The system and the transmitter
2.3 Transmitter
Figure 2.12 Typical station configuration. Twin-radar installations usually use two scanners which usually remain functionally independent even when sharing a display. The display may contain other facilities, particularly when forming part of an integrated bridge system (IBS)
The remainder of this chapter details how radars illuminate targets. Reception is detailed in Chapter 3.
Figure 2.13 A modern trend in bridge workstation design. Reproduced by permis- sion of Kelvin Hughes Ltd, Ilford UK
A steady train of powerful, short, pulses of electromagnetic energy is required. The train is sometimes said to be uncoded, for it carries no data. In radio terminology, the emission type is PON. Equipment limitations preclude generation of truly rectangular pulses - and they would have undesirably wide spectral width - but it is convenient for most purposes to think of them as rectangular or 'square'.
2.3.2 Magnetron power source
Except in coherent systems, the generator is usually a cavity magnetron valve (tube).
The magnetron has always been the cheapest and most efficient power generator and is a transit-time high power oscillator. Within a sealed envelope it contains a central cylindrical heated cathode, surrounded by pairs of anode poles connected to parallel inductor-capacitor (LC) tuned cavities, the inductor centre points being earthed. The high-vacuum working space between cathode and the anode system lies in the strong axial field of a permanent magnet, the tube/magnet assembly thus forming a packaged magnetron. Application of a negative pulse of about 1 OkV to the cathode makes it emit electrons. They are subject to crossed fields, radial electric and axial magnetic, and take spiral paths. Some electrons return to bombard the cathode, increasing its emission to several amperes, but most fly to the anode poles.
When a random grouping of electrons causes Poles 1 to fall slightly below earth potential, the centre-tapped tuned circuit forces Poles 2 positive. Electrons hitting
—ve Poles 1 deliver more energy than absorbed by those hitting +ve Poles 2, so the oscillation builds up. After a microwave half-cycle the tuned circuits swing the pole polarities to Poles 1 +ve, Poles 2 —ve. This instant is arranged to coincide with the transit time of the spinning beam to Poles 2, feeding more energy into the tuned circuits.
There are usually four pole pairs and the electrons cluster into four spokes. Inter- vening C-shaped cavities form the tuned circuits of the cavity resonators within the copper anode block. Frequency is therefore determined during manufacture and is ordinarily not externally controllable. Frequencies lie within sub-bands, usually
30 MHz wide, often located near the centre of the IMO operating band. The out- ermost 15 MHz 'guard bands' are always avoided to minimise radiation of outband spectral components.
The power is coupled out by a short coaxial line launching into an integral waveguide. Magnetrons are simple and have long life, and have stood the test of time. Their remarkable efficiency makes for cool running and high reliability.
Permissible duty cycle or on/off ratio is low and despite the high peak power, mean output power is only around 1OW (and waste heat needing to be dissipated is not much more), similar to that of shipborne very high frequency (VHF) radio with its quite different modulated continuous wave transmission. Actual transmitted pulse shapes are not generally disclosed by radar suppliers. Because target detectability is improved when there are many pulses within the packet, occasionally prf is raised so far on short-range scales that, to keep within the magnetron's maximum permis- sible duty cycle, pulselength has to be reduced below that essential for good range resolution.
The scanner is never perfectly matched. When the feeder is long, quite minor frequency change sharply changes the phase of the mismatch power returned to the magnetron. To minimise risk of provoking unwanted 'long line effect' modes of oscillation (Section 2.6.1), with their poor spectra and low efficiency, usually either a ferrite circulator is used as duplexer (Chapter 3, Section 3.2.3) or a ferrite isolator (a non-reciprocal device; ordinary reciprocal devices and components behave equally to either direction of energy flow) is inserted at the magnetron output to improve load match. Similarly, the rate of rise of drive voltage has to be controlled. More expensive coaxial magnetrons have tighter spectrum control and are better suited to coherent systems.
2.3.3 Modulator
The magnetron is driven from a modulator, designed to introduce insignificant noise.
Typically the modulator contains inductors and capacitors in a pulse-forming network (PFN) - or formerly a coaxial cable containing distributed inductance and capaci- tance - which accumulates low-voltage energy between pulses. When a solid-state thyristor, silicon controlled rectifier (SCR) or high voltage insulated gate field-effect transistor (FET) switch is fired by a small trigger pulse, the PFN rapidly discharges in a controlled manner through a step-up transformer, whose high voltage secondary is bifilar wound to carry the magnetron heater current. In some designs, the induc- tors and capacitors in the line overswing to double voltage. Pulse length is changed by switching inductor/capacitor combinations. Widely varying lengths pose difficult design problems and a delightfully named tail biter diode is often used to suppress secondary short pulses; alternatively FET switches terminate the pulse in a more definite manner. Formerly modulators used hard-vacuum triodes, or thyratron valves containing gas at low pressure - the author's first radar task was production test of hydrogen thyratrons. Occasionally saturable reactors (pulsactors, Melville lines) were employed, where a control current pulse switched magnetically stored energy.
These were heavy, complex and best avoided!
2.3.4 Influence of transmitter on system
The modulator/magnetron arrangement of non-coherent transmitters constrains many characteristics of the whole radar.
• Only a few discrete pulselengths are available, modulator design constraints precluding smooth variation.
• Magnetrons are bang-bang devices. Transmitter power is either full or zero. Unlike radio, there can be no low-power mode, although peak power may be a couple of decibels less than nominal on the shortest pulselength.
