By exploring the limits of microstrip circuits and structures, GEC Hirst Research Centre helped advance the components and assemblies needed for microwave receivers.
Microwave receiver technology has undergone dramatic improvements since the 1940s, with perhaps no organization contributing as much to these advances as the General Electric Company (GEC) Research Laboratories. Later renamed GEC Hirst Research Centre (HRC) after Hugo Hirst, one of the co-founders of the company that would become GEC, the facility gained notoriety for its work on cavity magnetrons during World War II. It also helped progress microwave receiver technology through many decades to the technology of present times.
The HRC, in its former incarnation, was founded in 1919 to pursue research independently of product manufacturing, with purpose-built laboratories at Wembley, UK opening in 1922.1 The HRC operated continuously and productively until its closing in the 1990s. A photo of the facility from the 1960s is shown in Fig. 1.
In 1938, work performed at the HRC was pivotal to aiding the war effort during World War II. One example was in applying thermionic valves (vacuum tubes) to transmitting and receiving applications in both military radar and communications systems. Efforts to develop glass magnetrons can be traced as far back as 1931, but it was in 1940based on the work of Sir John Randall and Harry Boot at Birmingham Universitythat the HRC created a practical cavity magnetron, the high-power model E1189 3-GHz device. The HRC also developed glass thermionic valves for receiver applications. The last of these was used for British 50-cm ground radar systems, before studies showed the superiority of the semiconductor crystal diode as centimetric radar became practical, and the crystal mixer, based on the "cat's whisker" in contact with a galena crystal (used in 1930s radio sets) became adopted for frequencies above about 500 MHz.
For the crystal mixer, a tungsten wire in contact with a silicon chip was the favoured rectifying element. In the early stages of development, this was mounted in a range of housings for application in coaxial and waveguide equipment, leading ultimately to the preferred format, the cartridge (Fig. 2). An in-depth review of this era's device technology can be found in ref. 2. In the 1940s, the general practice in radar equipment was to tune each crystal mixer by the use of variable tuneable mounts. Later, the fixed-tuned mount came into use, implying more repeatable control of mixer diode characteristics.
To meet receiver requirements in the UK and provide better shielding from stray radiation, a coaxial package outline eventually took the place of the cartridge. By the end of the 1940s, GEC Laboratories together with BTH (which later became AEI Semiconductor, Ltd.) became involved in developing a coaxial encapsulation capable of serving both S-band (3-GHz) and X-band (10-GHz) applications, with opposite-polarity devices required for balanced mixer designs. During the time when the UK made the transition away from the cartridge outline, the US continued its development for these frequencies.
During the 1950s, research and development at the HRC focused on moving silicon point-contact-diode technology from the 1940s to production levels for frequencies to 40 GHz, and initiating studies into germanium (Ge) point-contact diodes. The coaxial diode developed at HRC and BTH was mainly specified for use at approximately 9.5 GHz, but also had application at 3 GHz.3 The rectifying contact was formed by a tungsten wire in pressure contact with a bulk p-type Si chip (with a carrier concentration, n, of about 1018 cm-3. The junction was formed using a mechanical vibration technique generally known as "tapping," which allowed monitoring of the current-voltage (I-V) characteristic and total capacitance (at 45 MHz).
The requirements of fixed tuned mounts and balanced mixers implied a demanding RF specification, imposing tight control of rectifying junction properties and mechanical constructional/dielectric parts.3 For example, the major RF requirements included overall noise figure2,3 at a specified intermediate frequency (IF) and IF amplifier noise figure, and an RF admittance window with respect to a specified coaxial line. The overall noise figure for these devices was typically 9.0 dB for an IF amplifier with noise figure of 2 dB at 45 MHz. The RF admittance was specified as a maximum VSWR of 1.43:1 centered at 0.8 + 0.2j with respect to 1/68 mhos.
Coaxial Receiver Diodes
The coaxial diode is shown in Fig. 2, with its associated single-ended mixer, designed for a 1950s X-band marine radar front end, shown in Fig. 3. Following their development, pilot production, and type approval at HRC, these devices were transferred in the late 1950s to a GEC Product Group for full production.
