Transistor
Amplifier
Linearizing
Baseband signal
Circuit
Frequency
Multi- Modulator
plicator
Noise
Oscillator Interference
Antenna
VCO
Filter
Spurious
Noise
Frequency Synthesis
Phase Variable
Detector Divider
Baseband signal
Detector IF
HF
EMC
00:06
Created by EBCCW 96:05
Created by EBCCW 00:06
Created by EBCCW 96:05
00:06
Radio School
Table of Contents
1 Introduction..................................................... 7
2 Block diagrams............................................... 9
2.1 System requirements..................................................... 9
2.2 Superheterodyne receivers ......................................... 11
2.3 Homodyne receiver...................................................... 15
2.4 Gain and selectivity distribution................................... 18
2.5 HF selectivity in tunable receivers ............................... 24
2.6 System selectivity requirement in the GSM................. 27
4 Mixers ............................................................ 35
4.1 General........................................................................ 35
4.2 Diode mixers ............................................................... 36
4.3 Gilbert-type transistor mixers....................................... 39
4.4 IQ mixer....................................................................... 40
5 Detectors ....................................................... 43
5.1 General........................................................................ 43
5.2 Coherent detection of linear modulation...................... 45
5.3 Noncoherent detection of angle modulation ................ 48
5.4 Coherent detection of angle modulation ...................... 60
Created by EBCCW 00:06
Introduction
1 Introduction
The choice of block diagram and the requirements on the blocks are largely
determined by the system selectivity requirements. Spurious signals and
intermodulation are discussed inRT2B and RT2C, and the results used in
this module.
b) Homodyne or
direct conversion
A/D
π/2 Detector
fr
fr
A/D
Figure 1.1
This modules covers those parts of the receiver, which reconstitue the
analogue baseband signal fed to the transmitter modulator. The additional
signal processing at digital transmissions (matched filters sampling and
threshold detection) is discussed in other modules e.g. DT2, DT3 and DT7.
2 Block diagrams
The receiver must also tolerate strong iterference on other radio channels
with only minor degradation of the transmission quality. This property is
described by the system selectivity, which is a collective expression for the
characteristics of adjacent channel selectivity, blocking, and the
suppression of spurious and intermodulation products (see Fig. 2.1b).
2. System selectivity
Figure 2.1 a
I dB
100
Spurious
Blocking
Spurious
50 Adjacent Adjacent channel selectivity
Spurious
channel fr
attenuation
attenuation
0 f
Figure 2.1 b
a)
HF IF
Detector
Channel
osc.
LO
b) Mixer 1 Mixer 2
Channel
osc.
LO2
LO1
Figure 2.2
Created by EBCCW 00:06
Normal mixers have the same mixing gain for the wanted receiver
frequency, fr, and the image frequency, fim, (relative mixing attenuation:
0 dB). If the wanted receiver frequency corresponds to the equation
fr = fLO + fIF (sum mixing), then the image frequency can be obtained from
the equation: fim = fLO - fIF (difference mixing) (see Fig. 2.3). Alternatively,
the receiver frequency can be obtained from difference mixing, and the
corresponding image frequency from sum mixing. From the above follows
that the separation between the receiver frequency and the image
frequency is 2fIF. All other mixing products can be neglected at a perfect
"linear" mixer.
HF IF
fr
fr f IF f IF
fsp
f LO
Attenuation
Equivalent receiver selectivity at mixer input
dB IF selectivity (transposition of IF selectivity)
HF
selectivity
f IF f im f LO fr f
Figure 2.3
The lower the intermediate frequency chosen, the narrower will be the HF
selectivity needed to fulfil the image-suppression requirement. Due to
component limitations, it may still be necessary to use a relatively low IF
(relative to fr) in order to obtain sufficiently sharp and stable IF selectivity. If
the ratio of fr to fIF is very high, it may prove to be impossible to realize an
HF filter that can provide both high image rejection and low passband loss
(in order to fulfil the requirement for a low noise figure for the receiver).
(Other factors also need to be taken into account when choosing the
intermediate frequency in the case of receivers with a wide tuning range
and a fixed-tuned HF filter. This is dealt with in section 2.5 below.)
The block diagram for the receiver is also influenced by the type of
modulation employed. The main characteristic of interest is whether the
modulated radio signal has a constant or a varying signal envelope. In the
latter case, it is not generaly permissible to limit the signal in the
pre-detector stages. The original modulation with a varying signal envelope
was amplitude modulation (AM).
