September 28, 2008 8:39 AM
updated: September 28, 2008 8:42 AM
Either way can be educational, in different ways. I think or friend is more interested in applying it in his research. From my perspective replacing the CPU is cheaper than fabricating a new board (budget sometimes being an issue in research), but replacing the entire board may make using it as a research platform easier (depending).
When I dissected the RoboBoa, I thought the joints in might be interesting to adapt to a simple robotic arm. Unfortunately this project has been sitting on my shelf for a while now (along with a few others) while I have been devoting my times elsewhere. You can read about it on my site, this link should take you to a listing of the related posts in reverse chronological order.

This image is probably the best representation of the steps relevant to this discussion. You can see on the notepad the simple pinout I made for the CPU, the notes underneath are each pins function, along with relevant details about the signal (all of which was done with just a multimeter). The original CPU I removed with a hot air gun normally intended for paint stripping, and a razor blade (for lifting the edges). I then attached wires to the pads to connect the relevant signals to my dev board. That large number of unattached pads almost entirely relate to RoboBoa's LEDs.
RoboGuide - Your guide to hacking all things WowWee