questions and answers in brief
ENGLISH VERSION
Updated Aug, 05, 03
WORK IN PROGRESS
What is the difference between search and tracking radars?
============
Search radars are systems devoted to the systematic exploration of a large volume of space (for typical air search radars, this is performed over 360° in azimuth and over elevation angles ranging from 20-30° up to almost 90°, processing the echoes over the whole PRI, i.e. over the whole observable range), using different scan techniques.
On the other side, tracking radars remains "locked" on a specific target to provide continuos information about its position and motion. Usually, a "pencil beam" (i.e., narrow in both azimuth and elevation) is used. Only a small "range window" correnponding to the target range and its immediate vicinity is processed.
External designation systems (such as search radars) are normally employed for target initial location.
Tracking radars use "closed loop" (feedback) control systems to keep the target aimed in both angle and range. For angle tracking (azimuth/elevation) different techniques, such as conical scan or monopulse are employed to detect if the target is off the antenna axis, and the direction and amplitude of this deviation. These error signals is used to drive the antenna pointing servomechanism (or the steering control system for electronically-steered antennas) to keep the antenna beam centered on the target.
In the same way, range tracking is performed detecting the position of the echo centre of gravity (or, for some applications, the leading edge) with reference to the observed "range window" (using, for instance the early gate-late gate technique, in which the amplitudes of two samples collected on the leading and trailing edge of the echo - which, after filtering, is approximately a triangle - are compared to generate the correction signal) to generate an error signal which shift in time the processed range window to keep the target centered.
How do conical scan and monopulse tracking systems work?
============
To perform the angular tracking of a target, it shall be measured how much, and in which direction, the target is away from the radar antenna axis. The first technique used for this purpose was the so-called sequential lobing.
In this technique, normally using multiple antenna feeds, the beam was sequentially pointed slightly away, in the 4 directions, from the antenna axis. Comparing the amplitude or right-left and up-down echoes, it was possible to determine the target off-axis in azimuth and elevation.
The conical scan is an evolution of this technique, in which the beam is continuosly moved (nutated) around the antenna axis (typically, this is achieved by nutating the feeder, at frequencies in the order of tens of Hz). An echo from an off-axis target will then be amplitude modulated (at the conical scan frequency). The modulation depth provides the error amplitude, while its fase is related to the direction of the deviation.
Demodulating the modulation envelope in its sin and cosin component, the azimuth and elevation error are then extracted.
Both sequential lobing and conical scan have the disadvantage of being sensitive to errors induced by the echo amplitude fluctuation (glint) during the scanning. To avoid these errors, the measurement must be performed on the basis of a single pulse: this is done with the simultaneous lobing or monopulse technique.
In the monopulse technique, 4 different off-axis beams are used simultaneously:
| A | B |
| C | D |
Using special combiners, 3 channels are generated from the above beams: azimuth delta = (A+C) - (B + D), elevation delta = (A + B) - (C + D), sum = A + B + C +D.
The sum channel is used for the transmission and, on receive, for the range tracking. The delta channels, referenced in amplitude and phase to the sum channel, provide the angular errors.
How is the scan of a search radar performed?
============
There are several possible scan techniques for search radars. The "classical" old fashioned search radars use a "fan beam" antenna, i.e. narrow on the azimuth plane and tall in the elevation plane, to avoid the need to scan in elevation, rotating over 360°. The limitation of this system is that it does not provide information about target elevation, and the target data are limited to azimuth and range (so called bidimensional, or 2-D radar). The elevation information, when needed, is achieved by external means: commercial aircraft, for instance, transmit the flight level info via their secondary-radar transponder; in air-defence systems, dedicated "higth-finder" radars, with a so called nodding beam (a fan beam rotated by 90°) scanning in elevation only, where used, in association with the main radars, to detect the flight level of the objects to be intercepted.
To extract the tridimentional information without rely on external means, capability of scanning the antenna beam in both azimouth and elevation is required. Generally, the scanning speeds required to effectively cover a 360° angle are not compatible with mechanical antenna steering.
