I suspended the roughing pump with four springs. The reduction of the 58 Hz peak is similar to what I got when the pump was sitting on a box. So most of the coupling is due to acousting noise.