These coaxial diodes were produced in the UK in quantities of about 5000 per year during periods of peak demand (with production shared between GEC and BTH). They were still being produced, although in lower quantities, until the late 1980s. The large production numbers were a reflection of burnouts caused by transmit/receive cell spike leakage (Fig. 3), and many systems required that the diode be routinely changed after a specified operating time. The mechanism that caused transmit/receive cell burnout was extremely complex, and the effect could be catastrophic. It could also occur over time at an energy level below catastrophic or be a recoverable temporary deterioration in sensitivity during the transmit pulse.
Numerous studies were conducted during the 1950s and 1960s in conjunction with the British Ministry of Defence (MOD) to improve the burnout performance, including tests that simulated the spike by a DC or coaxial discharge. Although these studies may have led to a better understanding of the mechanism, it was not until the application of solid-state devices, such as varactor-diode limiters, in conjunction with transmit/receive cells, PIN diode switch/limiter combinations, and PIN diode switches, that considerable improvements were made in receiver burnout performance.
Basic Si point-contact technology was also employed for mixer diode development to meet radar requirements at about 35 GHz (Q-band). The plug-in waveguide device structure shown in Fig. 2, which was proposed by the Telecommunications Research Establishment (TRE)4 and developed at HRC (and BTH), positioned the rectifying contact across WG22 (26.5 to 40.0 GHz) waveguide. These devices were developed to achieve type approval status at HRC before being transferred to production. They typically achieved an overall noise figure of 13 dB with an IF amplifier capable of 2-dB noise figure at 45 MHz. The use of a waveguide structure in the UK was a divergence from the choice made in the US to design coaxial units for frequencies between 12 and 40 GHz.
Mixer theory2 has predicted the need for semiconductor materials with high carrier mobility when fabricating mixer diodes in order to achieve low conversion loss, and research at the HRC was undertaken in the late 1950s into both Ge and gallium arsenide (GaAs) point-contact diodes with the goal of creating better mixers. These studies yielded improved devices for use at 9.5 GHz and at 35 GHz, using a titanium wire with a pulse-formed contact junction on a bulk n-type Ge chip with n of approximately 1018 cm-3.3
As shown in Fig. 4, the development of improving Ge diodes resulted in high-performance mixers with overall noise figures of 8 and 11 dB, respectively, at 9.5 and 35.0 GHz for an IF amplifier at 45 MHz with 2-dB noise figure. This represented approximate improvements of 1 and 3 dB, respectively, at 9.5 and 35.0 GHz, compared to mixers using silicon diodes.5
The Si point-contact activities had been largely phased out by the end of the 1950s, with its transfer to a GEC Product Company. During the 1960s, the main emphasis of HRC studies concentrated on advancing Ge point-contact device technology and researching new state-of-art features. For example, novel point-contact diode encapsulations, the Ge backward (tunnel) diode, and the initiation of studies into the GaAs Schottky barrier diode as well as into microwave integrated circuits (MICs) provided the MIC technology base for system exploitation in the 1970s.
Progress in Ge point-contact device development at HRC in the early 1960s led to the development of a miniature double-ended capsule (Fig. 2), with the objective to overcome the frequency limitations of the existing outlines and provide a versatile common package (reversible for balanced mixers) to cover all frequency bands through the millimeter-wave range,5 with application in waveguide, coaxial, and stripline equipment. As an example of the work from that time, Fig. 5 shows a compact X-band waveguide balanced mixer. The capsule's unique construction features an alumina ceramic tube insulator formed with high-temperature brazing techniques throughout, thus entirely eliminating the use of epoxy resins or soft solders. The result of this construction approach was a robust, high-temperature, hermetically sealed package. Mixers constructed in this manner supported applications to 140 GHz,6 believed to be the leading mixer performance at that time, with typical overall noise figures of 6.0 dB and 8.5 dB at 9.5 and 35.0 GHz, respectively, for an IF of 45 MHz and IF noise figure of 2 dB.
