NASA Technical Reports Server (NTRS) 19700026477: An FM optimization study for the Apollo unified S-band telecommunications system

Survival, Water, Medical Field Manuals

Military Manuals

Nasa Technical Reports Server (Ntrs)

Document text

FAC11 


L ^ -o ^ 

n?^ * -o* ] 

!*fe *> JO- t» U> t-i 
fit » «*■< 

8 ®&fsS 


; V? 


7 r 7 i 5 =yWT 7 >' 


L-k 'v; — 

i - 


& -/' ;L „\> 


NATIONAL AERONAUTICS AND SPACE ADMINISTRATION 


MSC INTERNAL NOTE NO. EB-R-68-11 




18 OCTOBER 1968 




AN FM OPT I /VI I ZAT I ON STUDY 
FOR THE 

APOLLO UNIFIED S-BAND 
TELECOMMUNICATIONS SYSTEM 


• • a s • • «a 


• ■ • ••••« 




By F. J Loch 


. J*- ■»/,■■■ ^ *T ■=• M 

h y/££?-l y y • •> 5 »*•■ V ■•-■■■. I 



* ^ * 


N70 -35 793 


ACCESSION N 


(NASA CR OR TMX OR AD NUMBER) 


MSC FORM 2026A (IUN 66) 


INFORMATION SYSTEMS DIVISION 

MANNED SPACECRAFT CENTER 

HOUSTON, TEXAS 


Ic.* 5 : - , »-•>> V‘ ■"* ^ u < vl> ' . 


(CATEGORY) 


Reproduced by 

•> NATIONAL TECHNICAL 1 
INFORMATION SERVICE 

^ Spnngfiold, Va~ Z2I5h " * * 




Kg 


N70-35793 


AN FM OPTIMIZATION STUDY FOR THE APOLLO UNIFIED S- 
BAND TELECOMMUNICATIONS SYSTEM 

anned Spacecraft Center 
ouston, Texas 



This document has been approved for public release and sale 






MSC INTERNAL NOTE NO. EB-R-68-11 


AN FM OPTIMIZATION STUDY FOR THE 
APOLLO UNIFIED S-BAND TELECOMMUNICATIONS SYSTEM 


Prepared by 


F 

Applied Analysis Section 


Approved by 


$ 3 . 0 . Qwjt 

G . D Arndt 

Head, Applied Analysis Section 



14 * 






R W Moorehead 
Chief, Systems Analysis Branch 


Approved for /^Tj)) /j / 

Distribution f 

P H Vavra 

Chief, Information Systems Division 


NATIONAL AERONAUTICS AND SPACE ADMINISTRATION 
MANNED SPACECRAFT CENTER 
INFORMATION SYSTEMS DIVISION 
SYSTEMS ANALYSIS BRANCH 
18 OCTOBER 1968 



A CKNOWLEDGEMENTS 

The author wishes to thank Mr William M Conrad of the Philco-Ford 
Corporation s Education and Technical Services Division j for his 
valuable contributions to the EM optimization study . Mr Conrad is 
credited with the circuit design of the click detection and elimi- 
nator threshold extension device discussed in this document Mr 
Conrad also made valuable contributions to the test program that 
resulted in the data and photographs presented in this document 


-li- 



TABLE OF CONTENTS 


Section 

1 INTRODUCTION 

1 1 Purpose 

1.2 Theoretical And Experimental Approach 

2 SUMMARY 

3 FM THRESHOLD 

3 1 General ... ... 

3 2 Click Event . . . . 

3.3 FM Demodulator Performance With 

Click Noise 

3 4 Click Symmetry 

4 THRESHOLD EXTENSION TECHNIQUES 

4.1 General . . 

4 2 Click Detection . . . 

4 3 Click Elimination 

4 4 Threshold Extension Device 

5 EXPERIMENTAL RESULTS 


5 

1 

General 

5 

2 

Signal-Plus -Noise Waveform Analysis 

5 

3 

Signal-to-Noise Ratio Tests 

5 

3 1 

CSM FM Modes 1 and 2 . 

5 

3 2 

CSM FM Mode 4 

5 

4 

Word Intelligibility Tests 

5 

5 

Television Picture Quality Analysis 

Appendix 




A BIBLIOGRAPHY 


B ABBREVIATIONS AND SYMBOLS 


Page 


. 1-1 

. 1-1 

2-1 


3-1 

3-5 

. 3-15 
. 3-27 


4-1 

4-1 

4-7 

. 4-10 


5-1 

5-2 

5-11 

5-11 

5-15 

5-21 
. 5-26 


A-l 

B-l 


-in- 



LIST OF FIGURES 


Figure Page 

3-1 Graphical Determination of FM Threshold For 

A Typical Demodulator SNR Transfer Curve . .3-2 

3-2 FM demodulator Output Noise Characteristics . . . 3-3 

3-3 Phasor Representation of A, X, Y, R and N .... 3-6 

3-4 Phasor Representation of Noise, Carrier, and 

Resultant R, Near FM Threshold 3-7 

3-5 Relationship Between ±2ir Phase Excursions and 

Resulting Impulses of Area 2m 3-9 

3-6 First Order Click Waveform 3-10 

3-7 First Order Click Waveform . ... 3-10 

3-8 Second Order Click Waveform . 3-11 

3-9 Second Order Click Waveform 3-11 

3-10 Third Order Click Waveform 3-12 

3-11 Third Order Click Waveform 3-12 

3-12 Fourth Order Click Waveform ... 3-13 

3-13 ' Fourth Order Click Waveform 3-13 

3-14 Effect of Carrier Offset on Click Noise 

Distribution . . 3-28 

3- 15 Output Noise Power Increase vs Center Frequency 

Offset 3-29 

4- 1 Demodulated Signal Plus Noise Waveform With and 

Without Postdetection Filtering (f m = 10 KHz). . 4-2 

4-2 Demodulated Signal Plus Noise Waveform With and 

Without Postdetection Filtering (f m = 10 KHz). . 4-2 

4-3 Demodulated Signal Plus Noise Waveform With and 

Without Postdetection Filtering (fjjj = 50 KHz). . 4-3 

4-4 Demodulated Signal Plus Noise Waveform With and 

Without Postdetection Filtering (f m = 50 KHz). . 4-3 


-IV- 



LIST OF FIGURES (CONT'D) 


Figure Page 

4-5 Demodulated Signal Plus Noise Waveform With and 

Without Postdetection Filtering (f m = 100 KHz) . 4-4 

4-6 Demodulated Signal Plus Noise Waveform With and 

Without Postdetection Filtering (f m = 100 KHz) . 4-4 

4-7 Click Detector Block Diagram 4-6 

4-8 Click Eliminator Block Diagram 4-8 

4- 9 Click Detection and Elimination Threshold 

Extension Device Configuration .... 4-9 

5- 1 Signal Plus Noise Waveform Analysis Test 

Configuration 5-3 

5-2 Signal Plus Noise Waveform Analysis 5-4 

5-3 Signal Plus Noise Waveform Analysis 5-4 

5-4 Signal Plus Noise Waveform Analysis . . 5-5 

5-5 Signal Plus Noise Waveform Analysis 5-5 

5-6 Signal Plus Noise Waveform Analysis 5-6 

5-7 Signal Plus Noise Waveform Analysis 5-6 

5-8 Signal Plus Noise Waveform Analysis 5-7 

5-9 Signal Plus Noise Waveform Analysis 5-7 

5-10 Signal Plus Noise Waveform Analysis 5-8 

5-11 Signal Plus Noise Waveform Analysis . 5-8 

5-12 Signal Plus Noise Waveform Analysis 5-9 

5-13 Signal Plus Noise Waveform Analysis 5-9 

5-14 Test Configuration For CSM Modes 1 and 2 Signal- 

to-Noise Ratio Tests With and Without Threshold 
Extension Device 5-12 

5-15 Output SNR vs Input SNR For 1 1 CSM Playback Voice 

Mode 5-13 


-v- 



LIST OF FIGURES (CONT’D) 


Figure Page 

S-16 Output SNR vs Input SNR For 32.1 CSM Playback Voice 

Mode ... 5-14 

5-17 Output SNR Improvement vs Input SNR For 1 1 and 32 1 

CSM Playback Voice 5-16 

5-18 Configuration For CSM Mode 4 SNR Tests 5-17 

5-19 Output SNR vs Input SNR For CSM FM Mode 4 5-18 

5-20 Output SNR Improvement vs Input SNR For CSM FM 

Mode 4 With Click Elimination . . 5-20 

5-21 Configuration For CSM FM Mode 1 Word Intelligibility 
Tests With and Without Threshold Extension 
Device 5-22 

5-22 Percent Word Intelligibility vs Input SNR For 1 1 

CSM Playback Voice Mode S-23 

5-23 Configuration For CSM FM Mode 2 Word Intelligibility 

• With and Without Threshold Extension Device . 5-24 

5-24 Percent Word Intelligibility vs Input SNR For CSM 

32 1 Playback Voice Mode 5-25 

5-25 Configuration For CSM Mode 4 Television Picture 

Quality Tests 5-27 

5-26 CSM Mode 4 TV Picture Quality Tests 5-28 

5-27 CSM Mode 4 TV Picture Quality Tests 5-28 

5-28 CSM Mode 4 TV Picture Quality Tests 5-29 

5-29 CSM Mode 4 TV Picture Quality Tests 5-29 

5-30 CSM Mode 4 TV Picture Quality Tests 5-30 

5-31 CSM Mode 4 TV Picture Quality Tests 5-30 

5-32 CSM Mode 4 TV Picture Quality Tests 5-31 

5-33 CSM Mode 4 TV Picture Quality Tests 5-31 


-vi- 



LIST OF FIGURES (CONT'D) 


