national radio astronomy observatory - … · with ever increasing bandwidth this brute force...
TRANSCRIPT
NATIONAL RADIO ASTRONOMY OBSERVATORY
TITLE: CLIPPER BANDWIDTH REQUIREMENTS
Distribution: GB LibraryCV LibraryVLA LibraryTU Libraries: Downtown File
Mountain FileJ. PayneH. HvatumR. LacasseS. WeinrebP. NapierM. BalisterC. Burgess
Under the p
clipper bandwidth has been uncertain for many years.The requi ement fo
essure of getting equipment built and in service it has been easier
CLIPPER BANDWIDTH REQUIREMENTS
R. E. Mauzy,
to ignore the question and build the clippers with what seemed a generous margin,
about 10 times the signal bandwidth. With ever increasing bandwidth this brute
force approach may not be so easily accomplished the next time around. A calcu-
lated solution was suggested but dealing with the nonlinearity was considered
oach seemed reasonable if a clipperfar more work than by measurement.
could be modified for narrow bandwidth and remain stable.
A clipper from the Model II correlator having about 200 MHz bandwidth per
stage was modified with four 3-pole low pass filters for 12 MHz, 3 dB cutoff
per stage. See Figure 1. A second clipper with full bandwidth was used for
comparison. The two units were measured in the Model III cor elator in the same
IF channel. The first tests showed excess rolloff for the modified unit begin-
ning about 2 MHz and increasing to 1.9 dB at 8 MHz. The modified clipper was
measured for flatness at a non-clipping level and the curve corrected. The re-
sult was a bandpass that dipped to -0.7 dB at 5 MHz and returned to _0.2 dB at
8 MHz. With errors this small it seemed prudent to improve the flatness of the
modified unit, check the reference unit flatness and retest. Small changes in
capacitor values in the filters eliminated the drooping output in the upper half
of the band and provided a flatness within 0.1 dB out to about 9 MHz. Late mea-
surements with a somewhat different test setup showed about 0.2 dB dip in the
center of the band. See Figure 2C. The test clipper flatness was also checked
at five input levels from -20 to -56 dBm. This data showed a very flat response,
0.05 dB at all levels.
The units were aga n checked in the co relator with a common input. The
new data showed a dip through the center of the band of about -0.7 dB and a re-
cove y to dB at 8.5 MHz with respect to the 1.5 MHz level. The results of
0 tests are shown in Figure 2A.. Because of the correlator filter shape it
is not possible to make accurate measurements above 8.5 MHz or below 1.5 MHz on
the 10 MHz band. See Figure 3. The error up to 1.5 MHz was measured by running
bandpasses at 5, 2.5 and 1.25 MHz. The average of several measurements showed a
flat response to 0.75 MHz, down 0.02 dB at 1 MHz and 0.06 dB down at 1.5 MHz.
These results are obtained by aking printouts of the bandpasses hrough both
clippers and determining the difference by scaling. The scaling is referenced
to a baseline that is predicted from the shape of the skirts not from the zero
line of the graph. An error in this line could contribute to a slope across the
band.
The curves of Figure 2A did not have the shape anticipated so were accepted
with skepticism. The next effort was to try a different test method. The band-
pass was checked at a normal input level with one, two and three CW signals,
none of which duplicated the correlator results. The following attempt was to
introduce a flat (t 0.15 dB) 10 MHz band of noise at -20 dBm with a CW signal
added in at -35 dBm. With this combination the results were similar as shown in
Figure 2B, the primary difference being a 0.25 dB rise rather than a 0. 3 dB drop
across the band. The band of the reference clipper used for corre atormeasure-
ments was checked for flatness. As a linear amplifier it had about +0.1 dB rise
across the band, as a limiting amplifier the rise averaged 0.18 dB from 1
MHz and 0.08 dB from 5 to 9 MHz. This reference slope may account for most
the negative slope in the correlator tests.
The next step could be another round of improving measurement accuracy and
clipper flatness. But the effort required to improve the results would increase
3
dramatically. Because of the relatively small effect that bandwidth has on the
spectrum, the extra effort is not considered justified. In conclusion, it is
sufficient to say that a clipper bandwidth somewhat wider than the video band
will contribute no more than 0.5 dB ripple to the spectrum.
REM/cjd
Attachments:Figure 1: SchematicFigure 2: Narrow Band Clipper ResponseFigure 3: Correlator Bandpasses
Clip
per
fit
7.5v
390
L.4
.7V
5.1V
-3.3
p91
03
K
3939
2KM
S12
4
5.1V
3K39
2KM
S124
NO
TE
: T
RA
NS
IST
OR
S A
RE
2W
49
96
UN
LE
SS"=
==
t S
PE
CIF
IED
OT
HE
RW
ISE
.
AL
L V
AL
UE
S A
RE
IN
OH
MS
a M
1CR
OFA
AD
S,62
UN
LE
SS
SP
EC
IFIE
D O
TH
ER
WIS
E.
2N
44
2.*
N.1
.1.2
6 S
iMIL
AR
TO
4-959 5.
1VZt
4390
4 2
N3
90
6
3.6
K
100
56r
.01
MA
,
MS
124
(MP)
.05
2K
W15124 o
tooE
s A
RE
TH
E H
OT V
ARIE
R
(sctiorr
xy 8
AciR
IER)
TYP
E B
Y S
OLI
TRO
N
200 p
F
1.5K 6
170q
CG71
- L
typi
l#
330
27
OU
TPU
T
R45
1 -6.0
0/5
60
1K10
4D
CB
AL
.
