University of Tennessee, Knoxville TRACE: Tennessee Research and Creative Exchange
Masters Theses Graduate School
3-1987
Conversion of the HP3577A Network Analyzer to a Spectrum Analyzer Mode of Operation with Low Noise and Cross-Talk
John E. Crowley University of Tennessee - Knoxville
Follow this and additional works at: https://trace.tennessee.edu/utk_gradthes
Part of the Electrical and Computer Engineering Commons
Recommended Citation Crowley, John E., "Conversion of the HP3577A Network Analyzer to a Spectrum Analyzer Mode of Operation with Low Noise and Cross-Talk. " Master's Thesis, University of Tennessee, 1987. https://trace.tennessee.edu/utk_gradthes/2895
This Thesis is brought to you for free and open access by the Graduate School at TRACE: Tennessee Research and Creative Exchange. It has been accepted for inclusion in Masters Theses by an authorized administrator of TRACE: Tennessee Research and Creative Exchange. For more information, please contact [email protected]. To the Graduate Council:
I am submitting herewith a thesis written by John E. Crowley entitled "Conversion of the HP3577A Network Analyzer to a Spectrum Analyzer Mode of Operation with Low Noise and Cross-Talk." I have examined the final electronic copy of this thesis for form and content and recommend that it be accepted in partial fulfillment of the equirr ements for the degree of Master of Science, with a major in Electrical Engineering.
David Rosenberg, Major Professor
We have read this thesis and recommend its acceptance:
J. R. Roth, T. V. Blalock
Accepted for the Council: Carolyn R. Hodges
Vice Provost and Dean of the Graduate School
(Original signatures are on file with official studentecor r ds.) To the Graduate Council:
I am submitting herewith a thesis written by John E. Crowley entitled "Conversion of the HP3577A Network Analyzer to a Spectrum Analyzer Mode of Operation with Low Noise and Cross-Talk." I have examined the final copy of this thesis for form and content and recommend that it be accepted in partial fulfillment of the requirements for the degree of Master of Science, with a major in Electrical Engineering.
sor
We have read this thesis and recommend its acceptance:
Accepted for the Council :
Vice Provost and Dean of the Graduate School CONVERSION OF THE HP3577A NETWORK ANALYZER TO A
SPECTRUM ANALYZER MODE OF OPERATION WITH
LOW NOISE AND CROSS-TALK
A Thesis
Presented for the
Master of Science
Degree
The University of Tennessee, Knoxville
John E. Crowley
March 1987 ACKNOWLEDGEMENT
The author is deeply indebted to Professor J. R. Roth for his infinite patience, financial support, excellent working condi tions, and good advice.
The author also extends his gratitude to the remaining members of his thesi s commi ttee : Professor T. V. Blalock and
Associate Professor D. Rosenberg.
Special thanks goes to Professor J. Googe who had answers to problems at a particularly difficult time.
Finally, the author wishes to express his appreciation to his parents for thei r support and encouragement throughout his academic career.
Thi s work was supported by the Air Force Office of Scien tific Research under Contract #86-0100 (ROTH).
i i ABSTRACT
A system has been designed which modifies the HP3577A Network
Analyzer so that arbitrary signals in the range from 5 Hz to 10 MHz can be compared in amplitude and phase. The system will be used to sample density or potential fluctuations at two points in a tur bulent plasma, and measure the frequency, amplitude, and phase of signals detected by the probes. This modification is totally external to the HP3577A Network Analyzer, and has been designed to minimize the cross-talk between the two channels, so that non linear mode coupling processes in the plasma can be distinguished from mode coupling (cross-talk) originating within the instru mentation itself . The composite has a dynamic range from 0 dBm to
-60 dBm, crosstalk of no more than -40 dB between channels, a maximum spurious response -40 dB below the maximum input signal, and an amplitude and phase accuracy of 1 dB and 5 degrees, respec tively.
iii TABLE OF CONTENTS
SECTION PAGE
INTRODUC TION.
I. PROBLEM STATEMENT 6
II. SYSTEM DESCR IPTION. 15
III. MOV ING FILTER .. 23
Introduction . 23
Design •... 25 Crystal filter. 27 Mixers. . . . . 31 Ampl ifier .. . 37 Output filters. 38 Experimental Resu l ts. 39
IV. PREAMPLIFIER .. 40
Introduction. . 40
Design •... 40 Experimental Resu l ts. 43
V. LOC AL OSC ILLATOR GENERATION SYSTEM. 48
Introduction .•... 48
Design .•...... 49 Composite Filters .. 51
Experimental Results •. . . . . 53
VI. EXPER IMENTAL RESULTS... 61
Introduction...... 61 Accuracy, Linearity, and Calibration. 74
Phase Coherency Measurements. . . . . 74 System Dynamic Range. 76 Conclusions . 83
REFERENCES. 84
APPENDIXES...... 86
A. HP3577A CONVERSION SYSTEM SCHEMATICS. . 87 B. THEORETICAL AND ACTUAL MEASUREMENTS OF THE HARMONICS OF A SQUARE WAVE...... 92 C. DESIGN DATA FOR COMPOSITE FILTERS 102
VITA. . 107 iv LIST OF FIGURES
FIGURE PAGE
l. HP3577A in Normal Mode of Operation. 7
2. Desired Configuration of the HP3577A 8
3. HP3577A Displ ay with 4 MHz Sine Wave Input 10
4. HP3577A Display with 4 MHz Since Wave Input and Spurious Response at 3. 5 MHz ...... 11
5. Demonstration of Spurious Signal s in 50 kHz to 625 kHz Bandwidth...... 12
6. Demonstration of Spurious Signal s in the 625 kHz to 1250 kHz Bandwidth ...... 13
7. HP3577A Receiver Block Diagram 14
8. Tracking Brickwal l Fil ter Frequency Response 16
9. "Moving Fil ter" with 10 MHz Effective Center Frequency .. 18
10. Ten Superimposed Frequency Bode Pl ots from to 2 MHz. . . 19
11. Ten Superimposed Frequency Bode Plots from to 10 MHz 20
12. Conversion System Block Diagram..... 22
13. Filter Board Block Diagram (1 Channel) 24
14 . PTI Supplied Matching Network .. ... 28
15. Ampl itude Characteristics of 4371 Crystal Filter . 29
16. Actual Cascaded-Crystal Filter Response. 30
