PC and to the AVR chip, emulating RS232 communication.
The bootloader we use emulates the STK500, an AVR development board from Atmel. You can download the bootloader and get more information here: http://hubbard.engr.scu.edu/embedded/avr/bootloader/index.html.
If you have an AVR development kit that you got from CCRMA, you probably have a bootloader installed. If you didn't, you may want to install a bootloader or use an alternate programming method, such as #ftisp.
avrdude is the "standard" and probably most widely-used open-source AVR programmer software. By default, the makefiles in the #avrlib-demos and #avrlib use avrdude. There have been problems with some versions of avrdude on the Mac, so support for uisp is included as well.
uisp is another open-source AVR programmer software. It does not appear to have been updated in awhile, and it is unknown whether its development is still active. It may be necessary on some Macs where some versions of avrdude don't work with the hardware and software we are using.
PonyProg is a gui-based closed-source programming software package for a variety of devices including AVR microcontrollers. It is free to download, and available for Windows and Linux only. PonyProg has its own hardware /software specification for interfacing with AVR microcontrollers, or it can use other standard protocols.
Setting up your Computer to do AVRLib / CCRMA-Style AVR Development
No matter what platform you use, you need to get the avrlib. If you are using a Linux machine at CCRMA, see the next section for the best way to do this. Otherwise, you can download the avrlib here:
and decompress it into your home directory. On a Linux or Mac system you can do this with:
~> tar xzf avrlib_ccrma20070527.tar.gz
A set of demo programs to demonstrate the basic functionalities of the AVRlib and AVRmini are also available:
Again on Linux or Mac, you can decompress them with:
~/somewhere> tar xzf avrlib-demos20070124.tar.gz
The avrlib-demos can reside anywhere, but they assume that the avrlib is in your home directory.
Installing the STKload Bootloader
The only "catch" with our method of programming the AVR is that the bootloader must first be installed on the AVR.
Linux Machines at CCRMA
The Linux machines at CCRMA already have the AVR development tools installed, and in your path. All you need to do is get a copy of the AVRlib. The avrlib-demos are helpful in getting started.
To get the most up-to-date versions, use subversion.
In a terminal, change into your home directory, and then type:
~> svn checkout file:///usr/ccrma/group/pid/avr/svn/avrlib ./avrlib
This will create an avrlib directory in your home directory. You can update it anytime from a terminal, by changing into the avrlib directory and typing:
~/avrlib> svn update
To get the avrlib-demos with subversion:
~/somewhere> svn checkout file:///usr/ccrma/group/pid/avr/svn/avrlib-demos ./avrlib-demos
You can update them with svn update as well. The avrlib-demos can reside anywhere, but they assume that the avrlib is in your home directory.
PlanetCCRMA Linux on your own PC
AVR Development Tools
If you have PlanetCCRMA installed on your PC, then you just need to make sure that you have the AVR development packages installed, using yum or apt.
AVRlib and avrlib-demos
Download the avrlib and avrlib-demos as described above. Note that you need to get the .tar.gz files, as the svn repository is not available outside of CCRMA.
Mac OS X / Linux
These instructions are excerpted from avr-libc: Installing the GNU Tool Chain.
Mac OS X: The Easy Way
1.You need the gnumake program, and you also need a symbolic link to it called make.
There are two options: i) Install Apple's developer tools, which include gnumake and much, much, much more. If you want to develop OSX applications you probably want or already have these anyway. They have a fancy easy-to-use installer.
Then, unstuff it, and install the resulting file like this:
~> tar xzf osc-gnumake_<platform>.tar.gz ~> sudo mv gnumake /usr/bin ~> sudo ln -s /usr/bin/gnumake /usr/bin/make
2. Download, un-stuff this archive and put the resulting avr directory in /usr/local, which you may need to create if it doesn't exist:
You will probably need to do this as administrator, i.e.
~> tar xzf usr-local-avr_ppc.tar.gz ~> sudo mv avr /usr/local
3. Add /usr/local/avr/bin to your Unix path. For example, if you use the tcsh shell, add this line to the file .cshrc in your home directory:
set path=($path /usr/local/avr/bin)
If you use the bash shell, add these two lines to the .bash_profile file in your home directory:
PATH=$PATH:/usr/local/avr/bin export PATH
Linux / Mac OS X (the not-so easy way)
This requires compiling the tools from source. It is not too difficult, but may take some time, depending on your bandwidth and CPU speed. On a Mac, this assumes you have the Apple's developer tools installed. On Linux, it assumes you have gcc and a basic operational build setup.
