Phase noise is the curse of many a communications system, and one of the key performance parameters that most oscillators strive to minimize. System designers have long recognized the yttrium-iron-garnet (YIG) oscillator for its broadband, low-noise, high-frequency capabilities. But YIG sources are also power hungry and physically large. The mechanical design, with a YIG sphere mounted in the air gap of an electromagnet, is not conducive to integrated-circuit (IC) integration. And the sphere placement and tolerances required do not lend to high-volume production. The size of the sphere and the windings used for the electromagnetic field become prohibitive at resonance frequencies below about 2 GHz. In addition, YIG oscillators are sensitive to thermal drift, vibration, lighting, electromagnetic interference (EMI), microphonics, phase hits, and frequency modulation, all of which have a detrimental effect in designing modern communication systems. Fortunately, there is now an alternative source for tunable high-frequency signals with low phase noise: the distributed coupled YIG-replacement oscillator or DCYR series from Synergy Microwave Corp. (Paterson, NJ). The new patent-pending voltage-controlled oscillators (VCOs) are currently available for frequencies from 250 to 6000 MHz with typical measured phase noise of -132 dBc/Hz offset 100 kHz from carriers 250 to 1000 MHz.

In spite of their low phase-noise levels, it is a YIG oscillator's sensitivity to drift, vibration, and other negative factors that can diminish a microwave radio's bit-error-rate (BER) performance. ^{1- 10} YIG oscillators are also limited in tuning speed in the range of millisecond frequency switching speed. For point-to-point and point-to-multi-point radio designers, the microsecond switching speed of a VCO supports frequency-agile solutions, although the phase noise of source agile sources has traditionally been limited. With the new DCYR sources, the phase-noise performance is comparable to the best YIG sources over wideband tuning ranges, but at a fraction of the size and power consumption, and with considerably faster tuning speed than YIG oscillators.

YIG oscillators require a significant amount of power (typically 200 mA at 12 V or more) in order to power the heater that stabilizes the internal temperature of the YIG sphere and its supporting electronics. Dissipating this heat often becomes a problem, especially when protecting the surrounding circuitry of a microwave radio. Even with the integrated heater, the operating temperature range of most YIG oscillators is limited to about 0 to +60°C and not suitable for all wireless systems. The design of one YIG oscillator is not readily scalable to a nonstandard package or frequency range, so custom YIG sources generally require additional non-recurring-engineering (NRE) costs along with the costs of the products.

Another limitation of the YIG oscillator is the nature of its tuning, by means of an applied magnetic field. Instantaneous changes to that field are difficult to control, and creating the structure to generate such a field is not readily compatible with low-cost monolithic-microwave-integrated-circuit (MMIC) fabrication techniques. A YIG oscillator is essentially a YIG sphere with high unloaded quality factor (Q) set in a resonant cavity. The resonant frequency is tuned according to the magnetic field applied by a main coil. A second magnetic field, applied from a second coil, provides modulation when necessary. And a heating element is often added to stabilize the frequency of the YIG sphere and resonant structure over wide temperature ranges. The traditional design of a YIG sphere and resonant cavity becomes significantly larger as wavelengths become larger, typically at frequencies below 2 GHz. Few advances have been made in this basic proportionality rule, resulting in few commercial YIG sources for frequency ranges below 2 GHz.

Due to demand for low-phase-noise sources, and knowing the limitations of YIG oscillators, the design engineers of Synergy Microwave developed a novel patent-pending oscillator topology based on multi-coupled-slow-wave (MCSW) planar resonators. The approach supports multi-octave tuning in a small package, and is compatible with IC fabrication processes. ^{1- 3} The novel topology allows for a substantial reduction in phase noise by dynamically optimizing impedance transfer function and coupling factor across the guided distributed medium of the planar multicoupled network. ^{2- 9}

An MCSW VCO is planar and broadband in nature, therefore well suited for cost-effective, monolithic-microwave-integrated-circuit (MMIC) fabrication. With the potential to enable wide operational bandwidths, eliminate discrete resonators (such as a YIG sphere), and produce high-quality-factor (high-Q) planar resonators for low-noise VCOs by means of a planar fabrication process compatible with existing IC and MMIC processes, the MCSW VCO is a promising technology for present and future broadband communications requirements. The MCSW, for example, is well suited for use in microwave communications systems, test equipment, radar, local multipoint-distribution systems (LMDS), and multichannel multipoint-distribution systems (MMDS).

Figure 1 shows a block diagram of a DCYR series MCSW VCO. The DCYR series is currently available at frequencies from 250 to 6000 MHz (various tuning ranges); all models are capable of providing stable performance over wide operating temperature ranges of -40 to +85°C. The tiny low-noise oscillators are housed in surface-mount packages measuring either 0.5 0.5 0.16 in. or 0.75 0.75 0.16 in. (Fig. 2). These miniature sources are a fraction of the size of the smallest YIG oscillators, which are housed in metal cubes measuring either 1 in. or 1.25 in. on a side.

As depicted in Fig.1, the active impedance created by the three-terminal active device (a field-effect transistor or bipolar transistor) in a MCSW oscillator has a negative real part with a real magnitude and an imaginary part with an imaginary magnitude. The real magnitude is a function of the imaginary magnitude. The imaginary magnitude is selected such that the real magnitude compensates for the loss of the MCSW resonator. The selection of the imaginary magnitude should also coincide with the maximum-slope inflection point of the oscillator's phase characteristic curve, in order to optimize group-delay performance. The mode coupling approach also includes a methodology for optimum dynamic coupling. Optimum coupling enhances the dynamic loaded Q, reduces or eliminates phase hits, diminishes susceptibility to microphonics (to an extremely low level), and minimizes phase noise while achieving a broadband linear tuning range. ^{4- 9}

The multi-mode coupled resonator network shown in Fig. 1 is capacitively coupled across the base and collector terminal of the three-terminal (bipolar or FET) active device. This arrangement can be characterized as a high-Q multiplier based on evanescent-mode progressive delay that eventually improves the time average loaded Q of the planar resonator over its multioctave operating band. A slow-wave and a progressive-wave-coupled resonator (coupled through hybrid resonance mode convergence effect) connected through phase-compensating network across collector and base of the three-terminal (bipolar or FET) active device, which supports self-injection locking mechanism over multioctave-band. In addition to this, a phase compensating network (capacitively coupled between the base terminal and the slow-wave and progressive-wave coupled resonators) also optimizes group delay dynamically for uniform and minimum phase-noise performance over the band. An RF output signal is coupled through a distributed coupled medium, which is coupled across the slowwave and progressive-wave resonator networks, therefore, uniform output power and improved higher-order harmonic rejection through out the operating frequency band. ^8-12