An exploration of active hard mount vibration isolation for precision equipment

Abstract

Vibrations due to environmental disturbances can cause a loss of accuracy in high-precision equipment. To reduce the vibration levels, the equipment is commonly mounted on soft mounts. However, the soft mount's low support stiffness may introduce difficulties in the response to direct disturbances and with the equipment levelling. In this thesis, an alternative vibration isolation concept is investigated, which uses stiff supports (hard mounts). The increased support stiffness circumvents the mentioned difficulties of soft mounts. Additional objectives for this vibration isolation concept are improvement of the damping in suspension and structural modes and achievement of floor vibration isolation performance as achieved by state-of-the-art soft mounts. The required additional damping is achieved by feedback control. Absolute motion sensors are shown to be preferred over force sensors and displacement sensors for this purpose. To reduce the transmission of floor vibrations further, feedforward compensation of measured floor vibrations is applied. The (adaptive) FxLMS algorithm (with several extensions), which is widely used in Active Noise Control applications, is used to find the optimal feedforward compensation. The standard Finite Impulse Response parametrization requires a large number of parameters to model the optimal feedforward controller with sufficient accuracy. Alternatively, an Infinite Impulse Response parametrization with fixed poles is proposed. Then, a similar performance can be achieved with less parameters, resulting in less demanding computational requirements. However, the performance depends critically on the choice of the fixed poles in the IIR filter. Experimental tests indicate that the vibration levels can be reduced by a factor of 20, to typical levels of 1 mm/s2 RMS, while maintaining a support stiffness 300 times larger than typical soft mounts. The residual vibration level is not yet comparable to typical soft mount vibration levels, as the performance is primarily limited by the sensor noise level and time delay in the signal conditioning.