First, you need to download the following sources (either .tar.gz or .tar.bz2):
- GNU Binutils (currently at version 2.17)
- GCC (currently at version 4.2.0)
- avr-libc (currently at version 1.4.6)
- avrdude (currently at version 5.3.1)
Next, you'll need to decide where to install the tools. It is usually a good idea to keep the avr build tools separate from your system build tools, so we recommend installing everything to /usr/local/avr. In order to do this, it is helpful to set an environment variable called PREFIX that points to this location. In a Bash shell (if your prompt is a $, this is probably what you're using), do:
$ PREFIX=/usr/local/avr $ export PREFIX
The equivalent in csh or tcsh is:
~> setenv PREFIX /usr/local/avr
3. You'll also want to set your path to include the location where you are going to install the tools. In Bash:
$ PATH=$PATH:$PREFIX/bin $ export PATH
In csh or tcsh:
set PATH = ($PATH $PREFIX/bin)
In either case, you should also add this to your default path, as in Step 3 of the previous section.
GNU Binutils for the AVR target
The binutils package provides all the low-level utilities needed in building and manipulating object files. Once installed, your environment will have an AVR assembler (avr-as), linker (avr-ld), and librarian (avr-ar and avr-ranlib). In addition, you get tools which extract data from object files (avr-objcopy), dissassemble object file information (avr-objdump), and strip information from object files (avr-strip). Before we can build the C compiler, these tools need to be in place.
Download and unpack the source files:
~> bunzip2 -c binutils-<version>.tar.bz2 | tar xf - ~> cd binutils-<version>
Note: Replace <version> with the version of the package you downloaded. If you obtained a gzip compressed file (.gz), use gunzip instead of bunzip2.
There are 2 patches that need to be run before you can compile the binutils. From this URL:
Download the patch files called patch-newdevices.txt and patch-avr-coff.txt, and save them in the binutils-<version> directory. Apply the patches with:
~> patch -p0 < patch-newdevices.txt ~> patch -p0 < patch-avr-coff.txt
Now you are ready to compile binutils. It is usually a good idea to configure and build binutils in a subdirectory so as not to pollute the source with the compiled files.
~> mkdir obj-avr ~> cd obj-avr
The next step is to configure and build the tools. This is done by supplying arguments to the configure script that enable the AVR-specific options.
~> ../configure --prefix=$PREFIX --target=avr --disable-nls
When configure is run, it generates a lot of messages while it determines what is available on your operating system. When it finishes, it will have created several Makefiles that are custom tailored to your platform. At this point, you can build the project.
Note: BSD users should note that the project's Makefile uses GNU make syntax. This means FreeBSD users may need to build the tools by using gmake. If the tools compiled cleanly, you're ready to install them. If you specified a destination that isn't owned by your account, you'll need root access to install them. To install:
~> sudo make install
You should now have the programs from binutils installed into $PREFIX/bin.
GCC for the AVR target
Warning: You must install avr-binutils and make sure your path is set properly before installing avr-gcc.
The steps to build avr-gcc are essentially same as for binutils:
~> bunzip2 -c gcc-<version>.tar.bz2 | tar xf - ~> cd gcc-<version>
At this point there is another patch for avr-gcc. Download the file patch-newdevices.txt (different from the patch for binutils of the same name) from here:
Move the patch file into the gcc-<version> directory and apply it with:
~>patch -p0 < patch-newdevices.txt
Now, you can compile avr-gcc:
~> mkdir obj-avr ~> cd obj-avr ~> ../configure --prefix=$PREFIX --target=avr --enable-languages=c,c++ --disable-nls --disable-libssp --with-dwarf2 ~> make ~> sudo make install
Warning: You must install avr-binutils, avr-gcc and make sure your path is set properly before installing avr-libc.
To compile and install the avr-libc
~> tar xzf avr-libc-<version>.tar.gz ~> cd avr-libc-<version> ~> ./configure --prefix=$PREFIX --build=`./config.guess` --host=avr ~> make ~> sudo make install
avrdude is the application that downloads the compiled code from you computer into the AVR's program memory. To compile and install it:
~> tar xzf avrdude-<version>.tar.gz ~> cd avrdude-<version> ~> mkdir obj-avr ~> cd obj-avr ~> ../configure --prefix=$PREFIX ~> make ~> sudo make install
Compiling and Downloading Programs
Communicating with Pd via OSC over USB
Linux Machines at CCRMA
PlanetCCRMA Linux on your own PC
Mac OS X
Communicating with Max/MSP via OSC over USB
Mac OS X
==Communicating via OSC over udp ==</pre>