If the radio signal has a constant envelope, the receiver can deal with strong
level fluctuations of the received signal by clipping or limiting strong signals,
so that the level of the input signal to the detector remains largely constant
(desirable for optimum detector performance) (see Fig. 2.4, a). The figure
also shows the possibility of using automatic frequency control (AFC),
which continuously adjusts the receiver frequency to that of the carrier of the
received signal. An advantage of using a limiter instead of AGC is that it has
a fast response, i.e. it can easily handle rapid fluctuations (due to multipath
propagation) in the level of the incoming wanted signal.
Created by EBCCW 00:06
a) Constant-envelope modulation
Frequency
detector
Limiter
Frequency
IF filter monitoring
Channel
osc.
AFC (automatic frequency control)
b) Amplitude modulation
AM Det.
Level
monitoring
Channel
osc.
Figure 2.4
It is also essential that none of the blocks before the detector is overdriven
by strong interference on other radio channels than the receiver frequency.
(If a stage is limited by the wanted signal, the requirement is that this signal
shall dominate over all unwanted signals occurring simultaneously.) If a
stage should be overdriven by interference that are larger than the wanted
signal, the gain for this signal would be drastically reduced and the receiver
subject to blocking. (See section 2.4). To avoid this, accurate matching of
the filter selectivity, gain and overload levels must be made stage by stage.
The overall gain possible up to a certain point in the amplifier chain will be
determined by how much the dominating unwanted signals have been
suppressed by the selectivity of preceding stages. This is discussed below.
It might be possible to block the d.c. voltage from the mixer by means of a
capacitor in the output. However, this means that the modulation sideband
close to the carrier will also be suppressed. A further complication in TDMA
is that the level of the d.c. voltage changes with each data burst received.
Created by EBCCW 00:06
Even if the d.c. voltage is blocked by a capacitor, a transient will arise and
this will have to decay before the subsequent amplifier will be able to
operate.
There are two block diagram options in which these problems do not arise.
One option is to introduce a conventional IF filter with high enough
selectivity to achieve considerable attenuation of all interferences on other
channels (see Fig. 2.5b). This allows high amplification to be introduced in
advance of the IQ mixers without the risk of blocking. The requirement for
good d.c. balance and low 1/f noise in the mixers is thus substantially
reduced.
Homodyne receiver
[
S = Re ( Ib + jQb )e jω 0 t ] Qb
A/D
+45° Detector
-45°
A/D
Channel Ib
osc.
Figure 2.5 a
HF IF
Qb
A/D Detector
Ib
Channel
Provides most
oscillator
of the system
selectivity including Same as in Fig. 2.5a
high adjacent
channel selectivity
Figure 2.5 b
Image-suppressing mixer
Channel oscillator
HF A/D Detector
B
A
f
f im f LO fr
f r1 f r2 f r3 f r4
Channel
Modulation spacing
bandwidth
A: Suppression of lower adjacent channel (image suppression in mixer)
B: Suppression of upper adjacent channel (filter selection)
Figure 2.6
With this IF one of the adjacent channels will end up at the image frequency,
which would appear to be totally unacceptable, as the HF filter would give
no adjacent channel attenuation. The solution is to employ an
image-suppressing mixer (see Fig. 2.7). (The same coupling can be used
to suppress the unwanted sideband in the generation and detection of sigle
sideband signals - see Fig. 5.2.)
The fast development in the areas of fast A/D converters and advanced
digital signal processing makes it now possible to implement wide-band
Created by EBCCW 00:06
receivers, see module RT1E. This means that the receiver blocks from the
antenna including the A/D-converter are wideband.
Image-suppressing mixer
Wanted receiver frequency: ω o + ω 1
Image frequency: ωo − ω1
Intermediate frequency: ω1
[
cos(ω ot + ϕ ) a1 cos(ω o + ω 1 )t + a2 cos(ω o − ω 1 )t = ]
a1 a
= cos(ω 1t − ϕ ) + 2 cos(ω 1t + ϕ ) + 2ω o − termer
a1 cos(ω o + ω 1 )t +
2 2
+ a2 cos(ω o − ω 1 )t cos(ω ot + ϕ )
From HF filter a1 cos(ω 1t − ϕ )
π π To IF filter
− −
2 2
sin(ω ot + ϕ )
a1 sin(ω o + ω 1 )t +
[
sin(ω ot + ϕ ) a1 sin(ω o + ω 1 )t + a2 sin(ω o − ω 1 )t =]
+ a2 sin(ω o − ω 1 )t
a1 a
= cos(ω 1t − ϕ ) − 2 cos(ω 1t + ϕ ) + 2ω o − termer
2 2
Suppression is obtained of input signal: a2 cos (ω o − ω 1 )t
Figure 2.7
Detector
Cmin S/N
BIF = 16 kHz
(42 dBHz)
0, 5 ⋅ 10 −12
C= = 10 −14 = −140 dBW
50
Figure 2.8
Interfering
Receiver
transmitter
P≈10W
Propagation loss, L
Figure 2.9
As discussed in section 2.2, we can assume that a C/I after the IF filter of
10 dB gives acceptable transmitter quality. When designing a receiver, we
need to take into account several mechanisms that can reduce the post-IF
filter C/I:
a Intermodulation between two strong incoming
unwanted signals (gain distribution and intercept
points)
b Reciprocal mixing with LO noise(noise spectrum of
the LO signal)
c Spurious signals(gain distribution and intercept
points)
There are two further mechanisms that can reduce the C/I after the IF filter:
d Inadequate IF filter selectivity and including leakage
due to insufficient shielding.
