--
----X.
Prof. E. J.
A.
FREQUENCY
Baghdady
a
---
MODULATION
Prof. J. B. Wiesner
K. P. Luke
F.
I. Sheftman
AN OSCILLATING LIMITER FOR THE SUPPRESSION OF INTERFERENCE
IN FM RECEIVERS
A study has been made of the locking and interference suppression characteristics
of an oscillating limiter (1, 2, 3).
A summary of the results follows.
It has been shown that interaction between two signals, Escos pt and aEscos(p+r)t,
causes a frequency disturbance in the form of frequency spikes, whose amplitudes are
a function of the interference ratio a and the frequency difference r between the two
signals.
When positive bandlimited
feedback is applied around the limiter,
the interference ratio a and,
sequently,
con-
the amplitudes of the
frequency spikes are reduced, as
predicted by Baghdady (1, 2, 3) and
(a)
K,=O
r=-60 kc
=0.85
(b)
Ks= 2.0
shown in Fig. X- 1.
It was also predicted (1,2,3)
that positive bandlimited feedback
decreased the limiting threshold of
a practical limiter through an increase
in the effective signal amplitude seen
(C)
Ks
=
0
r=-100kc
o= 0.8
by this limiter. Figure X-2 shows
typical cases in which the effective
(d)
Ks= 1.2
signal amplitude seen by the limiter
is increased by positive feedback,
and Fig. X-3 shows a corresponding
improvement of the output waveforms.
Figure X-4 shows the famous "noise(e)
Ks= 0
(f)
r=-14kc
= 0.75
a 0.75
=1.
2
Fig. X-1. Effect of positive feedback around
the limiter on the frequency spikes
that result from interaction between the two signals, Escos pt
and aEscos(p+r)t. K s is the feedback factor.
squelching" effect of an oscillating
limiter in the absence of an input
signal.
Interference generally has a detrimental effect on the locking range and
locking threshold of an oscillating limiter.
Figures X-5 and X-6 show plots
of the locking range
and locking
threshold of an oscillating limiter as a function of the interference ratio a.
plicity, and to illustrate the most unfavorable condition of interference,
For sim-
an unmodulated
=
Es= 600 pv
E s =300 y
(a) Ks= 0
Es= 600 v
Es 300y
(g) K= 0
(b) Ks= 10.0
(a) K6s 0
(h) K:= 2.6
(b) Ks= 10.0
=
Es 700 pv
Es=400Lv
(c)
KS= 0
(d) KS= 7.0
Es= 500 pv
(j) Ks= 2.0
Es= 1000pv
(e) Ks= 0
Fig. X-2.
(f)
Ks= 3.5
(c)
(I Ks=
4
Effect of positive feedback around the
limiter on the input-signal levels to the
limiter for different input-signal levels
to the i-f amplifier.
(d)
Kg=7.0
(a)
(i)
(j)
Ks=O
Ks=2.0
Es: IOOOIv
(e) Ks=,
.0
Fig. X-3.
3
Fig. X-4.
Ks=0
Es=500pv
(k)Ks= 0
(h) Ks- 2.6
Es=700Lv
Es: 400 jv
(i) KS=O
(g) Ks-0
(f) Kz=3.5
(k) K 0
(1)Ks .4.0
Effect of positive feedback around the
limiter on the quality of the output signals for different input-signal levels to
the i-f amplifier. Af = ±20 kc.
Es! 0
a=0
(b)
Noise-squelching effect of an oscillating limiter in the absence
of an input signal: (a) without feedback; (b) with feedback.
200
180
160
140-
K,= 1.0
120
Ks= I
100
*
0
Ks= 2.0
80
60
40
20
I
0
Fig. X-5.
O0
I
I
I
0.2
0.3
0.4
I
I
05
0.6
0.8
0.7
0.9
Locking ra nge versus a characteristics for an oscillating limiter.
kG= KsEs
=
1.0
kG= Ks Es = 3.0
S
0.1
I
0.2
0.3
I
I
I
0.4
0.5
0.6
I
0.7
0.8
I
I
0.9
1.0
a-
Fig. X-6.
Locking threshold versus a characteristics for an oscillating
limiter. Locking range, 120 kc p-p.
(X.
FREQUENCY MODULATION)
100
100
I
0
a
z
50
o
50
0HI-_j
@
1.0
Fig. X-7.
Capture characteristics of receiver.
100
100
0z
0
o
a
50
:
50
(n
o_
o
o
@
r-
Fig. X-8.
Capture characteristics of receiver.
signal at the center of the i-f passband was chosen for the interference.
The capture characteristics of the receiver are plotted in Figs. X-7 and X-8.
remarkable features of these capture curves when feedback is applied are:
(a)
The
the
reduction in total distortion, and (b) the significant increase in the desired signal output
and decrease in total distortion for the values of a that are near unity.
