Radio Frequency and Microwave Design and Development
These are a few areas in radio frequency (RF) and microwave engineering that demonstrate FCL's capability and the work it has been engaged with.
FCL has supported clients with many RF and microwave specialisations on projects ranging from circuit level (PCB) design through to major fixed and mobile deployable systems.
FCL has designed and developed several PCB-based circuits with architectures down to discrete components and ICs. The circuit functions have included: RF, digital and mixed signal, including field programmable gate array (FPGA) interfacing. The IC functions have included: attenuators, phase shifters, synthesizers, IQ modulators, Ethernet interfaces, amplifiers, diplexers, switches and a range of sampling ADCs and DACs.
Common client requirements for hardware include de-bugging, enhanced performance, smaller size, adding testability and reducing production costs. For example, there might be a particular issue, such as a noise, spurious or electromagnetic compatibility (EMC) problem, that requires a robust, fully verified and documented solution to support a product upgrade. FCL has successfully solved many challenges such as these.
FCL's experience also extends to synchronous optical transmission over single mode (monomode) optical fiber (λ = 1310 nm and λ=1550 nm), multimode optical fiber (λ = 850 nm) and RF over optical fiber (ROF). The single mode optical fibers were applicable to high capacity telecommunications infrastructure carrier services, usually with wavelength division multiplexing (WDM) or dense WDM (DWDM). Examples of these form parts of synchronous optical networks (SONET) or synchronous digital hierarchy (SDH) networks. SONET and SDH employ indirect optical modulation because, for each wavelength, the laser source has appreciable optical output power and is therefore a separate discrete unit from the Mach Zehnder intensity modulator.
RF over optical fiber (ROF) uses the same principle of intensity modulation to modulate optical carriers but using a direct modulation approach with cheaper and simpler components aimed at carrying smaller bandwidths. In this, effectively, the laser power source itself is modulated by a band of RF which carries the information to be transmitted. A SONET or SDH indirect modulator would normally be modulated by a digital baseband with a bandwidth of about twice the signalling data rate used for the data stream. So, for example, if the data signalling rate was OC192 (SONET) or STM64 (SDH) that would correspond to 9.953280 Gbit/s and require a baseband bandwidth of about 20 GHz. A typical ROF system, in contrast to a broadband telecommunications service, might only require a small fraction of this bandwidth. A significant advantage of ROF is that the optical modulating source normally comes directly from the frequency band that would traditionally have been amplified and sent to an antenna. So just 3 GHz bandwidth would be sufficient for a vast range of communications services. In summary, ROF services are much cheaper and simpler than their indirectly modulated counterparts. ROF is one method used to extend digital cellular capacity and coverage and terrestrial broadcast coverage, for example inside tunnels and large buildings.
FCL has also provided support to a project using multi-mode optical fiber connections. These were at λ=850 nm, just above the optical wavelength range at lower frequencies moving into the infrared band. Multimode have a much shorter range and smaller capacity than monomode but they are more robust with cheaper with less precisely toleranced parts, especially connectors.
FCL also has experience of the common optical components used in optical fiber telecommunications. Examples are laser diode sources, photodiodes, WDM filters (add/drop multiplexers), optical connectors and the fiber optic cables themselves. Experience is offered with dense wavelength division multiplex (DWDM) filters, circulators, attenuators, Mach Zehnder optical intensity modulators and erbium doped fiber amplifiers (EDFAs).
Some clients have engaged FCL to develop equipment-based units and sub-units. Generally these have included bought-in, commercial-off-the-shelf (COTS) connectorised components, the inter-connections being provided either directly using suitable adaptors or with short lengths of semi-rigid or other low VSWR coaxial cable.
At system level, FCL has been a key member of a client's team responsible for the implementation of a major communications subsystem, part of a large defence ministry procurement using frequencies from HF to microwaves. Other systems projects have been with point to point line-of-sight (LOS) microwave digital communications and satellite communications (ground and space based equipment).
One client was a terrestrial broadcast service provider with whom FCL's job was to visit antenna sites and deal with both planned and unplanned maintenance issues to improve the quality of services to all customers.
FCL has extensive experience with specifying, using and interfacing a wide range of both active and passive components.
The active components have encompassed amplifiers, synthesised signal generators, oscillators, VCOs and signal conversion ICs (ADCs and DACs). The active (discrete) devices have included silicon bipolar, field effect transistors (FETs), junction FETs (JFETs), metal semiconductor FETs (MESFETs), metal oxide semiconductor FETs (MOSFETs) and pseudomorphic high electron mobility transistors (PHEMTs). Several amplifiers have been developed: low noise (MESFET and PHEMT), class AB power amplifiers and general purpose examples. VHF and UHF oscillators have initially been developed as free running versions based on the Clapp and Colpitts architectures and then converted to voltage control (VCOs) by adding abrupt and hyper-abrupt varactor tuning diodes with suitable biassing.