e Blocking as a result of saturation of amplifier or mixer
stages (gain/selectivity distribution and saturation level
of active stages)
The first condition that the receiver must meet in order to cope with
interference on near-by channels is adequate selectivity in the IF filter.
As regards adjacent channel selectivity, the situation is further complicated
in that the selectivity is usually measured with a modulated interference
signal. If we disregard this, then the requirement for attenuation of the
A typical specification for an IF filter for LMR with 25-kHz channel spacing
(10.7-ΜHz crystal filter) is shown in Fig. 2.10. Attenuation of 90 dB for
adjacent channels is specified. In addition, 70-dB attenuation is required at
a frequency separation of 17.5 kHz. The reason for this is that the specified
adjacent channel selectivity must be obtained even if the adjacent channel
interference is frequency modulated (half modulation bandwidth 7,5 kHz).
Specification
110 110
100 100
Attenuation (dB) ref. lowest loss point
25 kHz
90 90
80 80
70 17.5 17.5 70
kHz kHz
60 60
50 50
40 40
30 30
20 20
15 kHz
10 10
0 0
-50 -40 -30 -20 -10 0 +10 +20 +30 +40 +50
Frequency (kHz) ref. 10.7 MHz
(Specification: The selectivity curve
must not enter the shaded areas)
Figure 2.10
Created by EBCCW 00:06
Another condition that must be met if the receiver is to comply with the
required system selectivity is that strong unwanted signals must not
saturate any of the active stages between the antenna and the detector.
This means that the gain for the wanted signal (and thus the interference)
must be kept down until the filter selectivity has sufficiently reduced the I/C.
5V
2.5 mW
0.5 mA
Figure 2.11
Given these conditions, we can develop Fig. 2.2 to the arrangement shown
in Fig. 2.12. It is assumed that the level of the wanted signal must be at least
-30 dBW (e.g. 1 V over 1.000 ohm) at the detector input. The detector will
usually incorporate semiconductor diodes or the equivalent with strongly
nonlinear characteristics for low voltages. This means that the signal
voltage to the detector will generally need to exceed 1 V to produce a linear
detector function.
A drawback with a low HF gain is the higher overall noise figure for the
receiver. When designing the receiver, we therefore have to make a trade-off
between sensitivity (noise figure/noise temperature) on the one hand and
blocking and spurious characteristics on the other. This is discussed in
RT2E.
Figure 2.12 also shows the requirements as above for the out-of-band
rejection, Lrej, and image suppression in the IF filter. To obtain 80-dB image
suppression (i.e. C/Iim = -80 dB), the image rejection in the HF filter must
be at least 90 dB.
The above analysis of how the block diagram for the receiver is influenced
by the requirement for system selectivity is based on much simplified
assumptions and assumng a typical LMR application. The last two items in
section 2 supplement the previous analysis in two respects:
a Requirement for HF selectivity in a wide-band
receiver; (section 2.5)
b Reduced system selectivity requirement in cellular
systems such as GSM (section 2.6)
Created by EBCCW 00:06
Detector
Image-frequency
attenuation,
90 dB
GHF =10 dB GIF =100 dB
Iadj = –60dBW
Iim = –60dBW
Ibl = –40dBW
(Lrej)bl = 110 dB GHF represents a trade-off
between sensitivity and
(Lrej)adj = 90 dB interference characteristics
C/Iim = 80dB L rej dB fadjl : Lower adjacent channel
Specified fadju: Upper adjacent channel
C/Iadj = 80dB system 110
fim : Image
selectivity
C/Ibl = 100dB 90 Lrej : Required IF-filter
selectivity
fr : Receiver frequency
f
fadjl fr fadju fim
Figure 2.12
Wide-band receiver
HF IF
fin = f IF
= fo ± m ⋅ ∆ fLO 2
1dB B≈ ∆
n 3
m≈
2 B >A ( fLO = fin ± fMF )
Receiver tuning range
"n" radio channels
with ∆ channel spacing
f in
fo
A: Tuning range
A = n⋅∆
Figure 2.13
Created by EBCCW 00:06
90
Difficult to implement the HF filter if:
A. fIF < 3A Tight form-factor specification
B. fIF << fo High Qo for low pass-band attenuation
30
fIF /2
90
Figure 2.14
b) Intermodulation c) Spurious
fI 1 = ±800 kHz fI 2 = ±1600 kHz C = -129dBW pb < 2%
Isp
C = −129 dBW I1 = I2 = −79 dBW f − fo MHz
C dB
pb < 2% 0.6 – 1.6 56
1.6 – 3 66
3– 76
d) Interference on adjacent channels
I/C dB
50 49
40 41
C=-115 dBW (>>C o )
30 p b < 4%
20
Adjacent channel
10 9 (f-fo)
e) Blocking
Ibl dBW C =-129 dBW
-50 pb< 2 %
-53
-60
-63
-70 Ibl
-73 = 59 dB
C0
(f-fo)
-3 -2 -1 0 1 2MHz 3
(Outside the 915-980 MHz band, Ibl = -30 dBW shall be tolerated.)
