Accelerometer Noise

ajward

Me again...

I was experimenting with the 3-axis accelerometer to build a very primitive seismometer. Kind of works, but I haven't managed to catch any hiccups on the local faults.

Anyhow, I was wondering... when sitting still on my desk the readings have a =lot= of fluctuation. Not big movements, but quite constant. So, if the unit was mounted on the famous "immoveable object", would it show any fluctuation?
I know sitting on my desk it's going to pick up a lot of vibration. But does the chip generate any sort of internal noise?

That question make any sense?

Appreciate any wisdom y'all can impart.

Amanda

GordonMcComb

All electronics generate noise. The type of accelerometer you're using probably isn't instrument grade, so its noise is going to be quite a bit more than an industrial component made for the job. Even a random cosmic particle passing through it, part of the natural background radiation we all live in, will cause an instananeous change in the output.

Earthquakes have sustained movements, so your software can look for that. Or you might add a second accelerometer of the same type, mounted a few inches away but in the same orientation, and compare the instantaneous outputs of both.

erco

Balance your accelerometer on top of a tall stack of mason jars on a rickety table. When the quake hits, you will definitely hear it!

Genetix

Amanda, which sensor are you using and how are you powering it?

xanadu

What about a kalman filter? http://en.wikipedia.org/wiki/Kalman_filter

I bet even the best seismic system rely on filters. High sample rate sensor raw data is a beast.

SRLM

A low pass filter is the appropriate class of filter to use in this case. Simple averaging works pretty well if you can handle the phase lag.

Here are some plots of an LSM303DLHC accelerometer/magnetometer and a L3GD20 gyroscope covering 10 seconds of recording on a concrete sidewalk outside my house.



And if we zoom in on the acceleration:x feature:



So, you can see that this raw sensor has a noise of about +-0.25m/s^2 for ~75% of the time (eyeball).

There's the slightly bigger originals in the attached .zip (the forum has a restriction of 1024px max width).

ajward

Yup... kinda what I thought. Not like I have anything even close to being a solid foundation for my project. (A desktop on the second floor of a wood frame apartment building.)

Geology is one of my many interests and I thought it would be cool to try to catch a jiggle on one of our many faults. Tho' I suspect if I caught a tremor strong enough to be noticeable on the LCD, the LCD might be the least of my concerns. =:-|

@