signal integration

[kungfucrazy]

I can't seem to figure out how to continuiously integrate an accelerometer signal twice to view the displacement. Ive used all of the different integration VI's and i can't seem to get the result i need. i don't want to see a waveform, i just the value of the displacement. Any one help? been at this a long time now.

[LocalDSP]

A true (perfect) integration is probably not what you want because even the slightest dc-offset in your signal will make it drift and most likely corrupt your measurement. Double-integration just makes it even worse.

The best approach really depends on your signal. If it is, for example, a pure (or at least relatively clean) sine tone, you may want to perform your integration in the frequency domain. Otherwise you'll have to choose a "cut-off" frequency under which your integration flattens out or become differentiation (highpass filtering or ac-coupling).

Finally, when you say you just want the "value of the displacement", what is it more precisely? The peak value? The rms value?

Could you post a VI with your signal (for example default values on a graph)
to be integrated and some few comments on what you want to measure?

[kungfucrazy]

hi there. i guess i wasn't very clear, i apologize about that. i am going to be using a shaker to vibrate a fixture to achieve very small diplacements. the signal out of the accelerometer is going to be a sign wave of some sort, i want to be able to view the displacement after integrating the signal twice. This will allow me to adjust the settings on the shaker to the desired displacement. I want to continuosly see the display on an indicator in labview. I don't have a VI of the signal becasue im not exactly sure what the signal will look like out of the accelerometer. we haven't ran the test yet. I suppose the displacement that i will need is the peak to peak value. hope this helps a bit. thanks.

[Tom]

kungfucrazy wrote:

> hi there. i guess i wasn't very clear, i apologize about that. i am
> going to be using a shaker to vibrate a fixture to achieve very small
> diplacements. the signal out of the accelerometer is going to be a
> sign wave of some sort, i want to be able to view the displacement
> after integrating the signal twice. This will allow me to adjust the
> settings on the shaker to the desired displacement. I want to
> continuosly see the display on an indicator in labview. I don't have
> a VI of the signal becasue im not exactly sure what the signal will
> look like out of the accelerometer. we haven't ran the test yet. I
> suppose the displacement that i will need is the peak to peak value.
> hope this helps a bit. thanks.

You nee
d

loop:
y1=y0
y0 = y1+k*u
end

where k is a constant, y0 is the output and u the input to the
integrator. Any slight dc offset will send it soaring high into
saturation however. Pure integrators are normally only used when there is
feedback around say a closed loop system. To implement the above you need
a while loop and a register to get y1. Feed the output y0 into the
register and get y1 from the past value. u is the current input. If you
want this in vector (array) format it is a little more complicated as the
register does not do the trick there. For the array case you have to be a
little smarter.

regards

Tom

[LocalDSP]

You mention that your acceleration signal is a sine tone of some sort. Assuming it is a sine tone, the integration(s) needed to get velocity and displacement are easy to do in the frequency domain if you know the exact frequency of your signal and the amplitude of your acceleration. There is a VI in LabVIEW that can extract that information even from a noisy and dc-offset sine tone signal.

I have created a sub-VI that computes velocity and displacement amplitude and phase from your acceleration signal. See attached llb.

I hope this helps.