17. Double Balanced Mixer Schematic. 32
18. Preamplifier Block Diagram ... 41
19. Preamplifier Output for Fundamentals at 5 MHz and 10 MHz 45
20. Bode Plot of Low-Frequency Section . 46
21. Bode Pl ot of High-Frequency Section. . 47
v vi
FIGURE PAGE
22. Local Oscillator Generation 50
23. Local Oscillator Generation Operation for 5 MHz . 54
24. Local Oscillator Generation System Schematic. . 67
25. Bode Plot of 45. 45 MHz to 55 MHz Bandpass Filter. 58
26. HP3577A Display for a 4 MHz Since Wave with System
Modification. . • • • . . . • • . . . . . • . . 62
27. HP3577A Display Showing Attenuation of Image Signal 63
28. HP3577A Display Showing Lowered Spuri ous Responses. 64
29. HP3577A Di splay Showing Lower Spuri ous Responses. . 65
30. Calibration System Block Diagram. 66
31. 2.954 MHz Sine Wave Fundamental . 67
32. Example of Second Harmonic (5. 89 MHz) . 68
33. Example of Third Harmonic (8. 84 MHz). . 69
34. 9. 015 MHz Sine Wave with 0 dB Attenuati on . 70
35 . 9. 015 MHz Sine Wave wi th 10 dB Attenuation. 71
36. 9. 015 MHz Sine Wave with 40 dB Attenuation. 72
37. 9. 013 MHz Sine Waves Fed Into Each Channel of System. • 75
38. Phase Relation Between Two 9. 013 MHz Sine Waves 77
39. Dynamic Range of High-Frequency Section . 78
40. Amplitude and Phase Response of Hi gh-Frequency Section. 79
41. Normalized Amplitude and Phase Response of High- Frequency Section ...... 80
42. Normalized Crosstalk Between Channels . 81
43. HP3577A Display of 300 kHz Sine Wave in Low-Frequency
Channe 1 • ...... • . . • • . . 82 vii
FIGURE PAGE
A-1. Filter Board Schematic (1 Channel) 88
A-2. Preamplifier S:hematic (1 Channel) . 89
A-3. Local Oscillator Generation System Schematic 90
A-4. Power Supply Block Diagram •.•. 91
B-1. Theoretical vs. Actual Spectral Components of Square
Wave • • • • • . . . • • • • • • • • . • • • 93
B-2. Fundamental of Square Wave Measured by Conversion
System . . . • • . • 94
B-3. Second Harmonic of Square Wave • 95
B-4. Third Harmonic of Square Wave ••. 96
B-5. Fourth Harmonic of Square Wave . 97
B-6. Fifth Harmonic of Square Wave. 98
B-7. Sixth Harmonic of Square Wave. 99
B-8. Seventh Harmonic of Square Wave •. • • 100
B-9. Ninth Harmonic of Square Wave •.• • 1 01
C-1. Composite Filter Low-Pass Design Equations .. . . 103
C-2. High-Pass Design Equations for Composite Filters . • 105 INTRODUCTION
This paper reports the modification of a piece of Hewlett
Packard test equipment. It is desired to modify the HP3577A net work analyzer such that the unit will accurately measure the amplitude and phase of arbitrary signals. The frequency range for the signals of interest is between 5 Hz and 10 MHz. The amplitude range for the signals of interest is between 0 dBm and -8 0 dBm.
The modification is to be totally external to the HP3577A and have two-channel capability.
The need for a two-channel device,·capable of measuring the amplitude and phase relationship between frequency components of broad-band spectra, results from the study of turbulence in plasmas. Plasmas are hot gases which contain electrons and ions.
These electrons and ions move about in the plasma creating varying electrostatic potentials. While studying the way in which the charged particles gain and lose energy, it is very desirable to be able to determine the fluctuating electrostatic potential at dif ferent points in the plasma. The frequency components of the plasma potential at two different points helps determine the way in which electrons, ions, and propagating waves move in a plasma.
For example, if two points of the plasma have a strong frequency component at a particulcr frequency, coherent structures or waves could be moving from one point to the second as a function of time.
The phase relationship between the spectral components determines if fluctuation-induced transport exists between the two points. 2
If the phase relationship of the spectral components between the two points is coherent, mode coupling is implied. If the phase relation between the two is random, then there is probably no coupling at that frequency.
In addition to finding out if there is mode coupling in a plasma, the engineer must be aware of the coupling between the channels in his test equipment. If the crosstalk between channels in the test equipment is -2 0 dB, then it will be hard to determine if any observed phase coherency below this level is due to mode coupling in the plasma or to the crosstalk between channels. The crosstalk in the HP3577A is -10 0 dB between channels. It is de sired that the crosstalk between channels in the modification be greater than -40 dB.
The need to measure the phase relationship between spectral components is the reason why a two-channel data-handling system must be modified to measure mode coupling in plasmas. There is no off-the-shelf test equipment capable of phase measurements over the frequency range of interest. There are several devices avail able that measure phase between input signals. A dual channel spectrum analyzer capable of phase measurements will accomplish this measurement task. However, most of these units work at low frequencies. The HP3562A, for example, has a high frequency of 3 dB point of 10 0 kHz (and a cost of $23,100 ), which is below the region of interest in the current research program in the UTK
Plasma Science Laboratory. Vector volt-meters, which measure the 3 amplitude and phase difference between two input signals could be used . However, these units operate on one frequency and cannot produce useful measurements on broad-band spectra. A signal digitizer with two channels could take accurate measurements over a very short period of time, but not on a steady state basis .
Since it is desirable to move the measurement point at will, a digitizer is not the answer. Indeed, without modification, the
HP3577A, a network analyzer, will not produce accurate results.