In the early 1960s, the backward diode became a subject of considerable interest at HRC for its potential use in microwave receivers. This was a modified tunnel diode in which the peak current associated with the current-voltage (I-V) characteristic is reduced. Both n-type GaAs and Ge materials were studied at HRC, with the conclusion that Ge was better suited to forming the appropriate p-n junction. The problem of producing an appropriate low-capacitance junction for microwave operation was resolved at HRC by the development of a junction-forming technique (in place of the lower-frequency tunnel diode etched MESA), which employed a gallium (p-type dopant) plated gold whisker wire, pulse bonded to the n-type Ge chip.7
Devices were developed for the standard Si-point contact outlines, which found microwave application as a zero-bias, high sensitivity detector. This detector offered -56-dBm tangential signal sensitivity (TSS) in a 1-MHz video bandwidth compared to -52 dBm TSS for a metal semiconductor device of that time with forward bias. The diode also showed properties that made it useful as a mixer, including low flicker noise (for use in Doppler radars) and low local oscillator (LO) drive (for receivers for which there was limited LO power). As a mixer, these diodes could achieve an overall noise figure of 8 dB at X-band (for a 45-MHz IF with an IF noise figure of 2 dB) with only about 100 W of LO drive power and a noise corner occurring at a 100-kHz IF. Unfortunately, although the diodes offered many features that were attractive when used as mixers, their limited upper dynamic range and poor burnout characteristics made them liabilities for many applications.
Further exploitation of point-contact devices at HRC was phased out during 1965 with the advent of the Schottky barrier diode, transferring the point-contact technology to a GEC Product Group. The R&D emphasis was redirected to planar technologies to study the potential of epitaxial n-type GaAs Schottky barrier diode technology and MICs. Although silicon was more established as an epitaxial material, GaAs was preferred for the diode studies for its higher electron mobility and because of HRC's in-house epitaxial GaAs capabilities. The Schottky barrier diode represented a breakthrough in mixer diode technology and offered numerous distinct advantages over the point-contact in both performance and fabrication.8
Early Schottky barrier diodes were aimed at 9.5 GHz, using 5-to-10-m-diameter gold or titanium barrier contacts (which were overlaid with gold to facilitate wire bonds) evaporated onto 2-m epitaxial GaAs with carrier concentration of n = 1016 cm-3. Figure 6 shows an example of the leadless inverted device (LID) ceramic carriers introduced at HRC for compatibility with planar transmission lines for development of MICs.8 Initially, the Schottky barrier diode chips were mounted into these carriers, before being adapted to standard microwave coaxial outlines for measurements and evaluation (a preferred technique to the multidot whisker contact option of the day).
During the early stages of Schottky diode development, considerable attention was given to the device as a retrofit for point-contact diodes. This allowed direct comparisons of RF performance in similar applications, including for burnout, and also facilitated immediate application of the new diodes in existing equipment. Early devices exhibited overall noise figures of about 6 dB at 9.5 GHz and about 10 dB at 94 GHz (Fig. 4), with the 94-GHz performance later improved to about 8 dB. The dynamic range of these new diodes offered a significant advantage over point-contact diodes, with approximate -20 to 0 dBm 1-dB compression points. Unfortunately, the Schottky barrier diodes showed little improvement over point-contact diodes in terms of transmit/receive cell spike burnout. As a result, HRC embarked on studies into barrier contact metallization systems that were more durable in terms of transmit/receive cell spikes. Their research led to a technology that could achieve levels of better than 0.5 e/s (compared to about 0.2 e/s for point-contact diodes9), although the technology was never transferred to production due to its complexity and cost.