Figure 


5-34 

CSM 

Mode 

4 

TV 

Picture 

5-35 

CSM 

Mode 

4 

TV 

Picture 

5-36 

CSM 

Mode 

4 

TV 

Picture 

5-37 

CSM 

Mode 

4 

TV 

Picture 

5-38 

CSM 

Mode 

4 

TV 

Picture 

5-39 

CSM 

Mode 

4 

TV 

Picture 

5-40 

CSM 

Mode 

4 

TV 

Picture 

5-41 

CSM 

Mode 

4 

TV 

Picture 


Page 

Quality Tests 5-33 

Quality Tests 5-33 

Quality Tests 5-34 

Quality Tests 5-34 

Quality Tests 5-35 

Quality Tests 5-35 

Quality Tests 5-36 

Quality Tests 5-36 


-v 11 - 



LIST OF TABLES 


Table Page 

1- 1 Marginal Apollo Down- Link FM Modes 1-2 

2- 1 Threshold Extension Device Performance Summary. . .2-3 

3- 1 Duration and Amplitude Characteristics of Nth 

Order Click Waveforms 3-14 

5-1 CSM Modes Used For Threshold Extension Device 

Test 5-1 


-Vlll- 



SECTION 1 


INTRODUCTION 


1 1 PURPOSE 

An analysis of the Apollo Unified. S-Band (USB) communications 
system^, performed by the Information Systems Division, Manned 
Spacecraft Center, has indicated that certain down- link FM modes 
exhibit negative circuit margins at lunar distance for worst case 
conditions A negative margin implies that the operational re- 
quirements will not be met for the affected modes The three FM 
modes that were found to have negative margins are listed m 
Table 1-1 with their respective channel characteristics 

As a result of these findings, an intensive study of optimization 
techniques was conducted by ISD/MSC for the purpose of improving 
the performance of the marginal FM modes This optimization study 
resulted m the development of an FM threshold extension device 
that could be used to provide improved performance for the Apollo 
Manned Spaceflight Network (MSFN) carrier frequency demodulator 
By extending the threshold operation of the MSFN demodulator, it 
is possible to obtain better circuit margins for the affected FM 
modes 

The purpose of this document is to present the results of the FM 
optimization study which resulted m the development of the thresh- 
old extension device This document also presents and evaluates 
test data representing the performance of the device when it is 
used m conjunction with an MSFN receiver 

1 2 THEORETICAL AND EXPERIMENTAL APPROACH 

An analysis of the FM threshold phenomena based on the earlier 
work of S. 0. Rice^ is presented. The results of this analysis 
show that the presence of impulse noise m the output of a de- 
modulator is a primary factor causing threshold m an FM system 


1 

C. Dawson, A Performance Analysts of the Apollo Unified S-Band 
Communications System for a Typical Lunar Mission (MSC Internal 
Note MSC-EB-R-67-1, August 1967) 

2 

S 0 Rice, "Noise in FM Receivers," Proceedings 3 Symposium on 
Time Series Analyses 3 ed M Rosenblatt (New York, 1963) 


1-1 



MODE 

CSM MODE 1 
CSM MODE 2 


SERVICE(S) 


SPECIFIED 
CARRIER FREQUENCY 
DEVIATION (Af) (KHZ) 


WORST NOMINAL BEST 

1 1 PLAYBACK 60 100 120 

OF VOICE 
1 1 PLAYBACK 

CSM 51 2 KBPS 510 600 69 0 

TLM 

32 1 PLAYBACK 60 100 120 

OF VOICE 
32 1 PLAYBACK 

CSM 1 6 KBPS 510 600 690 

TLM 

0 9 10 11 


CSM MODE 4 


TV 


WORST CASE 
MODE-AS-A-WHOLE 
CIRCUIT MARGIN (DB) 
215,000 NM 

MODULATION 

FREQUENCY 


300-3000 HZ 

-0 7 

1 024 MHZ 
(subcarrier:) 


9600-96,000 HZ 

-0 7 

1 024 MHZ 
(SUBCARRIER) 


-0 7 


0-500 KHZ 













1 2 THEORETICAL AND EXPERIMENTAL APPROACH (Cont'd) 


The threshold extension technique is based on the elimination 
of noise impulses m the demodulator output An amplitude de- 
tection technique is used to obtain information for the noise 
elimination circuits that operate on a delayed output of the 
demodulator. 


1-3 



SECTION 2 


SUMMARY 


The performance of an FM demodulator can be defined m terms of 
the threshold phenomena, which determines the minimum acceptable 
input signal- to-noise ratio for the system The presence of high 
amplitude noise impulses (click noise) m the demodulator output 
is a primary factor contributing to the occurrence of threshold 
m an FM system Certain distinguishing characteristics of click 
noise provide a basis for practical techniques that can be im- 
plemented at the output of an FM demodulator to improve its thresh- 
old performance 

A threshold extension device has been developed that operates on 
the principle of click detection and elimination The device was 
designed to be compatible with the MSFN ground station receiver 
and to improve the performance of the Apollo Unified S-Band 
Communications System 

Previous work m the area of FM demodulator improvement has been 
concentrated on the design of specialized demodulation schemes that 
provide optimum performance for one set of channel parameters 
The Apollo down- link FM modes, however, use several channels having 
different characteristics m terms of required channel bandwidth, 
maximum frequency deviation, and range of modulation frequencies 

The click noise eliminator offers an advantage m flexibility 
since it is not a demodulator but, rather, a device which can be 
implemented at the output of any FM discriminator to provide im- 
proved system performance The presence of unsymmetrical click 
noise m the demodulator output, resulting from offsets m carrier 
frequency, does not affect the performance of the device 

Photographs of demodulated Apollo television pictures comparing 
the unprocessed demodulator output with the click-eliminated 
output are presented to illustrate the effectiveness of the 
threshold extension technique 


2-1 



v2 SUMMARY (Cont'd) 


Additional data obtained from tests of the Apollo playback voice 
inodes reflect the output ,signal- to -noise ratio improvement m the 
form of increased voice intelligibility scores Improvements of 
15 percent to 20 percent m word intelligibility scores are obtained 
when the device is used at the output of the Apollo MSFN receiver. 

Experimental results using the Apollo television and playback voice 
modes have shown that the click elimination device provides an 
improvement m output signal-to-noise ratio of 6 dB to 10 dB 
This improvement m output signal-to-noise ratio corresponds to 
a 1.5 dB to a 2.0 dB extension of threshold 

The performance characteristics of the threshold extension device 
are summarized m Table 2-1. 

The values listed m Table 2-1 are based on the improvement ob- 
tained when the device was used at the output of an Apollo MSFN 
phase lock loop (PLL) demodulator. The threshold extension obtained 
was m addition to the normally expected improvement of a PLL 
demodulator over a standard FM discriminator. The device was also 
tested with a frequency modulation feedback (FMFB) discriminator, 
and similar values of threshold extension were obtained Therefore, 
the threshold extension technique is capable of improving the 
performance of any FM demodulation scheme that is degraded by the 
presence of click noise m the output 

A breadboard version of the threshold extension device was used 
for the tests referenced m this document Work is presently under- 
way to complete a refined click detection and elimination device 
that uses integrated circuits to improve performance. It is ex- 
pected that the finalized version of the threshold extension device 
will provide additional SNR improvement 

The following steps must be followed to insure optimum performance 
of the click elimination device 

A. The threshold level of the click detector must be set as 

close as possible to the modulation peaks to insure maximum 
click detection efficiency. 

B The click-eliminator cutoff time must equal the click dura- 
tion The click duration is determined by the bandwidth of 
the preclick-detection filter. An optimum low pass band- 
width of 2 MHz was used for the Apollo modes. 


2-2 



TABLE 2-1 


THRESHOLD EXTENSION DEVICE PERFORMANCE SUMMARY 


CSM MODE 

SERVICE 

MAXIMUM 
OUTPUT SNR 
IMPROVEMENT 
CDB) 

THRESHOLD 

EXTENSION 

CDB) 

INPUT SNR 
FOR MAXIMUM 
IMPROVEMENT 

EFFECTIVE 
INPUT SNR 
OPERATING 
RANGE 

COMMENTS 

1 

1 1 PLAYBACK 
VOICE 

10 DB 

2 0 DB 

3.3 DB 

0 TO 10 DB 

MAXIMUM WORD IN- 
TELLIGIBILITY 
IMPROVEMENT OF 
25% WAS OBTAINED 

2 

32 1 PLAYBACK 
VOICE 

7 5 db 

1 5 db 

4.0 DB 

0 TO 10 DB 

MAXIMUM WORD IN- 
TELLIGIBILITY 
IMPROVEMENT OF 4% 

WAS LIMITED BY IN" 
HERENT RECORDING- 
PLAYBACK DISTORTION. 

4 

TELEVISION 

7 5 db 

1 5 db 

TO 

2 0 DB 

3 0 DB 

0 TO 10 DB 

THRESHOLD EXTENSION 
IS OBSERVED BY COM- 
PARING OUTPUT VIDEO 
PICTURES WITH AND 
WITHOUT CLICK 
ELIMINATION 




2 SUMMARY (Cont'd) 


C The beginning of the click-eliminator cutoff must coincide 
exactly with the beginning of the click in the demodulator 
output 

D The recovery time for the click detection and elimination 
circuits must be made as small as possible to insure effi- 
cient operation at high click rates 


2-4 



SECTION 3 


FM THRESHOLD 


3 1 GENERAL 

By definition, the criteria for evaluating the performance of an 
FM demodulator is based on the ability of the device to provide 
a linear relationship between the output and input signal-to-noise 
ratios The useful operating range of all FM demodulators, however, 
is limited by the fact that this relationship, or transfer charac- 
teristic, becomes non-linear below a certain value of input signal- 
to-noise ratios This value of input SNR is called the "point of 
threshold" for the demodulator 

Since it is difficult m most cases to determine the exact value 
of input SNR that divides the linear and non- linear regions of the 
performance curve, it is necessary to define a reasonable criteria 
for determining the threshold point 

One accepted definition of FM threshold is based on the graphical 
determination of the specific input SNR value whose corresponding 
output SNR occurs exactly 1 dB below an extension of the linear 
portion of the transfer curve This method of defining FM thresh- 
old is illustrated m Figure 3-1 and will be used consistently 
throughout this document 

The occurrence of threshold m an FM system can also be defined in 
terms of the demodulator output noise characteristics In general, 
when the demodulator is operating with values of input SNR greater 
than 10 dB, the output noise spectrum is parabolic, and the amplitude 
distribution is Gaussian As the input SNR decreases ,. the output noise 
voltage is punctuated with occasional high amplitude noise spikes 
having either positive or negative polarity These noise impulses 
become more frequent as the input SNR is reduced to values below 
10 dB For the region of input SNR's between 0 dB and 10 dB, the 
output noise power increases at such a rate that the slope of the 
SNR transfer curve becomes much more severe than the slope of the 
linear portion This implies that a small change m input SNR 
results m a relatively large change m output SNR as illustrated 
m Figure 3-1 The difference m the relative contribution of 
Gaussian noise and click noise can be seen by noting their indi- 
vidual spectral characteristics m Figure 3-2 


3-1 



OUTPUT SNR (dB) 



6 8 10 12 
INPUT SNR (dB) 


Figure 5-1 Graphical Determination of FM Threshold for 
a Typical Demodulator SNR Transfer Drive 


GC 10 177 (A) 10 


3-2 


G(f) 



(a) BASEBAND NOISE SPECTRUM ABOVE THRESHOLD 

(b) CLICK NOISE SPECTRUM BELOW THRESHOLD 


GC 101771 A; W 


Figure 3-2 FM Demodulator Output 
Noise Characteristics 


3-3 



3.1 GENERAL (Cont'd) 


The energy contained in the individual noise spikes (called click 
noise) at the output of an FM demodulator contributes significantly 
to the total output noise power. In addition, the impulsive nature 
of click noise causes it to be much more degrading to the demodulated 
signal than the Gaussian output noise. Although click noise is not 
the only phenomena which causes FM threshold, it is definitely a 
primary factor contributing to the occurrence of threshold in an 
FM system. 


