Should I use 1-D digital filtering or lowpass filter ?

Hi All

which of the two is better for random signals ( i have 6 signals in 6 degrees of freedom : should I filter them together with the same values of filter ?)

and what does this mean :

lowpass uses a minimum-order filter with a stopband attenuation of 60 dB

It's also not clear how should I calculate wpass for a given random signal !!

with normalized passband frequency wpass in units of π rad/sample

what should I read from my signal to calculate the wpass?

Answer 1

[Star Strider]

Yes. Also be certain that the signals have uniform sampling intervals. If they do not, use the resample function to force them to be regularly-sampled.

The lowpass function uses the shortest (i.e. most efficient) filter that will give the appropriate results, and the stopband is 60 dB lower than the passband, or 1/1000 the passband amplitude.

EDIT — (6 Apr 2020 at 20:49)

The ‘wpass’ argument is the passband frequency in units of radians/sample, with π being defined as equivalent to the Nyquist frequency (one-half the sampling frequency).

If you use the ‘fs’ argument as well, you can instead designate ‘fpass’ in the same frequency units as ‘fs’ (for example, Hz). The normalisation will be done automatically.

[farzad]

dear Star ! first thank you so much ! great answer already.

So : If I have my signal, sampled with 512 (or 1024) hz and my range of interest be 256 Hz, a 2 column of data, first time and second amplitude. what operations should I do and how do I calculate the wpass and fs and fpass for this signal ?

[Star Strider]

As always, my pleasure!

If your sampling frequency is 512 Hz, and your ADC has an appropriate (Bessel) anti-aliasing filter, you need to do nothing, since the Nyquist frequency is 512/2 or 256 Hz.

If the sampling frequency is 1024 Hz, you need to design your filter accordingly, with the passband frequency of 256 Hz. The lowpass function will do everything else that it needs to do to create an optimal filter, and filter the signal.

So if ‘s’ is your signal, you want to pass everything below 256 Hz, and the sampling frequency is 1024 Hz:

s_filtered = lowpass(s, 256, 1024);

will do what you want.

Get the second output of the lowpass function to use the designed filter with other signals, if necessary. See the documentation for those details.

[farzad]

thank you very much, in this case of 1024 Hz, what will be wpass ? how do I calculate it ?

[Star Strider]

As always, my pleasure!

Please see my previous Comment. I provided code for exactly that for a sampling frequency of 1024 Hz and a passband of 0 to 256 Hz.