Brent DeWitt wrote:
> Byron has incorporated a roll constant for steady state correction, and well
> done, the correction is entirely different under, shock absorber damped,
> fast transitions than in sweepers. Admittedly, this is probably the
> smallest factor of the three, but it would take some serious chassis
> dynamics modeling to get that one right.
I have seen a model taken with MUCH more sophisticated
equipment than ours that showed that the roll was *almost*
perfectly a straight line relationship. The biggest
variance was still within about 3% of the straight line. So
this supports the roll per degree of g model that we use for
the math. However, as Brent points out, in the autocross
world we sometimes use damping values that are way out of
the norm. This can result in a quick turn, like a slalom,
not giving the chassis the time it needs to roll, while long
sweepers do. To be sure however, we can probably still
expect to be within a degree or perhaps a degree and a half
of the steady state roll values, just because so much of the
roll comes from the tires. Even shifter karts roll about
1.8 degrees per g! So yes, it's a real factor, but yes, as
Brent says, it's a small factor.
> From my investigations, one of the best
> cost/performance ways to improve the situation would be to have two
> accelerometer systems at a measured distance (x,y,z) at both ends of the
> car, but his still doesn't get you all the way there.
I believe Pi offers a system that uses two dual pole
acceleromters to compute yaw, oversteer and understeer.
I've also seen a system prototyped by GM that used three
poles of accel and three poles of yaw and it was extremely
accurate, within a few inches in an autocross course. But
the yaw sensors are not nearly as cheap as the
accelerometers, and the software gets more complex, too.
Our original idea for a "third sensor" was a single pole of
yaw. Be throwing yaw rate up against lateral g's we can
compute speed with very good accuracy. Unfortunately,
sometimes cars yaw without pulling corresponding g's, so we
don't know how the real world computations would work out.
It still looks good in theory, but you never know til you
get the pieces put together. So as of a couple years ago,
when we first noticed that the prices on GPS were dropping
quickly, we abandoned the yaw sensor idea, and have been
casually watching GPS technology.
--Byron
///
/// geez@autox.team.net mailing list
///
|