Posts: 23

Threads: 6

Joined: Sep 2023

Hi Jens,

I was wondering whether you're considering to add aButterworth filtet as an analysis module choosing the order, the cutoff frequency and th type (low/high). Such a filter is commonly used to soomth acc/gyro signals in movment analysis.

Cheers,

Pietro

Posts: 497

Threads: 22

Joined: Apr 2020

Institution: RWTH Aachen University

We have put it in on our list as it appears to make sense to have it as an independent module – we are not so sure about the effort yet that we have to spent (research the numerics and its stability).

Would technically FFT, multiply Butterworth characteristic, invFFT work? Then you could already play around with it…

Posts: 23

Threads: 6

Joined: Sep 2023

Hi Jens, thanks for replying. So you mean to use fft to get the frequency-domain representation of the signal, and then set the power of unwanted frequencies to zero, followed by ifft to recover the filtered data in the time domain? I'm afraid the approach you suggested would introduce latency and can create undesirable ringing. But I'll try to investigate it. Thanks, Pietro

Posts: 497

Threads: 22

Joined: Apr 2020

Institution: RWTH Aachen University

We thought of multiplying the (smooth) Butterworth weight function frequency-wise with the FFT. We are not yet sure if there are unwanted discretisation effects though…

Posts: 23

Threads: 6

Joined: Sep 2023

ok Jens let me know about your progress!

Thanks, Pietro

Posts: 23

Threads: 6

Joined: Sep 2023

06-07-2024, 08:59 PM
(This post was last modified: 06-07-2024, 09:00 PM by pietropicerno.)
Hi Jens, I hope you're still around. So: I followed your suggestion and I used FFT for removing high frequency noise and then I used invFFT for reconstructing the signal back. I did the implementation in Matlab and it worked as I succeded in obtaining something very similar to the result of a Butterworth filter.

Now the problem with Phyphox is that I do not have a "invFFT" module. Are you planning to release such a module for the editor? Is there any experiment I could refer to?

Thanks, Pietro

Posts: 497

Threads: 22

Joined: Apr 2020

Institution: RWTH Aachen University

FFT and invFFT are almost identical, see

https://en.wikipedia.org/wiki/Fast_Fourier_transform, so you could you apply FFT again with minor adjustments.

Posts: 23

Threads: 6

Joined: Sep 2023

ok, got it. I'll try. Thanks Jens!