Page 137 - Radar Technology Encyclopedia

detector, frequency detector model 127

The frequency detector response is the dependence of These frequency detectors require preliminary amplitude
output voltage U out out on signal frequency f given constant limiting of the signal. Therefore, changes are sometimes
input voltage amplitude. An important detector parameter is introduced into detector circuits so they will acquire the addi-
response slope: tional properties of a limiter. AIL
dU out Ref.: Chistyakov (1986), pp. 161–168.
S = --------------
d
df
f = f 0 linear detector (see detector model).
where f = carrier frequency. A lin(ear)-log(arithmic) detector is one in which the output
0
Figure D30 depicts frequency detector circuits. In a bal- signal is linearly proportional to the input signal when its
anced frequency-amplitude detector (Fig. D30a), one of the level lies below some required value, and log proportional to
circuits is tuned to frequency f = f + Df, while the second is the input signal when vice versa. SAL
0
1
tuned to frequency f = f - Df. As it increases, the signal fre- Ref.: Currie (1987), p. 498.
2
0
quency approaches f and moves away from f . The voltage in A logarithmic detector is a detector with logarithmic depen-
2
1
dence between input and output voltages. It usually comprises
VT N amplitude detectors of the series type in which detected
f +
1
in U output voltages are added (Fig. D31).
1
U Stage 1 Stage 2 Stage 3 Stage 4
out
f E i
2
U
2
+
Detector Detector Detector Detector
(a)
+
C r
VD 2
M Amp
S
VT +
E
o
U U 2
1 U
in out Figure D31 Basic configuration of a logarithmic detector
C C
1 2 (after Skolnik, 1970, Fig. 29a, p. 5.34).
m VD
1
Here, diode detector voltage-current responses are
+
L L
1 2
selected so that a logarithmic dependence is obtained when
+E
k adding N detector output voltages. To do so, regulation in
Choke
(b) each stage usually is used. This makes it possible to regulate
the slope and length of segments of each response for best
Figure D30 Frequency detectors: (a) frequency-amplitude;
coincidence with a logarithmic curve. Logarithmic detectors
(b) frequency-phase (after Chistyakov, 1986, Fig. 5.32,
are widely used in monopulse radars. They are also used in
p. 161, Fig. 5.35, p. 163).
devices suppressing clutter from rain or from the ocean sur-
the first circuit increases, while in the second it decreases. face, and other variable intensity interference. AIL
The opposite is true if the frequency decreases. The FM sig- Ref.: Skolnik (1970), p. 5.34.
nal becomes amplitude-frequency modulated. Voltage from A detector model describes the ratio between output and
the circuits pass to amplitude detectors. The voltage at the fre- input voltages in an amplitude detector. Typically it applies to
quency-amplitude detector output has this form: diode detectors, and if the signal lies on a linear portion of the
diode response, the term linear detector is used. In this case,
U out = U – U = K m U in y 21 R yx()
2
1
d
k
d
the output voltage U out is proportional to the input voltage
where K = gain, m = coupling coefficient, U = input volt- U . If the output voltage is proportional to the square of the
in
in
k
d
age, y = conductance, R = equivalent resistance, and y(x) input, U out = KU in 2 , the term square-law detector is used. The
21
d
= detector normalized response. voltage-current response typically follows the square law for
In a balanced frequency-phase detector (Fig. D30b), in weak signals (for diodes, a signal with amplitude U < 0.25V
the absence of modulation voltage U is shifted 90° relative to is usually considered to be weak), and a linear law for stron-
2
voltage U . Given frequency modulation, additional shift ger signals.
1
proportional to the frequency change appears between U and Detector models are used in the theory of optimum detec-
1
U . Voltages U and U are applied to the diodes of a bal- tion, where a square-law detector is shown to be optimum for
2
1
2
anced phase detector, with this voltage being found at the lat- noncoherent integration at low signal-to-noise ratio (many
ter’s output: pulses integrated) and a linear law for high signal-to-noise
ratio (few pulses integrated). The linear law is better until
U out = K m U in y 21 R yx() approximately 30 to 100 pulses have to be integrated, and
d
k
d