After careful study it was determined that with proper signal processing, the HP3577A could be used for this task.
A note on the literature is now in order . Much has been written in the plasma field on mode coupling. However, because of its recent appearance on the market, apparently no one has ever attempted to modify the HP3577A to measure mode coupling . There fore no literature was found on this subject. In addition, no literature was found that discussed the modification of the HP3577A for accurate measurement of arbitrary signals in two channels, with low crosstalk.
The modification, which is the subject of this thesis, is a na�row band preselector which tracks the source frequency of the
HP3577A . The modification consists of a moving filter or tracking filter . The literature contains much information dealing with filters. Two main types of tracking filters are state variable
and yttrium-iron-garnet ( YI G) (ref . l ) • State vari ab1 e filters are filters containing operational amplifiers. In their operation, some 4 ampl ifier characteristic can be vari ed, affecting some criti cal frequency in the fi l ter's operat i on. State variabl e fi l ters have litt le to do with the operation of the mov ing fi lter contained in the HP3577A modif i cation. YIG fil ters, on the ot her hand, are very an alogous to the situation present in the modifi cation. These filters operate on the principle that by chang ing the DC magneti c field in a ferri te, the associated resonance of the ferrite can be adjusted. Bandpass fi lters in the range of 2 to 18 GHz can be constructed with Q's of up to 500. Preselectors that hel p mini mi ze al iasing in mi crowave spectrum ana lyzers are YIG fi l ters.
Since the region of operati on for the YIG fi lter is much higher than the fi lter needed in the mod ifi cati on, YIG filtering cou ld not be considered.
The fi lter implemented in the modification is a compos ite fi lter which contains cryst al fi lters and mixers . An extens ive computer search was done at the U. T. library. Sixty references were found wh ich contained one of the key words. However, none of the abstracts showed any work discuss ing adjustable fi l ters con taining crystal fi lters and mixers . The key words used included: fi lter, variabl e, mixer, state vari able, control lable, crysta l, an d composite .
The thesis is divided into seven secti ons. Section I dis cusses the HP3577A network an alyzer's normal operation an d the probl ems associ ated with the measurement of arbitrary signals .
Secti on II discusses the approach tak en in this thesis for solution 5 and estimated system performance. The second section also contains some experimental results to aid understanding the operation of the modification. Section III discusses the design and operation
of tracking filter. Section IV details the design and operation of the preamplifier . Section V discusses the way in which the local oscillator that drives the filter board is produced. This part of the system will be called the local oscillator generation
system. The spectral purity of the local oscillator and its effect
on system performance will be discussed. Also in section V, the
concept of composite filters will be discussed. Composite fil
ters are contained in both the preamplifier and the local oscil
lator generation system, but their operation was particularly
important to the local oscillator generation system. Sections
III, IV , and V will also contain performance data. Section VI
discusses experimental results and the calibration of the entire
system. Section VIII, the appendix, contains schematic diagrams
and data sheets not appropriate elsewhere. I. PROBLEM STATEMENT
To completely describe the task of converting the HP3577A network analyzer, the operation of the unmodified unit must first be discussed. The HP3577A is a signal/response instrument designed to measure the magnitude and phase transfer characteristics of linear systems. A signal/response instrument measures a transfer function by comparing the signal fed to the device under test (OUT) with the signal returning from the OUT. This is accomplished in the following manner: The HP3577A unit, in the normal mode of operation (Figure 1). generates a signal of known frequency, magni tude, and phase. This signal is fed into the device under test and the output of the DUT is fed back to the HP3577A. With proper calibration and extensive digital processing, the HP3577A measures the magnitude and phase characteristics of the OUT to within 0. 01 dB and 0. 005 degree, respectively.
Note that the receiver samples only 40 1 frequency points across the frequency range of interest. The effects of this dis crete frequency operation on the system conversion will be dis cussed in the performance section. The rated frequency response of the unit is between 5 Hz and 200 MHz.
The mode of operation into which it is desired to convert the unit is shown in Figure 2. Notice that with the modification of any arbitrary s1gnal can be measured by this "spectrum analyzer" mode of operation.
6 7
HP3577A I
OUT IN
IN 0 UT
DEVICE UNDER TEST
Figure 1. HP3577A in normal mode of operation. 8
HP3577A
INPUTS
A R
CONVERSION SYSTEM
ARBITRARY SIGNALS UNDER TEST
rigure 2. Desired configuration of the HP3577A. 9
At first glance, the user might feed an unknown signal directly into the HP3577A. When this is done, the frequency compo nents of the signal being fed into the HP3577A are indeed measured
(shown as spikes on the display screen) . However, spurious internally-generated responses will be displayed at frequencies
500 kHz below each spectral component actually being fed into the
HP3577A . The amplitude of each spurious signal will be indis tinguishable from the corresponding real signal. Along with these obviously spurious signals, many other responses at unpredictable frequencies may be present within the band of interest.
These results are demonstrated with a simple experiment. A
4 MHz sine wave with amplitude of -25 dBm is fed directly into the
HP3577A. Figure 3 shows the display screen in the narrow region around 4 MHz. The sine wave is shown at the correct frequency and amplitude (-25 dBm) . Figure 4 is an expanded frequency sweep.
Notice the second signal at a frequency of 3. 5 MHz. The signal at
3. 5 MHz is not present in the input, but is being spuriously gener ated by the HP3577A. �igures 5 and 6 show the display screen from
50 kHz to 625 kHz and 625 kHz to 1. 25 MHz respectively. Notice the many spurious responses due to the one 4 MHz sine wave. The direct-input technique is useless for the study of broad-band signals associated with plasma turbulence . The generation of the spurious responses will now be explained .