Adoption of MICs was spurred by the development of phased-array radars, which depended on the availability of small, low-cost high-frequency devices in large volumes. HRC responded by exploring the use of open microstrip transmission lines (formed by metallized conductor patterns on a thin insulating material), which the laboratories felt offered the most versatile high-frequency transmission medium. Following basic studies of strip transmission linese.g., fabrication and analysis of microwave characteristics, then circuit design and active device embedding more advanced studies led through the logical stages of single-ended and balanced mixer designs for 9.5-GHz receiver applications.10
The first MIC balanced mixer was produced in 1966. It was based on a ceramic substrate from Sintox, wire-bonded chip GaAs Schottky barrier diodes, and a simple 3-dB branch arm coupler. The mixer's coaxial-to-stripline transition RF feeders were designed and fabricated by HRC. By 1967, a 9.5-GHz MIC balanced mixer had been designed with overall noise figure of about 6 dB (for an IF of 60 MHz and IF amplifier noise figure of 1.5 dB), which was marketed commercially. The circuit was fabricated on a 15 mm x 500 m high-purity alumina ceramic (or sapphire) substrate (Fig. 7), and featured two LID-mounted GaAs Schottky barrier diode single-ended mixers with quarter-wavelength diode DC returns for LO second-harmonic termination. The RF and LO signals were fed via a circular branch arm 3-dB coupler with extra quarter-wavelength line on one output port to convert the 90 coupler to a 180 coupler. The design incorporated commercial coaxial-to-microstrip transitions. The overall circuit design provided the basis for many follow-up component and subsystem developments.
In 1967, this balanced mixer circuit was incorporated into an integrated superheterodyne receiver, which was demonstrated in a short-range 10-GHz link at the 1968 Physical Society Exhibition. This was believed to be the first such demonstration of its kind in the world,11 and was a harbinger of the coming widespread acceptance of MIC technology. The integrated receiver also included, on separate ceramic substrates, a microstrip resonator varactor-tuned Gunn LO, which provided 5 mW output power and a 300-MHz electronic tuning range; a varactor-diode filter network power limiter capable of 20-dB protection at 200 W peak power (0.5-s pulses at 2000 pulses/s); and an IC wideband IF amplifier. The components were housed within a compartmentalized enclosure measuring just 8 x 3 x 2 cm.
By the following year, the receiver design was improved to minimize the effects of noise and instability of a low-Q microstrip LO and also overcome the difficulty in mechanical tuning. The microstrip LO was replaced by an external but integral pretested, high-Q coaxial cavity Gunn LO which provided 10 mW output power and 500-MHz mechanical and electronic tuning range. The original IF IC amplifier was replaced by a 45-MHz discrete-device-based circuit. These improvements yielded an overall noise figure of typically 8.5 dB for the receiver.12 This method of incorporating the LO was adopted for the majority of MIC receiver units and subassemblies that would be developed at HRC (Fig. 8).
By the end of the 1960s, all of HRC's point-contact/pulse bonded diode activities and most of their waveguide encapsulated GaAs Schottky barrier diode development activities had been phased out. In the 1970s, HRC concentrated on R&D related hybrid microstrip MICs and related activities. These activities included developing planar passive and active device technologies; thin-film and ceramic machining facilities; hermetically sealed metal boxes and plastic molded packaging; improving circuit techniques, such as mixer image reduction and termination; developing multiple-circuit designs on single substrates; production-engineering subsystems; and extending the frequency ranges of devices and circuits to the millimeter-wave range (30 to 300 GHz).
In device technologies, there were two noteworthy developments, believed to be unique to HRC and essential to the advancement of the HRC MIC technology. The first of these was the development of a ferrite disk insert device for integrated nonreciprocal isolator and circulator functions. This approach provided the means for embedding a high saturation magnetization (4pMs) ferrite disk into a substrate with a permanent magnet positioned appropriately beneath. The typical performance possible with this technology included 0.5 dB insertion lossbetter than 20 dB isolation for a 10% bandwidth.13
The second notable HRC device achievement was the GaAs Schottky barrier beam-lead mixer diode. Progressive development in a family of mixer diode devices, based on the structure of Fig. 9, resulted in a glass insert technique for minimizing all associated parasitics, which also provided a rugged structure. The final glass moat design, with a cutoff frequency (defined as 1/2pRsCjo) of approximately 2500 GHz, and conversion losses of 6.5 and 7.5 dB at 94 and 140 GHz, respectively, is still currently in production.14,15 Beam-leaded backward-type diodes were also developed at HRC,14 in which the p-type dopant was aluminium, evaporated with the contact area photomechanically defined and the junction then formed by localized heating. The performance of these backward diodes was comparable to that of the pulse-bonded devices; these backward diodes would later be incorporated into a range of broadband military video detector subsystems.