For this reason, the modern tridimensional (3-D) radars, use the so-called electronical scanning, exploiting the Phased Array technique.
These systems usually perform the azimuth scan in convenctional way, while using electronical scanning for the elevation.
How does a Phased Array antenna work?
In a classic parabolic reflector antenna, the reflector is illuminated from a single point (feeder) and the signal is scattered back in order to produce a plane wave. It is also possible to produce a plane wave by a flat antenna composed by an array of radiators, all fed in-phase with the same signal: the waves from the radiators will combine to form a plane wave (see figure below). In the phases of the signals feeding the radiators have an appropriate relative phase shift, their waves will still combine to produce a plane wave, but the direction of propagation will not be anymore orthogonal to the antenna surface, by an amount depending of the phase shift between the radiators..
By properly controlling the phases of the individual radiators, it is then possible to synthesize wavefronts propagating in different directions, thus steering the antenna electronically (typically up to +/- 45°).
This is achieved by way of special electronic devices, called phase shifters, or with other techniques (e.g. by feeding the radiators through different electrical lenghts and varying the frequency, in order to change the relative phases).
This beam synthesis technique can be used both on a single axis (typically, elevation) using a mechanical scan on the other (the antenna is then made up of several rows of radiator, each one fed with the same phase) or on two axes (and, in this case, the antenna is made by an array of radiators each one with independent phase control), in order to avoid the need for the mechanical scan, at least in the sector covered by eletronic steering.
The latter solution is, of course, far more complex, requiring continous control of the phase of hundreds or even thousands of radiators, and is generally limited to special applications.
What is the "staggered PRF" operation?
Staggered PRF operation is used on many radars (almost all, in different forms).
Staggered PRF are mainly used to cope with multiple-time-around echoes. In fact, as explained in radar basics, targets at ranges greater than Ru=c.T/2 (where T is the pulse repetition interval) appear as echoes of the following pulse at shorter range.
Apparent range Ra = Rr-Ru where Rr is the real range.
It is possible to remove this range ambiguity by changing the PRI during the time-on-target. With different PRIs, the target will appear at different ranges. Using a proper logic, it is possible to identify the echo as a second-time-around one, and assign to it the proper range. As a general rule, use of n different PRI allows to solve up to nth-time around echoes (normally, 3 or 4 are used). It is possible to change the PRI at each transmitted pulse, but, generally, in modern radars using "packet" processing, they are changed on a packet basis (some tens of pulses). Note that many modern air-search radars (the so-called "pulse doppler" radars) interntionally work with PRFs so high to have ambiguous range, in order to sample the return at frequency higher than the maximum expected doppler shift (no doppler ambiguity), and cannot work without range ambiguity resolution.
[avoiding range ambiguity requires low PRFs, while avoiding doppler ambiguity requires high PRFs. The trade off between these two needs is a big issue in radar design: normally, you have to accept and solve ambiguities in one of these field, or in both]
At least 3 PRFs are needed because, for target at range equal to Ru, the radar is blind (the radar is transmitting another pulse, and therefore the receiver is blanked ).
Having 3 PRIs, this happens only in one packed over 3, allowing a reasonable decision algorithm (e.g., 2 out-of 3) to be implemented.
PRF staggering can also be an ECCM technique. In fact, it makes difficult for the jammer to predict the arrival time of the next pulse, making, for example, uneffective the use of the "range gate pull in" deception technique. Anyway, if only ECCM is of interest, "PRF jittering" (random pulse-to-pulse variation of the PRF) is normally preferred.
Is it possible to discriminate details smaller than the angular resolution? If yes, how?
============
Yes and no.
It is NOT possible to discriminate details smaller than the
radar resolution cell, but, in presence of relative (tangential) motion
between radar and target, this angular resolution is NOT necessary the
same as the physical aperture of the antenna beam.
The improvement can be achieved using a technique called "synthetic aperture". If proper coherent processing of the echoes is provided, observing the target from different points can be considered like "sampling" different points of a "virtual antenna" as long as the distance traveled by the radar during the observation time.
