Damper sensors usefulness

[Belatti]

A new question here. It may be related to this thread, too:
viewtopic.php?f=6&t=7710

What kind of filters do you have to pass damper sensors data through when you want to use them in:
a) post rigs
b) computing simulations
c) statistical graphs like FFT, histograms, etc..

I admit Im having a problem in recognizing what is noise and what is not.
Also I have my doubts when it comes to sample rate selection.
My system can record at 200Hz max and thats what I use in damper sensors.

To add something else to the discussion, I plot these 2 graphs:



Here you can see 1 second of the same data recorded at 200Hz (WHITE) and exported to a 100Hz (RED) file. In blue I have pointed out the noticeable differences.




Here we have the same data as in the previous image but with a moving average filter applied (0,1 sec). The curves are smoother, there are no noticeable differences between both.

I doubt but I wonder if this curves are closer to the real suspension movement than the unfiltered ones.

[Belatti]

Another thig I did to see if I can sort it out was to use the band pass filter to decompose the signal in frequency bands:



There are barely noticeable amplitudes above 10Hz...

[Jersey Tom]

FFT and/or spectrogram would probably be much easier to visualize that. But yes.. generally as frequency goes up, content amplitude diminishes quickly.

Majority of frequency content should likely come from low-frequency (handling range) behavior.

[speedsense]

A new question here. It may be related to this thread, too:
viewtopic.php?f=6&t=7710

What kind of filters do you have to pass damper sensors data through when you want to use them in:
a) post rigs
b) computing simulations
c) statistical graphs like FFT, histograms, etc..

I admit Im having a problem in recognizing what is noise and what is not.
Also I have my doubts when it comes to sample rate selection.
My system can record at 200Hz max and thats what I use in damper sensors.

To add something else to the discussion, I plot these 2 graphs:



Here you can see 1 second of the same data recorded at 200Hz (WHITE) and exported to a 100Hz (RED) file. In blue I have pointed out the noticeable differences.




Here we have the same data as in the previous image but with a moving average filter applied (0,1 sec). The curves are smoother, there are no noticeable differences between both.

I doubt but I wonder if this curves are closer to the real suspension movement than the unfiltered ones.

I don't recognize the system software, what type is it?

One quick way to determine if it's noise would be to graph a second,third and even a fourth, lap on top (over plot) and check for consistency. Secondly, if you suspect noise, one of the "common" areas of noises are pot mounts that aren't secure. As most pots are able to read down to .002 of an inch, if the mounting has even the slightest vibration, it will appear in the data.

To me the durations of movement are too long to be noise,especially as they still appear when smoothed by half.

[Belatti]

I don't recognize the system software, what type is it?.

Pi Toolbox (club level dongle )

I´ll try your advice and also check what the sensor reads with the engine running even if the car is standing still, too.

[Caito]

Can't you check post rig programmed movement vs your detected movement. The difference would be your "noise". Although that would consider no tire deflection.