Figure 7 is a block diagram for the input receiver of the
HP3577A. The presence of the multiple spurious signals can be 0
- 10
-20
- 30 ' - E CD "C -40 \ - I w 0 -50 ::::> I, I- ...J - I 0... 60 I ::::!: II
-80 I lu II ) I ' � ! i ' !/IJ 1 h I I � I k � �.. - ! ��iIIll I! 1'� ,, r 90 )J\H rl ! �.1�;1 :I)!J' I �l 11 �� -'I'l v �ifir , J -100 �! 1: 111. •lv � , :' ., �� ,, I J 1 , ' 'INJ v\lrf 1 � U�1i, M:f�'1MJ,���}V l_ fi,i\i ���['liM���A lJ�v/�,�JtJ 3.90 3.92 3.94 3.96 3.98 114.00 4.02 4.04 4.06 4.08 4.10
FREQUENCY MHz
gure 3. HP3577A displ ay with 4 MHz sine wave input.
...... 0 0
- 10
-20
-30 E (]J -o -40 � w 0 - 50 :::>
-1- 'I ,. H ..J -60 (L �
FREQUENCY MHz
Figure 4. HP3577A display with 4 MHz since wave input and spurious response at 3.5 0
-·0
-20
-30 E Cil "0- -40 w 0 -50 :::::> +----++ ·1·--++ -1- ....J a.. -60 :::':!! <1: -70
-80 f---1�.-l r.... · K- -9 11 I t I: +., . . ; I ��I I L �l -- 0 lj �---��-1 _l 1 ) Ill. I II
50 I 07.5 165 222. 5 280 337.5 395 425.5 510 5 67. 5 62 5 FREQUENCY kHz
Figure 5. Demons tra tio n of spurious signals in 50 kHz to 625 kHz ba ndwi
N 0
-10
-20 -- -·
30 - E ll:l '"0 -40 l w 0 -50 :::J
-t- __j -60 ..... ·-· Q_ � <{ -70 ------
I -80 --1---- .I i ,11:! -90 � -100 625�l�n!JVAHilr··· 687.5 750 812 5 875 937.5 1000�[ I'J\Y,.Jllii111ID 1062.5 �tml 1125 1187.5���H� 1250 FREQUENCY kHz
Figure 6. strati on spurious signals in the 625 kHz to 1250 k ndwi
-'
w 14
'i<'�me as •nput 1'\IPUT A >------R
I"JPUT �}------same8 1npu1 as R �------
ti gure 7. HP3577A recei ver block diagram.
Source: Operati ng Manual Model 3577A Network Analyzer, Hewlett-Packard Company, P. 0. Box 69, Marysville, Washi ngton 98270, U. S. A., Microfiche Part No. 03577-90050, section 5, p. 8, November 19 83 .
In the normal mode of operation, the HP357 7A receiver mixes the si gnal from the device under test to the intermed iate frequency of 250 kHz. The local osci llator that drives the input mi xer re mains at a frequency 250 kHz above the si gnal fed into the OUT.
Since the HP3577A was desi gned to measure transfer characteristi cs of linear systems, the signal to the HP3577A rece iver should be at the same frequency as the signal from the HP3577A source. Here- in lies the operati onal strength of the un it and the problem for 15 accurate measurement of arbitrary-input signals . When the HP3577A is measuring the magnitude and phase characteristics at a certain frequency, normal operation stipulates that only one frequency (the test frequency) be present .
The arbitrary signal is fed into the unit in the steady state . Since mixers create sum and difference frequencies, 250 kHz signals will be fed into the intermediate filter bandpass any time the local oscillator is 250 kHz above or below all input spectral components . The unit will display two frequency components, only one of which is actually present . Also, mixers are not ideal linear devices . In addition to sum and difference frequencies, the nonlinear characteristics of the mixer will generate spectral components at frequencies of N*F (local osc.) � M*F (RF) which will be fed into the bandpass filter . If these spurious components fall into the IF bandpass, they will be indistinguishable from actual input signals . II. SYSTEM DESCRIPTION
In this section , a de scription of the signal processing necessary to eliminate the above mentioned problems will be dis- cu ssed. As shown in the previous section, contin ued in troduction of an arbitrary signal in to the HP3577A, during the time s the unit wa s not specifically tuned to the frequency of the signal, wo uld le ad to the display of in ternally-generated spurious signals. It is proposed to process the arbitra ry input signals such that each individual frequency component will be fed into the HP3577A only during the time the unit is tuned to that particu lar frequency.
The co ncept can be vis ualize d as a mo vin g ba nd-pass filter that tracks the HP3577A receiver as in Figure 8.
BANDPASS MOVING WITH TIME ! .. I HP3577A ...... ---r- SOURCE FREQUENCY
FREQUENCY
rigure 8. Tracking brickwa11 filter frequency re sponse.
16 17
This filter, with moveable center frequency, will pass a particular frequency component when the HP3577A receiver is tuned to that frequency, allowing accurate measurement of the spectral component at each frequency. Otherwise, the component will be attenuated.
In order to illustrate the concept of the moving filter more clearly, Figures 9, 10 , and 11 , which contain actual performance data, are shown . Figure 9 shows, at one frequency, the type of filter that will be used. Notice the high selectivity and high out-band attenuation of the filter centered at 10 MHz. Figure 10 shows 10 superimposed traces in the range between 1 and 2 MHz.
Each trace was taken with the effective center frequency 100 kHz above the previous trace . Figure 11 is of the same form as Figure
10 except that the range is between 1 and 10 MHz and the spacing is
1 MHz. The filter should follow the source frequency of the HP3577A , passing only the component in the 25 kHz band around the instantan eous source frequency.
The input signals will be divi ded up into two frequency ranges of interest: 5 Hz to 450 kHz and 450 kHz to 10 MHz. The design of the local osci llator generation system limits the filter response to about 450 kHz on the low end. Implementing the moving filter alone will not cover the frequency range from 5Hz to 450 kHz. This problem is avoided by the fact that in the lower fre quency range, the signal processi ng mentioned above , is not neces sary. The design of the HP3577A is such that if there are no 0
-10
-20 H E -30 rf co "'0 -40 w- 0 - :::> 50 f---- 0...__j -60 � <( -70 � ( -80 \
-90
- 100 9.9 9.92 9.94 9.96 9.98 10 00 10.02 10.04 10.06 10.08 10.1 FREQUENCY MHz
Figure 9. "Moving filter" with l 0 MHz effective center fr·equency.
