Radar Schematics

Part 3:
Basic Design Architectures

The Magic of Radar

            A peek under the lid of a modern radome-based radar on a recreational vessel might lead to the supposition that there's nothing complex about radar: a fancy antenna rotating above or around a small cluster of processing boards and power supplies; and, somewhere sheltered from the elements, a compact LCD system that displays radar data along with other useful sensor information such as GPS, echo-sounder, fish-finder, throttle positions, speed through the water and so on. Mid-market displays will include electronic charts, so that radar data can be overlaid; and up-market systems may well include automatic echo-tracking capabilities. In fact, with so many uses for the LCD, it is easy to lose sight of the processes by which that 'hatbox' overhead performs its magic. 

        What, then, is this magic? In previous pages, the fundamentals of the magnetron and the antenna have been discussed, and there can be few in the modern world remaining unfamiliar with the concept of rastered displays such as used to be common in TV and computer-video systems; so, although their inception was figuratively spell-binding, these would not qualify as 'magic' nowadays. Appreciation of today's radar magic requires a basic understanding of the transformation that must occur between the transmission of raw power and the speckling on the LCD that signifies a gorging pelican, the groundswell of a lee shore, a ferry, rock, or tow-line - anything that matters to the conscientious mariner: 

• Timing is everything: Signal-processing theory is beyond the scope of this guide, and it must suffice to note that there is substantial detection gain derivable from coherent processing. Sadly, this advantage is denied the MNR in the RF domain, since the magnetron is a non-coherent electro-magnetic oscillator.[1] 

• For the marine navigation radar, then, "coherence" equates to synchronicity, with all relevant components functioning in harmony; this is especially important in interpreting the reflected energy, where multiple processing periods may be compared and contrasted for object detection in the presence of random "cosmic noise." It is also generally invaluable to control all processes, so that randomness does not itself become a "noise source."

• The pulse-timing device, or "clock", initiates the magnetron drive-pulse, a negative-voltage pulse that stimulates the magnetron cathode into emission. During this process, the receiver must be protected from the high-energy pulse that would otherwise deafen it, possibly permanently, a critical matter because transmitter and receiver share a common antenna. In lower-power architectures, it may be sufficient to rely on ferrites to direct RF energy, much as in a one-way revolving door; in higher-power systems it is utterly essential to use fast-switching circuitry to isolate the receiver. 

• To be of use, the copious electromagnetic energy flowing from a physically stationary magnetron chamber must be connected to a whirling antenna system: the slotted-waveguide antenna, or the enclosed array. This requirement mandates a rotary joint, or "rojo", with a very demanding characteristic: there must be no electromagnetic impedance, in either direction - otherwise, there may be damage to both the magnetron and the receiver during transmission, while reflections may be attenuated and lost to the receiver. 

•  It is still too soon to exhale a sigh of relief: unless the receiver is adequately protected, it is equally as vulnerable to reflections from adjacent objects as to direct connection with the transmitter. To forestall this possibility, the receiver must be desensitized at the instant of transmission and progressively re-sensitized over a period of microseconds. Sensitivity time-control (STC) circuitry is needed for this, and clearly it must be synchronized with magnetron activation. 

            Only now, with an RF pulse successfully launched on its mission, is it possible for the receiver to step up to the plate. It must first translate the potentially variable RF returns into a relatively constant intermediate frequency, or IF. This requires: 

• "memorizing" the variable RF of each transmitted pulse and summing this with a much lower RF, to create a waveform that differs from the transmitted RF by a known amount;  

• generating a continuous-wave representation of the summed result to mix with received energy; 

• extracting a "difference"[2] signal for subsequent processing; and 

• basebanding this into a time/magnitude representation.

       If processing terminated here, and the raw output could be intensity-modulated onto a PPI synchronized to both the pulse-timing and the movement of the antenna, the result would be the US Navy's 1942-era SG radar; and radar would remain a labor-intensive device best-suited for coordinating collisions at sea. Instead, the modern-day radar designer sees this point as barely the beginning of the process. Any number of algorithmic target extraction stratagems will be employed to segregate tangible objects from that "cosmic noise", almost all of them considered proprietary, but all with several common characteristics:

• They must be synchronized exactly to the timing of the radiated pulses; else it proves impossible to correlate target detections to range.

• They will divide the available "range window" into a substantial number of equal time-periods, commonly referred to as range-cells or range-bins. This enables comparison of equivalent range-bins' magnitudes over successive detection opportunities, for discrimination of tangible objects from the more nebulous, such as rain, and discrimination of both types from the ephemeral returns caused by, for example, other radars' pulses or similar stray electromagnetic spikes. Once again, timing is everything: the more time-coherent integration that can be attained, the greater the probability of valid detection, the lower the incidence of false alarms, and hence the lower the inter-process and processing bandwidths required further downstream. From this, it may be seen that the clock used to trigger the magnetron is even more critical to target-detection processing - sequencing the data through processing, slicing it into incredibly small bins and scheduling the various comparator processes, to name but the most obvious functions. 

