Introduction

In the previous chapter, we saw that a simple passive mixer driven by 50% duty-cycle clocks converts a low-Q baseband impedance to a high-Q bandpass impedance through frequency translation. The center of this high-Q bandpass impedance is controlled precisely by the clock frequency, making it very attractive for reconfigurable receivers in which it is desirable to have high-Q bandpass filters with centers that can be tuned over a wide range of frequencies. Being implemented with just switches and capacitors, the resulting high-Q filter is exceptionally linear, and because the switches carry no DC, there is no major flicker noise issue. We saw, however, that this filter has the problem of image folding and cannot be useful in its current format. In this chapter, this filter evolves to a new high-Q bandpass filter that no longer folds the image. The resulting filter is still controlled by the clock frequency and is still composed of only switches and capacitors [23, 53, 54, 57–62]. The only complication of this evolved filter is the need for four arms of switches in series with baseband impedances plus a more complicated clocking scheme. The four-phase high-Q BPF requires four nonoverlapped 25% duty-cycle clocks that are progressively phase-shifted by 90°.

The four-phase filter offers more flexibility in the choice of four baseband impedances. For example, if all four baseband impedances are replaced with a single complex baseband impedance, the switching system would frequency-shift the complex baseband impedance to the LO frequency, resulting in a high-Q bandpass filter having a center that is offset from the LO clock by an amount that is dictated by the complex filter.