UNCLASSIFIED

An 2 6 3 500
Reproduced
L f . ike
AR M ED SER V ICES T EH NICAL INFO R M f f l O f i AG ENCY
AR LING T O N H ALL ST An O N
AR LING T O N 1 2 . V IR G INIA
UNCLASSIFIED
DISCLAIMER NOTICE
THIS DOCUMENT IS THE BEST
QUALITY AVAILABLE.
COPY FURNISHED CONTAINED
A SIGNIFICANT NUMBER OF
PAGES WHICH DO NOT
REPRODUCE LEGIBLY .
. t
..
NO T ICE: When govern men t or other drawi n gs, speci -
f i cati on s or other data are used f or an y purpose
other than In con n ecti on wi th a def i n i tel y rel ated
govern men t procuremen t operati on , the U. S.
G overn men t thereby In curs n o respon si bi l i ty, n or an y
obl i gati on whatsoever; an d the f act that the G overn -
men t may have f ormul ated, f urn i shed, or In an y way
suppl i ed the sai d drawi n gs, speci f i cati on s, or other
data Is n ot to be regarded by Impl i cati on or other-
wi se as In an y man n er l i cen si n g the hol der or an y
other person or corporati on , or con veyi n g an y ri ghts
or permi ssi on to man uf acture, use or sel l an y
paten ted In ven ti on that may In an y way be rel ated
thereto.
l i m
HI
H
''al l l l m l
M E M O R A N D U M R E P O R T N O . 1 3 4 9
A U G U S T 1 9 6 1
A CO M B FILT E R FO R U S E IN
T R A CKIN G S A T E LLIT E S
A R P A S atel l i te Fence S eri es
XER O X
R i chard L Vi tek
A S T M

R ep o rt N um ber 2 4 i n the S eri es
D ep artm ent o f the A rm y P ro j ect N o . 503 -06 -01 1
O rdnance M anagem ent S tructure Co de N o . 52 1 0. 1 1 . 1 4 3
BA LLIS T IC R E S E A R CH LA BO R A T O R IE S
I' ^KSmf ^f ^F.
^:^M^<-iwwirT';i/ar^.-^-'trc-_''.'%' ..,-. ;:-';>....>.-.r;.-*! ncvrm.w
A BE R D E E N P R O VIN G G R O U N D . M A R YLA N D
BALLIST IC R ESEAR CH LABO R AT O R IES
M EM O R ANDUM R EPO R T NO . 1 5^9
AUG UST 1 96 1
A COMB FILTER FOR USE IN TRACKING SATELLITES
ARFA Satellite Fence Series
Richard L. Vltek
Ballistic Measurements Laboratory
Report Number 2k In the Series
Department of the Army Project No. 503-06-011
Ordnance Management Structure Code No. 5210. 11. 1^5
ABERDEEN PROVING GROUND, MARYLAND
BALLIST IC R ESEAR CH LABO R AT O R IES
M EM O R ANDUM R EPO R T NO . 1 3 ^9
R LV l tek/bst
Aberdeen Provi n g G roun d, M aryl an d
August I96 I
A CO M B PILT BR FO R USE IN T R ACK ENO SAT ELLIT ES
ABST R ACT
T hi s report presen ts the desi gn an d eval uati on of a I80 el emen t comb
f i l ter. T hi s un i t was devel oped In con jun cti on wi th the AR PA Satel l i te
Fen ce program f or the detecti on an d tracki n g of n on -radl atl n g satel l i tes.
T he pri mary purpose of thi s comb f i l ter Is to detect an d measure the
f trequen cy of short durati on Doppl er si gn al s In the presen ce of n ol ae.
T he f i l ter el emen ts have a ban dwi dth of 1 0 cps an d are spaced 2 0 cps
apart to cover a 3 800 cps f requen cy ran ge. A mul ti pl e pen an al og
recorder Is used to record In di vi dual f i l ter outputs. T he eval uati on
In cl udes data on si mul ated as wel l as actual satel l i te si gn al s.
T ABLE O F CO NT ENT S
Page
H f PR O DUCT IO N
SY O T EM AND CO M B PAR AM ET ER S o
FILT ER CO NSIEER AT IO NS ^
FILIER DESIG N , % j k
CIR CUIT DESIG N
l 6
T EST S AND EV ALUAT IO N
51
APPENDIX I, BIBLIO G R APH Y O P R EPO R T S IN T H E BR L-DO PLO C SER IES. . . . 55
imODUCTION
The Ballistic Research Laboratories have been engaged in tracking
satellites since the first Sputnik on October k, 1957. This tracking
was accomplished with the DOPLOC (DOploc Phase LOCk) system. Tracking,
in this case, means the reception of radio frequency signals and the
recording of their frequency versus time characteristics. The frequency
shift is a result of the Doppler effect associated with a moving source.
The received signals were so weak that a phase-locked tracking filter
was required to separate the signal from the noise. The signals were
obtained either directly from a satellite borne transmitter (active)
or indirectly from a ground transmitter by reflection from the
satellite (passive). A very brief description of DOPLOC will be
presented here as a background for the comb filter requirements. The
block diagrams presented have been greatly simplified and equipment
not directly related to the comb filter has been omitted.
Figure la is a block diagram of the DOPLOC receiving system for
tracking active satellites. With the signals received from active
satellites, low gain broadbeam antennas were used with a manually
locked narrow bandwidth tracking filter. The filter had bandwidths
from 0. 5 to 50 cps which gave a slgnal-to-noise (S/N) improvement up
to 1+ 2 db. This unit provided a clean, continuous output Doppler
signal. The results obtained with this system were excellent.
Continuous Doppler signals were received throughout the passage of
satellites from horizon to horizon. Multiple satellites were tracked
by having a tracking filter for each signal. Figure lb shows a
typical Doppler data curve of frequency versus time.
With the advent of passive tracking, the DOPLOC system was extended
to include illuminating the satellite and receiving reflected Doppler
signals. Because of the relatively weak received signals in the passive
or reflection system, high gain, narrow fan beam antennas had to be
employed. The Interim DOPLOC Satellite Fence stations, as set up in
September 1959 under ARPA sponsorship, contained three, fixed, 8 X 80
0
beamwidth antennas with the same type of receivers and phase-locked
7
filters used with the active satellite tracking system. This system,
shown In Figure 2k, no longer produced continuous data but gave three
portions of the Doppler curve as shown In Figure 2B. Received signals
above the noise threshold lasted 5 to 8 seconds In the center beam and
15 to 20 seconds In the north and south beams. These signal durations
were so short that manual lock-on of the tracking filter was no longer
feasible. An automatic system was designed which located the desired
signal and locked the tracking filter to It. This automatic lock-on
system functioned reasonably well, but It did not provide maximum
sensitivity or acquisition capability.
The final system as proposed to ARPA Is shown in Figure 3A. This
system contained a 1. ^ x 40 scanning fan beam antenna. The scan rate
was high compared to the satellite rate of travel and data were to be
obtained at short Intervals throughout the curve as shown In Figure SB.
The phase-locked tracking filter was no longer compatible with this type
(short duration) signal, even with the use of an automatic lock-on system.
A new filter-recorder combination was required which could locate and
record short duration signals. This lead to the development of the comb
filter herein described.
8
SYSTEM AM) COMB PARAMETERS
System Considerations
The comb filter design Is related to the electrical noise and
signal environment In which It must operate. This environment can be
divided Into three parts; desired signal, noise, and spurious signals.
These signals are determined by the DOPLOC system parameters. Therefore,
a more detailed look at the DOPLOC system Is required to determine Its
output signal characteristics.
A major advantage of the DOPLOC Satellite Fence Is Its Inherent
ability to provide all of the orbital parameters from the data obtained
during a single pass of a satellite. A minimum of twelve data points
are considered necessary to provide an adequate orbital solution.
Therefore, the antenna configuration and motion, which determine the
number of data points, must be tailored to provide at least twelve
data points. In addition to this restriction, there are three other
basic parameters which determine the signal conditions. They are the
carrier frequency, the satellite velocity, and the slant range to the
satellite. A carrier frequency of 108 rac and satellite velocity of
300 miles per minute give a Doppler frequency shift of approximately
12 kc. All three parameters determine the rate of change of frequency,
which, at closest approach, Is 25 cps/s at a satellite altitude of 1000
miles and 100 cps/s at 100 miles altitude. The maximum rate of change of
frequency of 100 cps/s determines the minimum usable bandwidth of 10 cps.
This bandwidth In turn sets the time available to recognize a signal at
100 milliseconds. This time of 100 milliseconds determines the maximum scan
rate for a given antenna beamwldth. Since twelve data points are required
on each pass, there la also a minimm scan rate associated with the low
altitude passes. If the antenna seem Is made fast enough to obtain at
least twelve points, where each point Is 100 milliseconds In duration, an
antenna beamwldth of l. k degrees is required. To summarize, the desired
signal characteristics are, a 12 kc frequency shift, a maximum rate of
change of frequency of 100 cps/s, and a signal duration of 100 milliseconds.
Spurl ouBSi gn al e
In addi ti on to the desi red si gn al , there are other si gn al s whi ch are
presen t an d i mportan t. In a ref l ecti on system usi n g a hi gh power groun d
tran smi tter, f orward tropospherl c scatter gen eral l y occurs to some exten t.
Fi el d experi en ce In di cates that thi s f eedthrough l evel can be equal to
the recei ver n oi se l evel . T hi s si gn al , al though f i xed In f requen cy, must
be con si dered In the f i l ter desi gn .
M eteors al so presen t a probl em. H ere, f i el d experi en ce In di cates
that two types of echos are presen t, a head echo an d a trai l echo. T he
head echo wi l l cover the en ti re Doppl er f requen cy ran ge at a very hi gh
rate of chan ge of f requen cy. T he trai l echo Is rel ati vel y f i xed In
f requen cy an d of l arge ampl i tude. T he trai l echos gen eral l y occur n ear
the cen ter of the Doppl er f requen cy ban d. T he head echos are di scri mi n ated
agai n st to a l arge exten t because of thei r hi gh rate of chan ge of f requen cy.
Noi se Ban dwi dth an d Ampl i tude
Noi se ban dwi dth an d n oi se l evel con si derati on s are of pri me Importan ce.
T he n oi se ban dwi dth Is Importan t In that the overal l ban dwi dth reducti on
determi n es the dyn ami c ran ge requi red of the comb f i l ter. T he n oi se l evel
must be con si dered both f rom the stan dpoi n t of absol ute l evel an d vari ati on
of l evel . Wi th a gi ven , f i xed, ban dwi dth reducti on , the n oi se l evel
determi n es the mi n i mum usabl e si gn al l evel . T he vari ati on i n l evel i s
i mportan t i n the automati c gai n con trol ( AG C) desi gn . T he R 3 90/A
recei ver used i n the DO PLO C system has a maxi mum I. F. ban dwi dth of 1 6 kc,
whi ch i s adequate to cover the 1 2 kc Doppl er f requen cy shi f t. T he maxi mum
audi o si gn al , wi thout an y dan ger of cl i ppi n g, i s on e vol t r. m. s. T he n oi se
l evel i s a f un cti on of the recei vi n g an ten n a ori en tati on wi th respect to
cosmi c n oi se sources, as wel l as the recei ver gai n . An overal l vari ati on
of 5 to 1 i n n oi se vol tage output of the recei ver i s possi bl e due to the
combi n ed vari ati on s of cosmi c n oi se an d recei ver gai n .
1 0
FILTER CONSIDERATIONS
Ideal and Practical mter for Sine wv
g
Once the Input signal characteristics have been defined, the
performance characteristics desired for the comb filter ay be outlined.
The basic comb filter Is a multiplicity of narrow band elements dlstrlb-
uted uniformly through a given frequency band. This Is shown graphically
in Figure i*. The problem may be better understood by first examining
and Ideal filter and comparing Its operation with the actual filter
Consider an Individual Ideal element as shown In Figure kB to be 10 cps
vide with a flat top and having Infinite side slope attenuation. This
10 cps filter is placed In a recording circuit as shown In Figure i*
The circuit contains the filter, an amplifier of gain K, and threshold
circuit and a recorder. The recorder Is an go no-go device. The threshold
and recorder combination are such that the input to the threshold circuit
must equal or exceed a given level E^ for the recorder to operate. Let
us make one further stipulation that the circuit have unlimited dynamic
range. These conditions describe an Ideal circuit element. To further
simplify the problem, let us consider the effectiveness of the circuit
to separate only two sine wave signals; first, , the desired signal,
whose frequency is contained within the bandwidth and second, E, , the
undesired signal, with variable amplitude and frequency but outSde the
bandwidth. What, if any, restrictions does this ideal circuit present
to these two signals?
With an ideal filter with infinite side slopes at the input, the un-
desired signal may be as close to the band edges as possible and as large
in amplitude as desired and it will never present any interfering output
signal within the desired bandwidth. Thus, there is no minimum amplitude
restriction on ^ as a function of E^. Also, as long as we are willing
to make the gain K large enough to amplify ^ to ^ E
D
has no minimum
level. Ep also has no maximum level. The signals may be as small or as
large as desired and still be recognized, and each will operate only one
circuit element.
11
V
Let us replace the Ideal filter with one of finite side slope of
x db/octave, and center frequency f , as shown In Figure 5A, and consider
its effect on the frequency and amplitude characteristics of E_ and E .
We select some arbitrary minimum level for E_, say y db, and plot . _
as a function of frequency vhlch will provide a signal level of y db at
f . This Is shown in Figure 5B. Since we have an attenuation of x db/octave,
a signal of y + x db one octave above or below f will be attenuated to y db.
Similarly, for two octave separation, EL can be y + 2 x db and so forth.
Therefore, if the signal at f due to EL must always be smaller than y,
there is a limit to the value of E _ as a function of its frequency proximity
to f . In addition to the interference produced within a given bandwidth, a
large signal can activate a number of circuit elements in proportion to its
amplitude. This is shown in Figure 5C. A signal amplitude of y + nx db
will be attenuated x db/octave and will cover n octaves before going below
the minimum level of y db. This will operate all the circuit elements
contained in n octaves, which is undesirable when reading the data. If
only one circuit element is to operate per signal, some type of AQC system is
required. To summarize, a non-ideal filter requires AQC and places a
restriction on the ratio of E _/_ as a function of frequency.
Effect of Noise
Let us consider the final condition of separating signal from noise.
We assume first that the noise is uniformly distributed throughout the
frequency band. Then, no matter which type of filter is used or where it
is positioned in the frequency band, it will contain noise. Therefore,
there is alwayn some noise introduced into the system. The amount of noise
is a function of the total noise power and the ratio of the input bandwidth
to the filter bandwidth. In this case, the input bandwidth and the filter
bandwidth are fixed so that the noise output of the filter is directly
proportional to input noise. The minimum level of EL is a function of the
noise out of the filter. Let us assume for the moment that E_ minimum is
equal to the noise level out of the filter. This then makes E_ minimum
directly related to total input noise. With noise in the system we can
no longer define E_ arbitrarily or even as a given voltage level. It
must be defined in terms of a S/N ratio with respect to the input or output
12
noise. To be consistent, then, we should define EL maximum and E.
maximum in terms of the noise level as a Reference. Thus, we can see
that the Introduction of noise into a circuit using a non-ideal filter
places restrictions on all the input signals.
Recording Data
A number of methods are available for recording the output of the
comb filter. In the works here described, a multiple pen, paper chart
recorder was used. There is one pen for each filter element. Since there
is no scanning involved, no time is lost searching for the signal. This
gives the additional capability of recording multiple satellites
simultaneously.
13
FILTER DESIGN
Overall Comb Filter
Having defined the Input signal characterletlcs and the relation between
these signals and a general form of narrow band filter, the overall comb may
now be outlined In block form, starting with the minimum block, any necessary
Items may be added as needed. Figure fe shows the basic receiver with the
three variables, l^, B^, and ^ as output signals, and a recording element.
We denote the signals out of the receiver by the subscript R and the signals
out of the filter by the subscript F, and consider these three variables and
their effect on the design of the comb. First,we consider the noise voltage
V Since there Is always some noise out of the filter, the gain K Is limited
to a value which brings E^ Just below the threshold level. Now we assume a
given ratio of E^ which gives a ratio E^E of one. E^ then Is the
minimum signal that will just operate the recorder. We conslL now the effect
of the receiver gain decreasing by a factor of two. The Input S/N ratio,
Ej^/E^, remains fixed, but E^ has decreased by a factor of two. If K is
fixed, then E^ will no longer operate the recorder, it now requires a
larger S/N ratio than before. If the receiver gain increases, the noise will
operate the recorder. Thus, to obtain maximum sensitivity, that is minimum
S
/N
ratio, and no extra readouts due to noise - the noise level must remain fixed
Just below the threshold level. Therefore, a constant noise input must be
provided to the filter with some type of AGO system. The conventional AGO
maintains a constant r. m. s. output by adjusting the sum of the noise plus
signal to be a constant. A large undeslred signal will cause the receiver
gain to be decreased and produce the undesirable effect of reducing the noise
output. An AGC based on noise only, which will maintain a constant noise
output even in the presence of a large signal, is necessary. Figure 6B
Aows this noise AGC circuit added to the comb. Having determined previously
that the noise out of the receiver can vary as much as 5 to 1 and that the
maximum level is one volt r. m. s. , the noise AGC circuit must be designed to
handle from 0. 2 to 1. 0 volts r. m. s, input and produce a constant output.
14