The passive components encountered include: antennas, couplers, splitters, filters, diplexers, duplexers, circulators, modulators and transmission lines of various types. FCL provides excellent support for transmission line theory and many of the different propagating wave types including transverse electric-magnetic (TEM), quasi-TEM and non-TEM modes (TE and TM in conductor and dielectric waveguides). Examples include: coaxial cables, microstrip, co-planar waveguide, co-planar strips, stripline, conductor waveguide (rectangular and elliptical) and dielectric waveguides (optical fibre). One project was dedicated to 'beam' waveguide. This achieved particularly low loss transmission at millimetre wave frequencies above 30 GHz using Gaussian-Laguerre beams utilising dielectric lenses and ellipsoidally profiled reflectors.
FCL has a good resources in antenna and propagation theory and many of the common antenna-related parameters such as gain, directivity, beamwidth, sidelobes, efficiency, polarisation, near-field and far-field properties. It has been engaged to predict analog and digital radio channel propagation performance across many different paths: terrestrial (flat earth and multipath-rich), free space, microwave LOS, ionospheric and non-line-of-sight (NLOS). It is also understands the general propagation mechanisms that typically exist across a range of frequencies.
FCL's early work with antennas and terrestrial propagation was across the VHF and UHF bands. It worked with directional element antennas across both spectra and more recently with multi-element passive and active phased arrays. Power levels were modest with the earlier hardware, moving on later to analogue TV broadcast antennas rated at up to 40 kW (peak vision carrier). Propagation predictions have been made based on the criteria from Bullington's papers to account for free-space path loss, ground reflections, diffraction, refraction and atmospheric absorption. This was followed by more prediction work covering the high frequency (HF) frequency range, using the Earth's ionospheric layers, for a backup defence communications system, operating in both point-to-point and point-to-multipoint modes. This required familiarity with HF channel prediction charts related to the frequency band, propagation path, season and time of day.
Much of FCL's work has been with microwave (passive reflector) and active (phased array) antennas. The passive examples included cassegrain, Gregorian and parabolic offset feeds: designed, built and tested, operating at Ka-band. These used corrugated (scalar) horns, conical 'Potter' horns and quasi-optic (reflector and lens) beam waveguides to provide the feeds. The far field radiation tests gave excellent results for gain, directivity, cross polar isolation and sidelobe performance. FCL has also specified passive microwave antennas for ground-satellite links and line-of-sight (LOS) digital terrestrial microwave links.
More recent work has been directed at the many situations where radio wave propagation is by 'non-line-of-sight' (NLOS) or 'beyond-line-of-sight' (BLOS), such as those operating under the WiFi (IEEE 802.11), WiMAX (IEEE 802.16) or low rate wireless personal area networks (WPANs) to IEEE 802.15.4. These newer technologies are specifically designed to mitigate against multipath propagation which is an inherent result of transmissions close to objects which cause reflections such as buildings, bridges, tunnels and walls. Examples include those using the rake receiver in the 3G universal mobile telecommunications service (UMTS) and the orthogonal frequency division multiplexing (OFDM) techniques used in 4G, long term evolution (LTE), its advanced version (LTE-A) and digital terrestrial TV broadcasting (DVB). One report delivered to a client used algorithms developed around Rician and Rayleigh fading channel models in a heavy multipath environment, based on one of the WiMAX standards and potentially suitable for a new mobile communications system.
FCL is very well equipped to support clients' requirements for RF, microwave and millimetre wave design and development. FCL also has a good grasp of electromagnetic field theory, if necessary returning to Maxwell's equations with the associated mathematical derivations. Several projects have operated at frequencies extending into microwaves: C, X, Ku and Ka-bands, and into the millimetre wave bands approaching 100 GHz. The work at the millimetre wave frequencies was on passive components including horn antennas, harmonic mixers, dielectric antennas and beam waveguide feeds. One project was to design devices for performing lens surface phase and amplitude measurements at about 30 GHz, the output being fed to a Hewlett Packard HP 8510® vector network analyzer (VNA) for analysis. The VNA included the inverse fast Fourier transform capability to convert from the frequency to the time domain, thereby supporting analyses of reflections at specific discontinuities along transmission lines. The work also included investigations into the transmission of fundamental (Gaussian) and higher order (Gaussian-Laguerre) beam waveguide modes for two dimensional monopulse tracking information. FCL has a good appreciation of the challenges encountered at these frequencies. The properties addressed included: conductor and dielectric reflections, refraction, higher order waveguide modes, skin depth limitations, absorption (loss tangent) and diffraction.
