autox
[Top] [All Lists]

Re: Datalogger Story (now shorter)

To: Jay Mitchell <jemitchell@compuserve.com>
Subject: Re: Datalogger Story (now shorter)
From: Byron Short <bshort@AFSinc.com>
Date: Fri, 06 Aug 1999 22:30:56 -0700
The G-Cube uses a 2 axis accelerometer.  Roll, pitch, and original 
non-level positions are all mathematically compensated.  Accuracy and 
repeatability is typically <0.01g, which you can easily confirm yourself 
when you get your own G-Cube.  Calibrate your G-Cube, the test each pole 
in a make believe run on your desktop.  It's a simple test.  

BTW, my G.Analyst is pretty far off these days, but it's an ancient 
thing and I've abused it pretty badly over the years.  I'd love to see 
what the Edelbrock records in this test...might give a clue to Dennis 
readings...

Also, regarding the G.Analyst's supposed 3 axis accelerometers.  The 
literature that came with the G.Analyst talked this up, and that all 
sounded pretty good, but the results that you get when you use the 
G.Analyst are exactly the same as if the third pole is not used at all. 
 It's possible that the third pole is not used in the manner that 
several of us have independently figured that it should be.  If it were, 
for instance, the g.Analyst would not be fooled by a simple tilt for 
acceleration, as the third pole would show that it was tilt, not true 
lateral acceleration, and the internal mathematics would cancel the 
readings.  But they don't.  I understand that there is very definitely a 
third pole in there, but I can't for the life of me figure out how it is 
used.  We use exactly the same mapping algorythm with the G-Cube that we 
do with the G.Analyst, too, which also shouldn't work if they were 
getting readings in a different way.

BTW, as a matter of interest, the G-Cube is designed to use it's two 
accelerometers differently in G-Dyno mode.  This mode is aimed at 
straight line performance, and in that case we turn one of the poles 
sideways, essentially measuring earth gravity in g's.  This allows us to 
parse out the forward g component while perfectly canceling the false 
g's from the rise of the nose of the car under acceleration.  This is 
how the third pole should have been used, IMHO, in the G.Analyst.  
However, this is programmatical because while the third pole can 
determine how much it is leaning, it can't know in which direction, so 
the value is suspect in a full 360 degree system (one in which you both 
turn and accelerate).  This may be why the G.Analyst seems to not use 
the third pole.   In GEEZ mode, roll and pitch are canceled 
mathematically by entering the roll and pitch degrees/g.

--Byron

Jay Mitchell wrote:
> 
> Wes wrote:
> 
> >What does Byron's Geez logger use, 2 axis?
> 
> That's my understanding, but Byron's the one to give a definitive
> answer.
> 
> Jay


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