Figure 2.15
generator is the current mirror, the principle of which is shown in Fig. 3.2. If
the transistors are identical and the collector current is almost independent
of the collector voltage, the currents flowing through the transistors will be
the same if the base-emitter voltages are the same.
Vbe1 Vbe2
Vi Vi
-4VT-2V T 2VT 4V T
(see Fig. 3.2)
I EE
current generator
V
− be 1
Ic 1 = I s e VT
kT I +I ≈ I
V =
Vbe 2 T
≈ 0, 026 volt C1 C2 EE
−
VT q
Ic 2 = I s e
(q = Electron charge)
−V
∆V = Rk αIEE tanh i
2VT
Figure 3.1
I0 ≈ I1 ≈ I2
if the transistors are identical
Io I1 I2
I1 I2
Re Re Re
Figure 3.2
Without emitter feedback, the linear operating range is largely limited to the
range defined by -VT to +VT. VT = kT/q ≈ 26 mV (where “q” is the electron
charge). The output range can be increased through emitter feedback but
at the cost of a reduction in the gain (see Fig. 3.3). If the curve for
IEERE = 10 VT is chosen, for instance, saturation will occur at an input
voltage having a peak-to-peak value of 25 VT ≈ 0.66 V, i.e. an effective
value of 220 mV. If the impedance is 200 ohm, this will correspond to an
input power of -6 dBm (approximate blocking level).
VCC ∆V
RE =0
∆V
RE RE Vi
Vi -20VT -10VT 10VT 20VT
IEE R E =20VT
IEE R E =10VT
Figure 3.3
The drawback with emitter feedback is that the input impedance is very high
- probably deviating widely from the generator impedance that gives the
best noise figure. This can be remedied by introducing a feedback loop
between collector and base (a shunt circuit) as shown in Fig. 3.4. Because
of the Miller effect, Rf ends up as an effective resistance across the input of
Rf
------------- , where “A” is the voltage gain of the stage. (The figure is an extreme
1+A
simplification as high frequency effect have been disregarded).
Rc
-AVin
Rf
R in Rf
Rin ≈ ( Re ⋅ β)
1+ A
Om
If β storthen:
ReβRehigh, :
Vin Re Rf
Rin ≈
1+ A
Figure 3.4
3.2 IF amplifiers
A receiver for constant-envelope modulation usually includes a limiter to
provide a constant signal level to the detector over a wide range of signal
inputs to the receiver. The higher the level of the receiver input signal, the
earlier the limiter needs to be placed in the IF amplifier stage. A symmetrical
limiter, i.e. one generating a square output wave, is critical to good detector
performance. This is facilitated by balanced design of the IF amplifier.
4 Mixers
4.1 General
This section discusses mixers for both transmitters and receivers.
• Mixer gain, i.e. the ratio of the level of the wanted output signal
to that of the input signal
The higher the intercept point, the better will be the spurious characteristics.
