Talking about calibration and GEEZ ratings.
>maybe your cube wasn't calibrated right?
<< I'd love to say this was true, but we had two different
cubes running at the same time in the car at Harrisburg!
GEEZ Cube ^2 ! :-) One was capturing at 10 samples/second,
the other one at 20. The only explanation is that the shiny
primer on my new car somehow warps either space or time. ;-) >>
>I cannot imagine how you'd get 99% usage on two runs that
>are 2+ seconds apart.
I use GEEZ, and I understand it perfectly. 99% usage means you used 99% of
the cars peak usage value. The systems is all relative on the peaks it saw.
And by using sustained peaks, it makes the max 100% reading the max it saw
for a given number of samples. So if you are very smooth, but only driving at
.7G, you can still score a 99%. If you hit 1G for a peak, but then only
maintain .8G, you get lower useage rating, even though you wen't fatser
through every turn. So now in the rain with a reduced max grip, it is easy to
see where a 99% usgae is much easier, and to get it you will have high
smoothness. Aggressiveness is a stranger one as it has to do with how fast
you can change from turning to braking etc. But once again, it is still
relative, so if you are always slow in transition rate, you can still score
a 99%.
<< That's 99% overall on slow runs in the rain, less on the
faster runs in the dry. Byron can (and I'm sure will) explain
the math ad nauseum. >>
The relative reading makes it far easier to set up, but does limit the
performance of the system as a comparison tool between cars and drivers. If
one car scores a 99, but it's sustained peak G's are 10% lower, it will be
slower, GEEZ ratings are no help at all for classing.
<< Going back to the beginning of this thread, Kevin Wallace
suggested using GEEZ ratings in place of PAX to equalize
drivers. My only point is that *anyone* can drive
to fool GEEZ. If a little rain can knock off the peaks
in the data and inflate your stats, then obviously sandbagging
could do the exact same thing. >>
Now, if Byron made it so 1G is 100% (ALWAYS) and 100% transition was some
mythical X G's per second (what car is THE standard here) and peak
acceleration had some "standard" we could make it compare cars, but it would
have to be able to do over 100%. And make the length of time for a sustain a
constant also. A car rating could be achieved. But this is an awful lot of
if's.
The same car on Concrete or Asphault would have seriously different ratings.
The rating would still be very subjective, if you weight the trans rate high,
it will score better for super light cars, if you rate accel high, powerful
cars would score better, If the course favors acceleration it skews the
results.
The software would need alot of re-working, and it is going backwards from a
driver training standpoint. GEEZ was written to teach the driver to use the
car. The current rating system shows you how well you are using a particular
car. But the ratings are only 1/10th of the picture. You need to look past
them a bit and use the actual data screens, especially the G vs G plot and
even the strip G charts. How fast do you get the G's up, how smooth do you
hold it for the curve, Did you hold speed in the curve, or give up too much
before entering, or dive too deep and have to scrub it off?? I am finding
turns where my wife gets a higher rating, but I covered the curve in less
total time. Looking further at the data is showing why and where. It is
helping her alot to get closer to me. I wish I had constant access to a
better drivers runs in the same car to see where I am off. Jeff Cashmore did
send me a couple if his runs from an even I was at with him, and it helped
some, but his calibration was a bit off, and he drives so loose, it made it
look like he was gaining speed when it was really just the tail drifting out
a bit, I had to greatly change the map settings to make it look like the
course and my map. And I have been tuning my car looser to get there, it sure
seems to work for him.
Gary Meissner
Going on 13 years, and the biggest improvements are still on the driver.
|