FCL has experience with several RF low noise amplifier types (Si bipolar, GaAs MESFET and PHEMT with input noise matching), power amplifiers (class B and AB push pull) and general purpose class A types for buffering and matching applications. The frequencies covered were mostly VHF/UHF but also extending into the microwave spectrum around Ku-band. FCL is familiar with many of the combining techniques used widely for solid state power amplifiers and low noise amplifiers to improve linearity, simplify redundancy provision and suppress harmonics. FCL has designed these products to requirements for stability, isolation, spurious, gain-frequency, gain-temperature, return loss-frequency, P1dB, IP3, NF, IMDs and AM-PM conversion.
FCL has developed double balanced mixers including characterisation and optimisation of the internal ferrite transformers for local oscillator isolation, conversion loss and noise figure. This work has spun off into the development of 'traditional' and transmission line transformers, using high frequency ferrite formers, for use in combiners and splitters as well as mixers. At microwave frequencies on microstrip FCL has developed balanced push-pull mixers using branch line couplers and packaged Schottky diodes, single ended mixers and active mixers .
Early work performed included the development of PIN diode switches on alumina microstrip operating at Ku band for a space hardware product. Later assignments included similar configurations of PIN diode switches but used in series and shunt configurations at VHF and requiring high bias voltages to achieve the necessary isolations.
FCL has designed, built and tested many lumped element LC filters based on filter tables (Williams and Taylor) for operation in the HF and VHF spectra. These included lowpass, bandpass (narrow and wide) and an absorptive notch. The applications were roofing filters, IF passband, transmit rejection, image rejection and discrete spurious rejection. The absorptive notch was used to improve the useable dynamic range of a spectrum analyzer by absorbing an unwanted high level carrier nearby in frequency to those being measured. FCL has also recommended UHF filters for procurement including waveguide, ceramic resonator and surface acoustic wave (SAW) types.
Much of FCL's experience has been with oscillators of various types: fixed ovenised high stability crystals, free running Colpitts and Clapp architectures with provisioning of frequency control using hyperabrupt varactor diodes (VCOs). These have been developed from component level into complete synthesisers designed around dedicated synthesizer ICs whilst overcoming the challenges of jitter-dominated phase noise, lock time, frequency pulling, spurious, stability and control interfacing. Most of the circuit level design and development work has been at VHF, with higher frequencies tending to be implemented in the more modular forms.
Many digital receivers and transmitters today have some form of quadrature modulation or demodulation process included, referred to as in-phase/in-quadrature (IQ) mixers or modulators. A recent project was based around an IQ modulator whose baseband I and Q components were supplied by a dual output DAC. It was possible to control the digital input to the DAC to marginally deviate the analog output signals from perfect quadrature, and thereby allow some control over levels of image (unwanted) signals output from the modulator.
One client's requirement was to investigate the input matching of a low noise amplifier (LNA) designed around a high electron mobility transistor (HEMT) for optimum noise figure. LNAs, normally placed at the front end of a receiver system, will significantly influence the signal to noise ratio further along the cascade and therefore the bit error rate (BER) after the demodulator. Furthermore, the physics of the HEMT means that its optimum noise performance occurs at a particular input match which, in general, differs from the match required for maximum power. Unfortunately though, the addition of matching components at the front end also incurs a small loss, dependent on their quality, which adds to the overall noise figure of the chain, so the low noise matching does not always provide a net improvement. FCL was able to improve the quality factor (Q) of the matching components and revise their values to achieve an improvement in noise figure compared to previously.
Software defined radio (SDR) has become more and more popular over the last 20 years or so in line with the availability and price reductions of high performance, fast sampling ADCs. As these have become more affordable, the analog/digital interface in receivers has gradually moved closer to the antenna and the receivable frequencies have increased. DSP processing speeds have also increased to allow delivery of near-real time data output even after significant digital domain management such as downconversion, filtering and demodulation. This 'DSP heavy' approach has allowed appropriate software code to be developed and simulated offline in an integrated development enviroinment (IDE) appropriate to the processor used, typically using languages such as C, C++, Python® or Matlab®. Upgrades and re-configurations, for example to use a new type of demodulation, may be developed offline and then simulated, tested, compiled and the binary output used for re-programming. Also, applications such as Matlab® have developed model and code (script) based SDR support packages for many popular and low cost SDR platforms. An important advantage of SDR is that, once the hardware is set up, it may be repeatedly re-programmed for upgrades, changes of demodulation, filtering etc before it becomes obsolete. The hardware and associated software therefore have a much longer life.