Let us consider the effect of the other variables _ and EL In the
non-Ideal filter. Large values of the sine waves will not be controlled by
the noise AOC, but they must be controlled If they are to operate only one
filter. At this point we consider the signal AGO as acting on all the
filters. If the signal AGC circuit operates on the basis of Input amplitude,
then pre-fliter AGC Is not possible. This Is so because the Input signal
can be very small compared to the noise and still be capable of operating
more than one filter. This small signal cannot be recognized In the noise
on the basis of amplitude. The post filter signal Is always larger than
the post filter noise and It Is at this point that we must develop the AGC
voltage. If the signal AGC Is developed from all the filters on the basis
of total signal and applied to a common amplifier, then only the largest
signal will be read. This Is not desirable. AGC must be developed on
each board and applied to only those boards which would have given an
ambiguous output, and preferably In an amount which Is proportional to the
received signal In each board. Fig. 6C shows the complete comb with
Individual signal AGC on each element. The methods of applying this AGC
will be covered later.
Wvunber of Elements
The next step Is to determine the number of elements needed In the
comb. This is a question of how many filters are needed in the 12 kc
frequency band. Considering that there are a minimum number of data points
available, and each lasts the minimum time, each point must be read.
Complete coverage of the 12 kc band would require filters at 10 cps spacing.
Since each filter element is fairly complex, a complete system would represent
a considerable investment. It was decided to try a token number of filters to
prove the system before expanding to the total 1200. This trial comb was
selected to cover only 5600 cps at 20 cps spacing with l80 filters. Its in-
put can be switched three times to cover the entire 12 kc band.
Having determined the overall block diagram for the system the
individual blocks can be developed in detail.
15
R ecei ver Noi se AG C
cn tcurF i ssi ai r
A need for an AQC circuit which is activated by noise only has been
discussed. Actually, it is Inpossible to make a circuit which is sensitive
to noise only. The best approach appears to be one which gives the maxi
mum rejection to the interfering sine waves. This rejection can be on the
basis of amplitude, tlrj, or frequency. Amplitude rejection is not feasible
because the interference can be larger than the noise, thus we must depend
on a time or frequency discrimination.
Let us consider first time discrimination. Up to this point consideration
has been given to the total change in noise level but not the rate of change of
level. The change in noise level is basically due to effective sky tempera-
ture change with antenna orientation. An antenna scan from horizon to
horizon requires fifteen seconds. Thus, a time constant of one second for
the AOC will permit the gain control voltage to follow the noise. However,
the desired signals of 0. 1 second duration will have little effect on the
AQC because they are not present long enough. Meteor trail echos and
signals received via scatter propagation from the transmitter which are
present for long periods will not be rejected by an AGC system with a 1 second
time constant.
We now consider what frequency discrimination will do. Feedthrough
(signals arriving via tropospheric propagation), is always present at a
fixed frequency. If the noise is sampled in a narrow frequency band it is
possible to exclude the feedthrough altogether. The probability of receiving
trail echos Is also reduced considerably.
Thus, it can be seen that time discrimination will eliminate short
duration signal interference, and frequency discrimination will eliminate
most of the feedthrough and meteor interference. Figure 7 shows a block
diagram of the circuit. The noise plus signal go through the gain controlled
amplifier to two outputs, the output to the filter bank and the output to the
control circuit. The amplifier is followed by a filter at a center frequency
of 6 kc with a bandwidth of l60 cps. The total noise bandwidth is 16 kc, so
a reduction of noise in the filter of 10 to 1 occurs. However, a signal in
16
this bandwidth Is not reduced, at all. Therefore, a S/N ratio of 1 at the
Input to the l60 ops filter would result In a S/N ratio of 10 at the output
of the filter, which would give too much weight to the signal. It Is
necessary to follow the l6o cps filter with an anpllfler-llalter combination.
The noise level Is adjusted to approximately one-half of full output and any
signal Is then limited to twice the value of the noise. This procedure still
gives more than adequate range for variation In noise level to operate the
AGO amplifier. The limit er Is followed by a rectifier, to convert the signal
to a d. c. voltage, and a d. c. amplifier to provide loop gain. There Is
sufficient gain In the loop to handle a 20 to 1 variation of the Input with
less than a 10 percent change In the output of the controlled amplifier.
The integrator provides the time constant which is one second. The value
of the time constant Is chosen to minimize the effect of signals with a
duration shorter than one second. For example, a 0. 1 second duration
signal large enough to produce a d. c. output of twice the noise level would
try to charge to twice the noise voltage level. However, it is present for
only 0. 1 second and would not Increase the voltage more than 10 percent.
Without AGO, the decrease in sensitivity would have been at least ko percent,
for a S/N of 1 and greater for larger S/N ratios. The only type of signal
which could cause interference would be a long duration trail echo. By
reducing the bandwidth we have reduced the probability of this occurring.
All other signal effects from feedthrough, short duration desired signals,
and high rate head echos are reduced considerably.
17
Q MLTIPLIKR
Bic Feedback Clreuit
The 10 ope narrow bend filter la im the form of a Q multiplier
using standard L and C components. Since the frequency band to be
covered Is 2 kc to lU kc, Q's range from 200 to lUoo to provide the 10 eps
bandwidth at the various frequencies. With nominal coll Q*s In the order
of 100, multiplication la from 2 to Ik. The problem was to design a small
compact circuit capable of providing these Q*s with as much stability as
possible. An analysis was made of Q multiplier circuits to determine the
best design criteria. Basically, any LC oscillator circuit with the
positive feedback below the oscillating point will. If used as an amplifier,
produce an Increase In the Q. However, the question Is, how stable Is this
new Q value, assuming nominal stability of the variables Involved. Before
this can be resolved, the circuit action which produces the multiplication
must be understood.
Two types of feedback circuits were compared, one In which the feed-
back factor, B, Is positive only and the other, where B Is a composite of
positive and negative feedback. In both cases the positive B network
uses an LC resonant circuit.
Let us consider first the basic feedback circuit, as shown In Figure 8,
In which B Is fixed. E, Is the Input voltage to the whole circuit, E Is
the Input to the amplifier of gain K, and B Is the feedback factor. Equations
1, 2, and 3 can be written by Inspection.
E
1
- E
1