3.2 CLICK EVENT 


Consider an unmodulated signal at the input of an ideal FM 
discriminator where the carrier is represented as G(t). 


G(t) = A cos w c t ( 1 ) 

The input noise to the discriminator can be represented in 
quadrature form by the following expression: 


N(t) = X(t) cos m^t + Y(t) sin w^t (2) 

where X(t) and Y(t) are independent random variables. Therefore, 
the total input carrier plus noise is 


G (t) + N (t) = [a + X (t ) ] cos to t + Y(t) sin a> t 

c c 


(3) 


which can also be expressed as follows 


G (t) + N (t) = R cos tu t + <J>(t)] 


(4) 


where R is defined as the magnitude of the resultant carrier plus 
noise waveform, w c is the carrier frequency, and <{> is the phase 
of the resultant R with respect to the carrier. The FM demodulator 
detects the frequency of the signal by differentiating the phase 
of the received signal plus noise. The phase of the signal -plus-noise 
waveform can be obtained from equation (3) . 




<f> 


tan 


Y 

A+X 


(5) 


3-5 


3.2 CLICK EVENT (Cont'd) 

For high signal- to-noise ratios, we can assume that A >> X. 
Therefore, 

- 1 Y Y 

<P = tan - = - (6) 

The following phasor relationship exists between A, X, Y, R, N 
and <J>, as shown in Figure 3-3. 



Figure 3-3 Phasor Representation of A, X, Y, R, and N 


X and Y are Gaussian distributed random variables whose fluctuation 
causes the angle <j> to change accordingly. 

The click event takes place at the input of the FM discriminator 
as an interaction between the randomly varying noise envelope and 
the carrier amplitude that results in a sudden 2ir phase excursion 
of the carrier-plus -noise vectoral resultant, R. 

Both plus or minus 2 t r phase excursions can occur with equal proba- 
bility, which results in a corresponding positive or negative noise 
spike in the demodulator output. 

Using the phasor representation of Figure 3-4, the click event 
can be defined in terms of a phase excursion of ±2tt radians by 
the angle <f> . The probability of such an excursion increases as 
the input SNR to the discriminator decreases below a value of 
10 dB. This figure shows that, for low values of input SNR, a 
small change in the phase angle between the noise and carrier wave- 
forms can result in a relatively large change in the resultant phase 
angle . 


3-6 



Figure 3-4 Phasor Representation of Noise, 
Carrier, and Resultant R, Near 
FM Threshold 


L D JO 1 79 ( A ' 


3.2 CLICK EVENT (Cont'd) 


Since the output of an ideal discriminator is proportional to 
d<J>/dt, the 2 tt phase excursion causes a noise spike of area 2ir 
to occur at the output as shown in Figure 3-5. 

It is also possible for the angle <f> to experience a phase excursion 
of 4tt, 6tt, or even 8ir radians. For phase steps exceeding 2n radians, 
the resulting output noise spike will have a proportionally greater 
duration prior to postdetection filtering. Figures 3-6 through 
3-13 show the click waveform before and after postdetection filter- 
ing. Several photographs are provided to illustrate each of the 
higher order clicks since each unfiltered waveform is unique. Fig- 
ures 3-6 and 3-7 represent clicks resulting from a ±2ir phase ex- 
cursion. These noise spikes will be referred to as first order 
clicks since they are a result of the minimum click-producing phase 
excursions. Some higher order clicks are also present in these 
figures, but the clicks referenced in the figure titles are shown 
in the extreme left portion of the photographs. Figures 3-8 and 
3-9 represent second order clicks (4tt radians) . The duration of 
the second order clicks is approximately twice that of the first 
order clicks at the unfiltered output of the demodulator. The 
noise contribution of the second order clicks is correspondingly 
greater than that of the first order clicks. 

The difference between first and second order clicks can easily 
be distinguished at the output of the demodulator postdetection 
filter by observing the relative amplitude of the spikes. The cut- 
off frequency of the low pass filter determines the duration of 
the output noise spikes so that both first and second order clicks 
have the same duration at the filter output. Therefore, the dif- 
ference in duration between the first and second order clicks at 
the unfiltered output is translated to an amplitude difference in 
the filtered output. Table 3-1 summarizes the relative amplitude 
and duration relationship between the different orders of filtered 
and unfiltered click waveforms. 

Figures 3-10 through 3-13 represent third and fourth order click 
waveforms at the demodulator output. These higher order clicks 
occur only for very low values of input SNR, and for practical 
considerations, their contribution to the total output noise power 
can be neglected. In most cases the degradation of the demodulator 
output below threshold will primarily be caused by the occurrence 
of first order clicks. 


3-8 





GC10177CA) 9 

Figure 3-5 Relationship Between ±2tt Phase Excursions 
and Resulting Impulses of Area 2tt 


3-9 




Figure 3-6 First Order Click Waveform 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 ^SEC/CM 



Figure 3-7 First Order Click Waveform 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 *iSEC/CM 


G C 10 177 (A) — 


3-10 







Figure 3-8 Second Order Click Waveform 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 m SEC/CM 



Figure 3-9 Second Order Click Waveform 
UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 /iSEC/CM 


GC 10 177 (A) — 19 


3-11 






Figure 3-10 Third Order Click Waveform 

UPPER TRACE:* FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 /xSEC/CM 



Figure 3-11 Third Order Click Waveform 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 fiSEC/CM 


GC10t77(A)- 18 


3-12 







Figure 3-12 Fourth Order Click Waveform 
UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 ^SEC/CM 



Figure 3-13 Fourth Order Click Waveform 
UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 1 m SEC/CM 


GC 10 177 (A) — IG 


3-13 



TABLE 3-1 


DURATION AND AMPLITUDE 
CHARACTERISTICS OF Nth ORDER CLICK V/AVEFORMS 


CLICK ORDER 

UNFILTERED 
DURATION 
CIO- 6 SEC) 

FILTERED 

RELATIVE AMPLITUDE 
CCM) 

FIRST 

0 5 

0 5 

SECOND 

1 0 

1 0 

THIRD 

1 5 

1 5 

FOURTH 

2 0 

2 0 


3-14 










3.3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE 


The click event was described m Paragraph 32m terms of the 
interaction between an unmodulated carrier and a quadrature-carrier 
representation of the input narrow-band noise However, m order 
to discuss the effect of click noise on the performance of an FM 
demodulator, it is necessary to consider a modulated input signal. 

An FM signal can be represented by S(t) m the following form 


S(t) = A cos [w c t + <J>(t) ] , 


( 7 ) 


where A is a constant representing the amplitude of the signal, 
o) c = 2 it f c , which is the carrier angular frequency, and <j>(t) is 
the modulation term 

The modulation information can be utilized to determine the in- 
stantaneous frequency of the carrier m the following manner 

= io c + ^ = w c + Atom(t) (8) 

where is the instantaneous frequency of the carrier, to c is the 
carrier rest frequency when modulation is not present, and Am is 
the carrier frequency deviation, and m(t) is the modulating signal 
waveform 

The magnitude of the frequency deviation Am is determined by the 
amplitude of the modulating waveform m(t) and by the sensitivity 
of the transmitter modulator 

For sinusoidal modulation, equation (8) can be written as follows 


to = to + Atocos to t (9) 

ic m 


3-15 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 


Since 


a + d<£ 

Aw cos to t = — 
m dt 


Then 


<f> = / Aw cos w t dt 

J m 


Therefore, equation (7) can be rewritten as 


S(t) = A cos [v + / Aw cos w^t dtj 

F Aw 1 

w t + — sm w t 
c w m 

L m -1 


= A cos 


Aw 


Cio) 


(ii) 


By definition, — is designated as the modulation index, £ 


w 


m 


The final form of equation (1) is 


S(t) = A cos j^w^t + g sin w m tj (12) 

The input SNR to the demodulator is determined by considering a 
rectangular predetection bandpass filter having a bandwidth 


BW Ip - 2(Af * f m ) 


(13) 


where Af is the peak frequency deviation and f m is the 
modulating frequency 

The input signal power can be found from equation (12) to be 



(14) 


3-16 



3.3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 
and the input noise power is 


N = KT BW Ip , (15) 

-23 -1 

where K is Boltzman's constant (1 38 X 10 Watts-sec °K ), 

T is the effective system temperature m degrees Kelvin, and BW^p 
is defined by equation (13) 

Therefore, the input SNR can be found by dividing equation (14) by 
equation (15) 


S i /N i 2KT BW 


IF 


(16) 


The output signal power at high levels of input SNR is independent 
of the input carrier amplitude A and is a function of the frequency 
deviation Af 


Aw 2 _ Tr (2-rrAf) 2 

d ~ ' h 


(17) 


The output noise power of the discriminator for the unmodulated case 
can be determined by considering the noise disturbance about the 
carrier frequency w c The input noise can be represented m quadrature 
form by the following expression 


N = X(t) cos (0 t = Y(t) sm w t (18) 
0)1 c c 

where X(t) and Y(t) represent the in-phase and quadrature components 
of noise and are defined by a summation of Gaussian random variables 
having zero mean. 


