Perhaps our team hasn't been setting it up properly, then? I distinctly recall trying just that, and ending up with an arc depending on the degree to which it was powered. I'm not sure how one would explain this, though.

As for Cartesian vs. polar, most implementations of Cartesian processing I've seen simply convert to polar coordinates, then use sines and cosines. I guess it would be possible to do so otherwise, but it seems less simple and/or flexible, at least at first glance. That could definitely be wrong, though.