Nope. Works the same for all motor types I know of, including series wound. Increase current and you'll increase the torque, no black magic involved.
Na, it just complicates things.
Not much that hasn't been said already. The AVR is powerful enough to run a DC-controller and it's a reasonably simple little microcontroller to understand and write safe code for if you know what you're doing. There's a lot of code on the net to study and there's the avrfreaks-forum that's usually very friendly and useful if you run into problems.
Constructing the power electronics' gonna be a bitch though. Not that it's theoretically complicated, but in real life weird things start to happen when you try to harness high power. Going for tens of kW, like in a golf cart controller, isn't very hard (and that's why those controllers aren't very expensive) but when you aim at hundreds of kW or the glorious MW you'll start to blow things up. A lot. See the threads about constructing the ReVolt in the Ecomodders forum or the Soliton 1 here. They have one thing in common, they blew up a lot of transistors in the prototype runs...
Combining software and hardware is gonna make things even more interesting. A small software error can easily blow up a few hundred dollars or more (been there, done that) a hardware problem is likely to blow up more or less all of the electronics (at least the expensive parts, Tesseract knows all about that from empirical experience) and in the end building your own controller from scratch and just making a single one is probably going to be more expensive than you think.
I've seen a few people boost here in the forum that they're gonna build their own controllers. In most cases we hear a lot about it until they start to run tests and blow up their first set of transistors, then they usually give up and go quiet. A few do actually finish their controllers, but that's usually guys that (excuse me if I'm blunt here) do have experience from similar constructions already so they're not getting stuck at the basics.
Personally I think you have two realistic options:
- Go for ReVolt and help improving that controller rather than trying to reinvent the wheel.
- Buy something that's been proven to be reliable and spend your energy at converting a car rather than blowing up silicon.