Читать книгу Handbook of Microwave Component Measurements - Joel P. Dunsmore - Страница 71

Another major class of components is mixers and frequency converters. Mixers convert an RF signal to an IF frequency (aka down‐converters) or IF frequencies to RF (up‐converters) through the use of a third signal, known as the local oscillator (LO). Typically, the output or input that is lower in frequency is called the IF (for intermediate frequency). The LO provided to the mixer drives some non‐linear aspect of the circuit, typically diodes or transistors that are switched on and off at the LO rate. Using the second definition, frequency converters and mixers would not be considered linear. In fact, in their normal operation, they are linear (under the first definition) with respect the desired information signal, and ideally the frequency conversion does not change the linearity of the input/output transfer function.

The input is transferred to the output through this time‐varying conduction, and the output signal includes the sum and difference of the input and the LO. Mixers tend to be fundamental building blocks that are combined with filters and amplifiers and sometimes other mixers to form frequency converters. In practice, frequency converters are typically filtered at the input to prevent unwanted signals from mixing with the LO and creating a signal in the output band, and they are typically filtered at the output to eliminate one of the plus or minus products of the mixing process. Some converters, known as image reject mixers, have circuitry that suppresses the unwanted product, sometimes called the sideband, without filtering. These are typically created by two mixers driven with RF and LO signals phased such that the output signals of the desired sideband are in phase and added together to produce a higher output, and the undesired sideband is out of phase; thus they cancel and produce a smaller output. Monolithic microwave integrated circuits (MMICs) sometimes blur the line between converters and mixers as they may contain several amplifier and mixing stages, but they typically don't contain filtering.

Mixers have fundamental parameters that include conversion loss (or gain in the case of MMICs with amplifiers), isolation (of which there are 12 varieties, to be discussed in Chapter 7), compression level, noise figure (which for passive mixers is typically just the conversion loss), input and output match, and most of the other parameters found for amplifiers.

Mixers are sometimes referred to as passive mixers or active mixers. Passive mixers don't contain amplification circuits, are typically constructed of diodes (with a ring or star configuration being most common), and have baluns in some paths to provide improved isolation and reduced higher‐order products. Higher‐order mixing products, also called spurious mixing products or spurs, refer to signals other than the simple sum and difference of the input and LO frequencies. They are often referred to by the order of harmonic related to creating them; for example, a 2:1 spur is created at the frequency of two times the LO plus and minus one times the input. The level of the mixer spurious products, which are sometimes called mixer intermod products (even though the two tones are not applied to the input), change with respect to the RF drive signal at the rate of the order of the RF portion of the spur. For example, a 2:1 spur will increase 1 dB in power for each dB in power the RF signal increases. However, since the LO is creating the non‐linearity in the mixer, as well as the spurious signal, it is difficult to predict how the spurious power will change with respect to LO power. In many cases, driving the LO power higher produces a higher spurious signal, as the relative magnitude of the RF signal to the LO power is reduced; in other cases, the transfer impedance of the non‐linear element becomes more consistent across the RF drive level. The spurious higher‐order products of a mixer are sometimes defined as a spur table, which shows the dBc values of higher‐order products relative to the desired output. Chapter 7 has more details on measuring these behaviors.

Mixers with baluns can suppress some of the higher‐order products, with baluns on the RF port suppressing products that have even‐order LO spurious, and baluns on the LO port suppressing products that have even‐order RF spurious. These are called single‐balance mixers, and it is typical that the LO port is balanced. Double‐balanced mixers have baluns on the RF or IF ports. Refer to Chapter 7 for more details on mixer configurations. Triple‐balanced mixers are usually comprised of a pair of double‐balanced mixers, adding a balun to the IF port. Their main advantage is to divide the RF signal power between the two diode quads, lowering the RF relative to the LO, whereby the spurious signals created will be lower; then the outputs are combined to recover the power. This provides a mixer with the same conversion loss and lower spurious products at the same output power level. The disadvantage is that since the LO drive is also divided, higher LO drive power is needed to achieve the same linearity for each diode quad.

The creation of spurious mixer products is a key aspect of system design, with the goal of eliminating spurious signals from the IF output. Unfortunately, some frequency plans are such that the spurious products must fall into the band of interest. In such cases, system designers move to multiple conversion stages to create a first stage, which produces an output free of spurious signals over the range of input signals of interest, and then has a second conversion stage that produces the frequency of interest at the output. This multiple conversion or “dual‐LO” system is typically called a frequency converter and often contains additional filtering and amplification, as shown in Figure 1.41.

Figure 1.41 Dual‐LO frequency converter.

Because of the multiple components used, the frequency response of converters often has gain ripple and phase ripple, which can distort the information signal. Key figures of merit for converters are gain flatness, group delay flatness, and the related phase flatness, which is also known as deviation from linear phase, and represents residual ripple after fitting the phase data to a straight line. Modern systems employ equalization techniques that can remove some of the flatness effects provided they follow simple curvatures; as such, another specification found on converters is deviation from parabolic phase, which is the residual ripple in the phase data when it is fitted to a second order curve.

Mixers often have quite poor input or output match because of the switching nature of their operation, so their effect on system flatness when assembled into a converter can be quite dramatic. Until recently, it was difficult to predict the effects of output load of a mixer on its input match, but the mathematical tools to model mixers as system elements have been developed (Williams et al. 2005) that describe these relationships. Mixers that produce an output that is the sum of the input and output signals are relatively simple to describe, but mixers that produce an output that is the difference between input and LO have a more complicated behavior in the case where the input frequency is less the LO frequency. These are sometimes called image mixers, and their unusual characteristic is that as the input frequency goes up, the output frequency goes down; this also applies to phase: a negative phase shift of the input signal results in a positive phase shift of the output signal. How these special cases affect the system performance will be described in more precise mathematical terms in Chapter 7.