+
E^
(1)
E^ - E
2 (2)
E
2
B
=
E
fb
( 5 )
Substituting (2) and (5) into equation (1) gives
E
2
/K = E
1
+ E
2
B
CO
18
R earran gi n g equati on (h) gi ves the f orm
E ( 1 -B) - E
R

V
E
1 " V ( l -BK ) - K ( 5)
Thl Is the fort of the standard feedback equation where K Is the closed
loop gain. Honnally In this type of circuit, B Is fixed with frequency
and K Is Investigated with respect to changes In K. However, In the Q
multipler It Is B which Is variable. Figure 9 shows the basic Q multiplier
In which B Is positive and the B network Is a resonant circuit. If we
consider the B network with E
2
as constant In amplitude but variable In
frequency, then E^ will vary as the LC circuit goes through resonance. Since
B Is defined as being the ratio of E^ to Eg, B will also vary with frequency.
The variation of B with frequency Is shown In Figure 10. The 3 db bandwidth
is determined by the loaded Q of the resonant circuit.
Q Multiplier as a Function of the Output Voltage
If the Q multiplier can be evaluated in terras of the output voltage
then it can later be defined in terms of B and K. In Figure 9 , If the
feedback circuit Is disconnected at point x and a curve is drawn of output
versus frequency, curve A in Figure 11 would be obtained. Then, upon re-
connecting the circuit, draw a second curve of output versus frequency and
curve B would be obtained. Since the feedback is positive, curve B would
be larger in amplitude. It is noted that curve B has been adjusted in
amplitude so that the intersection of the two curves is at E which is the
3 db point on curve A. Also, the 3 db bandwidth, Afg, of curve B is smaller
than the 3 db bandwidth, Af^, of a curve A. If we denote the Q of curve B,
Q
2
, and the Q of curve A, Q
1
, then the Q multiplier, M, is the ratio of the
Q's as shown in equation ( 6) .
M = Qg/^ = A^/Af, , ( 6)
Let us denote the peak value of curve A, E^, and the peak of curve B, IL .
If the assumption can be made that R
1
and R
2
in Figure 9 are greater than
the Z of the resonant circuit, then the voltages are directly proportional
19
to the Q* a and:
By substituting E-. i-
1
* E Into equation (7):
This defines the multiplier in terms of the change of output voltage from
E
o * > ^2
Q K ul ti pl i er i n terms of Ban d K
Curve A In Fi gure 1 1 an d the Beta curve of Fi gure 1 0 are both drawn
f or the same ci rcui t Q ; theref ore E represen ts the output E_ wi th 0. 7B
O e.
an d E-2 the output wi th f ul l B. Equati on s 9 an d 1 0 gi ve the cl osed l oop
gai n wi th 0. 7Ban d B, respecti vel y.
K - K /( l - . 7BK ) ( 9)
q - K /( l - BK ) ( 1 0)
Equations 11 and 12 give E-p and E in terms of the closed loop gains.
E
1
KJ - E
0
(U)
E
l " l - ^ ^
Substituting equations 11 and 12 into equation 8 gives:
KJ/IA iq, - M ( 13 )
Let
K
i/ ^ '
A
^-^
Then A/l.4 - M ( 15 )
" A" is then the variable part of the multiplier M and will determine the
stability of the new Q. " A" can now be expressed in terms of B and K by
substituting equations 9 and 10 Into equation 14 .
l4 / ir^c "
1

+
TS- (
16
'
Figure 12 shows a plot of A versus BK. As BK approaches 1, A approaches
infinity and the circuit will oscillate. It will be noted that the useful
20
values of A lie In the range of BK from O. 95 to O. 999. This means that the
circuit Is always relatively close to the oscillating point. Even with a
low value of A of 6, a 5 percent change in B or K would cause oscillations.
B, which Is made up of passive elements, can easily be held to one percent
or less. However, K which Is the open loop gain is not nearly as stable,
and In transistor circuits would be especially difficult to maintain. A
circuit configuration which Is relatively independent of K would be much
more desirable, and this is the second circuit to be discussed.
The second circuit contains two loops, one positive and one negative,
and is shown In block form In Figure 13. In this configuration the negative
feedback Is fixed and the positive feedback varies as before. The total
feedback, B, is the algebraic sum of the negative and positive B's. The
equations at resonance and at the 3 db point are as follows:
at resonance the total B B. B + B- (17)
K^ - K/{1- B
t
K) (18)
at the 3 db poi n t B . - 0. 7B
+
+ B- ( 1 9)
K ^ - K /( l - B; K ) ( 2 0)
By rearranging equation 1? and 19 and substituting into equation 20 , KX may
be expressed in the following form:
W , - K/( l- B
t
+ 0 .5 B
+
) ( 21)
A, which is the ratio of Kj/K^, may be expressed in terms of BK as:
0 .3 B K 0 .3 B.K
A
'
l
-
B
t
K

+
-LTl^K ' 1 TT-B^K (
22
)
At this point a restriction must be made that B may never be positive. Then
rearranging equation 22, by multiplying the second terra by B /B ,
0 .5 B B K
A
-
: i

