Analysis and Design of Highly Linear Capacitive Stacking Mixer-First Receivers
Stef van Zanten is a PhD student in the department Integrated Circuit Design. (Co)Promotors are prof.dr.ir. B. Nauta and dr.ir. R.A.R. van der Zee from the Faculty of Electrical Engineering, Mathematics and Computer Science, University of Twente.
A radio receiver ideally combines flexibility to serve various wireless standards with high linearity to handle interference. Although acoustic wave filters are widely used to add selectivity and thus relax the linearity demands on the receiver, their inflexibility obstructs further CMOS integration. This thesis presents a highly flexible and highly linear, wideband capacitive stacking mixer-first receiver (MF-RX) architecture that does not rely on acoustic wave filters for higher order filtering at the mixer baseband (BB) node.
Now, the traditional capacitive stacking MF-RX designs combine voltage gain through capacitor stacking to relax noise demands with good linearity due to their passive nature. However, they suffer from a limited maximum radio frequency (RF) range of operation due to parasitic loading at the RF input, as the RF input capacitors define the bandwidth (BW), constraining their value and typically resulting in a large parasitic capacitance.
This thesis covers an alternative low-loss capacitive stacking N-path filter/mixer (CSNPFM) architecture in which the capacitors at BB define the BW, allowing for smaller RF capacitors, extending the RF input range. A full characterization of this architecture through intuitive explanations, simulations, and analyses reveals major benefits over the traditional design aside from the extended RF range, such as a 4x smaller total capacitor area and elimination of the switch resistance selectivity bottleneck.
However, this characterization also reveals that the low-loss CSNPFM exhibits a stronger BB response when excited with interferers around even harmonics of the RF fundamental. This thesis shows that this is due to a second system response in the low-loss architecture that not only limits rejection around even harmonics, but also complicates its analysis.
To address this, a novel analysis methodology based on the adjoint network is presented. Using a discrete-time state-space model and eigendecomposition, the effects of the individual system responses can be analysed in isolation, increasing circuit insight. The resulting closed-form expression for the transfer function enables derivation of simple design equations for use of the circuit in the mixing region.
Measurements on a 22 nm FDSOI CMOS prototype show support for a decade of RF input range, up to 10 GHz. With excellent far out-of-band linearity across the entire RF range and a reasonable NF while consuming only very limited dynamic power, the low-loss CSNPFM shows great promise in serving as a flexible wideband receiver building block.
To further boost linearity, this thesis presents a state-of-the-art BB network that enhances the CSNPFM’s filtering slope, covering its design procedure and circuit implementation. This BB network uses a higher order capacitive feedback loop to realize near-third-order closed-loop filtering at the mixer BB node. Measurements on a 22 nm FDSOI CMOS prototype show that the loop indeed improves the filtering slope at this node to -15 decibel per octave, boosting near out-of-band linearity without having to rely on the use of inflexible acoustic wave filters.
