Position from accelerometer

[GeoPar]

Hello, I am reading the values from an accelerometer in m/s^2 and I would like to extract the displacement (in mm) . Should I double integrate my values in order to find the position at a certain time or taking into account that I have a vibration system(which does oscillations) can I just use a=ω^2*x where a=the acceleration aplitude,ω=frequency(2*pi*f) and x=position? 

[Bob_Schor]

If you are measuring acceleration and want to get displacement, you should integrate twice.  If you know your signal is a pure sinusoid of one frequency, without any other frequency components (including "noise"), then you could, in principle, integrate twice.  If you did this correctly, you'd get the correct formula a = - ω^2*x (signs are important!).

Bob Schor

[mcduff]

Integrations are very susceptible to offsets and low frequencies. Two methods:

1. Remove the mean from your signal and use a low pass filter before integrating; you may have to do this for both integrations.
2. FFT the data. Divide each point by (2 pi f)^2, where f is the frequency of each point of the FFT. Now you should have the displacement in frequency space. If you need to go back to the time domain you will probably need to remove the low frequencies as stated earlier.

mcduff 

[Coryman]

Also, you will need to put in a correction factor.  You are going from m/sec^2 to mm, so divide your answer by 1000.

[GeoPar]

Thank you all for your useful opinions, 

I will try the above and I will provide you with a reply.