s'r f
sb ωo
Figure 4.1
RF Order Harmonic
1 33 - 35 18 44 27 45 38 1 32 - 36 17 45 25 46 38
2 69 57 52 56 61 52 48 63 2 73 66 62 68 71 61 58 67
3 59 47 56 45 56 46 74 51 3 >73 68 >73 66 >73 67 73 70
4 77 82 >83 75 80 72 77 70 4 >73 >72 >73 >74 >74 >73 >74 >77
5 >83 >84 >84 73 83 72 78 73 5 >74 >74 >73 >74 >74 >74 >72 >74
6 >77 >83 >83 >85 >84 >84 <84 >84 6 >69 >75 <75 >73 >73 >74 >75 >74
7 >77 >77 >84 >84 >84 >83 <83 >85 7 >67 >67 >73 >74 >74 >73 >73 >74
8 >78 >78 >77 >84 >84 >85 <82 >84 8 >68 >66 >68 >74 >75 >74 >73 >74
9 >78 >79 >77 >78 >83 >84 <83 >83 9 >68 >68 >67 >68 >74 >75 >74 >73
10 >79 78 >78 >78 >76 >84 <84 >84 10 >69 >68 >68 >69 >68 >74 >73 >73
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Harmonic LO Order (n) Harmonic LO Order (n)
test conditions: test conditions:
RF. = IN: FREQ. 1000. 1 MHZ DRIVE -4DBM RF. = IN: FREQ. 1000. 1 MHZ DRIVE -14.07DBM
LO. = IN: FREQ. 970.01 MHZ DRIVE +9.98 DBM LO. = IN: FREQ. 970.01 MHZ DRIVE +9.98 DBM
IF = MEASURED: FREQ. 30.09 AMP. = -11.23 DBM IF = MEASURED: FREQ. 30.09 AMP. = -21.46 DBM
Conversion Loss
(dB)
RF LO LO LO
MHz MHz + 7dBm + 10dBm
10 40 7.09 6.53
20 50 7.15 6.60
100 70 7.10 6.55
200 170 7.14 6.56
400 370 7.36 6.74
Figure 4.2
At the bottom of the sheet we can find the mixing (conversion)loss for
different receiver (RF) and LO frequencies and two levels of LO signal
power. As can be seen, the loss increases with an increase in the input
frequency and is slightly lower for a 10-mW LO signal than for a 5-mW
signal. The figure also shows the mixing loss for different mixer products for
two levels of the input signal. The attenuation increases sharply with an
increase in the order m of the input frequency, the increase is much less for
increased order n of the stronger LO signal. Balancing results in some
suppression of the mixing products with m = 2 or n = 2.
Vout
t
VLO
Figure 4.3
Created by EBCCW 00:06
It is appropriate to use balanced amplifier and mixer stages for the wanted
input signal to the receiver, in which case two cells, as shown in Fig. 4.3, are
combined (see Fig. 4.4). To improve the dynamic properties, it is expedient
to introduce emitter feedback. This improves the linearity and operating
range in the same way as shown in Fig. 3.3.
Vcc
Rc Rc
Vout
VLO
V in RE RE
I EE
Figure 4.4
Here, too, the mixer does not have an ideal polarity-switching function. This
is mainly because of the finite switching times due to the LO signal not
having a square waveform, and the high-frequency limitations of the
transistors.
4.4 IQ mixer
The IQ mixer effects the conversion from the IQ baseband up to the radio
channel on the transmitter side and the reverse procedure on the receiver
side. The additional requirements as compared with a single mixer
(linearity, low LO leakage) are for gain balance between the I and Q
channels and 90° phase shift between the two channels. The two LO
signals must therefore be in quadrature. A possible complication is that the
90° phase difference must be maintained across the tuning range of the
local oscillator.
5 Detectors
5.1 General
This section serves to complement the general study of detection in
analogue transmission contained in Module G3. The following three
aspects are discussed:
a Coherent detection of linear modulation
b Noncoherent detection of angle-modulated signals
c Coherent detection of angle-modulated signals
There is no need to discuss the basic operation here. Instead, this section
deals with elimination of one of the sidebands in a double mixer
arrangement (in SSB modulation), and how the correlation between the
sidebands in DSBAM can be exploited to generate the local reference
signal for detection (Costa's loop). In connection with an ideal coherent
detector, the choice as to whether the filter selectivity should be placed
before or after the detector is arbitrary. A low-pass filter characteristc after
the detector can be transformed to the other side of the detector and be
combined in the IF selectivity (see Fig. 5.1). Alternatively, the IF filter can
be replaced by an equivalent low-pass filter after the detector. Moving the
IF-selectivity after the detector has the draw back that the detector can be
overloaded by strong close-in interference (i.e. on adjacent channel).
Created by EBCCW 00:06
Hr ( f + fo ) Hb ( f )
0 f
H r (f) H b(f)
b) With pre-detector filters
fo
|H r (f)] |H b(f)]
H r (f) H b (f-f o)
1 1
Hr ( f ) ⋅ Hb ( f − fo )
fo f 0 f
Hr(f): HF or IF filter
Hb(f): baseband filter 0 f
Figure 5.1
The most effective way to lower the threshold for a signal in a noise
background, is to use a coherent (“phase-sensitive“) detector. However, a
phase detector cannot be implemented direct, as practical systems will use
modulation index much higher than the operating range of a phase detector.
