Prof.ir. H.M.J.R. Soemers,
Prof.dr.ir. J.B. Jonker,
Dr.ir. J. van Dijk,
Ir. G.W. van der Poel (contact person)
April 2005 – March 2009
mechatronic design principles, control, adaptive, feedforward, feedback, identification
For over 30 years, Moore's law has been the driving force for the semiconductor industry, resulting in a continuing demand for higher accuracy and increased throughput. Similar trends toward higher resolution exist in related industries, e.g. nanotechnology and electron optics. However, the finite structural stiffness − although very high − results in structural resonances, which are often poorly damped. These vibrational modes can be excited by floor-transmitted vibrations or direct disturbances − e.g. cables, air flow or internal acceleration forces − resulting in a reduced accuracy. Hence, vibration isolation systems become increasingly important for high-precision equipment. Currently, most vibration isolation systems aim at reducing the effects of floor vibrations, by using so-called soft mounts (e.g. pneumatic isolators).
Figure 1: Vibration isolation concept:
An actively supported machine, subjected to direct disturbances and floor vibrations. The active mount is controlled by feedback (FB) and feedforward control (FF).
In this project, a vibration isolation concept is being developed which offers adequate vibration isolation from both direct disturbances and floor vibrations. The machine mounts are designed for high mechanical stiffness (typically 100−200x higher than soft mounts), resulting in a low compliance to direct disturbances. The transmissibility of floor vibrations is actively reduced, using sensors, actuators and a digital control system. The concept is illustrated in figure 1.
As shown in figure 1, the control strategy combines feedback control (FB) and feedforward control (FF). The feedback control is used to add artificial damping to the suspension modes and relevant structural resonances of the machine. A lag filter based on acceleration measurements is used for this purpose.
The feedforward controller is then used to further reduce the transmissibility of floor vibrations. The feedforward controller is a Finite Impulse Response filter whose coefficients are updated using a preconditioned Filtered Reference Least Mean Squares (FxLMS) adaptation algorithm. Due to its adaptive nature, the controller can cope (to a certain extent) with time-varying inputs, non-linearities and modelling errors.
Figure 2 shows a photograph of a laboratory setup with a single active vibration isolation channel. The other directions of motions are constrained by leaf springs. Figure 3 shows measured frequency responses of the machine when the floor is excited by a pseudo-random force. A reduction of approximately 20 dB is achieved with feedback and 30 dB when using feedforward control combined with feedback. It has been shown that the isolation performance in this setup is limited by actuator saturation, time delay and the limited number of feedforward parameters (limited by available calculation time).
Figure 2: Photograph of the single channel laboratory setup
Figure 3: Acceleration frequency responses to a pseudo-random shaker excitation
CURRENT RESEARCH ACTIVITIES
Further improvement of the adaptive control algorithms & review of absolute motion sensors
This research project is sponsored by the Innovation–oriented Research Program (IOP) Precision Technology, which is carried out by SenterNovem by order of the Dutch Ministry of Economic Affairs