In terms of MIC packaging, the favored type of packaging was a single-cavity, machined metal box. This form of packaging, however, could form a significant part of the overall cost for a component/subassembly. During the 1970s, HRC undertook studies into the feasibility of plastic encapsulation using transfer molding techniques, with the aim of reduced packaging cost and weight. The results indicated the possibilities of the technique, as demonstrated by the fabrication of balanced mixers at 17 GHz, which provided comparable performance to the conventional metal box approach, with an overall noise figure of about 7 dB for an IF of 60 MHz and IF noise figure of 1.5 dB.16 The studies, however, did not advance to more complex MIC structures, as a break-even cost occurred above several hundred units (i.e., in excess of the volume requirements of that time). Also, the technique implied a throw-away policy, when many applications required packaging which was accessible for circuit repair. Ultimately, HRC developed several metal packaging epoxies and solder-sealed techniques, the final choice being dependent upon application. For example, military packaging requirements were solder-sealed packaging back filled with a helium/argon atmosphere for leak testing.
Receiver noise figure was of prime importance in early systems and, although the theoretical concept of receiver overall-noise-figure enhancementby the recovery of signal power normally lost to the mixer image terminationhad been known since the 1940s,2 it was not until the arrival of MICs that the concept became practical. The early 1970s thus saw the revival of considerable interest in this low-noise technique by a circuit in which two mixers were appropriately phase coupled to improve receiver noise figure, and HRC had a leading involvement in these studies.17 Such studies included both branch arm and compact quad balanced mixers (the latter implying development of quad diode mixers).18 With advances in three-terminal device RF amplification, much of the work was ultimately applied to image-reject mixersi.e., for single-sideband (SSB) receivers.
Development of multicircuit technologies embraced many types of subsystemse.g., broadband video detector units (incorporating beam-lead backward diodes), frequency upconversion and downconversion units (including SSB receivers), built-in-test-equipment (BITE) delay-line modules, low-noise front-ends, transmit/receive units, etc. In addition to the MIC technology, these subsystems implied incorporation of associated low-frequency circuitry, such as IF and video amplifiers, control devices/circuits (with PIN and varactor diodes), and BITE facilities.
A 16.5-GHz (Ku-band) compact low-weight transmit/receive unit for hand-held radar applications provides a good example of how multicircuit subsystems were applied19: the unit incorporated signal and automatic-frequency-control (AFC) channels driven by a common LO, with overload protection and IF and AFC head amplifiers incorporated in the package housing. In 1971, the first experimental unit used six separate but linked sapphire substrates (15 mm x 15 mm x 250 m) in a package approximately 6 x 8 x 3 cm and weighing about 150 g. Microstrip circuit functions included beam-lead-diode signal and AFC rat race coupler balanced mixers, a ferrite disc insert circulator, a varactor-diode filter network power limiter, and low-frequency printed-circuit-board (PCB) circuitry; the pretested mechanical and electronically tuneable coaxial cavity Gunn LO was connected externally.
In 1972, the unit was redeveloped and later engineered to military environment specifications (Fig. 10); it was believed to be one of the most advanced MICs of its time.20 The circuit functions were produced on a single alumina substrate (45 mm x 30 mm x 500 m thick), the power limiter was a pre-encapsulated PIN device (PIN diodes with bias derived from SBD), the low-frequency circuitry employed PCBs mounted with discrete components, and the pretested mechanically and electronically tuneable Gunn LO (the latter by means of a varactor diode coupled into the resonator), was incorporated by a separate integral coaxial cavity within the overall housing (referred to as "partially integrated"). Essential performance included an overall noise figure of 10 dB, tunable frequency range of 700 MHz, peak power of 100 W (at a duty ratio of 0.001) at the aerial input, and nominal transmit capability of 3 kW peak power (at a duty ratio of 0.001).
