Is it possible to design a comb (notch) filter without removing frequencies near 0?

Context: I am trying to analyze some data from a system that was vibrating a lot while the data was being acquired. A bandstop filter and a moving average filter were both helpful but did not completely eliminate the noise on their own. If this is not possible, maybe you can suggest a different approach for filtering periodic noise.

[Star Strider]

Much depends on the filter design you are using, and how you choose the passbands and stopbands. Chebyshev and elliptic filters may offer the best band-rejection with the lowest loss on adjacent frequencies.

Data, and a plot of the time-domain and frequency-domain representations of your signals, would be helpful.

Using the filtfilt function (rather than filter) is preferred in almost all situations.

[Owen Cail]

I've attached pictures from MATLAB's signal analyzer app showing the signals in time domain and frequency domain. One image is of the entirety of my data, which demonstrates the peaks at different frequencies. The other image is zoomed in on the last trial to give a better idea of what the signal looks like in time domain.

I will look into designing a Chebyshev filter like you suggested. I am still curious, you did not mention if a comb filter that does not attenuate 0Hz is possible or not.

[Star Strider]

‘... comb filter that does not attenuate 0Hz is possible ...’

It would have to have very steep (short) passband-stopband transitions to work, to not include the 0 Hz d-c component. Chebyshev designs are usually preferable in this region. They are a challenge to design, regardless.