            Even now, a sigh of relief would be premature: there is an entire gamut of post-detection processing to follow, discriminating between the discrete (far-off ships, buoys, spars etc) that may require algorithmic tracking, and the continuous: coastlines, swells, cloud formations etc., which require an entirely different tracking strategy. The mariner who would become ancient wants to know more than just "what is"; knowledge, "what will be," or information processing, is even more important. 

           Hence a new customer-pressure on MNR designers: a self-preserving desire for discrimination between the new and the known; and for tools, visualizations and representations that allow rapid comprehension of the maneuvering and navigational situation surrounding the user. These pressures have introduced capabilities that would be truly magical to the original SG user - processes synchronized not to the pulse-timing clock, but to the sweep of the antenna, processes known today as ARPA (Automated Radar Plotting Aid) or its smaller siblings, the Automated Tracking Aid, the Electronic Plotting Aid, and the Mini-ARPA. Only when these memory-intensive algorithmic processes have completed their cycle does the user get to see the magic in action: a radar display refreshed and overlaid onto an electronic chart system, along with vectors indicating target motion and recent history. 

           Viewed in this manner, it is easy to see that most of the capabilities of the modern MNR lie not in the radar per se, but in the data processing and information extraction that surround it; and that is how, typically, the MNR supplier represents it. A product brochure nowadays will focus very heavily on the data-processing, track-handling and visualization aspects, and may barely mention the transceiver system. Indeed, it is fairly common to find that a supplier buys the transceiver on a business-to-business basis, applies its own brand name and then adds the components considered necessary to make it a marketable system. 

           Marine navigation radars, then, are becoming systems-of-systems: an antenna and a transceiver[3]; sometimes-collocated control and data-processing units; possibly an information-processing unit (ARPA or suchlike), which might be combined with the data-processing unit; and a display unit that commonly incorporates the control units and sometimes also the processing components within its chassis. Not all of these elements need necessarily be provided by a single supplier. They have also become much more flexible in the characteristics that they may exhibit: the scheduling of pulses is nowadays driven almost entirely by the data-processing logic, and the "clock"is typically to be found on the control board. Nowadays, just as the mariner may visit a convenient chandlery to obtain the latest electronic version of navigational charts on an industry-standard medium, a software or firmware upgrade to the radar logic may be obtained on a CD-ROM or USB flash-drive, just as one might update a Windows® operating system, changing the behavior of the radar to meet some new processing objective. As hinted at in the opening words to this manual, MNRs have become as potentially dynamic as modern software-engineering practices allow, and the behavior of a system could conceivably vary from voyage to voyage."[4]

 
Design Variations

All-in-One Radome:

            Undoubtedly the simplest design is the all-in-one "hatbox" radar with a radome which encloses the antenna, transceiver, data-processing unit and most system control-logic, and which is most like the traditional view of a radar system as a self-contained entity. Peak power outputs of the transmitter range between 1.5 and 4 kW, with a marked bias towards the lower end of the output scale. Like those early radars, the simplest "hatbox" requires only a power supply and a visual display system that incorporates basic controls for range-scale selection and sensitivity adjustment, all other control logic and processing taking place within the scanner unit. Only a very few[5] of these most elementary designs are in present-day production.

Modern Radome

            More complex in layout, but still comparatively simple, the modern "hatbox" system keeps antenna and transceiver under the lid, but may house the data-processing within its display-unit chassis, along with the control logic: the pulse-scheduling circuitry, receiver sensitivity controls, a "sleep" mode scheduler, antenna turn-rate controls and, in truly high-end systems, some information-extraction capability such as Mini-ARPA or EPA. In these latter cases, coordination between antenna, a compass, geo-fixing device, and the information processor is essential. Some systems incorporate automatic antenna-speed selection, geared to either the radar-range selector or to the vessel's speed-log. Most of the high-end complexity of "hatbox" systems lies in their information-fusion capabilities, with charts, GPS, sonar, weather-broadcast and radar information integrated into a single user console - topics of pride beneath the gin-pennant.

Up-Mast Configuration:

            Radars with slotted-waveguide antennas, at their simplest, are virtually identical in layout to the more modern radome-based system. Their difference lies in the pedestal upon which the antenna rotates: as well as containing the drive-motor, these units contain the transceiver - connected to the antenna array via a rotating joint and possibly a visible length of waveguide with a U-shaped bend at one end[6], connected to one end of the antenna; all other units (data processing, information processing and displays) may be either at or below the mast-head.
 
      This configuration is the modern-day default; only where alternative configurations are available is it designated as an up-mast configuration by the supplier. The scanner unit receives power for transceiver and antenna system, pulse-timing signals and a code defining the required PD; it outputs a continuous stream of magnitude data ("radar-video") from the receiver, periodic "bearing pulses" signaling the progressive rotation of the antenna, a "heading flash" that signals alignment of antenna with the ship's bow and, in some systems, a "stern flash" indicating antenna alignment with the stern. 