X(t) = Y(t) = 0 


(19) 


and 


X 2 (t) = Y 2 (t) = N 2 

v v (DC 


( 20 ) 


3-17 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 

3 

By definition , 

- 0 ) ) t + y sin(nu - w )t1 (21) 

c 'n o c J 

- w )t - y cos(nio - w )tl (22) 

c. n o c J 

where x n and y n are Gaussian random variables with zero mean 

The total expression for the instantaneous carrier plus noise can be 

written as follows 

Total input voltage = A cos w t + X(t) cos co t + Y(t) sm co t 

c c c 



= [A+X(t)] cos co t + Y(t) sm <o t (23) 

c c 

The phase error caused by the noise disturbance about the carrier 
frequency co is represented by <|> 

C 6 


<i> 


e 


tan 


Y (t) „ Y(t) 

A+X(t) A 


(24) 


This expression is similar to equation (6) for the angle <J> 

Similarly, the frequency error, or noise disturbance about the 
carrier frequency to , at the output of the discriminator is repre- 
sented by 

' e 


e e 


~ Y(t) 


Substituting equation (22) for 


co 


- “5 It S [* n 

n=l L 


Y(t) into equation (25), we get 

- co )t - y cos (nto - to )t 
c J ; n v o c l 


(25) 


(26) 


John J Downing, Modulation Systems and Noise 
1964), p . 52 


(Englewood Cliffs, 


3-18 



3.3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 


n-1 


x (nco - co )cos (n<o -w_)t+y (no) -a.)sm(nw -w_)t (27) 

o c o c' ' n o c J o c ' ■ 


2tr 

« = - 7 — 

e A 


00 

[x n Cn£ o -f c ) cos(nw o -t») c )t+y n Cnf 0 -f c )sm(nw o -a) )tj (28) 


n=l L 


The average output noise power is proportional to a) 


w 2 = ~ V[x 2 (nf -£ ) Z cos 2 (nw -w )t+y 2 (nf -£ ) 2 sm 2 (nco -u ) t 
e . 2 Z— . L n o c v oc n v oc o c 

A n =i 


o c 'no c 


+ 2x y (nf -f ) L cos (nw -w )t sm (nw -w t 
n' n o c o c o c 


“ 


(29) 


Since the mean value of the cross product term, 

2 

2x y (nf -£ ) cos fnu -w )t sm (nw -w )t, 
n 7 n o c o c ^ o c 7 7 


is zero 


0) 


2 4tt 


2 2 


/ (n£ -£ ) 2 x 2 cos 2 (nw -w )t+y ^sm (nto -w )t 

e . 2 L-J L o c 7 n o c 7 7 n v oc 

A n =l L 


(30) 


but, x = y = 0 
n n 


(31) 


2 2 KT 

anci n / n T 


(32) 


Where KT = equivalent noise spectral density and r is the period of 
noise under consideration 


3-19 



3.3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 
Therefore , 


o > e 2 = — “T^ n£ o" £ c ' 12 [ cos2 (nw o -u) c ]t+sin 2 Cnw o -u c )t J 

Since {"cos (nw -w )t+sm 2 (nu) -co )tl = 1 we have, 

L o c o c J ’ 


7- £ E 

A n=l 


By making the period t relatively long, the summation can be 
changed to an integral m equation (34) . 


f +f 

2 r c m 

co 2 = ~ I KT(f -f J 2 df 

e A 2 J o c o 


f -f 
c m 

where f is the maximum modulating frequency (the cutoff 

frequency of the postdetection low pass filter, BW ) 

o 


2 „ 2 
w = 4 tt 
e 


(s) 


(f -£ V 

u o cr 


f = f +f 
o cm 


f » f -f 
o cm 


f +f -f 
c m c 


f -f -f 
c m c 


f 3 +f 3 
m m 


87r 2 fm 3 KT 


3-20 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont‘d) 
Finally, we get 


N 

o 


K n w 
D e 


2KT (£ ) 3 4tt 2 K^ 
m' D 


3A 


( 37 ) 


The output SNR can be found by dividing equation (17) by equation 
(37). 


S /N 
o o 


3A f 2 A 2 


4KTf 


m 


(38) 


The output SNR can be expressed m terms of the input SNR by com- 
bining equation (16) and equation (38) 


5 /N = / S /N \ 
o o 1 l 1/ 


3Af 2 BW 


IF 


2f 


m 


(39) 


By assuming that the cutoff frequency of the postdetection filter 
is the same as the maximum modulating frequency f m we can rewrite 
equation (39) as follox\rs 




3Af 2 BW Ip 

2BW 3 
O 


(40) 


where BW is the bandwidth of the postdetection low pass filter 


3-21 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 
Remembering that the modulation index 3 is defined as 


_ Aw 2-?rAf _ Af 

w 2irf BW 

m o 


we can express equation (39) another way 


S /N 
o 


o 



3B 2 BW 

2BW 

o 


IF 


(41) 


Equations (39), (40), and (41) express the SNR transfer character- 
istic of an FM discriminator operating m the linear region above 
threshold This region corresponds to the portion of the curve 
(see Figure 3-1) for values of input SNR greater than 10 dB. 

The output noise represented by equation (37) has a Gaussian 
distribution and a parabolic spectrum This noise determines the 
behavior of the SNR transfer curve m the linear region above 
threshold The occurrence of click noise m the output, however, 
causes the transfer curve to deviate from its linearity, and the 
FM threshold effect results Therefore, equations (39) , (40) and 
(41) are valid only at values of input SNR that are large enough 
(generally > 10 dB) such that noise clicks do not appear m the 
output 

In order to describe the performance of an FM discriminator for 
values of input SNR below 10 dB , it is necessary to consider the 
addition of click noise to the Gaussian component m the output 

The noise power contribution of click noise increases as the 
number of clicks per second increases 


3-22 



3.3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 

Following the analysis of S 0 Rice4, the output noise power 
below threshold can be found by adding the contributions of the 
Gaussian and click noise as shown below 


N" = N + N 
o o c 


( 42 ) 


where Nq is the total output noise power, N Q is the Gaussian noise 
contribution, and is the click noise contribution 

Rearranging equation (39), we get 


N 


o 


2S f 3 

o m 

(y Ni )3Af 2 BW IF 


(43) 


Substituting equation (17) into equation (43), we have 


K 47r 2 Af 2 f 3 
D m 

° (V» x )3tf 2B »i F 


„ , 2_ 3 
K n 4ir f 
D m 



(44) 


The noise power contribution of click noise N has been found by 
Rice to be 8^ 2 (N + + N')f m c 


• N C ■ K D 8,r2(N+ + 


(45) 


+ — 

where N and N represent the number of positive and negative clicks 
per second, respectively 


4 


Rice, 


PP 


375-424 


3-23 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 

Rice further shows that the expression for the click rate N+ is 
given by 


N 


+ 



( 


1-erf 



( 46 ) 


where r is defined as the radius of gyration of the power spectrum 
w e (f) about its axis of symmetry, f = £ c 


r 


_1 

2ir 


(27 j) 2 J (f-f c ) 2 to(f )df 
*_0 


1/2 


/ 


0)(f )df 


(47) 


r 


27T 


tt 2 0) BW_ 3 
o IF 

3w BW__ 
o IF 


1/2 


BW 


IF 


2\/3 


(48) 


Substituting equation (48) into equation (46) we obtain 


BW t „ / , \ 

N + = ( 1-erf Vs /N ] (49) 

4vr \ 1 x ; 


The click rate given by equation (49) is valid only for an un- 
modulated carrier Assuming that the number of positive and 
negative clicks are equal, the_n equation (45) can be written as 


N = K T ,167r 2 (N + )f 
c D m 


(50) 


3-24 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont ' d) 
Substituting equation (49) into equation (50) we get 


K 4it BW / \ 

N = — — [ l-er£ VS /N If 

c vr \ 1 1 / m 


Therefore, the total average output noise power can be written as 

4 ' n ' 2 f ^K-, K f 4 tt 2 BW / , \ 

■ V» e ■ mZfty ( l -^) ( 


vr 


From equation (17) we get. 


K_(2irAf) 

S = -J 4 — 

o 2 


Dividing equation (53) by equation (52), we obtain the following 
expression for the SNR transfer characteristic of an FM discriminator 


S /N" = 
o o 


4 *V r, 3 bw if 2 ( : 

3BW if /VM vr f 2 


27T 2 Af 2 

3BW if 2 ( S x /N x 


1- erf VS /N 

i l 


Substituting BW^ in equation (54) for f and rearranging, we obtain 


Af BW., 


(v n i) 


S /N ^ = L _ J 

00 Vr/BW IF \ 2 /s i /N i Vl-e r fVs7N i \ 


Equation (55) can be expressed m the following form by substituting 
$ for Af. 


3-25 



3 3 FM DEMODULATOR PERFORMANCE WITH CLICK NOISE (Cont'd) 


( 56 ) 


Plotting equation (56) gives results similar to the curve shown 
in Figure 3-1 The deviation from the linear portion of the curve 
is due to the click noise contribution as expressed by equation (51) 
For values of input SNR greater than 10 dB, the click noise term 
becomes negligible, and equation (56) reduces to the form of equa- 
tion (41) . 