+
B
t
l-B
t
K ( ')
It will be noted that as K approaches infinity, the terra B.k/( 1- B.K) approaches
unity as a limit. A then approaches the value 1 0 .3 B /B. . Figure Ik shows
a plot of equation 23 as A versus KB. for various values of B /B. . One can
see that A in this case is relatively independent of KB. as long as KB is
21
.
V
kept above 25. A Is, however, very dependent on the ratio B /B . If the B
network. , which re nade ^ of passive elements, are held to
+
close tolersnces
the circuit will be very stable, it Is virtually Independent of K as long as
K Is large.
Description of Final Circuit
Figure 15 shows the schematic of this second circuit, which Is the
circuit a. used in the filter b*. The negative loop Is made up of R
fl
and
R
6
. The positive loop Is ^ and R
2
In series, plus the resonant circuit.
The basic amplifier Is d. c. coupled and RQ and R
6
provide negative d. c. as
well as negative a. c. feedback. This helps to, stabllze the d. c. operating
point considerably. The circuit Is very tolerant of any variations In
transistors. The Zener diode was added Instead of a dropping resistor to
provide additional gain. R^ d C, form a necessary phase correction network
to stop oscillation caused by the large amount of negative feedback. Frequency
is adjusted with the choice of coil and capacitors, and the Q is adjusted by '
setting the positive feedback level with R .
Measurement of Q is difficult at the higher frequencies when one attempts
to measure 5 cps at 12, 0 0 0 cps. A method was devised to use an oscilloscope
and speed UP this measurement. If the response of a resonsnt circuit to a '
step function is calculated, the envelope has the form,
Amplitude of Envelope - S - E e "
at
( 24 )
where a - R/2L, R is the series resistance of the coil and L is the Inductance
of the coil. Since Q - 2 fL/R, at resonance,
a - R/2L -

f/Q . Af, ( 25 )
where Af is the bandwidth between 3 db points.
This equation Is the same as the standard equation expressing the charge of a
capacitor. The time constant In this case would be equal to l/a. The time
constant can then be expressed as:

T = time constant l/nAf ( 26)
If a signal at the center frequency of the resonant circuit Is applied
whose amplitude Is in the for* , of a step function and a measurement is made of
the time response, we have a measure of the bandwidth. The time constant is
3 2 milliseconds for all of the filters, irrespective of the frequency of the
22
of the filter. This method has the advantage of permitting measurement of
bandvidths less than 1 cps with ease and speed compared to the 3 db method.
Tests Indicated that 1/2 cps bandvidths could be obtained before approaching
the oscillating point. This (1/2 cps bandwidth) gives a multiplication of
approximately 200 as the ultimate capability. Since the largest multiplication
needed vas 14, all of the filters were operating well away from the
oscillating point.
25
SIGIIAL AOC
Signal Contirol and AQC Pletrllmtlon
The need has been previously determined for a post-filter signal AQC
circuit. We have also seen that the filter used In this case will be a
resonant circuit. This section will deal with the gain controlled amplifier,
the generation of the control voltage and the distribution of this voltsge
smong the filters of the filter bank.
Assume an Ideal delayed AQC which starts to operate with a signal
Just above the threshold level and has an output d. c. voltage which Is
directly proportional to the input a. c. voltage. How consider the distri-
bution of this d. c. voltage to the other filters. To do this visualize a
group of filters with an input signal in the center filter. For ease of
description, let us call the center filter A and the other filters as they
progress away on either side of A, B, C, D, and so forth. This is shown in
Figure 16k. There will be some signal voltage at all the filter inputs due
to the signal at the Input of filter A. Since all the filters are resonant
circuits, and of the same bandwidth, distribution of the voltage at each of
the Inputs Is in the form of a resonance curve as shown in Figure l6B. The
dashed line with an amplitude of X represents the threshold level. All the
filters contained within the resonance curve above the threshold level would
be actuating the recorder.
Since the d. c. output voltage is directly proportional to a. c. Input
voltage. Figure l6B also represents the AGC d. c. output of each filter.
Filter A, which contains the signal at its frequency, has the largest d. c.
output. This voltage can be distributed to the other filters with a resistive
network, always to be larger than the AGC developed in the other filters.
This is shown by the dotted lines in Figure 16B. The circuit can also be
arranged to have the larger of the two AGC voltages control the gain in each
filter. This will then cut off all of the filters except filter A. The
resistive network is made in the form of a symmetrical ladder attenuator
with equal sections. Since all of the filters are connected together with
identical sections, the distribution of voltage is not peculiar to a
2k
particular set of filters. The voltage will be distributed as shown on
both sides of the strong signal, no matter which filter the signal
frequency matches. This type of circuit has the advantage of keeping a
minimum number of filters cut off. As an example, Figure 17 shows two
main signals X and Yof different amplitudes and their AGC curves. The
frequency covered Is a small portion of the total h kc band. Each
signal cuts off a number of filters In proportion to Its amplitude. Yet,
a signal such as A contained within the cut off region but larger than
the distributed d. c. voltage will still be read. Any signal not In the
cut off region can be read with maximum sensitivity. Thus, It Is
possible for meteors, feedthrough, and many desired signals to be
present simultaneously and still read each one by the operation of a
single pen per signal.
Circuit Analysis
In the general description of the AGC action, several points were
mentioned and assumed to be true without giving any proof. These were;
the linearity of a. c. input to d. c. output, the cut off action of the
distributed voltage, and the design of the ladder attenuator. These
items will now be taken up in turn.
Linearity of a. c. Input Versus d. c. Output
Figure 18 shows a combined block and schematic form of the AGC
circuit. The d. c. control voltage E^, is fed back across the series
connected resistor R
1
and diode D^^ This voltage determines the current
and thus the resistance of D^^ ^ is a very low impedance to the
operating frequency and acts only as a d. c. blocking device. D , in
series with R
2
to the input voltage , determines the voltage E
available to the fixed gain smpllfiers which follow.
The forward drop across a diode is relatively independent of the
current through it and for prsctical purposes can be considered a constant.
If I represents the current through the diode, and R the resistance of the
diode, then this constant voltage drop may be expressed as:
I lU = constant = A (27)
25
!
It can also be expressed In the form:
V
(R
1
+
V
I
(28)
If R
1
is much larger than R- then equation 28 reduces to:
^^l "
A
(29)
Substituting equation 29 in equation 27 ve have:
hchfil '
A
( 3 0 )
Since R
1
is fixed in value equation 3 can be written in terms of a new
constant as:
Ej^ Up - AR
1
- B constant (31)
If it is further assumed that there is sufficient loop gain to maintain
the output of the a. c. amplifier constsnt, then:
E con stan t - E G
D (52)
E
2
G
2 *
E
IH R ^T T g
0
2
If R
2
is l arger than R^ then:
E
o "
E
IN ^-
G
2 -
C
(55)
Since 0
2
, R
2
, and E
o
are all fixed in value,
E
o
R
2/
G
2 * constant
E
IN'^ ^
D
(5^)
Substituting the value of IL from equation 31 Into equation 34 gives:
a
8
^ '
D
E
IN '
D
/
BE
DC
=

