Bessel versus Butterworth filter

Giuseppe, Your considerations respect to overshoot and attenuation for the two filters are correct. In my opinion, for an accelerometer filter you must consider the posibility of overshoots that will distort the shape of the waveform (or the pulse). A Butterworth filter, because the phase shift response is not linear (not proportional to the frequency), will produce different time delays when trying to filter pulses in particular. This, as you know, will produce overshoots in the output. (If you are trying to filter signals where all frequencies are at the same gain (not pulses), then the Butterweorth is excellent). A Bessel filter is linear in its phase response (phase shift is linearly proportional to the frequency aat every point). This fact produces no overshoot, and the output signal will not have any distortion at all. The Bessel, as you indicate, has a "tough" behavior (to say the least) close to the critical frequency. The Bessel gives you less attenuation at fc than a similar Butterworth filter (with the same number of poles). You can, however, you can design your Bessel filter with a little higher cut-off frequcny (not too much, though). In the case of accelerometers, I consider overshoot avoidance more important than attenuation close to fc. Hope this second opinion will strengthen your ideas. Abe

Abe, Thank you for your answer. It is indeed very helpful. About the possibility of designing the Bessel filter with a little higher cut-off frequency, such a move would "make up" for the less attenuation close to the original fc. But the risk is to set the new cut frequency a bit "too" high and introduce undesired attenuation on the signal. I believe this is what you meant by saying "not too much, though". On the bottom of your answer, you consider overshoot avoidance versus attenuation in the case of accelerometers. May I ask you what type of professional experience (design, applications, etc.) you gathered with accelerometers? Again, thank you for all your comments and suggestions. Giuseppe

Giuseppe, I am an electrical engineer (more in the electronics side) that also teaches at a local Community College. I teach Electronics (Digital and Analog) and (among others) a course in Electromechanical Systems where we give special emphasis to the design of signal conditioning systems for sensors and data acquisition. Eventhough I can assure you that I am not an expert in accelerometers, I do teach in this course a chapter in general accelerometers theory and signal conditioning.