FCL recently purchased a quantity of the Analog Devices AD6645®, a 14 bit, 105 MSPS ADC, from an online auction site. This ADC uses AD's 'pipeline technology' architecture and was available very cheaply as it had been marked as 'becoming obsolete' and not recommended for new designs. A check of the history of the device confirmed that it had been a successful part for several years and simply had to make way for a faster, and no doubt more expensive replacement. It is proving to be very effective for developing an SDR receiver to demonstrate the principles involved. So far, the IC has been mounted and powered and RF transformers have been fitted. The next tasks will be to interface the (parallel) data output to a memory to store the data possibly using a fast interface like DDR. The data will of course also need to be read from the memory and analysed to verify it is as expected, possibly again with Matlab®. The initial sampling speed might be quite slow but hopefully it will verify that the correct techniques have been used. FCL has some experience of using Microchip® microcontrollers for the DSP functions but will take this opportunity to consider an ARM® microcontroller device, initially an entry level one such as the Cortex M0® or M0+®.
An alternative to the AD6645, but high level approach is also being investigated. This uses Simulink® which is a model-based feature of Matlab®. Simulink® allows the interconnection of model symbols on a Simulink® schematic diagram to build devices for simulation. The Simulink® symbols are available from a library and have been designed to represent the typical blocks, analog, digital and mixed, which would be required to construct, for example, a digital receiver or transmitter. The advantage of using something like Simulink® early in the design procedure, as opposed to a component-based schematic simulator/design package, is that it allows the relatively high level performance of devices to be verified in good time. A useful Matlab® system object known as comm.SDRRTLReceiver()® was used to simulate the performance of a SDR device known as NooElec R820T2SDR & DVB-T®. As its name implies, the NooElec® device was originally designed as a low cost consumer quality digital video broadcast terrestrial (DVB-T®) SDR receiver with a USB interface. So it was quite cheap and offered an ideal development platform.
FCL has significant skills, knowledge and experience related to scattering (S) parameters and transmission (T) parameters, used in RF and microwave design and test. There are of course many parameters used in electronics engineering but perhaps S parameters are the most commonly used in many aspects of RF and microwave engineering. Although not normally measured directly, T parameters may readily be derived from S parameters and are very useful in removing (de-embedding) the effects of cables, connectors and adapters in a typical measurement system. FCL has developed detailed knowledge in the associated theory, test equipment, measurements, their limitations and the interpretation of the results. This is not limited to the frequently specified requirements of gain and return loss but many other derived properties including Rollet stability analysis, isolation, phase frequency response and delay.
FCL has used many of the latest types of vector network analyzer based test equipment capable of measuring S parameters very accurately, not only swept continuous wave (CW) but also swept and pulsed CW. The design features of equipment for CW and pulsed CW operations may differ significantly. Pulsed CW measurements are applicable to, for example, radar systems.
Much of the work performed by FCL for clients has been 'hands-on' circuit design and development. It is often necessary to investigate and resolve hardware bugs, to interface with test equipment and to test upgrades and other improvements such as potential replacements for obsolete parts.
Especially at high frequencies and signalling speeds, we need to assess how well a particular modification or interface will work. There is sometimes the risk that the interconnections and components used for a very valid proposed modification are not appropriate at the operating frequencies concerned. Once the frequencies involved are known, FCL has a good understanding of what types of components are suitable. For example, the component Q factors, loss parameters (tan delta) and self resonant frequencies must be chosen correctly.
Another challenge is in choosing the correct grounding method, depending on whether it is required to monitor differential or single ended lines. In recent years differential transmission lines have become very popular for high speed baseband transmission due its good common mode EMC performance. Examples in copper are: Ethernet, LVDS, RS-485, USB and PCI-e. Unfortunately, it is often difficult to measure true differential voltages on a differential transmission line, especially at high frequencies. Either the test equipment must have a true and correctly matched differential probe of some kind, calibrated across the required frequency range, or it may be possible to use a 4 port vector network analyzer (VNA) after calibration with a 4 port electronic calibration device (eCal). Test equipment manufacturers like Keysight Technologies® and Rohde and Schwarz® have various solutions which might be suitable. Each of the 4 VNA ports is a precision (unbalanced) 50 Ω connector such as APC-7® or APC-3.5®. The VNA provides several different calibration options using the eCal depending on the configuration of the device under test (DUT). Assuming that the impedance of each normal unbalanced VNA connector is 50 Ω then the common mode reference impedance becomes 25 Ω and the differencial mode reference impedance becomes 100 Ω. If a VNA to DUT connection to an unbalanced port was required, this would be achieved with a simple coaxial cable connection similar to the traditional (50 Ω unbalanced) approach. If the VNA was required to connect to a balanced DUT port, it would be necessary to provide a connection to each leg of the balanced port with a separate unbalanced (50 Ω) connector. Therefore the differential impedance measurements would be referenced to ground. The result of performing a mixed mode S parameter measurement like this for a 2 port DUT is to produce a 4 x 4 matrix of S parameters instead of the more common 2 x 2 matrix. The resulting 16 complex elements results a stimulus and response term for each port including common and differential modes.