KE
DC (
5
5)
Equation 55 shows E__ to be a linear function of E .
DC IN
26
Cutoff Action of AQC
^^____ ^
It is readily apparent that we can feed back a voltage which is larger
than the AOC voltage on any given filter, but it would appear that this
voltage would merely reduce the gain and not cut off the filter. The
following discussion will show how the cutoff action takes place. Figure 19
gives a plot of two curves, curve A, the a. c. input versus the d. c. control
voltage, and curve B, the a. c. input versus the threshold gain. By threshold
gain is meant that gain which is required to give an output at the threshold
level from the AQC controlled amplifier. Let S represent the signal in filter
A, and P represent the signal in filter B. At signal level P, Q represents
the AOC voltage developed and R is the gain required to bring P up to the
threshold level. However, T represents the AOC voltage at filter A and U
shows that portion which is fed back from filter A to filter B. If the
voltage U controls filter B then it will produce gain V which is lower than
gain R, and filter B will no longer have adequate gain with input P to
reach the threshold level. This feedback voltage will work in a similar
manner for all the other filters.
Ladder Attenuator
Figure 20 shows the network and filter connections. Let us consider
only a part of the network as shown In Figure 20 b. Let E. be the Input
voltage and E
2
, Ej, and E^ be the voltages down the attenuator. Since each
section has an impednace Z, then:
E^ - Z/( R
1
+ Z) - Ej/Eg E
u
/E
5
. X ( 5 6)
There is a fixed loss ratio of X as one progresses down the network. Equation
3 6 can be rearranged to give R as a function of Z and X .
R
1
= Z ( 1- X ) /X ( 3 7)
Using only one section of the network as shown In Figure 20 c, Z can also be
written equal to the parallel combination of R
2
and IL plus Z In series:
( 3 8 )
( " a; (
K
i z )
R
2 + (1^ + Z)
2 7
Substituting equation 37 In equation 58 gives:
(59)
- R
2
f/Rs * | )
R
2
- Z/(l-X)
Thus, equation. JT and 59 give ^ d ^ In terms of 2 and X, where Z Is
determined by circuit Impedances and X Is a function of the distribution
desired. Pigure 21 shows the resonance curve (a) as well as the distribution
curves for several values of X. The final choice of X Is somewhat arbitrary,
however, there are some circuit considerations. To be sure that the larger
voltage always controls the AOC, a diode D
2
was added as shovn In Figure 18.
To obtain the best circuit operation, one must alwsys maintain a few tenths
of a volt back bias across Dg. This requires that X must be chosen so that
the curve Is high enough at Its extremities. X Is not a fixed value, since
the resistor R
2
la actually the parallel combination of R and the filter
resistance, as shown In Figure 2QA. This Input resistance Is variable with
signal level and will vary from 2QK with large signals to 1 Megohms with
small signals. This condition is helpful since it produces larger values
of X with small signals which will occur at the extremities of the curve.
In this particular case, linearity of the signal in the AGC amplifier
is not too important. However, it is important in the noise AGC amplifier
The input voltage across the diode must be kept below 50 millivolts to keep
distortion at a minimum.
28
Threshold Circuit
This section covers the threshold circuit and the relation of the s/w
ratio to the threshold level. Up to this point the minimum usable signal
was considered as one which produced a S/H ratio of unity past the 10 cps
filter. Actually this Is the theoretical limit which we would like to
achieve, hut never quite realize In practice. This Is best explained by
reference to the block diagram and the various signal levels which occur
as shown In Plgure 22. The signal and noise are rectified, filtered, and
then go to a differential amplifier. The other Input to the differential
amplifier Is a reference level which determines theltoreshold level. When
the rectified Input exceeds the threshold level the recorder writes.
The signals are presented plctorlally below the block diagram of
Figure 22. At the input to the rectifier shown In (A) we have the signal
whose frequency Is f
c
, and the noise which contains the frequency components
between f
c
- 5 cps and f
c
+ 5 cps. The next set (B), shows the full wave
output of the rectifier. Up to this point the signal is in the frequency
range between 2 kc and 6 kc. Set (C) shows the output of the low pass
filter. It is seen that the signal d. c. component is clean because all of
the ripple components are high compared to 10 cps and are removed. The noise,
however, consists of all frequencies between f -5 cps and f + 5 cps and
its rectified output contains all the difference frequencies between the noise
components in this band, that is, 0 to 10 cps signals. The rectified noise
output then contains a d.c. component and ripple in the 0 to 10 cps region.
The next set, ( D) , shows two conditions, one, when both signal and noise are
present and Just equal to the threshold level, and the other when the same
amount of noise is present but without signal. In the first case the noise
ripple causes the voltage level to go below the threshold which causes a
miss in the output when the signal is present. The second case shows the
noise ripple exceeding the threshold and giving a false data pulse when
the signal is not present. The total number of false alarms which can be
tolerated depends on the read out system. The general approach to eliminate
these errors is to lower the noise and increase the signal, as shown in
29
Figure 22 Z, to olnlalse the occurrence of misses and false alaras. This,
of course, calls for a better S/H ratio and lowers the overall sensitivity
of the system. The actual loss can only be determined by experience with
field data on a particular system. Since the final system has not been
constructed, a firm figure Is not available. However, general tests Indicate
that a loss of 6 db to 9 db could be expected.
50
TESTS AND EVALUATION
General
A brief description of the tests performed on the comb filter Is
given In this section. Initial tests for stability of the frequency and
Q of the filter disclosed a need for temperature control. The relay
rack used to house the filters and power supplies was equipped with
blowers, heaters and a thermostat to provide temperature control. A
temperature range of from 92 to 98 degrees P was maintained with this
system. This was adequate to maintain the frequency of the filters
within + 2 cps and the Q within 10 percent of design values. However,
the threshold level and threshold gain had a 3 to 6 db variation in
sensitivity between individual filters. As the following figures,
5 3 through 3 ^, show not all of the filters begin to write at the same
signal level. This sensitivity effect is most noticeable when viewing
the noise background, which shows light and dark lines instead of uniform
gray lines. This effect can be eliminated by a modification of the d.c.
amplifiers in the filters and recorder, but was not done on the prototype
because of the time required to make the change.
Since the main use of the comb was for tracking satellites, tests
were made for specific satellite tracking problems, as well as general
problems. The following figures show simulated satellite signals as
well as actual satellite signals obtained from field station magnetic
tape records. In Figures 23 through ^\ the paper was moving vertically
at 0 .1 inch per second, so the vertical axis represents time. The pens
are spaced horiz ontally across the paper and each pen is connected to a
filter, the filters being spaced 20 cps apart. Each presentation is
then a frequency versus time graph. A total of 120 filters were used in
the tests, giving a total range from 2 kc to 4 .1+ kc.
Simulated Signals
As mentioned in the previous section, the relation of the noise level
to the threshold level is determined by the type of recording system to be
used with the filter. The first test was to determine the optimum noise
3 1
level for maximum signal-to-nolse ratio, as Is shown In Figure 23* Signal
was mixed with noise and applied to the Input of the filter. The noise
level was maintained fixed while the frequency of the Input sine wave was
varied from 2 kc to k. k kc. During the sweep, the amplitude of the sine
wave was varied in amplitude in three steps, which divide the sweep into
three equal parts. The sine wave amplitude steps used were - 27 db, -30 db,
and -33 db with respect to the noise in all cases. The noise levels were
from 100 millivolts to 250 millivolts as shown in the figure. By inspection
of Figure 23 it is easy to see that the 150 millivolt noise level is optimum.
These tests were made using a 20 kc bandwidth noise source and the 10 cps
filters. With this combination, and input signal-to-noise of -33 db would
provide a post filter signal-to-nolse ratio of one. It is seen that at the
150 millivolt noise level some points are still readable at the -33 db sine
wave level. With noise levels above 150 millivolts the system becomes
saturated with the noise, and at lower levels the system becomes less
sensitive. However, in the scanning beam antenna tracking system, which
provides fewer data points, it would not be possible to read the data at
the optimum sensitivity noise level. There would be hundreds of noise
points for each data point. In the Interest of trying to anticipate the
loss of sensitivity associated with fewer data points. Figure 2k shows
higher post filter signal-to-noise ratios. Figure 2k is similar to Figure 23
except that the noise levels vary from 100 mv to 70 mv and the signal levels
vary from -21 db to -30 db. In the scanning beam system, data points would
have been obtained every 15 seconds or 1. 5 Inches apart on the graph paper.
If one used the criterion that there should be approximately the same number
of data points as there are noise points then the 70 mv noise level with a
-21 db sine wave level would be about the right choice. This is 9 db less
sensitive than the optimum case which was anticipated.
One very important feature is the response of the filter to a signal
with a high rate of change of frequency. Tests were run on the comb filter
to indicate the minimum usable signal for a given rate of change of
frequency in cps/s. The results are shown in Figure 25. It is interesting
32
to note that a loss of sensitivity did not occur until a rate exceeding
Af was resched, and that rates in excess of 100,000 cps/s could be seen
with a signal-to-noise ratio of one at the input to the filter.
The next test, shown in Figure 26, was with various signal levels to
test the action of the signal AOC. Signals from -30 db to + 6 db with
respect to the noise were used. The AOC performed well up to 0 db. At
+ 6 db the system began to overload as indicated by the dark traces in
the background noise. Figures 27 and 28 show the action of the AOC
with two signals present. Figure 27 shows a strong fixed frequency
signal at -6 db level. By noting the background noise it can be seen
that this signal is strong enough to cut off the noise from 18 filters
and the chart appears clear on both sides of the strong signal. The
second signal is a sweeping frequency with amplitude levels from -12 db
to -30 db. It can be seen that the stronger signal at -12 db overrides
the cutoff action of the -6 db fixed signal, but the weaker does not.
Figure 28 shows two signals crossing simulating the data from two
satellites tracked simultaneously. Since they are at the same amplitude
all of the points are readable on both curves.
Satellite Signals
During the time that the satellite fence was in operation, many
active and passive satellite signals were recorded on magnetic tape.
These magnetic tapes were used to produce the chart records shown in
Figures 29 through j k. As a means of comparison with the methods used
at the APRA stations, chart records are shown of the same satellite
passes that were tracked with the ALO-tracking filter combination.
Section " A" of each figure is the record made with the comb filter, and
section " B" is the ALO-tracking filter record. Figures 29 through 53
show satellite i960 Delta revolutions 12i*, iko, 156, 165, and 172.
Figure 3^ shows an active track of Jupiter C during the launch phase.
The arrows on the figures indicate the frequency limits of the tracking
filter records. In general, the comb filter tracked the signal for a
longer period. This was expected since the best sensitivity of the ALO
3 3
was -21 db, and could not pick up the signal as soon. However, after a
lack was obtained the tracking filter was Independent of the ALO and
Its sensitivity was Identical to the comb filter.
The signals on the tapes had a frequency range from 2 kc to 12 kc.
It was necessary to mix the tape signals with a signal from an oscillator
to provide the comb filter with frequencies In the 2 kc to U kc range.
Tests with simulated signals on the mixer Indicated a 6 db loss In the
S/N ratio as a result of the mixing. Therefore, In comparing the two
types of records, the signal level for the comb filter should always
be taken as 6 db less then that for the ALO record.
8uary
The test records of simulated and actual satellite signals show that
the comb filter prototype has met all of the design requirements. If this
unit had been expended to the full compliment of 1200 filters it would
have provided an effective tracking capability for the scanning antenna
system. This model has proven the advantages and disadvantages of this
type of system. It provides an excellent answer to tracking short duration,
high rate of change of frequency signals.
RICHARD L. VTEEK
54
BRL-DOPLOC REPORTS
Ho. 1
Ho. 2
Ho. 3
Ho. 4
Ho. 5
Ho. 6
Ho. 7
Ho. 8
Ho. 9
Ho. 10
No. 11
No. 12
No. 15
HRL Mono Report Ho. 1055 - October 1958, Doppler Signals and
Antenna Orientation for a Doppler System, by L. P. Bolglano, Jr.
CONFUEHi'IAL
BRL Nemo Report Ho. llj - January 1959, First Semi-Annual Technical
Summary Report, Period 1 JMly 1958 - 31 December 195, by L. 0. deBey,
V. W. Richard, A. H. Hodge, R. B. Patton, C. L. Adams. COHPHKHTIAL
BRL Tech Hote Ho. 1265 - June 1959, Orbital Data Handling and
Presentation, by R. E. A. Putnam. UHCIASSIPIED
BRL Tech Hote Ho. 1266 - JHily 1959, An Approach to the Doppler
Satellite Detection Problem, by L. 0. deBey. COHPnEHTIAL
BRL Memo Report Ho. 1220 - JUly 1959, Second Seml-Annual Technical
Summary Report, Period 1 January - 30 June 1959, by L. 0. deBey,
V. W. Richard and R. B. Patton. COHFITgNTXAL
BRL Tech Hote Ho. 1278 - September 1959, Synchronization of Tracking
Antennas, by R. E. A. Putnsm. UNCLASSIFIED
BRL Memo Report Ho. 1237 - September 1959, A Method of Solution for
the Determination of Satellite Orbital Parameters from DOPLOC
Measurements, by R. B. Patton, Jr. UHCIASSIPIED
BRL Memo Report Ho. 1093 - March i960. The Dynamic Characteristics
of Phase-Lock Receivers, by Dr. Keats Pullen. UNCLASSIFIED
Station Geometry Studies for the DOPLOC System, Stanford Research
Institute. UNCLASSIFIED
Final Report, Part B, Stanford Research Institute - July i960, DOPLOC
System Studies, by W. E. Scharfttan, H. Rothman, H. Outhart,
T. Morita. UNCLASSIFIED
Philco Corporation - k May i960, Polystatlon Doppler System.
UHCIASSIPIED
Space Science Laboratory, General Electric Co. - October i960.
Orbit Determination of a Non-Transmitting Satellite Using Doppler
Tracking Data, by Dr. Paul B. Richards. UNCLASSIFIED
Final Technical Report - University of Delaware - June 15, i960,
Quantum Mechanical Analysis of Radio Frequency Radiation, by L. P.
Bolgiano, Jr. and W. M. Gottschalk. UNCLASSIFIED
Ho. lU Plnal Report F/157, Columbia Italverslty - Pebruary 11, i960,
SunMtry of the Preliminary Study of the Applicability of the Ordir
System Techniques to the Tracking of Passive Satellites.
UrciASSIPIBD
Ho. 1? BRL Report Ho. 1110 - June I960, Precision Frequency Measurement
of Holsy Doppler Signal, by V. A. Dean. UNCIASSIFIED
Ho. 16 Third Technical Summary Report - Period July 1999 thru June 30, i960,
BRL Nemo Report Ho. 1287, by A. L. deBey. UHCLASSIPIKD
Ho. 17 Columbia University Tech Report Ho. T-l/157 - August 1, 1959,
The Theory of Phase Synchronisation of Oscillators with Application
to the DQPLOC Tracking Filter, by E. Krelndler. UHCIASSIFIED
Ho. 18 BRL Tech Hote Ho. 13U5 - August i960, DQPLOC Receiver for Use with
Circulating Memory Filter, by K. Patterson. UHCIASSIFIED
Ho. 19 BRL Tech Hote Ho. 13U5 - October i960. Parametric Pre-Amplifier
Results, by K. Patterson. UHCIASSIFIED
Ho. 20 BRL Tech Hote Ho. 136? - December i960, Data Generation and
Handling for Scanning DQPLOC System, by Ralph E. A. Putnam.
UHCIASSIFIED
Ho. 21 BRL Report Ho. 1123 - February 1961, The DQPLOC Instrumentation for
Satellite Tracking, by C. L. Adams. UHCIASSIFIED
Ho. 22 BRL Memo Report Ho. 1330 - March 1961, DOPLOC Observations of
Reflection Cross Sections of Satellites, by H. T. Lootens.
UHCIASSIFIED
Ho. 23 BRL Memo Report Ho. 1362 - August 1961, Satellite Induced lonlzatlon
Observed with the DOPLOC System, by H. T. Lootens. UHCIASSIFIED
Ho. 2k BRL Memo Report Ho. 13^9 - August I96I, A Comb Filter for Use In
Tracking Satellites, by Richard Vltek. UHCIASSIFIED
Ho. 23 Final Summary Report on the BRL-DOPLOC Project - July I96I, by
A. H. Hodge and R. B. Patton, Jr. UHCIASSIFIED