It should be noted that the click noise term expressed by equation 
(51)’ is valid only for an unmodulated carrier at the input to the 
discriminator For a modulated signal, the click rate increases 
substantially so that the portion of the curve below threshold in 
Figure 3-1 exhibits a more severe slope 

The previous analysis indicates that the threshold effect m an 
FM system is primarily determined by the click rate in the demod- 
ulator output This implies that a substantial extension of thresh- 
old could be accomplished by reducing the click rate at the output 
for a given input SNR Experimental results have verified that, 
indeed, the threshold performance of an FM demodulator can be im- 
proved by reducing click noise The following sections describe 
a threshold extension technique that is based on the detection and 
elimination of click noise m the demodulator output 



3-26 



3 4 CLICK SYMMETRY 


For a properly aligned FM demodulator, both the positive and negative 
noise spikes m the output below threshold are present m equal 
quantity However, certain conditions can exist that cause the 
output noise voltage to be unsymmetrical as represented by the pre- 
dominance of either positive or negative clicks 

The most common cause of unsymmetrical click noise is an input 
carrier that is not centered properly m the passband of the 
demodulator predetection filter. Either a frequency drift m 
the transmitter or an unaligned predetection filter can result m 
offsets that cause unsymmetrical clicks The effect of unsymmetri- 
cal clicks is to increase the output noise power and, hence, degrade 
the performance of the demodulator Figure 3-14 illustrates the 
effect of input frequency offsets on the output noise power, while 
Figure 3-15 represents measured data that shows the output noise 
power increase as a function of carrier offsets 

The problem of unsymmetrical click noise is significant since 
the CSM FM transmitter can experience offsets of ± 500 KHz from 
center frequency which adversely affects the performance of the 
FM channels 

The amount of degradation is a function of frequency offset, input 
SNR, and the shape of the filter response In most cases a degra- 
dation of 1 dB to 2 dB was measured m tests of the CSM modes 
using offsets of - 500 KHz The problem of frequency offsets can 
be solved by implementing Automatic Frequency Control (AFC) in the 
MSFN receiver so that the IF frequency is always centered m the pre- 
detection filter passband It is also necessary to use a predetec- 
tion filter that has a symmetrical frequency response 


3-27 



-28 





MINIMIZED 


OUTPUT NOISE POWER 

INCREASED FROM MINIMUM 


SIGNAL OFFSET ABOVE CENTER 
FREQUENCY IN IF BANDWIDTH 

(C) 



2Af 


t 


OUTPUT NOISE VOLTAGE 

PREDOMINANTLY POSITIVE 
IMPULSES 

OUTPUT NOISE POWER 

INCREASED FROM MINIMUM 


Figure 3-14 Effect of Carrier Offset on Click Noise Distribution 


GCI0I77(A>-E 



OUTPUT NOISE POWER INCREASE (dBm) 



-1.6 -1.4 -1.2 -1.0 -.8 -.6 -.4 -.2 0 +.2 +.4 +.6 +.8 +1.0 +1.2 +1.4 +1.6 

OFFSET FROM f (MHz) 

0 GC 10 1 77(A) — 6 


Figure 3-15 Output Noise Power Increase vs Center Frequency Offset 




SECTION 4 


THRESHOLD EXTENSION TECHNIQUES 


4 . 1 GENERAL 

The discussion of FM threshold in the preceding section recognized 
the presence of high amplitude impulse noise (click noise) in the 
demodulator output as a primary factor contributing to the occur- 
rence of threshold in an FM system. 

Certain distinguishing characteristics of click noise provide a 
basis for practical techniques which can be implemented at the 
output of an FM demodulator to extend the threshold performance 
of the system. The threshold extension technique that will be 
described in this section is based on the accomplishment of the 
following two steps: 

A. Detection of the click producing noise impulses in the 
demodulator output by distinguishing them from the demodu- 
lated signal and low-level Gaussian noise. 

B. Utilization of the detected click noise information to per- 
form a click elimination operation on a delayed version of 
the demodulator output. 

4.2 CLICK DETECTION 

The problem of click detection can be simplified by observing that, 
in most cases, the peak amplitude of the click-producing impulse 
noise is greater than that of both the modulation and Gaussian 
noise in the unfiltered demodulator output. Figures 4-1 through 
4-6 are presented to illustrate this point. 

The upper waveform in each photograph represents the signal plus 
noise output of a 500 KHz low-pass postdetection filter whereas 
the lower sweep in each figure represents the signal plus noise 
present in the unfiltered demodulator output. The sweeps in each 
photograph are aligned such that the relationship between unfiltered 


4-1 


f = 10 KHz 
m 

Af = 1 MHz 

BW = 500 KHz 
o 


Figure 4-1 Demodulated Signal Plus Noise Waveform With 

and Without Postdetection Filtering (f m = 10 KHz) 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 20 ^SEC/CM 


f = 10 KHz 
m 

Af = 1 MHz 

BW = 500 KHz 
o 


Figure 4-2 Demodulated Signal Plus Noise Waveform with 

and Without Postdetection Filtering (f m = 10 KHz) 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 20 ^SEC/CM 

GCIO 177 (A)— 14 




4-2 



f = 50 KHz 
m 

Af = 1 MHz 

BW = 500 KHz 
o 


Figure 4-3 Demodulated Signal Plus Noise Waveform With 

and Without Postdetection Filtering (f m = 50 KHz) 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 10 m SEC/CM 


f = 50 KHz 
m 

Af = 1 MHz 

BW = 500 KHz 
o 


Figure 4-4 Demodulated Signal Plus Noise Waveform With 

and Without Postdetection Filtering (f m = 50 KHz) 

UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 10 m SEC/CM 

G C 10 177 (AJ — 12 




4-3 





Figure 4-5 Demodulated Signal Plus Noise Waveform With 

and Without Postdetection Filtering (f m =100 KHz) 


UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 5 mSEC/CM 


Vs 
* ^ 


‘ W 


i I ! 


f = 100 KHz 
m 

Af = 1 MHz 
BW q = 500 KHz 


•//'I W 


Figure 4-6 


Demodulated Signal Plus Noise Waveform With 
and Without Postdetection Filtering (f m = 100 KHz) 


UPPER TRACE: FILTERED 
LOWER TRACE: UNFILTERED 
HORIZONTAL SCALE: 5 ^SEC/CM 


GC 10 177 (A) — 13 


4-4 



4 2 CLICK DETECTION (Cont'd) 


and filtered waveforms may be observed for a particular click event. 
It should be noted that the 500 KHz postdetection filter acts to 
attenuate (as well as to broaden) the click waveform. 

It is important to note that a click occurs on the 500 KHz low- 
pass filtered output only when the peak amplitude of the corres- 
ponding noise impulse m the unfiltered output exceeds the average 
peak amplitude of the modulation plus Gaussian noise There are, 
however, a certain number of high amplitude noise impulses present 
on the unfiltered waveforms shown m Figures 4-1 through 4-6 that 
reach the amplitude of the click-producing noise spikes but do not 
result m a corresponding click on the unfiltered output waveform 
These noise spikes belong to the Gaussian portion of the unfiltered 
output noise whose energy is concentrated primarily m frequencies 
above the 500 KHz cutoff of the low-pass postdetection filter 
The parabolic spectrum of these occasional high amplitude excur- 
sions of the Gaussian noise do not produce clicks on the filtered 
output waveform since most of the energy is concentrated m fre- 
quencies above the cutoff frequency of the low-pass postdetection 
filter . 

The probability of detecting these non-click-producing noise 
impulses can be minimized by implementing a pre-click detection 
filter to attenuate the higher frequency Gaussian noise components 
The relative high amplitude characteristic of the click-producing 
impulse noise will be preserved if the cutoff frequency of the 
filter is high compared with the 500 KHz postdetection filter A 
2 MHz low-pass filter was used for this purpose as shown m Figure 
4-7 


Experimental results have shown that amplitude detection of the 
unfiltered click noise m the demodulator output is a simple and 
efficient means for obtaining the desired information indicating 
the occurrence of a click The actual detection process is accom- 
plished with a pair of conventional Schmidt triggers It is neces- 
sary to use two triggers since both positive and negative clicks 
must be independently detected 

These devices are configured to detect noise spikes that exceed a 
predetermined positive or negative voltage level The reference 
(trigger) voltage level is selected so that only the click-producing 
noise spikes will be detected Figure 4-7 shows the system config- 
uration for the click detection process 

The output of the amplitude level detector is used to trigger a 
pulse generator which provides gating pulses for the click elimi- 
nation circuitry 


4-5 



FM 

DEMODULATOR 


TO CLICK 
ELIMINATOR 



Figure 4-7 Click Detector Block Diagram 


BD 10 178 (A) 



4.3 CLICK ELIMINATION 


The detected click information is used to perform a click elimi- 
nation operation on a delayed version of the demodulator output 
The elimination process is accomplished by first passing the delayed 
demodulator output through an amplifier whose output can be gated 
off for a predetermined time. 

Gating pulses from the click detection system are then supplied 
to the amplifier so that it is turned off at the beginning of a 
click event The turnoff time of the amplifier is preset to coin- 
cide with the duration of the click 

A "holding" circuit is used m conjunction with the amplifier to 
provide a constant output voltage from the system during the time 
that the amplifier is biased off The output voltage of the "holding" 
circuit corresponds to the amplitude of the demodulator output just 
before the beginning of the click The net effect of the click elim- 
ination circuitry is to provide an estimate of the modulation as a 
substitute for the high amplitude noise spike m the demodulated 
output A block diagram of the click elimination system is shown 
m Figure 4-8 

Sance the click elimination process is performed prior to postde- 
tection filtering, the turnoff time, or click duration, is small 
compared with the average modulation frequency The 500 KHz post- 
detection filter provides a smoothing effect on the click-eliminated 
output, which enhances the performance of the device 

A block diagram of the complete threshold extension device is shown 
m Figure 4-9 


4-7 



i 

CO 



FROM CLICK 

DETECTOR 

OUTPUT 


Figure 4-8 Click Eliminator Block Diagram 


BQIO I7S(A)-2 





Figure 4-9 Click Detection and Elimination Threshold Extension Device Configuration 




4 4 THRESHOLD EXTENSION DEVICE 


The following steps summarize the operation of the threshold 
extension device 

A The output of an FM demodulator, which consists of signal plus 
noise, is split into two separate channels The first channel, 

A, is passed through a pre-click detector filter and is then fed 
to a series of circuits that detect the presence of high amplitude 
impulse noise. This is the noise that is a primary factor contri- 
buting to the degraded performance of an FM system. 

