I’ve picked up some representative AVR based boards to supplement the ARM based RPi board I have around the house. Between the two, I expect that a pretty capable solution can be put together. The ARM running Linux and providing high level business logic and user interface while the AVR boards provide hardware level control and real-time ‘fast twitch’ capability.
The ATMega2560 is a high-end device with generous resources in pretty much all relevant areas. It is quite a bit more expensive than the lower tier parts and I’m inclined to look at a finer grained approach initially.
The ATmega328 is the ‘common’ controller in this line. This is the part that powers most of the Arduino line of boards. It has relatively limited resources but with multiple timer/counters and an I2C link should suffice for most of the sorts of tasks I’m looking at. This is a fully featured board, but I have a package of five free-standing boards on the way. These boards run less than $5.00 a piece but have no integrated USB capability and thus need either an ISP programmer or a separate USB to RS232 converter to load with code.
The last three small boards I picked up for sandbox purposes are ATmega23U4 boards. These use the AVR controller that has built-in USB capability. I’m very interested in what can be done with this support as it should allow coding of various sorts of USB devices using the native hardware on the parts.
I’ve had these pololu stepper controller break-out boards sitting around for some time now. They’re destined for a RepRap machine (assuming I ever find the time and budget to finish buying components). They’re H bridge, step and direction style stepper controllers with an Allegro A4988 device. I’m expecting to push these together with a stepper motor and at least see it responding to commands in the near future.
Once I’ve got a stepper running at least a bit, the next step needs to involve more mechanical parts…probably makes sense to get the real RepRap implementation rolling at that point.
I am a bit curious about ARM based possibilities…something like a cortex M0 with a free/open source RTOS running on it. This would certainly allow for more to be done with fewer independent controllers…though the relevant pieces remain to be worked out. I’m suspicious that the RPi SOC is too poorly documented to work here…would need a lower end, more dedicated part with a board support package for something free and capable…more to come.