1
1
1
s
1
i
a
o
i Z
I i
00
2
Q.
A 3 N 3 no 3 dj vai ddo a
^
1
i
i
hl
B
hl
A B
- "
III tk
E
.
I
s
s
u
3
w
s
9
a.
i
i
f
U

U J
W <0
a c
A0ManO3IJ N 3 -IM 00
3 9
8
s
I
II
ui
u
t
5
* i.
B S
2
0
s
I
8
s
2
->
g
I
z
Ik
U I
K M
H 0.
<

5
o
>
.
-1
y <
9 U I
bl
3
a
O
U I
ao ni i i rf Hv

0 N O U VO N aXi V
1
UI
t
3

5
o
3
UI
8

a.

K
U
o
.
o
UI
u
S
2
c
!

s
o
o
ae
UI
o
o c
o
o
UI
o
<
ae
ui
^o
S - u
III
i i
a N o uvnnai i v
5
I
X
i
o
to No uvnNaxiv
>-
i
o o
z
U J
E
3
o
U J
a
K IT f c
u.
(A %
M ^
>
*m
M^r ^
Z
>-
1
i
U l
- >-
!
>
* yS
0
M >
8
i
o
i
z
U J
z
U J
3
o
U J
u.
55
U J
o
z
U J
3
O
^
^N v
w
i t
U J
o
<
z
(9
H U J A
3 55
s o c - ! H
b.
E
-1 *
3 X
-1
M
s
2
i
<
>
X
<
U J s
o _i
i
<
z
z
o
o
z - 1 1
W)
ao N O U Vf l N ai i V c i
>

i ,l
8
o
8
8
I
T
H
R
E
S
H
O
L
D

C
I
R
C
U
I
T

1 -
1 3
1 1 1 bl hi
bf h? hi
|
I
(
t
u
p

s
i s
t o
w m
tt
x u
5-
u S
o E
w o
5 ^
S w
bl
>
bi
O
w
1
!
I
T
N
R
C
S
H
O
L
O

C
I
R
C
U
I
T

T
H
R
E
S
H
O
L
D

C
I
R
C
U
I
T

*
u u
N Ii
I
O
C
.
F
.
S
.

F
I
L
T
E
R

I
O
C
.
F
.
S
.

F
I
L
T
E
R

m
1

1

!
<9
U
o
oc
hi
u.
0
2
O
o
<
hi
<0
2
1 . 2
<
o c
CP
<
a
o
o
u
o
6
<
J
(0
( 9
hi
i
o
Ul
f t!
u
<
o
tZ
3
o
ac
a
S
<
z
o
M
Ul
o
i Z
$ _
I
1 atN O dS JM 3 A U V1 3 II
* * ** * ** **"
w
bi
K
b.
^ 3 S N O d6 3 M aA l i V1 3 M
^5
i
M
(0
Q.
O
o
o o
a I
u o
< z
O D <
Q
Ul
u
C
1 2
o
M 01 9 VJ N O IXVO I1 4 IXi nN 0-V
I
o
o
" Q Q .
5
^
=i Q
o >
<n E
> O T
O
X 0.
D
< a
CD o
Q b.
Q
j S
(VJ
8
i Z
^6

+1 "
CO
u
3
%
M
o
K
s
I
( 9
vr
>-
H
I vSAr
8
y^x.
^\
1 :'
o U
n
I
2
-WA r
-A A A r
s
tM i maamm
u
I
2
u
X

M
oe
UJ
3
2
O
j O
o
i Z
1 . 8
i
o
5

u



-f -u 9
-- w
2
I
5

S
u
S
i
c
1
I
/
^Z
I
s
o
m
w
a
o
w
5
bi
5
o
>
S
-
hi
Z
O
z
-I
<
z
( 9
i Z

aani ndN V
ao m nwN V aA i i vi aH
5
U9
^-'
^
"^.
V
I \
5
m
o
w
7
A
\
>
o
z
Ul
I
It
o