      Radars with this basic architecture will typically have peak power outputs of 4 to 30 kW, and antenna lengths of 4 to 10' (1.2 - 3.0 m) for 3-cm wavelengths, and 10 to 12' (3.0 - 3.7m) for 10-cm wavelengths; a few systems may incorporate automatic antenna-speed selection, especially those designed to satisfy IMO regulations governing high-speed craft.

Down-Mast Configuration:

            Some high-powered systems are offered in two configurations, either "up-mast" or "down-mast"[7] - the latter signifying that the transceiver may be bulkhead-mounted somewhere below the mast, connected to the antenna pedestal by waveguide "plumbing." A survey of present-day designs indicates that the "down-mast" architecture is offered principally for high-powered systems with peak powers of 25kW or more. Most down-mast configurations incorporate the basic data processing element into the transceiver chassis, separating it from the control and display modules; more-modular versions may have the transceiver located separately from the data-processing. 

            Down-mast systems are most likely to include ARPA or similar target-tracking capabilities and hence to need the same level of interaction with the antenna system as the up-mast systems.

Black Box Configuration

            Many open-array radar systems are now being offered by their manufacturers as "Black Box" systems[8] - architectures designed to be integrated with pre-existing wheelhouse display facilities, mostly in vessels obligated to register with the IMO. Typically, Black Box systems are highly networked, with their facilities accessed remotely through software; their control, data-processing and information-processing functions may be located separately from the general-purpose display, which is likely to be some variant of the computer-industry standard VGA. They may be offered in both up-mast and down-mast configurations, and will have similar control/antenna-interface requirements.

            Somewhat confusingly, one manufacturer (Furuno) has a "black-box" architecture, known as NavNet2, which has a networked architecture in which the radar system is central, and other sensors are connected to the data network; the NavNet2 structure exists in parallel with its other Black Box systems, and differs in only one important respect: all devices on the network must be NavNet2 or -Version2 (vx2) systems.

Processing Architectures

            The extraordinary improvement in microprocessors over the past few years has moved many of the control and processing aspects of MNR systems within reach of high-performance PC-based architectures, potentially undermining the dominance of the two key alternatives: the application-specific integrated chip, or ASIC, and the field-programmable gate array, or FPGA. The former, as a hardware-like implementation of algorithmic processing, provides unmatched throughput at the expense of flexibility - ASICs simply may not be reprogrammed, and their development is time-consuming; consequently, nowadays they are not favored in MNR architectures. The FPGA, on the other hand, sacrifices some of the speed in return for programming flexibility; some FPGA architectures, in fact, can change their logical configuration in milliseconds, leading to the concept of reconfigurable computing systems. 

            Most MNR processors today use FPGAs, not with instant re-configurability in mind, but with longer-term re-programmability as an intentional trait of the processing architecture. However, a significant trade-off in FPGA designs is time-to-market: the logical firmware, developed in a hardware-specific language[9], must be extensively tested before release, especially in safety-of-life systems, and the upgrade cycle is therefore slow. FPGAs continue in favor with the manufacturer, but microprocessors are slowly emerging as alternative data-processing elements; with software-development maturity and continuing acceleration, they may reduce the upgrade-cycle substantially. 

            Whichever approach is used, the data-processor is the key driver for the transmitted waveform, controlling all timing characteristics, from pulse to scan; it would be reasonable to project that ever-greater flexibility will emerge. One consequence of this performance growth is the emergence of "radar manufacturers" that may make no radars. They offer products that are "Black-Box"-like, often with names based on the terms "radar" and "PC." There appear to be two fundamental adaptations: one provides "slave" processing, where the radar continues to function as designed by its manufacturer, and the PC-based processor augments the normal processing; the other assumes full control of the transceiver, often as a convenient way of upgrading a radar installation with minimal hardware expenditure. 

Footnotes:

[1] Some magnetron devices are phase-controlled. Use of these in MNR designs is cost-prohibitive. 
[2] The down-converted frequency favored in most MNR designs is centered at 60 MHz, with a PD-related selectable bandwidth of 15 MHz or less. 
[3] Because of the need for careful matching between antenna and transceiver, these two are most commonly considered as a single entity, a "scanner" in manufacturer parlance; some of the more powerful transceivers may be housed separately from the antenna unit. 
[4] A usually-stable VHF oscillator. 
[5] The GMR-20, -21, -40 and -41, manufactured by Garmin Electronics; possibly some Raymarine systems. 
[6] Many open-array antennas are center-fed, rather than end-fed, and the waveguide is not visible. 
[7] Furuno systems may be identified as down-mast configurations by the suffix '-W' after the system enumerator. Possibly this signifies "Wheelhouse."
[8] Not to be confused with the so-called aircraft "Black Box", for which ships' Voyage Data Recorders are synonymous.
[9] VHDL, VHSIC Hardware Description Language, is the most common.