The solution is to introduce modulation feedback to reduce the phase
difference between the input signal and the reference signal to the extent
needed to make phase-locking possible. (Section 5.4).
The threshold is in this case determined by the fact that the phase locking
drops out if the noise density in the detector input is too high. The critical
design parameter is the bandwidth of the loop. A narrow loop bandwidth will
reduce the influence of input noise but, on the other hand, the bandwidth
must be wide enough for the reference oscillator frequency to follow the
variations in the frequency of the wanted input signal, with only a moderate
phase error being incurred.
(Module G3, Fig. 3.6). This has been the only way to achieve adequate
sideband suppression.
sb (t)
cosω οt
s(t)
+ SSB signal:
Gilbert
90° 90° sb (t ) cos ω ot + sˆb (t ) sin ω ot =
filter
ŝb (t ) sinω οt [
= Re (sb (t ) + jsˆb (t ))e jω o t ]
Figure 5.2
Modulator
(DSBAM-sc) sin(ω ot + ϕ )
sb2 (t )
sb (t ) sb (t ) cos(ω ot + θ) -π/2 sst = sb2 (t ) sin[ 2(ϕ − θ)] ≈ (ϕ − θ)
4
Input
cos(ω ot + θ)
baseband VCO
signal
cos(ω ot + ϕ ) If ϕ − θ is small:
sb (t )
≈
2
(Detected baseband signal)
sb2 (t )
Control voltage to VCO: sst = s (t ) sin[ 2(ϕ − θ)] ≈
2
b (ϕ − θ) If ϕ − θ is small:
4
Figure 5.3
Created by EBCCW 00:06
Phase modulation
Above the threshold, i.e. with an adequate signal-to-noise ratio in the input
to the detector, a reasonable approximation is that:
a the noise level in the detector output will be the same
with or without modulation.
b the level of the wanted signal in the output will not be
influenced by the signal-to-noise ratio in the detector
input.
This calculation is shown in Fig. 5.4. The result is that the modulation gain
2
m
above the threshold in phase modulation is G M = ------ .
2
No
S˜ ϕin → sout
[
s = Re Ae jϕ.e
jωo t
] B
ϕ in
Phase
sout = ϕin
W
Sout
1 2 fo comparator
Nout
C= A UN
2 ρw = C
NoW
+
ρ = Signal-to-noise ratio normalized to baseband width, W
w
ϕin = Phase angle of the complex envelope of the wanted signal + noise
ϕ = Phase angle of the complex envelope of the wanted signal
Output after the baseband filter
a) Modulated noise-free input signal: sout = ϕ = m ⋅ sin ω t ⋅ t (voltage)
Sout = ϕ = 21 m 2
2
Output power of wanted signal:
Figure 5.4 a
Created by EBCCW 00:06
sδ = 2 No cos(ω o + ω b )t =
Wanted input signal
power = C
Pδ =No
= Re
Re 2N [[14243⋅e⋅e []
2 Nooee
jω
jωb t t
b
jω
jωot t o
No SS˜δδ
fo
f-f o S˜δ = 2 No
1 Hz
fb S˜δ cos ω b t
ϕδ ≈ =
A
S˜ + S˜δ S̃δ = 2 No
cos ω b t = No
cos ω b t
A2 C
ϕδ ωb ⋅ t
ϕ δ2 =
No
S̃ = A S̃δ << A 2C
U N = 2ϕ s2 W
= ∫ U N df =
NoW
No After baseband filter Nut C
out
C o
Sout m ⋅C2
1
f ∴ = = m 2 ⋅ ρw
W B/2 Nout 2 No W 2
1 C
∴ GM = m2 ρ=
2 No ⋅ W
Figure 5.4 b
Frequency modulation
does not change the level of the wanted signal (test tone) and the signal
power will still be S = 1/2m2.
We now need to calculate the noise power, N, after the treble-boosting filter
and the subsequent low-pass filter. Multiplication by a factor of (f/ft)2 means
that the noise spectrum will be parabolic, with the power density increasing
with rising baseband frequency. Owing to the effect of the low-pass filter, we
need to integrate the unilateral noise spectrum only between 0 and +W. The
result is that for frequency modulation the modulation gain will be:
GM= 3/2.(∆F/W)2.
