The MICROMIXER: a highly linear variant of the Gilbert mixer using a bisymmetric Class-AB input stage

Abstract

This paper outlines the basic theory of a development of the Gilbert mixer. The bipolar junction transistor (BJT) differential pair widely used as the RF input stage is replaced by a bisymmetric Class-AB topology based on translinear principles. It does not have inherent gain compression, affording a greatly extended signal capacity. The linearity of variants of the basic form is excellent, providing two-tone intermodulation intercepts as high as +30 dBm, without the expenditure of high bias currents. It can operate on supplies as low as 2.2 V, with a power consumption of under 5 mW. The input impedance of this mixer is accurately controllable (typically 50) and provides a true broadband match. The noise figure depends on design details and is generally not as low as in mixers specifically optimized for noise performance, although acceptable for many receiver applications. Inductively degenerated variants can be tuned to a narrowband match at microwave frequencies and provide fullmixing SSB noise figures as low as 6.5 dB. Practical realizations are in use in applications to 1.9 GHz.

A 85-mW Multistandard Multiband CMOS Mobile TV Tuner for DVB-H/T, T-DMB, and ISDB-T Applications with FM Reception

“…2, the SDT adopts a bisymmetric class-AB input stage that converts a single-ended input signal into a differential output signal and increases the P1dB performance [8]. The SDT is suitable for wideband input matching as well as a flat gain response over a wide operating frequency due to the common gate input stage.…”

“…The following RF PGA and quadrature down-conversion mixer are shared in all bands to reduce silicon area. To cover the FM and VHF bands, a common source (CS) cascode LNA with 3-bit controlled gate-tosource additional capacitance is used [8]. A shunt resistive feedback LNA is used for the UHF band.…”

Complementary High-Speed SiGe and CMOS Buffers

“…A problem with this implementation is the reported -5 dB loss performance per branch, despite the 0.3-µm GaAs MESFET technology used. Nevertheless, this balun form has been known for decades, and it has also been used as a class-AB mixer input stage by Gilbert (1997), and in noise-canceling LNAs originally proposed by …”

Preamplifier effect on the performance of distributed active mixer

“…In contrast, distributed mixers have uniformity of gain due to the widespread use of artificial transmission lines [6], [7]. In order to improve mixer linearization, the flow injection method and class AB are presented in [8] and [9], respectively.…”