In the early 1970s, there was significant system-level interest in extending MIC technology through millimeter-wave frequencies. When the early study phase into microstrip technology showed that it could be applied to frequencies to at least 100 GHz, the roles of the GaAs beam-lead diode and ferrite insert technologies became even more vital. These high-frequency studies followed a similar pattern to those pursued at lower frequencies, in that they examined microstrip characteristics as well as single-ended and balanced mixer designs.
The significant differences at millimeter-wave frequencies included the use of lower-permittivity substrates, such as fused silica (a rigid substrate being preferred to a plastic circuit-board material) and later Z-cut quartz (with improved thermal expansion properties for ferrite insert technology and ground-plane soldering), and incorporation of waveguide-to-microstrip transitions. For example, such transitions would include tapered then later multiple stepped ridge designs; they also included combining a two-step ridge within a box wall, with the option of hermetically sealing the waveguide aperture with a low-loss glass window; and the E-plane probe in which the circuit ground plane is removed where the microstrip line protrudes into the waveguide,21, 22 but later also providing the option of forming an hermetic sealed waveguide window.25 Early four-step and probe transitions are illustrated in Fig. 11; the latter would also be incorporated in a 35-GHz receiver demonstrator that provided an example of some of HRC's millimeter-wave design and fabrication capabilities.
By 1973, two sub-assemblies were initially produced which demonstrated the feasibility of using microwave technology at millimeter-wave frequencies: a 35-GHz multiple-circuit integrated front-end (Fig. 12) and a 35-GHz superheterodyne receiver. The first receiver front end incorporating signal and AFC channels (mixers and head amplifiers) and LO isocirculator with built-in stepped ridge waveguide-to-microstrip transition feeders. The early development included four interconnected fused silica substrates (10 mm x 10 mm x 250 m) with one additional ferrite substrate for the isocirculator. Later on, the unit included all circuits on a single Z-cut quartz substrate (35 x 25 mm x 250 m) with a ferrite insert circulator. The 35-GHz superheterodyne receiver incorporating a bolt-on pretested waveguide cavity, second-harmonic Gunn LO, separate linked quartz substrates containing a ferrite insert isocirculator and a balanced mixer, both with probe transitions (Fig. 11), and a PCB IF amplifier.
Following 1975, the frequency range of three-terminal device RF amplifiers was extended to millimeter-wave frequencies and attention was given to SSB mixer circuit techniques based on microwave frequency studies. Figure 13 shows an early 35-GHz SSB receiver circuit. This was rapidly followed by the development of SSB receiver units incorporating appropriate microstrip RF circuitry, ridge waveguide-to-microstrip transitions, and 90 IF output combiner with thick-film IF amplifiers in the same package.
By 1977, component developments based on lower-frequency designs and technologies had progressed to 100 GHz. A 75-GHz (W-band) balanced mixer was produced and marketed by HRC in 1978; it was believed to be the first commercial W-band integrated product (Fig. 14). By 1978, 94-GHz mixers were developed, and together with advanced ferrite insert technology, the basis was being applied to multi-circuit subsystems with both ridge and probe microstrip transitions.
By 1979, single-substrate multicircuit technology had been established and military environmental standards were becoming an essential build criterion with the development of integrated subassemblies engineered with production in mind. Such units included thick-film, low-frequency circuitry and were hermetically sealed, incorporating a pinch-off tube to provide means for back filling with an argon/helium mixture for leak testing. These multicircuit designs included SSB receivers for both 3- and 10-GHz applications.
The circuits were designed for image suppression following RF amplification, but were essentially image-recovery mixer circuits.23 The 3-GHz unit (Fig. 15) incorporated two compact quad diode mixers and exhibited an overall noise figure of about 5.0 dB. The 10-GHz unit (Fig 16 showing both top and base views) used two branch arm mixers and exhibited an overall noise figure of about 6.0 dB (compared with about 8.0 dB for a previous unit designed for image suppression only).
