Another thing to check is the airplane's rigging. I spoke with the S-TEC folks at Oshkosh a couple of years ago about a similar oscillation (except in the pitch axis) and they handed me a card with a bunch of autopilot troubleshooting steps that they hand out, because they get a TON of autopilot computers back to their shop for "repair" that test out just fine. When they get the whole aircraft and can check everything, they find that the #1 problem is aircraft rigging, and that you don't start getting to problems with the autopilot computer until #6 or so...