B The output of the impulse noise detection circuits consists 
of a series of positive pulses that are fed to a gated 
amplifier 

C The input to the gated amplifier is channel B of the de- 
modulator output Channel B is identical to channel A 
except that it is time delayed by a preset value 

D The gating pulses from the noise detection circuits are 

used to turn off channel B whenever a noise impulse occurs 
This creates a "hole" m the channel B output that is 
smoothed over by additional circuitry which provides an 
estimate of the modulation during the turnoff time 

E. The output of the gating amplifier is fed to a low-pass 
filter that acts to provide additional smoothing to the 
output modulation during turnoff time 

F The resulting output consists of signal plus low-level 
noise with fewer high amplitude impulses to degrade the 
signal- to-noise ratio 


4-10 



SECTION 5 


EXPERIMENTAL RESULTS 


5 1 GENERAL 

The performance of the click elimination device has been evaluated 
m terms of its ability to improve the demodulated output of an 
Apollo, MSFN, phase-locked loop carrier frequency demodulator Sev- 
eral tests have been conducted with input signals having parameters 
representative of the Apollo Block II downlink television and play- 
back voice modes The signal characteristics of these modes are 
listed m Table 5-1 


TABLE 5-1 


CSM MODES USED FOR THRESHOLD EXTENSION DEVICE TEST 


MODE 

BASEBAND 

SERVICE 

Af 

f ' 
m 

BW 0 

CSM FM 

1 1 PLAYBACK 

100 KHZ 

3 KHZ 

70 KHZ 

MODE 1 

VOICE 




CSM FM 

32 1 PLAYBACK 

100 KHZ 

70 KHZ 

70 KHZ 

MODE 2 

VOICE 




CSM FM 

TV 

1.0 MHZ 

409 KHZ 

500 KHZ 

MODE 4 






The performance evaluation tests can be outlined as follows 
A Signal plus noise waveform analysis 
B Signal- to-noise ratio tests 

1 CSM Modes 1 and 2 

2 CSM Mode 4 

C. Voice intelligibility tests 

D. TV picture quality tests 


5-1 




5 2 SIGNAL-PLUS -NOISE WAVEFORM ANALYSIS 


A qualitative estimate of the click eliminator performance was 
obtained by observing the effect of the device on the output noise 
and signal-plus-noise waveforms as displayed on an oscilloscope 
The test configuration is shown m Figure 5-1 The results of 
this test are shown m Figures 5-2 through 5-13 

The upper trace m each figure represents the unprocessed demod- 
ulator output whereas the lower trace shows the click eliminated 
output 

Figures 5-2 and 5-3 show the effectiveness of the click elimination 
process when the demodulator output consists of noise without 
modulation The device eliminates 100 percent of the noise spikes 
for an input signal- to-noise ratio of 3 dB as shown m Figure 5-2 
For an input SNR of 1 dB, the elimination process is more than 
90 percent efficient. In both cases the postdetection filter 
was 70 KHz, which corresponds to the CSM Modes 1 and 2 demodulator 
configuration 

Figures 5-6 through 5-13 show the improvement that is obtained 
when both sinusoidal signal plus noise are present at the output of 
the demodulator. The modulation frequencies and deviations (Af) 
used for these photographs are compatible with the Apollo CSM FM 
modes listed m Table 5-1 

Several combinations of modulation frequency and output bandwidth 
were used to demonstrate the spike noise elimination process for 
various ratios of click duration to modulation frequency 

Figures 5-4 and 5-5 illustrate the click elimination process for a 
modulation frequency of 30 KHz, a peak frequency deviation of 1 MHz 
and a postdetection low-pass filter bandwidth of 500 KHz 

For Figures 5-6 and 5-7, the modulation frequency is 1 KHz, the 
peak frequency deviation is 100 KHz, and the postdetection low-pass 
filter bandwidth is 70 KHz 


5-2 




B D 10 178 (A) — 4 


Figure 5-1 


Signal Plus Noise Waveform Analysis Test Configuration 





WITHOUT CLICK 
ELIMINATION 


*r**y w 'r 


WITH CLICK 
ELIMINATION 


Figure 5-2 Signal Plus Noise Waveform Analysis 


BW 0 = 70 KHz 

SNR... = 3dB 

IN 


NO MODULATION 



WITHOUT CLICK 
ELIMINATION 


WITH CLICK 
ELIMINATION 


Figure 5-3 Signal Plus Noise Waveform Analysis 


BW 0 = 70 KHz 

SNR... = ldB 
IN 


NO MODULATION 


GC 10 1 77 (A) - 38 





1 

1 



1 

1 

1 

wwvv 

WITHOUT CLICK 
ELIMINATION 

■ 

1 

wvwv 

WITH CLICK 
ELIMINATION 

Figure 5-4 Signal Plus Noise Waveform Analysis 

R f = 30 KHz 

m 

Af = 1 MHz 

. BW 0 = 500 KHz 

1 SNR |N =2dB 

1 

1 

1 

WVWx? 

WITHOUT CLICK 
ELIMINATION 

1 

1 

VvWA 

WITH CLICK 
ELIMINATION 

1 

1 

1 

Figure 5-5 Signal Plus Noise Waveform Analysis 

f = 30 KHz 
m 

Af = 1 MHz 
BW 0 = 500 KHz 
SNR = ldB 

GC 10 177 (A) — 37 

1 0 

5-5 



WITHOUT CLICK 
ELIMINATION 



WITH CLICK 
ELIMINATION 


Figure 5-6 Signal Plus Noise Waveform Analysis 

f = 1 KHz 
m 

Af = 100 KHz 
BW 0 = 70 KHz 
SNR |N = 4dB 



WITHOUT CLICK 
ELIMINATION 


WITH CLICK 
ELIMINATION 


Figure 5-7 Signal Plus Noise Waveform Analysis 

t = 1 KHz 
m 

Af = 100 KHz 
BW 0 = 70 KHz 
SNR m = 3dB 

G C 10 1 77 (A) — 36 


5-6 




WITHOUT CLICK 
ELIMINATION 



WITH CLICK 
ELIMINATION 


Figure 5-8 Signal Plus Noise Waveform Analysis 

f = 30 KHz 
m 

Af = 100 KHz 
BW 0 = 70 KHz 

SNR|N = 3 dB 



WITHOUT CLICK 
ELIMINATION 


WITH CLICK 
ELIMINATION 


Figure 5-9 Signal Plus Noise Waveform Analysis 

t = 30 KHz 
m 

Af = 100 KHz 
BWo= 70 KHz 
SNR |n = 2 dB 

GC tO 1 77 (A)— 35 


5-7 





WITHOUT CLICK 
ELIMINATION 


WITH CLICK 
ELIMINATION 


Figure 5-10 Signal Plus Noise Waveform Analysis 

f = 30 KHz 
m 

Af = 100 KHz 
BWo= 70 KHz 
SNR |N = 2 dB 



WITHOUT CLICK 
ELIMINATION 


WITH CLICK 
ELIMINATION 


Figure 5-11 Signal Plus Noise 

f = 30 KHz 
m 

Af = 100 KHz 
BW 0 = 70 KHz 
SNR (N = 1 dB 


Waveform Analysis 


GC 10 177 (A) — 34 


5-8 




WITHOUT CLICK 
ELIMINATION 



WITH CLICK 
ELIMINATION 


Figure 5-12 Signal Plus Noise Waveform Analysis 

t = 30 KHz 
m 

Af = 100 KHz 
BW 0 = 70 KHz 
SNR )N = 1 d B 



WITHOUT CLICK 
ELIMINATION 


WITH CLICK 
ELIMINATION 


Figure 5-13 Signal Plus Noise Waveform Analysis 

f = 30 KHz 
m 

Af = 100 KHz 
BW 0 = 70 KHz 
SNR 1N = ldB 

G CI0 177 (A) — 33 


5-9 




5.2 SIGNAL-PLUS-NOISE WAVEFORM ANALYSIS (Cont'd) 


These figures represent the case where the click duration is 
small compared to the modulation period. The individual noise 
spikes do not significantly distort the overall shape of the 
signal because of their relatively short duration. However, the 
contribution of the clicks to the output noise power is still con- 
siderable because of their relatively high amplitude. 

An important observation to be made from Figures 5-4 through 5-7 
is that the click eliminator is capable of removing both low-level 
(those spikes having a peak amplitude less than the maximum modu- 
lation amplitude) and high-amplitude spikes from the demodulated 
waveform. This observation verifies the validity of the criteria 
for amplitude detection of click noise in the unfiltered demodula- 
tor output . 

In Figures 5-8 through 5-13, the modulation frequency is 30 KHz; 
the peak frequency deviation is 100 KHz; and the postdetection 
low-pass filter bandwidth is 70 KHz. These figures represent the 
case where the click duration is approximately equal to half the 
modulation period. This means that a single noise spike will cause 
considerable distortion to the shape of the demodulated signal 
waveform . 

Figures 5-8 through 5-13 illustrate the ability of the click elim- 
inator to provide a modulation estimate as a substitute for the 
distortion caused by the occurrence of a noise spike on the output 
waveform. It should be noted that, in most cases, the noise spikes 
occur when the modulation amplitude is at a maximum. 


5-10 


5.3 SIGNAL -TO -NOISE RATIO TESTS 


5.3.1 CSM FM Modes 1 and 2 

A quantitative evaluation of the click eliminator's ability to 
improve the performance of an FM channel can be obtained by de- 
termining the threshold extension that results from the suppres- 
sion of spike noise. The threshold extension can be determined 
by measuring the output signal- to-noise ratio of an FM demodulator 
before and after the click elimination process. 

These tests were performed with a simulated Apollo Block II CSM-to- 
MSFN playback voice channel in the Electronics Systems Compatibility 
Laboratory (ESCL) of ISD/MSC. The ESCL contains an Apollo MS FN 
ground station receiver that was used in conjunction with a simu- 
lated RF path and CSM transponder as shown in Figure 5-14. The 
click elimination device was inserted between the output of the 
carrier frequency demodulator and the 71-KHz low-pass postdetection 
filter . 

