混频器设计

Figure 4-2
Additive mixing in a BJT [1].
638
MIXER DESIGN来自Figure 4-3Additive mixing in a single-gate MOSFET [1].
when the R. and LO signals are applied to the same input port, as in .igures 4-2 and 4-3. Multiplicative mixing occurs when the R. and LO signals are applied to separate ports, as in .igure 4-4. As a rule, multiplicative mixers afford better isolation between their LO and R. ports than additive mixers, and this enhanced interport isolation is their principal merit. Multiplicative mixing does not in itself suppress unwanted products; the spurious response of a basic multiplicative mixer cell is poor unless it is used in a pushpull or quad configuration. Let us now consider the basic theory of mixers. Mixing is achieved by the application of two signals to a nonlinear device. Depending on the particular device, the nonlinear characteristic may differ. However, it can generally be expressed in the form I = K(V + v1 + v2)n (4-1)
4
MIXER DESIGN
4-1 INTRODUCTION Radiocommunication requires that we shift a baseband information signal to a frequency or frequencies suitable for electromagnetic propagation to the desired destination. At the destination, we reverse this process, shifting the received radiofrequency (R.) signal back to baseband to allow the recovery of the information it contains. This frequency-shifting function is traditionally known as mixing; the stages that perform it are known as mixers. Any device that exhibits amplitude-nonlinear behavior can serve as a mixer, for, as we saw in Section 1-6-2, nonlinear distortion results in the production, from the signals present at the input of a device, of signals at new frequencies. Even a rusty screw or bolt on an antenna element can act as a mixer, producing unwanted IMD products that appear at the receiver input. Although mixers are equally important in wireless transmission and reception, traditional mixer terminology favors the receiving case because mixing was first applied as such in receiving applications. Thus, the signal to be frequency shifted is applied to the mixers R. port, and the frequency-shifting power or voltage [from a local oscillator (LO)] is applied to the mixers LO port, resulting in two outputs at the mixers intermediate frequency (I.) port. If the wanted I. is lower than the R. signal, the mixer is a downconverter; if the wanted I. is higher than the R., the mixer is an upconverter. Converter may also be used as a term for a single stage that simultaneously acts as mixer and LO. .or a given R. signal, an ideal mixer with a perfect LO (i.e., an LO with no harmonics and no noise sidebands) would produce only two I. outputs: one at the frequency sum of the R. and LO, and another at the frequency difference between the R. and LO. .iltering can be used to select the desired I. product and reject the unwanted one, which is sometimes referred to as the I. image. The simultaneous generation of LO + R. and LO R. outputs results not from a departure of mixer performance from the ideal, but from the mathematics of mixing itself. Another unavoidable mixing artifact, the R. image response, also results from the mathematics of mixing rather than mixer nonideality. Just as a given R./LO combination produces two I. outputs (LO + R. and LO R., the I. and I. image), the mixer will produce output at the
R./Microwave Circuit Design for Wireless Applications. Ulrich L. Rohde, David P. Newkirk Copyright © 2000 John Wiley & Sons, Inc. ISBNs: 0-471-29818-2 (Hardback); 0-471-22413-8 (Electronic)
636
4-1
INTRODUCTION
637
Figure 4-1 Relationship between a mixer’s image and desired-signal responses. The image is 2fIF away from the desired signal.
desired I. (LO + R. or LO R.) in response to two possible R. inputs: one at LO + I. and another at LO (I. (.igure 4-1). The undesired response, the R. image (traditionally referred to merely as the image), is 2fI. removed from the desired response. Even if no humangenerated signals exist at the R. image frequency, reducing a mixers R. image response can be important because noise at that frequency, including that produced by circuitry between the mixer and antenna, will still be mixed to the desired I., degrading the signal-to-noise ratio. .iltering and phasing techniques can be used to reduce the R. or I. image responses: filtering if the image is sufficiently removed from the desired response that filtering will provide the necessary rejection; phasing if the desired and image responses are insufficiently spaced for filtering to work, as in the case of a double-conversion receiver in which signals at a high first I. (e.g., 5070 MHz) must be converted to a very low first I., such as 25 kHz. The output of every real mixer includes a vast number of additional unwanted products, including noise, the fundamentals of the mixers R. and LO signals and their harmonics, and the sums and differences of the R. and LO and their harmonics. Intermodulation distortion between multiple signals present at the R. port and I. output resulting from the mixing to I. of LO noise-sideband energy by strong adjacent signals (reciprocal mixing, Section 1-6-2) further complicate a mixers output spectrum and may compromise system performance. All mixers are multipliers in the sense that the various new outputs they produce can be described mathematically as the multiplicative products of their inputs. .rom an implementation standpoint, however, a given mixer circuit can be characterized as additive or multiplicative depending on how R. and LO signals are applied to it. Additive mixing occurs