hJ
l am ndnv aA i xvi i i i
50
<
III
X
M
O
I
w
Jo
1
'"
i S
u
i V NA/ i
u
a
N
o :
i WV
^ I s
o o
>
U Q
2
i
3
o
z
o
CO
v .
<
5
<
m
a
-i
o
CO *
w ~
i - c
o
z
<
9
S
o
Q C
o
u
o
( 9
<
3 0f ur
>i s o v
52
i
M
N
< w
N
< M
S i h^VA r " M WN 4 CU
4 4 +-r
CZK^
< M
1 I^A HN
LJ
^N A r| l
w t-A A A rj Il
!
8
55
I
M
<9
l i i
i
O
Q
<
Noi i vn Naxi v l Auvi au
54
w
>
u
V)
_l
I
o
I
I
a
i
55
o rrwuM N O i ti LE VE L
mMtmm WIH IT IV IT V
VM IIA ILI N O Itl
M
If N A L Lf VE L
. - , .
N M K Lf VIL : M O M V M M ' 'M U I
-
tmtMP R A T C : M CYCLKS /M C*
nrccp LE VE LS : A S S N O VN
I
I
<,
-S M k . . j ^
i
NO ISE LEV EL. 'ISO M V R M S
SWEEP R M E : SO evCLES/SEC. ' M " v
l ''. 'V* *' ' ' ' : ' i i*. . 'a. i j v! i
H ,-} N O IS E LE VE L : S O O M V R M S Jf |
3 g WE E P R A T E : S O CYCLE S /S E C* | | j j
0
:
J S WE E P LE VE LS ! A S S HO WN i | f
re|ls t m -- uwlm
st'W
'I i " * I W ' i
m m m m
MjWR ||fe|j. ;ii
> :" : l 'i . 'tf l i f f ! ri -
1
i ! i i
ml
HI
^- N O IS E LE VE L: 2 9 0 M V R M S
y I ] S WE E P R A T E ! S O CYCLE S /S E C.
S WE E P LE VE LS ! A S S HO WN
I r
. 1
im si
FR E QU E N CY S CA LE
FIS . IS
O P S .
56
1 tl Wtm VIT Y CM A WT
VWA i LE N O IS E A N D S I8N A L LE VE L
D E T E R M IN E O P T IM U M S E N tl T IVIT V
T O
N O IS E LE VE L : ICO M V R M S
WE i P R A T E : 1 00 CYCLE t/S BC*
S M N A L LE VE L ! A t S HO WN
-
1 7
O
:
N O IS E LE VE L : S O M V R M S
S WE E T R A T E : N X) CVCLE S /S E C*
S N HA L LE VE L : A S S HO WN
!
[
N O IS E LE VE L : S O M V R M S
S WE E P R A T E : 1 00 CVCLE S /S E C*
S IS N A L LE VE L : A S S HO WN
I'
N O IS E LE VE L : 70 M V R M S
S WE E P R A T E I 1 00 CVCLE S /S E C*
S IS N A L LE VE L ! A S S HO WN
-tin
4 0
I
2
i
2
i
I
2
i
5
!
Cf M
i
I
S
5
FR E O U E N CV S CA LE - C. P . S .
Fl . 2 4
57
CM
O
qp - ao m ndwv wnwm i w
58
D YN A M IC HA N S E CHA N T
N A N CE O P M A XIM U M T O M IN IM U M LE VE LS
CIVIN C S IN CLE P E N O P E N A T IO N
N O IS E LE VE L : ISO M V RMS | . , . ,
S WE E P R A T E : S O CYCLE S /S E C* *] J j j ,
i F.
i .
S IS N A L LE VE LS '. A S S HO WN
IT 1
'. s .
H I' ,. ;. ,; .
5 ssr.1 f rfs i-M i''''
:
^i . ::r!-
:
:;
:
:: r>; !'
i.!
:
;:
'
;
.
. j . ,
-IS O
5 ill li : : : .<{., * < . nsj
!lj: ;:> . |.
!
:i..- j'. . ...5
,
I.
.! i
:i ' f t
i
: : I:I
rk
. i .
:
- ; p .
i ' . i . "
. 1 '
:
-.
<:
i
. i
:i-
lin
: i l ! " I,
i
!
. fc =
ui r
1
- *
P h :
:
' : . i ' ! . . :
; ' t;
. ! 'if '
!';!..; I
! ' !
. 1 43,
!
.:
. i i r. . 1 1 t ' i ^ . 1 1 IL v
i *
h+]| :::; :i !
'
!
i f 1 1 I
I i ::::
i ! !
l i
1 1 i n
9
'I
H
n
N
I
H
o
1 1 ^
M J , I,
' (j i i !
1 . . C .
I i
I
K
5
FREQUENCY SCALE - C.P.S.
FI6. 26
59
A O C A CT IO N CHA R T
LA R G E A M P LIT U D E FIXE D FR E QU E N CY S IG N A L
V8
S WE E P S IG N A L LE VE L
N O IS E LE VE L : 1 50 M V RMR
FIXE D S IG N A L I -6db
1
S WE E P R A T E I 77. 8 CYCLE S /S E C.
1
. . S IG N A L LE VE LS : A S S HO WN
6
i n
to
o
z
o
r
-I2 *b
>:: i :
:
!l |'V i !i
* . . * ' !
; T . >
: .
. .
l * j ' i 1
S l 1
i ! i
I
' tt.
; *
' ' I-
":
i '
::. f i
. 1 :
I 'I
. I
I
i
a:
i l .
i
. . . j .
. 5
1 .
i|
>:
i
l i : i
. 1
1 :
^P
1 . 1
. t
. l
:: f t.
*
:
. i *
:
l
:!

1 * !
t
l l :
I.
I i .
- I
i i .
I
I-, ;' :
: I'.
: r
1 .
T
. ' I "'
I"- ' ' 1 .
r
8
o
M
* . i
. i I :. I i f ;
* I . . i i r I* I.
. . i r :
i z. !. .
il:,!: -so db
i i r-.:.!! ...
; . I
f "
. ;
.
i :.
i ; .
.
'i '
*
ih
:
i !
!': n
!.
j ^^ >
,
i . . . .
eg
T
8
p
*
'. i '
" ; ! .
; ! 1 :1 M
. :F
u
'". 'i *-: l i
i

i !i *
i
. i ,
::riii
i
1
:
:

1
i . .

:
i d
r
o
s
FREQUENCY SCALE - C.P.S.
FIG. 27
6 o
j i i :!:. , . . . . , . ,
' 1 1 N r. i !
T WO CR O S S IN G S IG N A LS
WIT H V AR IABLE SWEEP R AT ES
19
ii
f cl f f l i t
1
snE
| M i S WE E P LE VE LS :-l 2 db j P f l H
i . | M - WE E P R A T E S : A S S HO WN g{y|
i f di 'f ! . . . . . . . ^ T S . r'V ! ; . !?. : f -! :. . j ' J f c
i ^S tJ CYCLE 8/8E C ! i j -:" [ i s-j ^ ! |
a Sh^JNs. : U M'll . :' '^ I:! I*; i::: ! fc
CYCLE S /S CC*
5 R l ?@S l . ^a ^00 CYCLE . .
2 &!> "J 1 f rl | :
:
!; 1 1
:
Ij . Is; :
1
1 r; -!
:
5:: l J r
7 IJ ^
g iwci im
m
fe*U!i! . ;
!
'iJir-. !!. :!
H
i l r
r
'i i i 't E^i tJ 'l " . : ' 'I. H l ;'
-
!
:!
'
Wkm it". ': tJt
!
;4. IX
S
:;i STv
:
- u'lig
r V >^S M CYCLE S /S E C* ; ; K
H
I , : . 1 U I j ' i
J
k
M ; i l
:
"'
:
"%:-
:
--;
,
:!| :. !. :; J i i i : -^
M 2
CYCLE S /S E C
i f - '; . i :' \''' * . ' :. '; ^ - . : ;:'; ^
1
8
2
FR E QU E N CY S CA LE - C. P . S .
FI6 . 2 8
6 1
i
60 DELTA REV. 124
NORTH - CENTER - SOUTH AMTENMA8
MORTH - SOUTH PASS
Vu
!
". t : ! J. . M l ; .
! ] '>
.ii!r
:
i l -. i . S'i -. r
sRtl^iii^t
l
1
:
ff. ''
!i
< i3
?!
t
:
H
i,

|
1
If i IS
l l l l i l l l
i
*" i ' ' j B ' , . . , . . ; , . . i , 1 . i | I i l l ! H*
m
%
ii'jH ijiif^i
l ; IHIl M - %'. '.
illlH iMiMiii!
mm m
i
i in
i i j
i
a
s
5 5
u
O O P P LE R FR E QU E N CY - C. P S .
FIG . 2 9 A
6 2
? t i t
UM i reM U M T IM O O
I I I II
K Wi n O M IAi n BM
i v / 1 :1 M rua M
e
i
J
" '8
1 a"3 '
I > ? sS
58
6 5
0 OCLTA REV. 140
NORTN - CENTER - SOUTH ANTENNAS
NORTH - SOUTH PASS
;l 53 i
i :i
rm M '
' . i * ' I *
. .?:; i
t -3 1 ' !. . . * iilifl
v . *
f
i :
'III:
ll
i
I!
I
V r. . > f <
t
I '^T f Hi -v. ": :v ? j
! I : -t M t tH . W
. M um
!
* ^H
i
i j KT
:
' S i
;
' ^U '

