long acceleration in g calculation

[tiburonski]

I tried to verify the long acceleration data (g) by differentiating the speed in m/s versus time giving me the acceleration in m/s^2 and dividing that result by 9.81 to get the long acceleration in g.

then I compared the derived g with the long acceleration in g provided as a standard variable and noticed that they do fairly match but with some exemptions.

any ideas why? the standard vector g also seems to be much smoother and sometimes higher.

[tiburonski]

yes I found the reason:

the difference is due to the slope of the track. the negative slope reduces the long g and the positive slope increases the long g.

the difference between the long g and the derived long g is the vertical g....finally I found out how to derive the vertical g. great!

[faraday]

Please be aware that there is a lot of noise in accelerometer data and GPS altitude is much less accurate than (X,Y) position.
Think about the effect of applying calculus to noisy data.
It is easy to smooth and filter data, but often not without introducing innacuracy.

It is easy to generate data with lots of decimal places, but this does not make it accurate. All physical measurements have innacuracy and variability => uncertainty.

Sometimes our mathematical tricks can actually increase the uncertainty.

[tiburonski]

no problem. I have data of 6 laps which are quite simular concerning the altitude data. besides they match fairly close the offical altitude data of the race track.

the difference of measured altitude compared to the offical altitude is minmal. at max 1 meter only!

look over here where I posted the comparison graphs:

http://www.race-technology.com/forum/vi ... .php?t=224

[faraday]

I get nothing like that accuracy with DL1s, but I'm not suggesting there is a difference AX22vDL1.
Your comparison looks great, not just because the accuracy is exceptional, but because of the significant elevation change.
Long circuit too; name and location?

I'm used to quite flat circuits with less than 10m change. Its easy to get +/- 6m. A trend is visible in a 6 lap overlay, but...

Please prepare your 6 laps overlaid. Zoom into the best and worst deviation areas. Oops... we can't do six in Analysis; five will have to do.
Send me a run file or so, if that's possible: rotoflow@optusnet.com.au
I'll swap you some really interesting stuff!

[tiburonski]

as I am quite busy at the moment (I have to drive to nürburgring ) I'll e-mail you the complete run file. the track is brno circuit in the chech republic which is rather close to vienna/austria where I live:

http://www.automotodrombrno.cz/go/main. ... 3&langid=2

when you make a comparison don't forget that the GPS receiver is mounted on the top of my car roof so you will have to subtract about 1,33 meters.