Fi9
uve
---If '
i',_
-11-1- I
l l-
- IIII
IIII
IIII
IIII
IIII
IIII
IIII
IIII
."_
i111
1111
111
: fl,_
-'----
I-- -1-
--i- ',---
H1
111111
1_
III 111
111111
11111
_111
11,
-_
___._ ...______ _ ii___ __
1_ _7"
..--:
gi
imito
ui-i-
iiim
ilt.
.... -...-:.
4.1
110til
l! d
illi
149
---.:
gilli
in°
"I".1
1111
11"
-_ -1 r
i.
----t
i-U - f-
III
_.,
iii„,,,„
„bila
nim
m..-
- isi
iiiag
. ---'lii
iiit
tnu
nin
imin
iiii
llu
zi
}-
- -f ----f-
-
-11
1111
1,
,-[-r
-r iH t-- --
•-,
...
1+i - !
- 1_ .
t--
-4 -
IIII
1111 11
----ft
1111
111
---III
IIIIII
IIIIII
III:1
1111
11.___._-_
--Nil
I.'
- il -ii,; 11 -,,
.1
-il.-t-1
''1
1111
111P
:III
IIII
IIP
„4
.•
I
H-t-
-,---,--
f:17.-1-
,1! ,
-f
-1,
1 1-4
7I-1
'i,_F-_-.1
I:ii
i'
' 111
11 I
'-III
II111
- II
.I
_ i__ 111
1111
1111
1111
1111
1
_,:„..___iiiii--
i]4
-
r-t- i-
.
:1-._
._ -._--[4.-.r‘
--F-q
l-.4
.. 71
rT.
. __
--.-1 -- -
4 -
--.
-----7 1111
1-1 -iii7:1
111i
ddim
il-.p
illow
a-till-
11M
M-lip
ilm
iraii
-----.
•-
radii
•
ril-
4 -1 4-1i-p
-t- illi
livilliril
litif
usif
ig[
_---
----
-TE-1
-I
I-t
4
fi--- 1
-,
nil
*Fill
L11
11-4 '
110
------
1111
1111
111
--
LI-
,.
-
1r
wil
'•
A r
---:---1
1i..'li t-
f lit '.-111
-1--.4
-- ---ri': -
ill
riil
tair d ,it
.er li
ffir
itil
iiii
iiii
a iiii
ll11
I"1 11
1111
1111
- 1"--
-----f_ -__. - . --
-t--I
t-I- ----
r--,-- -t - -
14
1-4
-t -+1
I-, 1-
-1-
f Th
•
.11
1111 i1
1111
1111
1111
1111
,III
IIIIII
IF II0::_
_ _. 1
1111
"...
1 III
1T-4
,-5
-----
-1-
-•
-I:
1-
11loilo
iloilioili
iionnou
1-
- r_-- -- --
.4-iti
ii-Ti
m1
-I-1
1 .
1-1-
r'
Ialij
ilri
llP
llii il
lgII
IIII
IIII
iN--- II
IIIM
II.
-.---
+I
-r -1
HI
_.,
a ....
.___
.1 il
__,_. t_ r
,_'--
--
.,..„
.. ..6...i
mmom
ouR
imm
is".
"--l
i g
ftilia ii
iiu
nil
isw"r
e--li
iii
iiii
"-
q---
--- -I-
._ _., i
117--
':--11
1111
Pilli
iii11
1111
1 LI
IIII
IIII
I- 1111
1111
111
11 11
1111
11:
- _-I
,711 .,_ ,
---).:
i-,:ii
iiii
r: 111P
1111
11II
L11
1111
1111
1P11
1111
1111
1111
1111
111il
l 111
1111
1111
1110
1:-.
11111 _ ,,f
11i
, I , Il
l
_111
1111
11IL. -- -t'-
1--P
i-a
- -
- ----
n'-- 11
11IIN
IIIiii
ihil1
1111
1110
11i10
1111
1111
1111
-11
11,11
1101
1111
1111
1111
11 I
IIII
IIII
III
Nol
se 0
0'1
wv.
t..w
ide
bak
tdii
CP 'rr
e /
CA t
O r
CW
w I
tAN
ois
e
cW
jie
/t')
hkyl)
t;ki,
03
46
7
Fre
e d
) M
Hz
NA
RR
OW
BA
NE
) C
LIP
PE
R R
E5P
ON
SE
- - s••■••••••••••,.... -
D clitawr 1-4
d. tipp.ele Info ..S.4014ya
1178
62164TEST
*O.o,FERFERFERFER*0.0
0.000 _
4 7 4:119111.M.1.0• ••■■• •■• • Y.. .•••
46948 *000.0 5301.7_U.
S603•
15
5: Fif-e,r '714r-p, iico.hr;t;"?..er
*0.000 r__
tv4:1,
frf.o
JaMii
*0.000
-,o/r
—0-31
_ 1
0.00040 ..94 .7
' . •"000 5301.7 56:33. -20
efit'Pe".
„111*0,000FER /AU4FER SG /AU4FER B ,AU4FER 6 B /AU-414temX
*0.000 0 0 7
ep ,1 -K=7 1;04.chrpertr to 4-cIS &k
V!
/*err f4r 4. 40 P17) afri pot'
44a 1 ; et ok,g1 4.4 e ile/et.A.7.C.,