THEORY Mixer, as it names suggests, “mixes” or multiplies two signals to get a resultant output signal. A time domain multiplication is a convolution in frequency domain so, in the transceiver, mixers perform frequency translation or frequency up-conversion and down-conversion by multiplying an RF input signal, with frequency wRF, and an LO signal, with frequency wLO, present at mixers the RF and the LO ports, respectively. Ideally at the mixer IF output two distinct signal frequencies are present, one at wRF – wLO and the other at wRF + wLO as shown in Figure 3-1. The difference term is classified as the down-converted IF frequency and the summation term is considered the up-converted frequency. In transmitter, the mixer is used for up-conversion and in the receiver, the mixer is used for down-conversion. The mixer, aside from performing frequency multiplication or translation, can provide amplification or attenuation. A passive mixer generally attenuates signal amplitude when translating RF signals to IF whereas an active mixer is generally used for signal amplification or to provide gain. The selection of mixers are application based and depends on other mixer performance parameters such as noise figure, linearity, and power consumption. Those are discussed in detailed later in this Chapter. In physical realization mixing and frequency multiplication are nonlinear operations by nature. Generally in DCR for maintaining high input dynamic range the active mixer is used to supplies noise. Although the nonlinear characteristic is essential to mixers, their signal handling capacity is limited due to their nonlinearity at large signal amplitudes, which in turn generates undesirable odd and even harmonic distortion products and other fractional intermodulation distortion that can reduce the SNR of a receiver. Furthermore, in radio receivers, in which nearby channels are also occupied, the nonlinearity can lead to the aliasing of the power from the nearby channels to the receiver passband. The RF receivers and their building blocks require large input dynamic range and therefore mixer must be sufficiently linear. Generally passive mixers have good large signal handling capability while for active mixers achieving high linearity is always a challenge. Mixer Linearity Linear circuits are defined as those for which the superposition principle holds, i.e., if input x1(t) and x2(t) are applied separately to a circuit having responses y1(t) and y2(t), respectively, the response to the excitation ax1(t) + bx2(t) is given by expression ay1(t) + by2(t), where a and b are arbitrary constants By this definition as long as the system response is a linear combination of the input, the system is linear. However, all electronic circuits are nonlinear, some more weakly nonlinear in their actual ranges of operation than others. The noise of the mixer is generally measured by the noise factor and/or noise figure. The noise factor is the ratio of output SNR to input SNR, while, the noise figure is simply the noise factor in logarithmic scale. For direct down conversion receiver architectures the mixer noise is measured by double-sideband (DSB) noise figure as opposed to single-sideband (SSB) noise figure in image reject heterodyne receiver architectures. In theory, the noise power of the down-converted signal is the same for both SSB and DSB but in the DSB case, the signal power is double when compared to single sideband in heterodyne, and therefore, the SNR and the noise figure is better for a DSB system compared to a SSB system by 3 dB. Generally mixers have quite poor noise performance, usually the noise figures are over 10 dB. This is due to a combination of different noise sources that includes signal loss in the switching, noise contributed by harmonics and other distortion existing in the mixer output, supply noise, and device noise. The analysis of the mixer noise is not very trivial as in the case of LNAs, due to its time varying characteristics and other non-integer harmonic conversion products. However, in terms of mixer design for a DCR, one should focus on device noise, such as 1/f or flicker noise, thermal noise, common mode noise and supply noises. The flicker noise or 1/f noise is a major concern in CMOS mixer design since flicker noise are low frequency noise (generally <1 MHz) and exist in the desired baseband IF frequency band. While the specific cause of flicker noise are not completely understood the general consensus is that it is caused by carrier generation and recombination due to surface traps in the silicon and silicon oxide interface. The flicker noise is therefore a function of surface area and is inversely proportional to FET geometry (W, L) and frequency (f) MIXER design papers free download
A Low-Power SiGe-HBT Active Double-Balanced Mixer for Compact Integrated C-Band Wireless-LAN Receivers, Corrado Carta, Rolf Vogt, Werner Baechtold, ETZH.
RF Mixer Design, L.M.Devlin – Paper presented at the The IEE Training Event – How to Design RF Circuits , 5th April2000.
The MICROMIXER: A Highly Linear Variant of the Gilbert Mixer Using a Bisymmetric Class-AB Input Stage, Barrie Gilbert, Fellow, research, research JOURNAL OF SOLID-STATE CIRCUITS, VOL. 32, NO. 9, SEPTEMBER 1997 page 1412
A high Dynamic Range 3.4 GHz CMOS micropower mixer, Vancorenland P., K.U.Leuven, proceedings ESSCIRC 2000
A Self-Calibrating 900-MHz CMOS Image-Reject Receiver, Razavi B., University of California, proceedings ESSCIRC 2000
Using Capactive Cross-Coupling Technique in RF Low Noise Amplifiers and Down-Conversion Mixer Design, Zhuo W., Texas A&M University, , proceedings ESSCIRC 2000
A 1-GHz CMOS Up-Conversion Mixer, Peter R. Kinget and Michiel S. J. Steyaert, research JOURNAL OF SOLID-STATE CIRCUITS, VOL. 32, NO. 3, MARCH 1997
A direct-conversion BiCMOS mixer, Tiiliharju E., Helsinki University of Technology, presentation slides ESSCIRC 2001, September 2001
An Image-Reject Down converter with Sideband Selection for Double-Conversion Receiver, Stadius K., Helsinki University of Technology, presentation slides ESSCIRC 2001, September 2001
Low Voltage Performance of a Microwave CMOS Gilbert Cell Mixer,”P. J. Sullivan, B. A. Xavier * and W. H. Ku, University of California, San Diego, CA, 92093-0408, USA, * Pacific Communication Sciences Incorporated,San Diego, CA, 92121, USA
Active Doubly Balanced Mixers for CMOS RFICs,” P. J. Sullivan, B. A. Xavier, and W. H. Ku June 6, 1997
A 40 –45 GHz Monolithic Gilbert Cell Mixer, Andrew Dearn and Liam Devlin – Paper presented at the IEE Colloquium on MM-Wave Circuits and Technology for Commercial Applications , held at Savoy Place, London on 24th March 1999.
A Shunt Feedback Technique for Improving the Dynamic Range of a Balanced Mixer, Bernard A. Xavier, Pacific Communication Sciences Inc.