Discussion of Ackermann steering seems to get a lot of the racer-types all hot under the collar.
When a car turns, the outer front wheel goes around a larger radius than the inner wheel. To track properly without skidding, the tires should turn at different angles to match the radius of the turn. Effectively, the front tires toe out slightly when they're turned.
During the 1960s there was a fad for zero Ackermann or even anti-Ackermann. The proponents of zero Ackermann said Ackermann was of no practical use in the first place, so just dispense with it. Those favoring anti-Ackermann claimed what you really wanted to match wasn't the tire's path, but its slip angle, which could be quite a bit different from where the tire was actually pointing.
At the practical level, front-steer cars - those with a rack or tierod forward of the front axle centerline - tended to have little or no Ackermann, and rear-steer cars - those with the rack or tierod behind the centerline - had some Ackermann. I expect this is mostly due to the packaging problems with getting much Ackermann into a front-steer setup. Basically, the brake rotors and outer balljoints want to share the same space, which requires considerable compromise, usually at the expense of steering geometry.
Though the subject of Ackermann can cause hot debate, I'm only aware of one group who ever do much about it - the Fox Mustang guys. They often relocate the suspension bits to get more Ackermann, and claim less tire wear and less understeer even on open, fast courses.
With the idea that Ackermann is a Good Thing comes another complication - most Ackermann geometries possible with conventional steering linkages are only accurate at one specific steering angle (and often at a specific roll angle!) and can diverage quite sharply from true Ackermann. As for slip angles... nobody knows.
It occurs to me you could vary the length of one of the steering rods and adjust Ackermann on-the-fly. For example, the center link of a rear-steer car, though even a tierod would work. If you put an actuator such as an electric ballscrew in the center link, you could change the toe to anything you wanted - zero for minimum rolling resistance while going straight, or whatever was the optimum slip angle for both tires when turning.
The mechanical bits, oddly enough, could be simple cheap surplus-catalog items. A Stamp or similar small controller could move the actuator according to speed and steering angle according to a table.
A fancier system could use a strain gauge in the center link or
tierod, or possibly a lateral G sensor. I haven't done any figuring on
that the control algorithms would look like for that, though.
Return to Dave's home page