The same effect can be understood by thinking at a radar (supposed operating in continuos wave for simplicity) moving wrt a point target at about 90 deg from its velocity vector. The echo will have a positive Doppler shift when the target is at <90 deg (closing) and negative Doppler shift when target is at >90 deg (receding).
Analysing the target Doppler history, it is then possible to localise the target with a resolution better than the antenna beam aperture.
Independently from the approach used to model this effect, the conclusion is that, for a so-called "stripmap" (i.e., side-looking with no antenna steering) synthetic aperture radar (SAR) the resolution (in m) is independent from the target range and is proportional to the antenna dimensions. For an antenna of lenght L, the resolution is L/2.
This because a smaller antenna provides a larger beam, which allows longer illumination times and then a larger "synthetic antenna", thus improving the angular resolution. Increasing the target range also produces an increasing of the synthetic antenna aperture and then an improvement of angular resolution, compensating for the degradation due to the increased distance.
Even if the principle is simple, practical implementation of this technique
is very complex. The correlation to be performed on each individual
pixel requires huge computing resources, and several disturbing effects
need to be accounted for and compensated. Anyway, several systems of this
kind are currently operational on both airborne and space platforms.
A deeper description on how a Synthetic Aperture Radar works can be found
here
What is the "Coherent-on-receive" operation?
============
In the times of earlier radars, the only available device capable to produce
an high-power output at microwave frequencies, and therefore, suitable for
use in radar transmitters, was the Magnetron tube.
Unfortunately, the magnetron is not an amplifier, but a power oscillator:
when a high-voltage pulse is applied to the cathode, it generates at its output
a corrensponding pulse of RF energy, with a random initial phase.
This is not a problem if coherent operation is not needed: if you don't
have to discriminate the phase of echoes, the only thing you need to make
this device working is an Automatic Frequency Control (AFC) circuit
to ensure that the transmitter and the receiver are working at the same frequency
(usually, tuning the STALO - in this case, used only for the downconversion
in the receiver - frequency).
Things changes if the radar has to exploit the echo phase information,
as in MTI radars. In this case, you may either:
- Force the magnetron to start oscillating at a given phase, or
- Keep memory of the trasmitted phase and compensate for it in the receiver.
The second approach, named "coherent-on-receive", was by far the most widespread used
In this approach, the STALO is used for the receiver local oscillator,
and it is locked to the magnetron frequency by means of an automatic frequency
control circuit (same as for non-coherent systems).
The Tx pulse is coupled in the Rx chain, and at IF it is used to phase-lock
the COHO (phase detector reference).
Two different types of COHO were used:
- delay line: the coupled IF signal was injected in a loop with a delay
line (often, a long cable) with delay equal to the pulse lenght. The output
was fed back to the delay line imput (recovering the losses with an amplifier)
to cover the whole PRI.
- locked oscillator: the oscillator loop gain was reduced below the unit
(stopping oscillation), than the oscillation conditions was restored while
the reference pulse was applied. In this way, the oscillation started with
the same phase of the reference pulse.
A third way (I dont know if ever used in practice) is to sample the I/Q
component of the reference pulse to detect its phase, than to compensate
for it by means of a phase shifter (at IF or in video by cross-multiplying
the I/Q components)
Note that coherent-on-receive techniques recover the coherence only over
a PRI (the system keep memory only of the phase of the last transmitted pulse),
i.e. you cannot cancel multiple-time-around clutter.
To overcome this limitation while using magnetron or similar devices (such
as EIOs - Extended Interaction Oscillators) the transmitted oscillator
must be forced (or "primed"), by injecting a signal (derived from the COHO
+ STALO upconverted chain to ensure phase coherency) in the cavity while they
are starting oscillating, in order to 'lock' their phase exactly as done
for the COHO (but here is much more tricky, due to the high power levels involved).
In this way you get a fully 'coherent-on-transmit' system.