00 0
-10
-20
- E 30 CD -c -40 w 0 ::::> - 50 1- ...J a.. -60 ::;E <( -70 i � � ""Y'"""'1 -90 - - 100 --- �--·- 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 LS 1.9 2.0 FREQUENCY MHz Fiqure 10 . Te n supe rimposed frequency Bode plo ts from 1 to 2 MHz. '--0 0 - 10 -20 +> E -30 Q) "'0 - i - 40 w 0 => -50 1-- ___J I a... -60 � H <[ j' 11 - 7 � I H I 0 • _....J I II!I . ->L -- rr I/ ��?...- -80 � �� ··�4R: '4"--���� w"'- "··� - � 90 � � · l. tltJ • � i � · �� t�. -100 '� . n.l..ITtL ._._ __:,· :r rr �;&!.1�!W.J.ii: �: . ill � tlliV''1: f• �i· \i 1itr{l: L.I�nn�M�Il����r�� �,l��� � 0.5 1.5 2.5 3.5 4.5 5.5 �J� 6.5 7.5 8.5 ·l�� 9.5 10.5 FREQUENCY MHz Fi gure 11. Ten Superimposed Frequency Bode Plots from l to 10 MHz. N 0 21 frequency components in the arbitrary input signal above 500 kHz, the unit will operate alone with the same accuracy as the unit with the modification. In order to eliminate the frequencies above 500 kHz in the low frequency section of each channel, special filters were designed . The block diagram of the modification is presented in Figure 12. Again, notice the three major system blocks. The function of each major system block is explained in the next three secti ons. HP3577A Dl SPLAY NPU TA 5Hz TO 450kHz n EX PHAS 5Hz TO 0450 TO T lOMHz R LO CKING 10fl1 lz FILTER HP3577A PREA!v1P 0,45 TO IOMHz BOARD INPUTS A INPU T B 5Hz TQ 450kHz � SOURCE 5Hz TO -- '---··· 45.45 TO 55MHz I OM --iz + 26dBm LO 0.450 TO IO MHz GEhlE RATION -7dBm + 15dBm IO MHz TH. E'-- Fiqu re 12. Conversion sys tem block d1 agram. N N III. MOV ING FILTER Introduction The moving filter board is the heart of the conversion system. The functional purpose of the moving filter is to selec tively filter the arbitrary signal input and pass only the desired components . A list of important specifications that the filter board should approach is given in Table l. The moving filter block diagram is shown in Figure 13. TABLE l FILTER BOARD SPEC IFICATIONS Frequency Range of Operation 450 kHz-10 MHz Passband ± 15 kHz F(instantaneous) Attenuation < 10-15 dB Dynamic Range 0 dBm to -60 dBm Impedance 50 ohms Crosstalk -50 dB Out of Band Attenuation > 45 dB Output Spurious Response > 40 dBc 23 FILTER BOARD MIXER MIXER I 2 LOW PASS LOW PASS LOW PASS CRYSTAL FILTER FILTER FILTER FILTER = 45MHz Fe= 10 F 50MHz Fe:::IOMHz L L 15db AMPLIFIER SIGt�l1L UNDER TEST To Input 50kHz - IOMHz 3577A DRIVE BOARD 5 WAY A � POWER } TO MIXERS / 1------CHANNEL 2 o{:POWER AMPLIFIEH 50.11 Fi gure 13. Fi lter board block diagram (l channel). N -+:> 25 Design The moving filter can best be visualized as a bandpass brickwall filter with variable center frequency. When one thinks of a bandpass-brickwall filter, a crystal filter comes to mind. However, a crystal filter has a set center frequency and cannot be vari ed. Ixers therefore are used to convert information from the frequency of interest to 45 MHz and back to the ori ginal frequency. The pri nci ple of operati on of the system is straightforward. All si gnals are mixed to 45 MHz, filtered by a crystal filter with a 25 kHz bandpass, then mixed back to the origi nal frequency. This process passes the frequency components whi ch satisfy the following equation: F(local osci llator) - F(spectral component) 45 MHz . Eq. l It is easily seen by the preced ing equation that by varying the local osci llator frequency, the effective passband of the system is controlled. Duri ng actual system operati on the local oscillator remains at a frequency 45 MHz above the HP3577A source frequency. The operation of the movi ng fi lter will now be di scussed. The output of the preampli fier is fed into the input mixer's inter mediate frequency port. Thi s si gnal can be vi sualized as any spec trum that has been bandli mi ted to 10 MHz. When the si gnal is mixed up, it is centered around the local osci llator frequency. The local osci llator is set so that its frequency will remai n 45 MHz above the instantaneous HP3577A source frequency . While the source 26 frequency sweeps from 450 kHz to 10 MHz, the local oscillator sweeps from 45. 450 MHz to 55 MHz . The output from the input mixer •s radio frequency port is fed into the crystal filter configuration with center frequency of 45 MHz. Any signal that passes the crystal filter configuration will be amplified and mixed back down to its original frequency . An example of the moving filter•s operation is now illus trated. Suppose a sine wave at a frequency of 2 MHz is fed from the preamplifier into the input mixer. Also, assume the HP3577A source frequency is sweeping from l MHz to 3 MHz . When the source frequency is at l MHz the local oscillator is at 46 MHz . The in out mixer will produce sum and difference frequencies of 44 and 48 MHz . Neither of these signals will pass the crystal filter and no signal will be present to drive the second mixer . Indeed, no signal is exactly what the system should produce when the source frequency is at l MHz. The input signal is at 2 MHz and should be fed into the HP3577A only when the source frequency is at 2 MHz. Next, assume the source frequency is at 2 MHz. The local oscil lator frequency will be 47 MHz . The input mixer will produce signals at 49 and 45 MHz. The 45 MHz sisnal will pass the crystal filter, be amplified , and then be mixed again. Since the local oscillator Jf the second mixer is still at 47 MHz, the output of the second mixer should have si gnals at 10 2 and 2 MHz. The HP3577A receiver is tuned to 2 MHz and will correctly measure the input 2 MHz sine wave. It is necessary to filter the output of the 27 second mixer to eliminate local oscillator feed through and all other high frequency components. In the following, each of the components in the filter board is described and the design