The output s ignal - to-noisc ratio was measured with and without 
the click eliminator in the system for a wide range of received 
signal levels. Figure 5-15 represents the measured data for the 
1:1 playback voice mode test. The threshold point is approximately 
-90.6 dBm for the unprocessed channel and -92.7 dBm for the click 
eliminated channel. The difference between these two values of 
total received RF power represents the effective threshold exten- 
sion obtained with the click elimination device. From Figure 5-15 
it can be determined that a 2-dB extension of threshold was obtained 
for the Apollo CSM Mode 1. However, the improvement in output SNR 
is significantly greater than indicated by the 2-dB threshold ex- 
tension. A maximum improvement of 8 dB to 10 dB is obtained for 
certain values of total received power that are below the threshold 
points found in Figure 5-15. The significance of this SNR improve- 
ment will be shown in the results of the word intelligibility tests. 

Figure 5-16 represents the measured data for the 32:1 playback 
voice mode test. The threshold point is approximately -89.0 dBm 
for the unprocessed channel and -90.4 for the click eliminated 
channel. A threshold extension of approximately 1.5 dB was achieved 
for this mode, while the maximum output SNR improvement was 10 dB. 

A 1-KHz tone was used as modulation for the CSM Mode 2 tests. 


5-11 


CSM 



*NOTE: The test configuration MSFN receiver does not 

include a paramp and, therefore, has a noise 
figure of approximately 10 dB. 

Figure 5-14 Test Configuration for CSM Modes 1 and 
2 Signal- to-Noise Ratio Tests With and 
Without Threshold Extension Device 


5-12 


OUTPUT SNR (dB) 



INPUT SNR (dB) 

Figure 5-15 Output SNR vs Input SNR for 1:1 
CSM Playback Voice Mode 


5-13 




OUTPUT SNR(dB) 


toi to 
55 
a:\cz 


TOTAL RECEIVED POWER (dBm) 

3 4 5 6 7 8 9 10 11 12 13 14 15 

INPUT SNR (dB) 



G CIO 1 77 (A) — 5 

Figure 5-16 Output SNR vs Input SNR for 
32:1 CSM Playback Voice Mode 


5-14 



5.3.1 CSM FM Modes 1 and 2 (Cont’d) 


It should be noted that in both Figure 5-15 and 5-16 the measured 
data does not agree with the theoretical one-to-one output vs input 
SNR relationship expected for large values of input SNR. Instead, 
the measured curves flatten such that the output SNR remains almost 
constant for values of input SNR greater than 12 dB. This limiting 
effect is due to the presence of three scientific subcarriers which 
are part of CSM FM Modes 1 and 2. The subcarrier frequencies are 
close enough to the playback voice bandpass to limit the output SNR 
at large values of input SNR. 

Straight lines have been drawn tangent to the measured curve with 
a slope of 1 for the purpose of determining the threshold point 
for Figure 5-15 and 5-16. The threshold point was then determined 
graphically according to the procedure discussed in Section 3.1. 

A plot of the output SNR improvement versus input SNR is shown in 
Figure 5-17. The maximum improvement occurs for values of input 
SNR between 3 dB and 5 dB for both modes. The click eliminator im- 
provement increases linearly as the input SNR decreases from approx- 
imately 9 dB to 5 dB. For values of input SNR below 3 dB , the 
improvement drops off rapidly due to saturation of the click noise 
detection circuits. It is possible that this saturation effect 
could actually degrade the output performance of the demodulator. 
This would occur when the number of clicks per second is such that 
the output of the click eliminator is turned off most of the time. 

5.3.2 CSM FM Mode 4 

The system configuration for the CSM television mode test is 
shown in Figure 5-18. A 300-KHz tone was used to modulate the 
50-MHz FM test set. A threshold improvement of 1.5 dB was obtained 
as shown in Figure 5-19. The maximum improvement in output SNR was 
found to be approximately 8 dB. It should be noted that the shape 
of the plot representing the performance of the demodulator with 
click elimination is considerably more linear than the curve repre 
senting the performance of the same demodulator without click elim- 
ination. 


5-15 



-1 0123456789 10 


INPUT SNR(dB) 


Figure 5-17 Output SNR improvement vs Input SNR 
for 1:1 and 32:1 CSM Playback Voice 


5-16 


GC10»77(A)- 1 


LI- 


CD 



TEST POINT: SNR 0|JT 


BD!0I78(A)-S 


Figure 5-18 Configuration for CSM Mode 4 SNR Tests 



INPUT SNR(dB) 

Figure 5-19 Output SNR vs Input 

SNR for CSM FM Mode 4 


5-18 


5.3.2 CSM FM Mode 4 (Cont'd) 


The implication of this observation is that the rapid deterioration 
of the output signal which is normally associated with the operation 
of an FM demodulator below threshold can be significantly reduced 
by the application of click noise elimination techniques. There- 
fore, an input SNR of 3 dB would probably result in a "useless" 
output signal for a demodulator without click elimination whereas 
the corresponding output signal from the demodulator using click 
elimination might be useful. 

A plot of output SNR improvement versus input SNR for CSM Mode 4 
is shown in Figure 5-20. The maximum improvement is obtained for 
values of input SNR between 1 and 3 dB. 


5-19 



Figure 5-20 Output SNR Improvement vs Input SNR 

for CSM FM Mode 4 With Click Elimination 


o C 10 177 (A) -8 


5-20 


5.4 WORD INTELLIGIBILITY TESTS 


Experimental results from the signal- to-noise ratio tests indicate 
that a threshold improvement of 1.5 dB to 2 dB was obtained for 
CSM Modes 1 and 2, while the maximum output SNR improvement for 
the modes was 8 dB to 10 dB . 

A more realistic evaluation of the click elimination performance, 
however, can be obtained by determining the improvement in output 
word intelligibility. Several tests were conducted, using the 
configuration shown in Figure 5-21, for the 1:1 playback voice 
mode. The pre-recorded voice tapes used to modulate the CSM FM 
transmitter were composed of 150 words separated into three groups. 
Each group of 50 words was spoken by a different person. A 
different word list tape was used for each test run representing 
a specific value of total received power, and the demodulated 
information was recorded for later evaluation of the output word 
intelligibility. 

The results of the word intelligibility tests are shown in 
Figure 5-22. The maximum improvement in word intelligibility 
was approximately 20 percent for an input SNR of 4.3 dB. Referring 
to Figure 5-17, this value of input SNR lies within the region of 
maximum improvement (between 3 dB and 5 dB input SNR) for this 
particular mode. 

The results of the word intelligibility tests for the 32:1 play- 
back mode are shown in Figure 5-24. The improvement obtained for 
this mode is considerably less than that obtained for the 1:1 
playback mode. A maximum improvement of only 4 percent in word 
intelligibility occurred for an input SNR of 6.3 dB . The differ- 
ence between the improvement observed for these two modes is re- 
lated to the distortion inherent in the spacecraft playback system. 
It should be noted that even at relatively high values of input 
SNR the intelligibility is limited to 81 percent for the 32:1 
playback mode, whereas 91 percent intelligibility is obtained for 
the 1:1 playback mode for the same input SNR. The 32:1 playback 
voice tests were performed with the test configuration shown in 
Figure 5-23. 


5-21 


CSM 



Figure 5-21 Configuration for CSM FM Mode 1 Word 
Intelligibility Test With and Without 
Threshold Extension Device 


5-22 


BD10178(A) 9 



INPUT SNR (dB) 

Figure 5-22 Percent Word Intelligibility vs Input 
SNR for 1:1 CSM Playback Voice Mode 


5-23 


GC10177(AJ — 3 



CSM 



Figure 5-23 Configuration for CSM FM Mode 2 Word 

Intelligibility Tests With and Without 
Threshold Extension Device 


5-24 


B D 10 1 78 ( A) — 7 



Figure 5-24 


Percent Word Intelligibility vs Input SNR 
for CSM 32:1 Playback Voice Mode 


GCI0177(A) 7 


5-25 



5.5 TELEVISION PICTURE QUALITY ANALYSIS 


The test configuration shown in Figure 5-25 was used to obtain 
a series of photographs representing the demodulated video signal 
output of an Apollo type FM demodulator. The demodulated signal 
was displayed on the CRT of a slow scan monitor operating in the 
CSM-LM Block II, 10 framc-per- second mode. The photographs were 
taken with equal exposure times of 1/10 second and with a constant 
camera aperture. Therefore, each picture represents one complete 
frame of information. 

A grey scale signal was used to modulate the FM test set with a 
peak frequency deviation of 1.0 MHz. Figure 5-26 shows the de- 
modulated signal displayed on the slow scan monitor for a 20-dB 
input SNR. This picture is used primarily as a calibration refer- 
ence to which the following data will be compared. 

Figure 5-27 shows the demodulated output for a 10-dB input SNR. 

The presence of Gaussian noise results in a "fuzzy" picture as 
compared with Figure 5-26. The absence of click noise in these 
pictures indicates that the demodulator is operating above thresh- 
old. 

The presence of click noise, which appears as white or black spots 
on the photograph, is first noticed in Figure 5-28 for an input 
SNR of 8 dB. Although only two or three clicks can be found in 
this figure, one of them has resulted in a momentary loss of line 
synchronization as indicated by the torn segment of the vertical 
white bar. The synchronization circuitry in the MSFN slow scan 
monitor is particularly sensitive to the presence of click noise. 
Figure 5-29 shows the demodulated output for an 8 dB input SNR 
with the click eliminator in the system. Notice that the click 
noise has been eliminated and that the line synchronization has been 
preserved. 

As the input SNR is decreased to 7 dB, the demodulator approaches 
threshold, and the number of clicks, as well as synchronization 
perturbations, increases accordingly. 