Another example of the trend in integration at HRC was embodied by a 16.5-GHz duplexer-receiver subsystem designed for hand-held radar systems (Fig. 17). Based on the design described earlier, a higher-power system developed for about 50 W peak power was constructed with additional circuitry for transmit injection locking and performance monitoring. It included five ferrite insert nonreciprocal devices, a 50-W PIN limiter and signal/AFC mixers. The initial development used three substrates in separate compartments for appropriate isolation of the receiver from the high transmit power. Ultimately, however, it was produced on a single alumina substrate with novel power screening techniques.24
In another integrated subsystem, operating at 94 GHz, a SSB modulator serves as a frequency upconverter, in which both upper and lower sidebands were accessed separately and used as LO feeds for two balanced mixers (Fig. 18). Microstrip circuits on Z-cut quartz substrate (26 x 24 x 0.125 mm) included E-plane probe waveguide-to-microstrip transitions, which formed hermetically sealed waveguide windows when the substrate ground plane was soldered to the box package. Surface modes were suppressed by the deposit of a lossy material on the substrate.25
By the end of the 1970s, HRC had demonstrated the foundations of an MIC business. These skills and capabilities were transferred to a GEC-Marconi Group for further business development, together with the majority of the hybrid MIC technology. HRC concentrated on a major program on monolithic microwave integrated circuits (MMICs).
Following technology transfer, MIC development continued at GEC-Marconi Product Groups based on the HRC technology, with the development of many advanced components and subsystems for commercial exploitation. Much of the emphasis was on millimeter-wave frequencies to meet increasing system demands for miniature integrated systems at these frequencies. This may best be demonstrated by a 94-GHz FM-CW transceiver contained within a volume of about 1 cubic inch26 and a 94-GHz dual-channel radar receiver/duplexer with a high level of integration. This 94-GHz design incorporated some 15 circuit functions on a single 18 x 18 x 0.12 mm quartz substrate, with externally connected oscillator functions.27
A major MMIC program had been initiated at HRC in the late 1970s, resulting later in the formation of GEC Monolithics operating within GEC Research, Ltd. As part of this program, studies were undertaken on diode-based mixer circuits. In the early 1980s, the diode was preferred to active three-terminal semiconductor devices for mixer applications due to the high flicker noise exhibited by the three-terminal devices. Later, as the result of process compatibility with other MMIC components, the resistive three-terminal device became the preferred option for mixers. Several diode balanced mixer circuits were demonstrated in the 1980s, in general based on the MIC design of two Schottky diodes with LO and signal inputs combined by a 3-dB coupler. These designs included the 8-to-12-GHz circuit (with IF of 10 to 500 MHz) of Fig. 19, which featured two-finger (2 x 20 m) interdigital geometry diodes and a 3-dB Lange coupler on a 3 mm x 3 mm x 200 m chip. This mixer circuit achieved a conversion loss of 6 dB.28 Another design was an 18-to-40-GHz mixer circuit with IF of DC to 5 GHz formed of a diode configuration of three 2 x 30 m air-bridged fingers each contacting five 1.0-m-diameter Schottky contacts and a 3-dB Lange coupler with Schiffman phase shifter.
Finally, in Fig. 20, the 94-GHz circuit incorporates a rat race balanced mixer and LO and signal waveguide coupling E-plane probes on a chip of 4 x 1 mm, with the diodes formed by a single air-bridged finger (1 x 5 m); the chip conversion loss was typically 7.5 dB.28
In conclusion, this historical report only briefly outlines some aspects of research and development that the GEC Research Laboratorieslater the GEC Hirst Research Centreplayed in its internationally recognized role towards the advancement of mixer diodes and associated circuit technologies, for the enhancement of microwave receivers. Apologies are offered in advance for the many unavoidable omissions from this historical review.
Acknowledgments
Sincere acknowledgements are owed to The GEC Hirst Research Centre, particularly to all colleagues who contributed to the technology advances briefly outlined here; the support and encouragement given by many MOD establishments, particularly the role of the Common Valve Development (CVD), later to be named the Directorate of Components Valves and Devices (DCVD), then the Electronic Component Group (ECG); and the support and encouragement given by GEC Product Groups.
References