specificatio n that led to the selectio n of each discu ssed. To minimize costs, standard crystal filters, mixers , and amplifie rs were used. Crysta 1 Filter In the selection oi a crystal filter, cost and frequency of opera tion were the main considera tion s. Crys tal filters were readily available from Pie zo Tech nology, Inc. (PTI) for $45 each . The standard va lues ava ilable were at 30 MHz, 45 MH z, and 75 MH z. The 45 MH z filter wa s chosen for two reasons: Fewer spurious sig nals enter the 450 kH z-10 MHz passband from the 45 MH z filter than the 30 MHz filter (t his remark will be be tter understood after the mixers are discussed}. Fu rther, amplifica tion of the 45 MH z signal is easier than the amplification of the 75 MHz signal. Also, fewer second order frequency effects occu r with the 45 MH z sys tem than with the 75 MH z system. Fo r these reasons, the Model 437 1-45 MHz crys tal filters we re used. "The filter ign utilize s four re sonators on a single quartz wafter to provide four-pole res ponse. These models offer wider passband va lues than two -pole inductorless ta ndem-set units" (ref . 3). The input imp edance of the filter is 7000 ohms, therefore, a matching network is necessary. This matching network, the design 28 of which was supplied by PTI , is shown in Figure 14. Cl , C4 = 5 pF Ll, L2 = 0. 88 uH C2, C3 1-10 pF. Figure 14. PTI supplied matching network. Source: Single-Wafer Four-Pole Monolithic Crystal Filters (Brochure) , Piezo Technology, Inc. , ?. 0. Box 7859, Orland, Florida 32854-7859, p. 4. Figure 15 shows the PTI supplied attenuation-frequency plot of the Model 4371-45 MHz crystal filter. While this plot is very impressive , it does not tell the whole story. Note that the fre- quency bounds of the plot are 60 kHz below and 14 0 kHz above the center frequency. At greater deviations from the center frequency the attenuation decreases to about 25 dB. Also reducing the filter 's out-band attenuation is the varying load presented to th2 filters by the mixers. The varying local oscillator frequency driving the mixers changes their output impedance. Changing the load on the crystal filter reduces the out-band attenuation, since the matching networks are tuned for maximum out-band attenuation at a single impedance. Taking both these characteristics into consideration, 29 it was necessary to cascade three crystal filters to obtain 45 dB out-band attenuation for the entire frequency range of interest. The matching network was modified and the final configuration is shown in the filter board schematic (Appendix A) . � :::: r � / � r-... 70 I "'\ i'J \ \1/ 60 I(( rt/A � I �(I z 1/ 40 0 \ fi 7; Wl, '17 ::I 30 z vm\ w t:• '< \ I 20 \ /I 10 � J 0 - 60 -40 -20 fon 20 40 6000100 120 140 RELATIVE FREQUENCY (kHz) Figure 15. Amplitude characteristics of 4371 crystal filter. Source: Single-Wafer Four-Pole Monolithic Crystal Filters (Brochure), Piezo Technology , Inc. , P. 0. Box 7859, Orlando, Florida 332845-7859, p. 3. The frequency response achieved with this system is shown in Figure 16. The out-of-band attenuation of this system varied between 45-85 dB across the frequency band. This out-of- band attenuation 1s a limiting factor for the dynamic range of the in- strurnent. A few practical points should now be discussed. The varia- tion of the variable capacitors in the matching network has a marked 0 ------·------� --� ------.------··-----" ··- - - -�------10 r------f.---- --2 0 - ·;r::--�--;--- - 30 "-- ·---- -· E OJ "'0 - 4 0 ------·- - � w 0 - 50 ::J r------1- I _j - \ 0... -60 � c::t: -7 -- -, \ = 0 f-· f--· I I I -80 '------1/ 1\.� VJI"i,· ·�IVJ( �i{\1 ''·'·l'i· ·�v1--,'--·J llrlfl -9 ''\lt•i1'(\!-\.y� u;\��,' 0 "h'M"'1Yl""�liiM'W·� -100 '---··- ----'----- " 44.9 0 44.92 44:94 44.96 44.98 45.00 45.0 2 45.04 45.06 45. 08 45.10 FREQUENCY MHz .� Figure 16. Actual cascaded-crystal filter response. w 0 31 effect on the performance of the filter. It is possible , while covering the entire range of the capacitors, to transform the system from a workable filter with 10 dB loss and 2 dB in-band ripple to a system which will have 50 dB loss and 35 dB ripple. It is neces sary to use high quality high-Q air dielectric capacitors. To achieve the desired out-of-band rejection it was necessary to place a grounded metal plane 1/ 4 inch below the filter combina tion. This plane grounds out fringe fields that were produced by the inevitable impedance mismatch. While the filter has flat amplitude characteristics , the phase shift across the bandpass is large (500 degrees) . This phase shift and its effect on system performance will be discussed later. Mixers In each of the two channels of the filter board, two mixers are used. One mixer mixes the signal up in frequency , while the other mixes the signal down in frequency. The mixers should satisfy the following constraints: First , the frequency range of each of the three mixer ports must incl ude the frequencies that will be introduced to that port (RF = 45 MHz, Local Oscillator = 45 MHz to 55 M�z, and IF = 450kHz to 10 MHz). Second, the mixers must satisfy the desired ampl itu�e dynamic range from 0 dBm to -80 dBm. Third, it is necessary that the spurious responses generated by the mixers be as low as possible. Fourth, the cost of the mixers must not be excessive . 