Figures 5-30 and 5-31 show the demodulated signal before and after 
the click elimination process for an input SNR of 7 dB . Figures 
5-32 and 5-33 provide similar results for an input SNR of 6 dB 
(the click eliminator has been able to dispose of almost every 
click for values of input SNR down to 6 dB. For SNR's below 


5-26 


CSM 



Figure 5-25 Configuration for CSM Mode 4 

Television Picture Quality Tests 


5-27 


BDIOI 78 (A) - 


Af = 1.0 MHz 
BW |F = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-26 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 20 dB REFERENCE PICTURE 


Af = 1.0 MHz 
BW |p = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-27 CSM Mode 4: TV Picture Quality Tests 

SNR... = 10 dB WITHOUT CLICK ELIMINATION 




GCI0177 (A) — 32 


5-28 




Figure 5-28 CSM Mode 4: TV Picture Quality Tests 

SNR m = 8 dB WITHOUT CLICK ELIMINATION 
IN 



Figure 5-29 CSM Mode 4: TV Picture Quality Tests 

SNR... = 8 dB WITH CLICK ELIMINATION 
IN 


1.0 MHz 
= 4.9 MHz 
= 500 KHz 


1.0 MHz 
= 4.9 MHz 
: 500 KHz 


GC 10177 (A) - 31 


5-29 



Af = 1.0 MHz 

BW |p = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-30 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 7 dB WITHOUT CLICK ELIMINATION 



Af = 1.0 MHz 

BW |p = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-31 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 7 dB WITH CLICK ELIMINATION 


GC 10 177(A)- 30 


5-30 



Af = 1.0 MHz 

BW |p = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-32 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 6 dB WITHOUT CLICK ELIMINATION 


Af = 1.0 MHz 
BW |p = 4.9 MHz 
BW q = 500 KHz 


Figure 5-33 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 6 dB WITH CLICK ELIMINATION 




GC 10 177 (A) — 29 


5-31 





5.5 TELEVISION PICTURE QUALITY ANALYSIS (Cont'd) 

6 dB, the efficiency of the device decreases as a result of 
circuit saturation. The ability of the device to eliminate noise 
spikes is primarily related to the click rate and the recovery 
time of the detection circuitry. 

In Figure 5-34 the number of click-produced white spots (before 
elimination) is considerably greater than the number found in 
previous pictures, which indicates that the demodulator is begin- 
ning to operate below threshold. 

Figure 5-35 shows the effect of click elimination on the output 
signal for a 5-dB input SNR. The click rate is such that approx- 
imately 10 percent of the impulse noise perturbations remain 
after the elimination process. However, the improvement in picture 
quality due to the click elimination process is still significant. 

The effectiveness of the modulation insertion circuitry incorporated 
in the device can be seen by comparing Figures 5-36 and 5-37. In 
Figure 5-36 the demodulated signal information is almost completely 
masked by the perfusion of impulse noise. The line synchronization 

is also perturbed for most of the frame duration. Figure 5-37 
however, displays a demodulated signal that is considerably more 
intelligible than that of the previous figure. The number of missed 
line synchronizations is also drastically reduced. 

A similar evaluation of the modulation preservation action can be 
obtained by comparing Figures 5-38 and 5-39 for an input SNR of 
3 dB. The grey scale pattern is not visible in Figure 5-38 due 
to the presence of click noise whereas in Figure 5-39 it is possible 
to distinguish the individual vertical bars from the remaining noise. 
The improved line synchronization also contributes to the overall 
image enhancement obtained with the click eliminator. 

Figures 5-40 and 5-41 represent the output of the system for an 
input SNR of 2 dB. The grey scale modulation is completely masked 
by the impulse noise as shown in Figure 5-40. The click eliminated 
output shown in Figure 5-41 contains considerably less noise, but 
the grey scale vertical bars are still not intelligble. It appears, 
therefore, that the modulation insertion operation has become inef- 
fective for this low value of input SNR. This observation is correct 
since previous results have shown that the high click rate which 
occurs at low input SNR's causes the click eliminator to turn off 
the output of the demodulator at a rate approaching the modulation 
frequency . 


5-32 


Af = 1.0 MHz 
BW |p = 4.9 MHz 
BW 0 = 500 KHz 


Figure 5-34 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 5 dB WITHOUT CLICK ELIMINATION 


Af = 1.0 MHz 
BW |p = 4.9 MHz 
BW 0 = 500 KHz 


Figure 5-35 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 5 dB WITH CLICK ELIMINATION 




GC 10 177(A)— 28 


5-33 





Af = 1.0 MHz 
BW (F = 4.3 MHz 
BW q = 500 KHz 


Figure 5-36 CSM Mode 4: TV Picture Quality Tests 

SNR... = 4 dB WITHOUT CLICK ELIMINATION 


Af = 1.0 MHz 
BW |p = 4.3 MHz 
BW q = 500 KHz 


Figure 5-37 CSM Mode 4: TV Picture Quality Tests 

SNR| N =4 dB WITH CLICK ELIMINATION 



GC10177CA) — 27 


5-34 




Af = 1.0 MHz 



Figure 5-38 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 3 dB WITHOUT CLICK ELIMINATION 



Af = 1.0 MHz 

BW |p = 4.9 MHz 

BW rt = 500 KHz 
o 


Figure 5-39 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 3 dB WITH CLICK ELIMINATION 


GC 10 177 (A) —26 


5-35 




Af = 1.0 MHz 

BW |p = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-40 CSM Mode 4: TV Picture Quality Tests 

SNR |N = 2 dB WITHOUT CLICK ELIMINATION 


Af = 1.0 MHz 

BW |p = 4.9 MHz 

BW = 500 KHz 
o 


Figure 5-41 CSM Mode 4: TV Picture Quality Tests 

SNR |N =2dB WITH CLICK ELIMINATION 




GC 10 177 (A) — 25 


5-36 




S 5 TELEVISION PICTURE QUALITY ANALYSIS (Cont'd) 

This implies that the modulation tracking circuitry m the device 
does not have sufficient time to obtain an accurate estimate of 
the modulation amplitude between successive click events This 
situation is not so much a limitation of the threshold extension 
device as it is a consequence of having a click rate approach the 
modulation frequency m the demodulator output 


5-37 



APPENDIX A 


BIBLIOGRAPHY 


Clarke, K , R. Pickholtz, and D Schilling, A Space Communications 
Study--Final Report — September 19 6 6- September 1967 Prepared 
for NASA Electronics Research Center under NASA Grant NGR-33- 
006.020. 

Clarke, K , K, and D. T Hess, "Frequency Locked Loop FM Demodula- 
tor," IEEE Transactions on Communications Technology } Vol COM- 
15, No, 4 August 1967 

Dawson, C, A Performance Analysis of the Apollo Unified S-Band 

Communications System for a Typical Lunar Mission. MSC Internal 
Note MSC-EB-R- 67- 1 , August 1967, 

Downing, J. Modulation Systems and Noise Englewood Cliffs, New 
Jersey Prentice-Hall, Inc , 1964 

Enloe , L H "Decreasing the Threshold m FM by Frequency Feedback", 
Proceedings of IRE , Vol 50, January 1962. 

Hess, D T , "Cycle Slipping m First Order Phase Locked Loop", 

IEEE Transactions on Communications Technology 3 April, 1968 

Rice, S 0 , "Noise m FM Receivers " Proceedings 3 Symposium on 
Time Series Analysis M Rosenblatt, Editor New York John 
Wiley § Sons, Inc , 1963. 

Schwartz, M , W Bennet, and S. Stem , Communications Systems 
and Techniques New York McGraw Hill, 1966 


A-l 



APPENDIX B 


ABBREVIATIONS AND SYMBOLS 


A 

Carrier Amplitude 


AMP 

Amplifier 


ATTEN 

Attenuator 


BW 

Bandwidth. 


w 

1— 1 
*ri 

Predetection (input) Bandwidth 


BW 0 

Postdection (output) Bandwidth 


cm 

Centimeter 


CSM 

Command/Service Module 


dB 

Decibel 


dBm 

Decibel, referenced to 1 milliwatt 


DEMOD 

Demodulator 


DSE 

Data Storage Equipment 


ESCL 

Electronic Systems Compatibility Laboratory 


FM 

Frequency Modulation 


fm 

Modulation Frequency m Cycles per Second 


FMFB 

Frequency Modulation Feedback Discriminator 


f 0 

Reference or Center Frequency m Cycles per 

Second 

GEN 

Generator 


G(t) 

Unmodulated Carrier 


Hz 

Hertz 


IF 

Intermediate Frequency 


ISD 

Information Systems Division 


ISO AMP 

Isolation Amplifier 


K 

Boltzman's Constant (1 38 x 10"23 Watts-sec 

0 K _1 ) 


B-l 



KHz 

K D 

LM 

MHz 

MSC 

MSFN 

m(t) 

mv 

N + 

N“ 

*c 

Hi 

NM 

N 0 

No 

N ( t) 

OSC 

PA 

PLL 

PMP 

R 

r 

RCVR 

RF 

RMS 

sec 


Kilohertz 

Discriminator Constant 
Lunar Module 
Megahertz 

Manned Spacecraft Center 
Manned Space Flight Network 
Modulating Signal 
Millivolt 

Positive Click Rate 

Negative Click Rate 

Output Click Noise Power 

Input Noise Power 

Nautical Miles 

Output Gaussian Noise Power 

Total Output Noise Power 

Discriminator input Noise 

Oscillator 

Power Amplifier 

Phase Lock Loop 

Pre-Modulation Processor 

Resultant Carrier Plus Noise Magnitude 

Radius of Gyration of the Power Spectrum about 
its Axis of Symmetry 

Receiver 

Radio Frequency 

Root-Mean-Square 

Second 

Input Signal Power 


B-2 



SNR 

snr in 

SNR 


OUT 


T 

TRIPLX 

TV 

USB 

XMIT 

X(t), Y(t) 

3 

A£ 

A<j) 

Ah) 

psec 

T 

4 > 

<b 

T e 

h) c 


(0 


m 


h) r 


s 

Signal-to-Noise Ratio 

Input (Predetection) Signal-to-Noise Ratio 
Output (Postdetection) Signal-to-Noise Ratio 
Output Signal Power 
System Temperature 

Triplexer 

% 

Television 

Unified S-Band 
* 

Transmitter 

Independent Random Variables, Representing the 
Magnitude of the In-Phase and Quadrative Phase 
Components of N(t). 

Modulation Index 

Frequency Deviation 

Phase Difference between the Resultant Signal plus 
Noise Vector and the Carrier Vector at t^ and 

Radian Frequency Deviation 

Microsecond 

Time Interval 

Phase of the Resultant Signal-plus-Noise Vector with 
Respect to the Carrier Vector 

Phase Error caused by the Noise Disturbance about 
the Carrier Frequency. 

Radian Carrier Frequency 

Radian Frequency Error caused by the Noise Disturb- 
ance about the Carrier Frequency. 

Modulation Frequency m Radians per Second 

Refereace or Center Frequency m Radians per Second 


B-3