A passive 6db/octave hi-pass filter, which could consist of a single low value/high quality capacitor in series with the amplifier input might be quite transparent. However, the cut-off frequency would have to be too high up into the audio range (guessing around 50Hz) to be effective at the very low frequencies for "woofer pumping". To implement an effective rumble filter that cuts in only at sub-audio frequencies, e.g., below 20Hz, I think you need a slope of at least 12db/octave or higher. Then, if you do that, you are looking at insertion loss, several components needed to get the steeper slope, etc. To avoid insertion loss, you might need an active filter. All of the above leads to loss of transparency and fidelity that can affect the entire audio range. So in this instance my bias is the same as Raul's. Less is more. But this is just "in principle"; I am not about to say that someone else's system cannot sound better with a well designed rumble filter vs without it.
Right now I am using a pair of Transmission Line woofers that I built several decades ago as bass support for a pair of Beveridge 2SW ESLs. (The 2SWs were designed to operate from 100Hz up.) Long ago, when I built the TL cabinets and used them subsequently, woofer pumping was always an issue because the woofer in a TL is essentially undamped by the cabinet. I am rather surprised that I see zero evidence of unwanted woofer motion, and I wonder whether the Dynavector tonearm I am using is more resistant to the problem than most.