• Power builds up very rapidly ( ^ 10 ns) at the start of the pulse and pulse-end decay is nearly as fast, broadening the frequency spectrum and necessitating output filters to minimise interference to other spectrum users.
• Transmitter frequency is built into the magnetron cavity and cannot be adjusted by the operator or service engineer.
• Frequency changes by 10 MHz or so due to heating, pulling (load match change, VSWR preferably being held below 1.3), pushing (drive voltage change) and ageing. The receiver has to match drift using automatic frequency control (AFC) or manual retuning and bandwidth has to include a margin for error, letting through more noise and effectively reducing receiver sensitivity and SNR.
• Maximum duty or pulselength/prf combination is dictated by prf to prevent overload.
• The cathode may take several minutes to heat from cold, necessitating a heater-on hot-standby mode.
• Small size and high efficiency permit compact installation, enabling mounting at the scanner, obviating wasteful feeders.
2.3.5 Spectrum problems
It is difficult to tame magnetron output pulse edges and the output spectrum is unde- sirably rich in harmonics, often being especially dirty on short pulses. The anode tuned circuit is tightly coupled to the output to extract maximum power. Selectivity is perforce low, like a muffled bell, so the tuned circuit cannot fully suppress out-of- band frequency components. Other oscillation mechanisms become significant when the valve ages, especially when rate of rise of drive voltage is outside specifica- tion, and tend to cause spectral lines some tens of megahertz from centre frequency, called moding. Oscillation amplitude is limited by onset of saturation and cut-off effects which cause harmonics of the microwave carrier frequency to be generated.
A rounded Gaussian pulse (shape similar to Chapter 3, Figure 3.5) would deliver a cleaner spectrum, lying almost wholly within the marine band, but magnetrons are unsuited to this pulse shape. Successful modulator design demands particularly close liaison with the magnetron supplier.
At first the microwave spectrum was not intensively used. There were a few industrial and, later, domestic microwave heaters at 2.45 GHz and some low power industrial activities at 10.688GHz. Astronomical research receivers near 10 GHz demanded quiet conditions. Otherwise the civil and military radar fraternities had
the field to themselves. Although radar receivers are sensitive, the highly directional antennas invariably employed mitigate mutual interference. Civil marine sets employ prf stagger to break up 'running rabbit' interference - patterns of dots slowly travers- ing the display - from other sets employing similar frequency and mean prf. The military have to be prepared to counter hostile jamming, so can generally put up with considerable inadvertent interference. The situation had similarities to the spark transmitter days of early marine radio telegraphy.
For many years there was therefore little objection to transmission of rectangular pulses, with their profligate spectrum. Rectangular pulses are particularly convenient to generate by discharge of a delay line into the transmitter valve, and facilitate good range resolution.
As pointed out by Williams [1], the 1990s telecommunications explosion placed intense pressure on the lower microwave frequencies. Governments recognised the spectrum as a finite and valuable resource, to be auctioned to the highest bidder for billions of pounds. Covetous eyes focused on the centimetric bands. It is likely that radar will soon have to share the frequency bands with telecomms. While a good case can be made on safety and commercial grounds for pulse marine radar, together with sufficient spectrum for accurate range determination, it is hard to jus- tify the pollution of adjacent frequencies by unnecessary transmission of rectangular pulses, not to mention moding lines and harmonics. Occupied bandwidth must be minimised.
To a telecommunications receiver, it is peak power that matters, in other words the equivalent isotropic radiated power (EIRP), the power in the beam, the prod- uct of transmitter power and antenna gain (their sum when using dB). Radio circles prefer the term peak emitted power, PEP, which assumes radiation from a dipole of gain 2.15 dBi. Typical big-ship radars of say 2OkW peak power and scanner gain 1000-2000 (30-33 dBi) have very high EIRP, - 3 0 M W (75 dBW). As receiver noise in telecomm receivers is only a few decibels above the thermal noise floor of
—204 dBW/Hz, they may encounter severe interference. Since 2003, Appendix S3 of the ITU-R Radio Regulations2 in essence requires out of band shipborne radar emissions to be 6OdB below EIRP, for example, 30 W in the example, the 'relative' phrasing of the regulation (inserted at military behest) giving little incentive to reduce EIRP itself. VTS radars are required to be considerably better. This situation is likely to be tightened within a few years; current relative friendliness to high EIRP may be rebased to require out of band emissions less than some specified 'absolute' low number of watts per megahertz.
Some future spectrum control possibilities are outlined in Chapter 16. Currently, control is often by a bandpass filter at the magnetron output. Problems include space availability, designing to handle the peak power, provision of sufficient attenua- tion through to harmonic frequencies, energy loss in the passband and maintenance of good impedance match. If the filter appears highly reactive away from centre frequency, it may pull the magnetron frequency or provoke moding. Filtering has
2 ITU-R SM 329-7; Category A, Shipborne radar; Category B, VTS radar.
little practical effect on the shape, bandwidth or detectability of the received echo apart from introduction of ~ 1 dB insertion loss.
Radiated spectrum depends on several linked features.
• Modulator detailed design, particularly the slope (MOOV/ns) and dynamic impedance of its output pulse.
• Magnetron detailed design. New 'third generation' designs being developed claim sharply reduced out-of-band spectral components.
• Magnetron age. Spectrum deteriorates near end of life.
• In-band load mismatch presented by the duplexer or output filter. Four-port circulators with matched loads (Figure 2.11 and Chapter 3, Section 3.2) are superior to three-port types.
• Out-of-band load mismatch. It is difficult to design filters to retain good match in the stop bands.
• The out-of-band scanner efficiency. If poor, out-of-band radiation is reduced.