32 A description of the operation of the mixers used is now in order . The mixers used in the conversion system are shown in Figure 17 . This is a classic double-balanced mixer that suppresses the local oscillator frequency component from entering either the IF or RF port. schenncmc l 1 i Figure 17 . Double balanced mixer schematic. Source: RF/IF Signal Processing Handbook, Mini-Circuits, Inc. , P. 0. Box 16 6, Brooklyn, New York 11235 , p. 33, 1985-1986. In Secti on II, the spurious generation properties of mixers were briefly di scussed. However, the dependence of the mixer oper- ation on the local osci llator was not mentioned. The maximum input amplitude and purity of the local oscillator signal are very important system parameters. The amplitude of the local oscillator determines the maximum signal that can be fed into the RF or IF ports while maintaining linearity. Also , the local oscillator has much to do with the spurious output components. �or a given I� or RF signal level , the level of spuriously generated response will be inversely proportional to the local oscillator level . That is, the higher the local oscillator power level , the lower the spurious responses . This is best illustrated in Table 2. TABLE 2 XER HARMONIC LEVELS FOR TWO MIXERS TH DIFFERE DRI LEVELS ----·----· mixer harmonic intermodulation (r&lalive lo dolj(&CI f oolpull \ Rf Rf CAL 0 30 16 34 2:. 41 $1 � $ 0 0 ;.QO 10 a :.t 11 :l9 l1 41 4Q u .. <14 - 20 44 z z 30 ll 34 l2 $1 61 :.a $1 >90 tO J.4 40 4a 211 41 e.o 4a 0 t I (I 60 0 I � ' � 15 l2 :l9 45 34 � $1 2 11 11 41 11 1$ 0: 2 :14 � � :.9 60 0: $3 00 00 < < :X: 42 61 ::z: l>QO 13 $1 11 12 10 11 l « u 43 00 .. , $(> 6J 1>0 6-4 0: .9!f 0: � w � $] $1 4a � 41 10 ll 0 .. > >12 >It >62 11 11 • 90 12 � 00 Q() n >U 0: 0: 0 0 LL $ ·94 ll :l.>QO 11 11 > 12 >at 11 6'1 t>:. :.9 � :.9 :.9 6-4 f.t9 6() ..... >U >02 0: 0: -94 1Q •. 1(1 19 $1 61 11 6 >QO 16 Ia 11>12 >U :..82 4 .;-U 64 >42 u �8l ;� � 7l 61 13 1 >QO 1 11 11 11 >e2 >81 �w. -� 4 >82 -94 -84 ·42 >44 13 a 11 11 18 16 18 >42 >at • ·6<1 -·� 6() 6'1 ;;. oo .Q4 -u -u -n "u >44 >a.t at a >11:. 11 11 11 11 18 11 > 62 >Ill 9 -e..& 9::.-Q() �n >44 :.32 >44 >6$ IIQ 7a 11 11 111 11 11 16 > 82 IQ ·9l ·U .;.. 84 116 �>54 � 10 :. 6() 1 10 10 Q t a l .. � 6 • 9 0 t 2 4 � 6 1 I 9 tlarmonlc LO Or dar ftarmonlc lO Or dar Mil -4 100Mil hUI ------· ----- SOURCE: Ibi d., pp. 164-165. (..d w 34 Tabl e 2 co ntain s spurious response tabl es fo r two mixers that use different local oscil l ator levels but have the same RF leve l . Notice that the components of in terest (inside the lines) for the mixer with a 17 dBm local oscil l ator (l evel 17) have much lower leve l s than the mixer with a 7 dBm local oscil l ator (l evel 7) . No te that the fre quency components of in terest are the 2x2, 3x3, etc. These may fal l in the output bandpass of in terest. Another form of dis tortion, lower in the leve l 17 mixers , is two-tone in termodul ation dis tortion . Tw o-tone in termodulation occurs when two components of a spectrum (fl and f2) , fed into mixers RF or IF port, mix together to produce an output spectral component. The frequency of this spurio us response wil l occur at a frequency defined by the re l ation ((2*fl = f2) � f(l ocal osc)). Al l mixers present some degree of two-tone dis tortion. The two tone dis tortion leve l in the level 17 mixers wil l be 50 dB Je low the amp litude of the smal ler of the two spectral components mixin g together, the level 7 only 30 dB. The mixers chosen for the project were the Mini-circuits TFM-3. These mixers have the specifications shown in Tabl e 3. This component fit s the des cribed re quirements we l l. The maximum inp ut level is we ll above the desire d input specifications. The frequency dyn amic ran ge is much greater than needed . The local oscil l ator to IF isolation shoul d be suffici ent, so that with addi tional fil tering its effect on system performance should be smal l. 35 TABLE 3 MINI-CIRCUITS MODEL TFM-1 CHARACTERISTICS Frequency Response: LO/RF 0.1 to 250 MHz IF DC to 250 MHz LO Signal Level 17 dBm Maximum Input Level 14 dBm LO to IF Isolation > 25 dB LO to RF Isolation > 30 dB Impedance 50 ohms Conversion Loss 6 dB Cost $45 SOURCE: RF/IF Signal Processi ng Handbook, Mi ni-Circuits, Inc. , P. 0. Box 166, Brooklyn, New York 11235, pp . 164-165, 1985- 1986 0 Mixer performance specifications are contingent on proper loading of all ports and adequate local oscillator level . If the load on the RF or IF ports is not the desired 50 ohms, the levels shown in Table 2 will go up . Even if the mismatch does not occur in the band of interest, it can affect in-band performance . For example, directly cascading a fi lter after a mixer can cause a system to display th e above menti oned effect . Suppose the base band of interest extends to 10 MHz and signals abov2 are filtered with a multiple pole low-pass filter. The in-band impedance of 36 the filter is probably close to 50 ohms, but the out-band impedance may be far away from 50 ohms. It was observed that the mixer's third harmonic suppression was greatly reduced when loaded directly with a filter. The out-band mismatch of the crystal filter con figuration reduced harmonic suppression. However, the change in the impedance of the crystal filter configuration is not as radical as an LC filter. If the level of the local oscillator signal is not adequate, several undesirable effects arise. First, the con version loss increases as the local oscillator signal level de creases from the recommended level. If the level of the local oscillator signal on the level 17 mixers drops from 17 to 10 dBm, the conversion loss increases from 6 to ll dB. Further reduction in local oscillator level increases conversion loss exponentially. Second, reducing local oscillator level reduces the suppression of spurious outputs. Finally, it is essential that the packaging of the mixer follows good practice. If the mixer is improperly grounded, the local oscillator to RF isolation and spurious generation specifica tions will be compromised. Correct grounding, for a mixer packaged in an 8-pin dip can, means that the cas2 itself nust be soldered to the ground plane. It was necessary to use a ground plane on ooth sides of the filter board, and then to solder the mixer case directly to the board. 