The ATmega32 microcontroller has the following architecture:

• The CPU components are shaded blue.
• The memory components are shaded green.
• The clock components are shaded in orange.
• The I/O components are shaded in purple.

• Native data size is 8 bits (1 byte).
• Uses 16-bit data addressing allowing it to address 2¹⁶ = 65536 unique addresses.
• Has three separate on-chip memories
  • 2KiB SRAM
    • 8 bits wide
    • used to store data
  • 1KiB EEPROM
    • 8 bits wide
    • used for persistent data storage
  • 32KiB Flash
    • 16 bits wide
    • used to store program code
• I/O ports A-D
  • Digital input/output
  • Analog input
  • Serial/Parallel
  • Pulse accumulator

Before we get into the details, let's look at the programmers model to help motivate our desire to better understand the internals of the ATmega32 microcontroller.

• Using high-level languages like C, C++, and Java require little, if any, knowledge of the CPU's internal configuration.
• Assembly language requires knowledge of the internals of the CPU since we are operating at a lower level.
• Machine language is the native language of the CPU
  • Consists only of 1's and 0's.
  • Is what we actually download to the controller's program memory.
  • Is stored in the .hex file generated by AVR studio.
• Program hierarchy:
  • High-level language gets converted into assembly language by a compiler.
  • Assembly language gets converted into machine language by an assembler.
  • AVR Studio acts as our assembler for this course.
• For CE2800, we will be writing programs in assembly language.
• In CE2810 we will write programs/functions in C and use a GNU compiler to compile those programs/functions into assembly.

• Program Counter fetches program instruction from memory
  • The PC stores a program memory address that contains the location of the next instruction.
  • The PC is initialized to 0x0 on reset/power up.
    • When we placed the bootloader on the chip, we set a fuse that enables a reset vector to load the PC with the address where the bootloader program starts.
  • When the program begins, the PC must contain the address of the first instruction in the program.
  • Program instructions are stored in consecutive program memory locations.
  • The PC is automatically incremented after each instruction.
  • Note: There are jump instructions that can modify the PC (e.g., the PC must change when calling or returning from some other routine).
• Places instruction in Instruction Register.
• Instruction Decoder determines what the instruction is.
• Arithmetic Logic Unit executes the instruction.

The CPU clock determines the timing of when instructions are fetched and executed:

• Directly accessed by the CPU/ALU – very fast.
• Registers contain:
  • Address of the next instruction to fetch from program memory (PC).
  • Machine instruction to be executed (IR).
  • “Input” data to be operated on by the ALU.
  • “Output” data resulting from an ALU operation.

• There are 32 8 bit general purpose registers, R0-R31.
• X, Y, and Z are 16 bit registers that overlap R26-R31.
  • Used as address pointers.
  • Or to contain values larger than 8 bits (i.e., >255).
  • Only register Z may be used for access to program memory.
• Certain operations cannot be performed on the first 16 registers, R0-R15
  • LDI
  • ANDI
  • etc…
• CLR Rx does work on all registers.

• Stack pointer (SP)
  • Is 16 bits wide.
  • Stores return address of subroutine/interrupt calls.
  • May be used to store temporary data and local variables.
• Status Register (SREG)

• Is 8 bits wide.
• Contains information associated with the results of the most recent ALU operation.
  • Carry flag (C) set if CY from most signficant bit (MSB).
    • Often used when adding numbers that are larger than 8 bits. Here we need to use an add with carry instruction (ADC) to add the next most significant byte.
  • Zero flag (Z) set if result is zero.
    • BREQ instruction says branch if zero flag is set.
    • BRNE instruction says branch if zero flag is not set.
  • Negative flag (N) set if MSB is one.
    • Arithmetic instructions change the flags
    • Data transfer instructions do not change the flag.
    • The instruction set documentation identifies which flags each instruction may modify.
  • Overflow flag (V) set if 2's complement overflow occurs.
    • An overflow occurs if you get the wrong sign for your result, e.g., 64₁₀ + 64₁₀ = -128₁₀ (01000000₂ + 01000000₂ = 10000000₂).
    • Note: whenever the carry into the MSB and the carry out don't match, we have an overflow.
  • Sign bit (S) s = N EXOR V.
  • Half carry flag (H) is set when carry occurs from b₃ to b₄.
    • Used with binary coded decimal (BCD) arithemtic.

• The address bus is 16 bits wide.
• The data bus is 8 bits wide.
• Program memory is stored on Flash from 0x0000 to 0x3FFF (F_END).
  • Our boot loader is loaded in the last 1KiB of the Flash memory.
• SRAM is used for:
  • 32 General Purpose registers from 0x00 to 0x1F.
  • 64 I/O ports from 0x20 to 0x5F.
  • User data from 0x60 to 0x85F (RAMEND).
  • Note: the bootloader uses the last 32 bytes of data memory, we will consider RAMEND-0x20 as the end of data RAM.
• ROM for user data is stored on EEPROM.
  • ATmega32 has 1KiB of storage from 0x000 to 0x3FF (E_END).

You should read pp. 3-13 of the ATmega32 Specification.