S̃s̃ No
()
2
[
s = Re Ae jϕ.e
jωo t
]+ ϕ in UN Treble boost Un′ = U N f
ft Nout
Phase
11 2 ρω detector 6 dB/oct
Sout′ = Sout Sout
CC== 2AA2 Sout
W
2
1 dϕ
()
2 2
= fi = ∆F cos ω t ⋅ t U N′ = U N f f
W
ft Nout = ∫ U N df =
2π dt ft
( )⋅ sin= sinω ωt ⋅ t
o
∆F
ϕϕ == ∆F
t 1 W3
= ⋅ UN
fft t Nout
t
( )=m
3 ft 2
∆F f
0 W B/2
ft
No
1 ∆F
2 UN = enligt fig 5.4b
from Fig.
1 C
Sout = m2 =
2 2 ft
1 ∆F 2 3 ∆F 3 ∆F
2 2
Sout 3 ft 2 1 C
= ⋅ = ⋅ = ⋅
Nout 2 ft 2
W UN 3
2 W WU N 2 W No ⋅ W
3 ∆F
2
∴ GM =
2 W
Figure 5.5
Created by EBCCW 00:06
Sr = A cos(ω 0t + ϕ ) n(t) ñ
A
A
t t
(Probability density) A2
( )
p ñ Ar : e − 2σ 2 = Output signal
Phase jump
Sl
Pr
= e− No B = e−ρ B
σ 2 = No B Nd
A 2A 2
r ==
PP
m 2 2
t
Α
Origin Αr ñ
S̃r = A A phase shift of 2π occurs
if the sum vector S˜r + n˜
ñ circles the origin
Figure 5.6
If the unwanted signal in the detector input consists of Gaussian noise, the
phasor for the sum signal will hunt rapidly between random points for as
long as the noise envelope is greater than that of the wanted signal.
Because of capture, the detector outputs the noise modulation instead of
the wanted signal. The longer this phenomenon persists, the greater will be
the reduction in the level of the wanted signal as a result of the noise
modulation. However, the main reason for the sudden reduction in the S/N
Created by EBCCW 00:06
at the threshold is something else: namely, origin circling (see Fig. 5.7, a),
giving rize to “clicks“ (impulse disturbances from an FM-detector).
FM detector
a) Origin circling
Noise phasor
Below the
threshold
Above the
threshold
f
W
Figure 5.7
Origin circling occurs if the noise phasor is longer than the phasor for the
wanted signal. Since the phase difference between the two phasors varis
rapidly and randomly, they will often be opposed. If the noise phasor is also
longer, origin circling occurs, whereby the phase shifts rapidly by an amount
close to 2π. This results in an irregular sequence of voltage jumps in the
output from a phase detector or in a sequence of short pulses (white
spectrum) from a frequency detector.
At and below the threshold, the clicks from an FM-detector due to the origin
circling is much higher than the noise signal caused by noise peaks
exceeding the amplitude of the wanted signal (noise capture). The
threshold is usually defined as the point on the detector characteristic at
which the signal-to-noise ratio is diminished by an additional 1 dB as a
result of additional noise, above all the clicks caused by the origin circling.
envelope will exceed a certain peak level, above the effective value. If the
margin (peak factor) is higher than 8-10 dB, the probability will be relatively
low. For Gaussian noise, therefore, we usually assume that the threshold
will occur at a signal-to-noise ratio of 10 dB in the detector input. For good
threshold characteristics with a noise background, it isvital that the IF
bandwidth is not too great. A guideline is that the optimum IF bandwidth will
correspond to the bandwidth determined by Carson's rule: B = 2(∆F + W).
In the C/N region immediately below the threshold, the dominating cause of
detector noise is the clicks from origin circling. The phase jumps from a
phase detector generates a noise spectrum that diminishes by 6 dB/oct,
whereas the corresponding frequency pulses from a frequency detector
produce a nearly white noise spectrum (see Fig. 5.7, b). This added noise
is dominant below the threshold. Thus, the shape of the noise spectrum is
different above and below the threshold.
Created by EBCCW 00:06
So (1 − e −ρ B )
2
Phase-jump noise
Threshold
1 dB
C
ρB =
1 10 dB No B
ρ B ( dB)
B
Unmo- Test
W tone
dulated
mod.
4 7 7.8
8 8 9
18 9 9.9
40 10 10.8
170 11 11.8
Figure 5.8
Fig. 5.8 shows how the signal-to-noise ratio C/N in the detector input
influences the levels of the noise signal N and wanted signal Sl in the
detector output. The shift of the threshold is fairly small when modulation is
applied. The threshold is almost entirely caused by the rapid rise in the
noise level due to phase-jump noise (clicks) when the signal-to-noise ratio
in the detector input falls below ρB = 8-10 dB. The level of the wanted signal
starts to fall when the signal-to-noise ratio gets below ρB ≈ 5 dB
This is of no importance so long as the limiter stage and the detector have
an adequate bandwidth for the wideband sum signal. The short-term
average of the instantaneous frequency of the total signal is determined by
the instantaneous frequency of that one of the two signals that has the
highest level. If the amplitude of the wanted signal is highest, its
instantaneous frequency will largely determine the output signal. The
co-channel interference gives rise to a relatively small unwanted deviation
about this frequency. If the bandwidth is insufficient, on the other hand, with
the result that the peaks of the frequency deviation are cut off, this will affect
the short term mean value of the frequency. Much stronger co-channel
interference will occur at the detector output because the capture will be
ineffective. See also Fig. 5.9, which shows the conditions above the
threshold, and Fig. 5.10, which relates to the threshold characteristics.
