I was previously poking around this newsgroup for help deciding on a microcontroller to use. Along the way the CAN bus architecture was suggested several times over for linking the individual parts together.
This gave me an idea of a moderate paradigm shift with respect to the idea I had originally planned.
How many individual nodes can be placed on a single CAN bus (what practical impact does adding another node have on maximum bus length/speed), and how well does it handle multiple nodes trying to transmit at the same time (assuming a random source and target distribution)? I expect to have something in the range of 70 or so (worst case figure) nodes in total (instead of 10 in my original design) by the time I'm finished.
If I am to consider this new layout, my priorities become minimum component count to achieve uC + CAN (as opposed to modular flexibility). Being able to update the code via network is still important to me, but as each node will only be performing a single task that code will be substantially simpler. (Except for the "controller" nodes (8 to 10 of the above 70), which I may even implement using a different/bigger uC later on down the line, and will be carrying the burden of some of my more interesting plans.)
So, what's the minimum component count of, for instance, a CAN device able to monitor a switch and turn a power point on or off?
Fredderic