geez
[Top] [All Lists]

Re: Calibrating the Cube...

To: HUOA@aol.com
Subject: Re: Calibrating the Cube...
From: Byron Short <bshort@AFSinc.com>
Date: Wed, 06 Oct 1999 11:09:32 -0700
HUOA@aol.com wrote:
> 
> Sometimes I'll get a run that appears as if the cube wasn't correctly
> calibrated.  Typically, the cube will not auto stop.  When I look at the run
> afterwards I realize that the cube is getting a decel reading after I've come
> to a complete stop.  Sometimes this reading is great enough (around .1 G or
> so) to keep the palm from auto saving the run.  Strange thing is, prior runs
> recorded fine and subsequent runs also recorded fine.
> 
> Am I doing something wrong?

When you hit "AutoStart", the Cube does a 1/2 second calibration right 
at that moment.  In that 1/2 second, it sets it's "zero" position.  If 
at that moment your car was on an uphill slant, the cube would then see 
a level surface as deceleration.  My guess is that if you are getting 
this kind of error, but only occasionally, that's the type of problem we 
are dealing with.  The good news is that even relatively large amounts 
of error such as 0.10g can be compensated in the Adjustments screen.  
Therefore, if you find that your G-Cube is still recording after you 
stop, just hit the STOP button, and let the run save.  You'll have a 
bigger trim job and adjustment job to do, but it'll likely still work 
out.

> 
> Also, should I not expect my overall scores to be similar from event to
> event?  I've seen courses where an overall score in the 90% range felt like a
> really awesome run and other events where an overall score in the low 80%
> range felt equally as good to me.  Was I really driving that much worse on
> the second day?

Two comments on this.  First, the scores are definitely dependant to a 
degree on the course.  It's considerably tougher to score 90's on a 60 
second course which includes both slaloms and sweepers than it is to 
score 90's on a 30 second course with all similar turns.  Ideally, I'd 
love to be able to tune this all out of the ratings, but some of it will 
always be there, because, well, because we probably do perform better on 
a 30 second simple course than we do on a longer and more complex 
course.

Second comment.  When we were only supporting the G.Analyst, we took a 
couple years to get the ratings tuned to closely match reality.  I feel 
that our correlation (fast = highscores) is better than 90% with the 
G.Analyst.  However, I don't feel that strongly about the G-Cube...yet. 
 The G-Cube's lighter weight deflection arm (the actual micro-machined 
arm that bends under acceleration) makes the G-Cube much, much more 
responsive than the G.Analyst.  This feature alone has made our ratings 
less accurate with the G-Cube than with the G.Analyst!  Not a situation 
I like, but one that we are working on.  Now that we have a pretty good 
collection of a wide variety of runs from Nationals, and other good 
long solo courses, we are beginning the process of tweaking the G-Cube 
rating system to have better correlation.  I will obviously keep 
everyone informed when we think we have this ready.  It *will* certainly 
require different algorythms for G.Analyst than it does for G-Cube, as 
we don't want to hurt the high correlation we now see with the 
G.Analyst.

--Byron

> 
> Rich Wayne



<Prev in Thread] Current Thread [Next in Thread>