Carnegie Mellon University > School of Computer Science > Robotics Institute > Garth Zeglin > Research > Bipeds > MPC5xx > MPC565

Garth Zeglin: MPC565 Setup

Overview

In 2004 our lab began using a MPC565 controller as part of our biped robot research project. I am publishing my project notes on this page to help others who might want to use the technology, particularly in an academic context. If you find something unclear, feel free to send me a note. However, I don't have the resources for a lot of hand-holding, so please don't bug me for a tutorial, I am assuming reasonable expertise on your part. All information provided at your own risk.

My Test Setup

• Phytec phyCORE-MPC565 controller
• Phytec pC-565 phyCORE Development Board PCM-991
• Phytec Bare PCB Expansion Board
• PCAN-PCI CAN interface card
• generic Windows XP machine (1.4 GHz Pentium III)
• Cygwin 1.5.7
• GCC 2.95.3 PPC cross-compiler
• OCDLibRemote Wiggler BDM driver
• GDB 5.2.1 / Insight debugger
• newlib-1.12 (a version of libc)

Notes:

• The Phytec development board includes an integrated Wiggler interface that connects to the parallel port on the PC. I would have preferred to use Linux tools but haven't yet found a Linux BDM driver for this particular interface. It's not incredibly fast, my code seems to download at about 113 kbaud (about 14K/sec), but that's usable for me.
• Our Phytec development kit came with a demo version of Metrowerks CodeWarrior which allows a one-month trial license. This was handy for a quick functionality test. However, a full license costs $6000, which is too much for our academic budget.
• As delivered, our phyCORE-MPC565 was already mounted on the development board, and all jumpers were correctly set. I connect the supplied parallel port cable to the DB25, and the supplied serial port cable from COM1 to the upper jack of P1 on the development board.
• As unpacked, gcc doesn't quite work; one of the files is corrupted with carriage returns. I simply used emacs to delete all the CR characters (ASCII 13, control-M) from /usr/local/ppc/lib/gcc-lib/ppc-elf/2.95.3/specs, and then it worked fine.
• This gcc installation doesn't include any libraries or header files other than libgcc.a, which contains internal compiler subroutines, e.g. long long division.
• I compiled newlib-1.12 from source without any problems using the following commands:

  CC=ppc-elf-gcc
  sh ./configure --prefix=/usr/local/ppc --target=ppc-elf
  make
  make install
  

• Newlib requires a low-level target-specific support library. This is small, we wrote our own.
• Motorola supplies a lot of useful sample code, which I've started collecting into a small library.

Purchases For My Setup

Phytec phyCORE setup:

• Phytec phyCORE-MPC565. (Rapid Development Kit)
• PCAN-PCI (distributed by Phytec), PCAN-PCI (manufacturer page)
• Bare PCB Expansion Board

Axiom setup:

• Wiggler (BDM interface)

Downloads For My Setup

Tools:

• Cygwin
• GCC, binutils, GDB/Insight, and OCDLibRemote. Description on this ocdemon.com page.
• newlib

Source code:

• Motorola MPC5xx header files
• Motorola general TPU functions
• Motorola quadrature decode TPU functions
• Motorola interrupt handler prototypes
• Motorola C3F driver.

  This is a driver for the internal flash memory, supplied in a binary-only format (registration required). The download contains a Windows installer that unpacks the driver in several formats. I use the ELF format driver, which unpacks to the tree under c:/Program Files/Motorola SPS/General Market C3F/MPC56x/object_library_driver/. I installed the following files in /usr/local/ppc/ppc-elf/include/Motorola:

  gmd_include/gmd_C3F.h
  gmd_include/gmd_include.h
  gmd_include/gmd_type.h 

  I installed the following file in /usr/local/ppc/ppc-elf/lib:

  gmd_driver/ppcc3f310.a -> libppcc3f310.a 

• PEAK-System PCAN-PCI Linux driver source

Essential documentation:

• MPC565 User's Manual

Quadrature Decode TPU Module

Our application requires measuring the position and velocity of optical encoders. For a test setup I plugged the expansion board into the development board (only one way to fit it) and wired an optical encoder module to the expansion board. I soldered two 5x2 headers to the board, then used wirewrap wire to connect a US Digital S2-2048-IB encoder (which has a shaft), as follows:

TouCAN Controller Area Network Module

To test my TouCAN driver code I made a trivial cable to allow the TouCAN A and TouCAN B units to communicate locally: two DB-9 Females with solder lugs, with about four inches of wire connecting pin 2 to pin 2, and pin 7 to pin 7. I soldered a 100 ohm resistor bridging pins 2 and 7 at one end. CAN correctly requires 120 ohm termination resistors at each end, but this worked fine.

I had trouble finding good sample source on-line; Motorola application note AN1776 was somewhat helpful. In the end I wrote a simple driver from scratch to handle basic communication.

One point that confused me was the configuration of the bit-timing registers, I did not find the User's Manual very clear. AN1776 helped somewhat, and I found this website that computes a register value table. One point I still found confusing was that my bus appeared to be running slowly. I was using a RJW value of 3; apparently in my setup the timing resynchronization was consistently slowing the bus. For now I tightened up RJW to 0, which still allows a jump width of 1.

Miscellaneous Discoveries

I successfully ran the CPU at 56MHz while testing, but the spec sheet only guarantees 40MHz, so that is what my code now uses. This clock is synthesized by a PLL internally using an external 4 MHz crystal as a reference oscillator.

The phyCORE dev board BOOT switch controls the DSCK line, which is sampled by the MPC565 at reset to determine whether to enter BDM mode or not.

Links and Resources

• James' MPC555 Webpage
• Motorola MPC565 Product Summary Page

Page revision: 2006/03/26 20:01:44.