Created by EBCCW 00:06
Sd C ⋅ 6 Wme
If No is replaced by Ni: ( S N )d = = me2 ⋅ = 6 me3 ⋅ ρi
Nd I ⋅W
Modulation gain when co-channel interference present: GM = 6 me3
More-accurate analysis gives GM = 4me3
Output signal from detector
Sd
= 4 ⋅ me3 ⋅ ρi
Nd
l
na
sig
d
te
or
ist
D
C
ρi =
I
Figure 5.9
ρi <> 1 Ai e jω i t
ω zt ωi t
Origin
Ar- Ai A
i
a. ρi < 1 Mean value of instantaneous
dφ frequency determined by
ωz = interference
φz dt
4π
ω z = ωi
ωi
3π ρi < 1 ωi t
π 2π 3π 4π
2π
b. ρi > 1φ Mean value of instantaneous
−φ = ω it
frequency determined by
π
wanted signal
ρi > 1
ω z = 0 (mean value)
0 ωi t dφ
ωz =
π 2π 3π 4π dt
(ω )
z max
>> ω i om
If I ≈ C 0
π 2π 3π 4π
ωi t
Figure 5.10
Created by EBCCW 00:06
Phase-locked detector
HF Mixer IF A
LO
Phase Loop
A Limiter detector filter
θ
ϕ Bn W
VCO
π
2
π π θ−ϕ
− 2π
2
Stable static
operating point
Figure 5.11
The noise spectrum in the input to the detector is transposed down to the
baseband and, after filtering in the loop filter, is fed to the VCO for the
reference signal. The VCO is therefore modulated with Gaussian noise and
the resuling noise modulation is added to the VCO modulation by the
detected wanted signal. The instantaneous phase difference between the
wanted input signal and the reference signal therefore comprises two
contributions:
a The phase error due to the finite amount of feedback
of the angle modulation of the wanted signal.
b The noise modulation by the VCO.
a sudden phase shift of 2π will occur. This will give rise to a click of the same
type as that produced by origin circling in a noncoherent detector. The
position of the threshold can however be moved to a lower signal-to-noise
ratio. Accordingly, the phase-locked detector is also known as an extended
threshold detector.
As regards the design for adequate loop stability, the same considerations
apply as those discussed in RT1B for phase locked oscillators. To obtain an
acceptable phase margin, the loop gain must decrease relatively slowly
with increasing frequency. This means that the loop bandwidth, Bn, will be
much wider than the width of the baseband, W, if high modulation feedback
is required. The reduction in the noise bandwidth that determines the
threshold will be B/Bn, where B is the IF bandwidth. An upper, unattainable
limit value for the threshold reduction is given by B/W. It follows that the
threshold extension by means of a phase-locked FM detector can be
greater with wideband modulation (a high modulation index).
G
H= : Förstärkning för motkopplade slingan
1+ G
a. Bodediagram för öppna π
slingan |G(f)| Fashopp om |θ−ϕ| > 2
Fassfelet (θ−ϕ) härrör från brus-
modulering av ϕ och återstående
(log- signalmodulering på skillnads-
skala)
signalen (θ−ϕ). Val av B n blir
12 dB/oktav en kompromiss.
|G(f m )|
Motkoppling av
modulationsspektrum
( )
2
G ( fm ) ≈ fn
fm
(fm : mittfrekvens
fm fn för basband)
f
log-
W skala
(basband)
b. Slingans överföringsfunktion
|H(f)|
Brusbandbredd:
∞ 2
≈1
Bn ≈ ∫ H ( jω ) df
o
(Bn bestämmer
Bn ≈ 3 fn brusmodulering av ϕ)
f
fn
Figure 5.12
Threshold extension
4 10 14 dB
TH
40
2 6 8 dB
TH
1 dB TH 1 4 2 dB
30 TH
TH
TH
TH C
S
N = + GM dB
20 No W
TH Above the threshold:
2
3 ∆F
GM =
TH
TH 2 W
TH C
No W dB
10
5 10 15 20 25 30
fo
f f
B=2(∆F+W) W
Figure 5.13