Self-healing Narrowband MEMS Filter Design for RF Receivers
As the radio frequency (RF) spectrum is becoming increasingly occupied, receivers are designed to resist out-of-band interferers. Due to lack of on-chip high quality factor (high-Q) filters, conventional receivers pursue frequency translation using mixers and VCOs, which inevitably increases power consumption and area and degrades noise and linearity. The piezoelectric AlN contour mode resonator filter, with its high-Q, low insertion loss and compatibiliy with CMOS integration, has emerged as a viable option for a high performance channel selection filter as the first stage of a direct sampling receiver. However, maintaining frequency precision for any high Q filter is challenging in the presence of both random and systematic process variations. To address this, the extended statistical element selection (ESES) method is applied to compensate for the frequency variations using a combinatorial redundancy scheme. This research explores the feasibility, advantages, implementation, and testing of a heterogeneously integrated switchable AlN filter system that can minimize parasitics and “self-heal” to increase yield and to realize uniform loading. In addition, several design techniques are proposed to implement the selfhealing filter to a 1 GHz receiver frontend, including breaking LNAs into sub-LNAs to select sub-filters and using current summing to uniformly combine signal outputs. To control parasitics, 3-D integration is used to limit the interconnection lengths between the CMOS switches and the AlN resonators.