Unexpected overshoot in integrated value.

[shrekt]

Hi Guys,

I'm taking input through sound card of the pc, giving sine wave as input from signal (function) generator.

then i am plotting the trend of the intergrated value of the sine wave, for some time the trend is consistent in terms of

value but around after 1 min or so some unexpected overshoot comes, and its amplitude increases as i plot it for some

more time. and the peak repeat itself at regural interval but with increased amplitude.

Can someone suggest way to avoid it.

the vi is attached below,

thanks in advanced,

Regards

Shrek

[johnsold]

Shrek,

It would be helpful if you saved some data showing what you are seeing.  Run your VI.  When the unexpected data is displayed, stop the program, go to the Edit menu and Make Current Values Default.  Then save the VI with the data and post that.

I suspect that your data has an offset or DC component.  When integrated, the DC component becomes a ramp added to the integrated sine.

Lynn 

[shrekt]

Hi johnsold,

Sorry for the late reply,

I have attached the problem as below,

As  you can see in the image the normal value is 0.0079 its not increasing but the peaks are increasing

This is my problem, i dont want that peaks to come,even if they exist their amplitude should be less and not increase.

I think DC is not there as the value is steady at 0.0079 only the peaks exists.

Also you can see the table in the VI how the peaks are coming.

any solution for the problem.

Thanks

Regards,

Shrek

[johnsold]

Shrek,

You do have a non-zero mean in the data.  When that is removed, the integral no longer runs away.

I put Diagram Disable blocks around the portions of the program which are hardware related so I could run your data.

Lynn 

[shrekt]

Hi Lynn,

Thanks a lot , The problemwas solved quite a bit,

The peaks have disappeared, even if they come the amplitude is very small.

Now  I'm testing for all the frequency  and amplitude range, will reply after that if any problem comes.

I have been stuck up since days in it, Its a bit of relief to me

Thanks again.

Regards,

Shrek

[shrekt]

Hi Lynn,

I have tried all the possible combination for the above problem, its working fine for me.

But now i also want to do double integration of the same input signal, but the same peaks problem is there,

so can you suggest any solution of double integration  of signal.

Thanks

Regards,

Shrek

[johnsold]

Shrek,

Your first integration results in positive peaks above (an approximately) zero baseline.  A waveform like that will have a DC offset.  Have you tried removing the mean again between integrations?

Lynn 

[shrekt]

Hi,

Ya I tried to take mean after first integration but some error is coming, the error is the frequency Fc should be greater than Fs. I'm trying to resolve the issue.

I think the samples are not read that fast to integration block, what are your comments?

Regards,

Shrek

[Ujjval Shah]

Hi Shrek,

Can you please attach the vi with double integration that you are working on?

I tried replicating your problem by generating a simple single channel sine waveform of 1k with sample rate of 22k instead of taking from sound card, but i didnt get any error or peaks.

I dont think it has anything to do with reading speed of the integration block, in fact can you please explain what you mean by that?

Regards,

Ujjval

[shrekt]

Hi,

Sorry to reply so late,

The samples problems is solved. Actually by mistake the size of the bufer was very small.

The update is that, After single integration it is working perfectly but after double integration

i tried to remove mean then the ouput was better, but was not as expected.

The peaks in double integration now also haunt me, can suggest another method.

Regards,

Shrek