H
J
1 1
I
H
!
r! Bl
III
i ! i l l
If l !
!!
1 1
II
i r 1
t
M :!
1 1
Iti i
il
ig ill
1 :1
w w w
ifiii; itill\:illl;!
W'
iH P^iii 4 1
o
D O P P LE R FR E QU E N CY - C. P . S .
FIG . 3 0A
6k
AM l O O M i NIIM O O
tnM xnx
T
:
tltti
I
I J
Mils
' ^ < Z a:
58
o
e
It
6 5
6 0 D E LT A R E V. 1 56
N O R T H - CE N T E R - S O U T H A N T E N N A S
N O R T H - S O U T H P A S S
HI. i
M j ! M i I' i : IT ' "
' :
:
;:M:T:.:: ;:i|
\
\3
mm
i.
d :;;*:= ,; 'iP i. .:; ......II
;
|i
^ :: ..! * is. :! ...I'St.
i
o
HI
1 1
. 1
s
iii
!
'I
I
it
111 I
i
' i i /i - . I M i . i r i
I' II:. " I i :-|:
. T . , . ; . .
' i . . ' . ' . . * .
!
! :. ! " ' . I '^i . , . '
i.;i.:: ;fv'il!:.:i.JM| '
i;;,i'.
!
.:!i J.^: yr
:
'
I
i i
y: ; ' I . i f f
1
l i :!
:
; III!
Ii
' I
i
1
'
I
I!
I!
i i '
iS.s
i
l!
I
M I ri
I'. :.!: I- j
l. tplf. :>
ml
i
Mil::;.!!
r-IiVl Ii :,;ii
i r
o
K
Q
!
:
}-W\ x ill I!
^ri | i !;Eli
1 1
I;!! !
' 'l
'I I
. i i i i M ; i
Mm
u
1 0
DOPPLER FREQUENCY - C. P . S .
FIG. 3IA
6 6
N i M i tut -nrtM i t
tmuen i M
M O M M
3 I
. el
Q >r M
U C K H
K K H M W
S U W H
J W O =i 5
i s
1
?!
a
W -u
ma o
I -I < K CC
58
6 7
0 OCLTA REV. IM
CENTER - SOUTH ANTENNAS
SOUTH NORTH PASS
. ! -I
I:
I II
; M | ' 1 1
N
" :i :
!M
:
:.
i] ( .
i : :" ;
Ir 'i v
i -
:
. *
. . I 1 , 1
I
i I r u i i m 1 1 1 '
il
hi
a
s
i
I;: L
w . !:, .
s ;
, :
!' i
i
l
f
S i
I , 1 . 1
. ;
:
.
ri l l 'I
i d!. |j I
t-i'ljl
T ::! i . . . n I] '. v
3tii
iiM
1
:i;
1 ; .
:
li ttl ';
''! i '
. T I, '
;
!> : !
!
'!
|i i n V I ' > * i l f !. ! < I
I: r
I
i'.
':-i
IT "
'
i l
1 1 "' '
! !' '
.( , .
I
?i: ;,. ft! ;;. !; . !'
i.l'M!
'' I'l l , pi
5:
rli i
H u
' l * ''A'
i ! i ! I u " ' SBT ) :! -it H I"
1
1
mil'
?
iii' :
:
ni;:l!i
I. :
''i
ill-;
i i
': m. .
I'.rlM^'li
:
i .H 'i :;
;i
i nil
l i
:
; ; ,: .'.
f i
^ i . i
M i 'I. -'
-#:
i f .
* ' : .
li
n
. , i . i i
!:<t; "j
i :
2
i
2
. . Li :'
i'li
';;>,
. h n::
o
I ! Hi . -
1
. . ? f l 'i l I!:
i :i !t i ;; ;
I
if;. I
t'i
i . . i
f. i %
i \i
\\i
' ill
I
1
H
l O
II
!l
!
r-
H
p
ijl
l i 'i
I
Il M
i !. . t.
" !i
:lL'
i 'i f . l
i'i I
h
1
!" i l
D O P P LE R FR E QU E N CY - C.P.S
FIO . 92A
6 8
M NM O M M M T IM O O
a j
litil
?s If
Is
6 9
6 0 D E LT A R E V. 1 72
N O R T H - CE N T E R - S O U T H A N T E N N A S
N O R T H - S O U T H P A S S
. . . . H I ,1 . % \IM ; . : !;
I :;-d:
'l i t: ;
Il l l
i
Ul
3
u
to
! , ! r-< . *
. i
ii
. i
5
1 . .
."i
I ' ! ! K i i 'i i * .
*
!
: * ' ! Mi:- M - I
I..!... .1 ii- ! f il lir.:*;l4l9 ' |ii|i!-i M. 1 1
I I" 'I ' 'I.
-
'. '!' i * -
'f
i;
i l l l . .
I t. l .
l i m .
I'. -. J''.
m. am
'i ! ul
. . . 'If ! I ; i ! vl i '
I
. *
! I
I'.
I I. I
i i i i i | | ; j . ' !
[Km
1
. I . I
i
u
o
esi
. : l
. I i i
::
'
. . . . 'Hi
i i
S
O
IO IO
'' I
1 1 . :l |
o
illlitip
M l
i l -
&
'i

IT
till
Hi I
i i
. i
. 'i j !
U
j ! i
1
l l j
I ! !
j : ; i It!
DOPPLER FREQUENCY - C.P.S.
FIG . 3 3 A
70
H1 0N 3 U XS 1 VN 9 I8
80
M K
M o
II III
xndi no M 3 A i 3 9 aN
IV N /S Kl 01 3 8 80
J
S S N
I? pi rrr
r

1
o
I
ri | i ; U H
i
to . z 4
w in t"
mm
o at f c. - o ?
as 5
N
t S o
o
hUi =
O C K W
iilii
o . i n U J a
A - - S 1 U
' ui w h 5
J o a *
c o
to. N o
S | 3
71
o
u
(0
o
I
bJ
-I
<
o
in
I
Wi 4 HH i i i i 'i i i -T
I'M !-'
il'i
i i
l !
1 i
LA U N CH R E CO R D JU P IT E R C
JA N . 3 1 , 1 9 58
y i i ti i i : HIS T ! i :tf j l : i t :M ; i i f t tttt ! M :
;
:
- ! ::
:
t:'':
!
f f
;
M r
! :
r
j ?H i tiV
I I '
:. i
: :: :
r
i .
:
t; : :
' ' i . 1 :-
ti l l S ^Hr^i ; :: !
i
;
, 'U 'Kgi l -'-; -:;
1
*':. * ' ' .
(' / :. . : 'i l j i "'. . . : . ; ! . :| i :: :
i i * . 1 l i 1
u
O
:H, :
l :
!;;
i '
l
i If
:
. . i ;
I i
* r-
1
1
:|
; l .
1 ,

i
i "

I.
O
IO
m
ro
v n rn n n i f s m
1
I
. 1 1
. . . i
( i
!!
f l i . .
! ; i * .
D O P P LE R FR E QU E N CY - C. P . S .
FIG . 3 4 A
72
-v. . ^ f *
O O P P LCR R CCO N O O F T HE LA U N CH O F
CXP LO f tCR I (S S A LP M A ) T R A CKE D A T
A BCR O CCN nravi N G G R O U N D . M O .
-
0
?l
?
"1 r
T IM E
f t S E CO N D S
S T A G E FIR IN G
S T A G E FIR IN G
S T A G E FIR IN G
3 B .
8* 1
M
O S >-
I
i . o 8 2
i t
- K
51 JA N . 58
05S 5O Z
M HM M IIIM IM IM IIIIM M M M M IIM M *IM IM IM W<#<M M IIM M M M HIIM M IIIM M M WM M M M
>
. 1
S i
I
ki
-I
K
a
o
o
51 JA N 9 8
1 0 S E CO N D S 09 59 O I
| M >tl i i i i r-"'i i i i i i i ni wi i i nM Hf M ni i r t""*
1
S i i i wi i i i i i i i tf i M WWIM
=1 r
o
OOPPLER FREQUENCY RECORD OF TH E LAUNCH OF EXPLORER Z
( 5 8 A LP HA ) TRACKED AT ABERDEEN PROVING GROUND. MO.
73
DISTRIBUTION LIST
No. of
Copies
Organization
1 Chief of Ordnance
ATTN: QRDTB - Bal See
Department of the Army
Washington 25, D. C.
1 Commanding Officer
Diamond Ordnance Fuze Laboratories
AITN: Technical Information Office,
Branch Okl
Washington 25, D. C.
10 Commander
Armed Services Technical
Information Agency
ATTN: TIPCR
Arlington Hall Station
Arlington 12, Virginia
10 Commander
Air Force Systems Command
ATTN: SCTS
Andrews Air Force Base
Washington 25, D. C.
1 Commander
Electronic Systems Division
L. G. Hanscom Field
Bedford, Massachusetts
1 Commander
Ai r Provi n g G roun d Cen ter
APPN: PG API
Egl l n Ai r Force Base, Fl ori da
No. of
Copies
Organization
Commanding Officer
U. S. Communications Agency
The Pentagon
Washington 25, D. C.
Commanding General
White Sands Annex - BRL
White Sands Missile Range
New Mexico
2 Commanding General
Army Ballistic Missile Agency
AOTN: Dr. C. A. Lundqulst
Dr. P. A. Speer
Redstone Arsenal, Alabama
1 Director
Advanced Research Projects Agency
Department of Defense
Washington 25, D. C.
1 Director
National Aeronautics and Space
Administration
1520 H Street, N. W.
Washington 25, D. C.
75
I
'4 1
flfl
il
1
!
B n
558
8 i
S 4
iSl'li
4 % 4 1 1 1 4 i g
H lliJ!
i H
3 v. I
V.il\*V.
ifi!nl!
liil.li
3
!l! u
i!
&e3 s
& > h 3 &
m m So
Hill
a!
I-
II"
I
J8
JIJj
i
I!
SI*I
5
( B CO
ilft|!i!
1 g ? ^
ti
I I
si
III
I
K
H a a T 4
9 8 S
S, M ja " a
:l!i4fiil
tt
l;ij
I I o ^
I
t s-p 5 8
3 ) u i ) o <H H
a "i
c a-e, I
ill
!
J
1
"I jll
5
a
IJ
I
23
^l
1
*
8
I
is-
m _
41
8
Ilf
I! H 3 1
Hill
e
il.Jl I
i
<g
I|
5 K
I

8
Tl -rt U
t|
H
i
ii
P S H
5
g si
41 B K 9
IllH lll
Ijfilf]
Ii!
H llial!
flll'lfi
ilPllli
altliH I
i
m
i
& IH
1 1 a
s:
i
Ui : I
ihm
!
fe l
B > b 8 a 5.
C 9 S o -S T
ijiiJ
I
i l !
_9
8 i ^J " > * -!
4 > H o 8
llitl
I
A
K<9 CO
I
9||lg]|
I
4
n't
- s 6 "
1 1 '"
O *> > A 'H
,8S
0
e I
lie
O
i s
3
If
Al
5s ps
u 5" ^ ^ .
9 ti u bq -H H
hflKfl
Vi o tt < S
^299
t. '
I
5
I
|*|
88
II
iiit|!il
i i! il!J!!l
I . f l o bt S
d
ilfl!
[ii
I-
i !l s

p. +> B & 4
. . . Hi
Imltil
f > p B+>
M .
I
TI
0) K

I
|3
ill
%
K
H
I
liinlli
IJlfa M
l}iin|
"J O ti II p , -H
III!!
N
S3
2 1 | It

, , s
n .
5s
N IL
i <
si
8
-
Il
ui !!
. f l &. d f c
UNCLASSIFIED
UNCLASSIFIED