In the meas-
urements that led to the capture curves of Figs. X-7 and X-8, the two signals are at
the center of the i-f passband.
In Fig. X-7 the desired signal is modulated with a devia-
tion of ±5 kc, and the interference is unmodulated; in Fig. X-8 each signal is modulated
(X.
FREQUENCY MODULATION)
with a deviation of ±5 kc. The modulation frequencies are 1 kc for the weaker signal,
and 400 cps for the stronger signal.
K. P. Luke
References
1. E. J. Baghdady, Theory of feedback around the limiter, IRE National Convention
Record, Part 8, 1957, pp. 176-202.
2. E. J. Baghdady, Interference rejection in FM receivers, Sc.D. Thesis, Department
of Electrical Engineering, M. I. T., May 1956.
3.
K. P. Luke, The locking and interference suppression characteristics of an oscillating limiter, S. M. Thesis, Department of Electrical Engineering, M. I. T.,
Sept. 1958.
B.
CAPTURE OF THE WEAKER SIGNAL
A receiver capable of capturing the weaker of two cochannel FM signals has been
constructed and tested in the laboratory. It utilizes standard frequency-conversion
techniques and a fixed trap to suppress the stronger signal without materially distorting
the weaker signal (1).
In the experimental model the weaker signal can be received satisfactorily in the
presence of an interfering signal that is more than twenty times as strong. For voice
modulation, the output is perfectly intelligible and is useful for many applications,
although some noise and cross talk are present. Decreasing the relative strength of
the interfering signal improves the quality of the output but does not quite render it
z
80
so
a
a:
70
60
50
<40
1o0
I01
0.02
0.05
0.1
0.2
0.3
0.5
WEAKER-TO -STRONGER-SIGNAL-AMPLITUDE
Fig. X-9.
2
1
RATIO
3
4
5
10
a
Capture characteristics of conventional and fixed-trap receivers.
All curves are for cochannel signals for a deviation of ±75 kc.
Subscripts refer to strong and weak signals.
FREQUENCY MODULATION)
(X.
suitable for high-fidelity applications.
A quantitative measure of the receiver performance can be obtained by use of single
A plot of the fundamental
tones, instead of voice or music, to modulate each signal.
and of the distortion of each tone versus the input interference ratio a (weaker-tostronger-signal-amplitude ratio) gives the complete capture characteristic of the system.
Curves A
w
and A of Fig. X-9 show the capture characteristics of a conventional FM
s
receiver when both signals are centered in the middle of the passband.
The deviation
is ±75 kc for each signal, and the modulating tones are 400 cps and 1 kc for the strong
and weak signals, respectively. Curves B and B represent the output of the weaks
is now centered at a
w
It will be observed that the capture-transition region
signal receiver for the same input.
= 0. 04 instead of at a = 1, as in the conventional receiver.
The
quality of the weak signal is indicated by the total distortion of the fundamental
(curve Cw), which shows a good output for a = 0. 1.
The best performance of the present experimental receiver is obtained when the
stronger signal is unmodulated or when its modulating frequency is low and the deviation
Curve Dw of Fig. X- 10 gives the fundamental of the weak signal in the presence of an unmodulated stronger signal. The distortion curve E w shows that the outis small.
put is quite good for a z 0.01.
Curves F w and F s indicate the performance for a deviation of ±35 kc for each
50
40
30
20
10
0
0.001
0.002
0.005
001
0.02
0.05
0.1
Fig. X-10.
0. 5
02
WEAKER -TO -STRONGER -SIGNAL- AMPLITUDE
RATIO
a
Limitations on capture performance of the receiver.
Curves
Dw, E w , Hw, I,
H s represent a deviation of ±75 kc; curves F
F ,
and Gw,
All modulating frequencies
a deviation of ±35 kc.
are the same except for curves H
are interchanged.
,
H
,
and I
, in which they
(X.
FREQUENCY MODULATION)
carrier, with the same choice of modulating tones as before.
Curve G
shows the
quality of the output, which is quite good for a z 0. 03. The most serious limitation of
the present system is demonstrated by curves H and H which result from inters
w
changing the modulating tones. The strong signal is now modulated by the 1-kc tone;
and curve I clearly shows that the output is rather distorted for all a up to approximately 0. 5.
All of the foregoing data were taken for the maximum attainable trap attenuation
(approximately 0. 001) and for a constant trap bandwidth of approximately 10 kc between
the frequencies at which transmission is 0. 3 per cent of the maximum. Since ordinary
music and voice waveforms have an average value that is less than 50 per cent of the
peak value, curves Fw, Fs, and G describe the output of the system under typical
broadcast operating conditions.
F. I. Sheftman
References
1.
F. I. Sheftman, A fixed-trap system for capturing the weaker of two co-channel
FM signals, S. M. Thesis, Department of Electrical Engineering, M. I. T., Sept. 1958.