37 Amplifier The purpose of the amplifier in each channel is to prevent the signal to noise ratio reduction of the system from becoming too high. The sum of all losses in the channel is approximately 30 dB (excluding the amolifier}. The amplifier has a gain of 15 dB. Cascading the amplifier in line with the rest of the system results in a system that has a 15 dB reduction in signal to noise level. The input noise level of the amplifier is very small (-110 dBm) and does not affect system performance. The amplifier used was obtained from government surplus. The Avantek UTA-108 has the specifications listed in Table 4. TABLE 4 AVANTEK MODEL UTA-108 AMPLIFIER Gain 15 dB Frequency Response 5 MHz to 500 MHz 1 DB Compression Point 3 dBm Supply Voltage 15 Volts Signal to Noise Ratio 13 dB Impedance 50 ohms The amplifier was driven very close to the 1 dB compression point. The 1 dB compression point is defined by the output level where the gain has been r·educed by 1 dB. Due to nonlinearity. the 38 energy will go into the harmonics instead of the fundamental. As the output level approaches the 1 dB compression point, the level of the harmonics greatly increases. Fortunately, these harmonics are located around 90 and 135 MHz , far away from the band of in terest. Output Filters The filter cascaded with the amplifier is a low-pass filter with a cut-off frequency of 50 MHz. This filter eliminates higher frequency components generated by the input mixer and the amplifier. Since the output of mixer 2 will have many spurious compo nents outside the 450 kHz to 10 MHz band of interest , it is neces sary to filter out these components so that they do not create spurious signals in the final system. A four section standard low-pass filter was used . The attenuation at frequencies above 50 MHz is 60 dB, which effectively eliminates high frequency signals. The VSWR of this filter is typically 1. 5 in band and 17 out of band. The output filters have the same effect on mixer performance a5 described in the mixer section . In tris case, the higher level of harmonics, generated by an improperly loaded ou tput mixer , has little effect. Suppose a 1 MHz s1 gnal is filtered by the modi fica tion and fed into the HP3577A receiver when the receiver is tuned to 1 MHz ( normal operation ) . The third harmonic will be at 3 MHz . Since the receiver is not tuned to 3 MHz , the spuri ous response 39 should not affect the measurement of the l MHz signal greatl y. When the receiver frequency reaches 3 MHz the 45 MHz frequency driving the output mixer is attenuated so that its vol tage level is negl igible, so there wil l be no 3 MHz harmonic to measure. Experimental Results The most graphic experimental resu lts for the fil ter board are shown previ ousl y in Section II. These show the effective fi l ter passband being vari ed across the band of interest. The passband wi dth is 25 kHz and the ri ppl e is 4 dB. The frequency range of the fil ter is DC to 10 MHz. The ampl itude dynamic range of the fi lter is 0 dBm to -100 dBm . The -100 dBm specificati on is for an ideal loca l osci llator and is not obtai ned with the current local oscil l ator generation system . The outband attenua ti on is greater than 45 dB across the 450 kHz to 10 MHz band . The harmonics are greater than 40 dB bel ow the fundamental. IV. PREAMPLIFIER Introduction The two-channel , four-output preampl i fier is used to iso late the outsi de measurement devices from the mixers of the adjustabl e fil ter. Speci fic features incl ude the fol lowing: The preampl ifier should provi de proper 50 ohm loading for outside measurement devices as well as mixers on the adjustable fi lter. The preampl ifier should divide si gnals into two frequency bands of interest from 5 Hz to 450 kHz and 450 kHz to 10 MHz . The system should be band limited to 10 MHz and DC should be blocked on the high-frequency channel. The low-frequency section of each channel shoul d have a sharp rol loff so that no si gnal s of fre quency greater than 500 kHz pass into the HP3577A. The noi se level of the ampli fi er should be lower than the noi se level of the filter board. The maxi mum input level shoul d be greater than -6 dBm . Harmoni c distorti on for the system shoul d result in harmonics wi th power level s at least 40 dB bel ow the maximum power output over the enti re range of interest . Cesi gn The block diagram for the preamplifier is shown in Figure 18. The majority of the system is composed of operational ampl i fiers and standard components . A design using discrete devi ces was employed for the composite filters. The operational amplifiers 40 -- IOMHz 51 F IOdB n oo: I I - � LOW - PASS BUFFER � 5Hz TO FILTER 450kHz j_: - COMPOSITE COMPOS I :FFER BUFFER 450kHz TO FILTER FILTER 1 � IOMHz 51 .00564F I OdB IOM Hz n LOW -PASS BUFFER � E----o 5Hz TO ATOR �--_: FILTER 450kHz COMPOSITE f-----.1 COMPOSITE BUFFER BUFFER -----a 450kHz TO FILTER FILTER [ IOMHz Figure 18 . Preamplifier bl ock diagram. -Po 42 used were the Harri s Semi co nductor model s 2539 and 5002. The HA2539 is a device wi th a 600 MHz gai n bandwi dth product . One HA2539 was used in each chan nel with a linear gai n of 13.3 or 22 .5 dB. Since the gai n bandwi dth is 600 MHz, the 13.5 gain al l ows a bandwidth of 44 MHz. From a chart of phas e shi ft vers us loop transmi ssion, the phase marg in was determi ned to be ap proxi matel y 75 degrees. Thi s 75 degree phase margin was needed because of the suscepti bi lity of the op-amp to oscillate. The slew rate was speci fi ed to be 600 V/uS. At 10 MHz maxi mum frequency the maximum ampl i tude should be 9.5 vol ts or 29 .6 dBm . The noi se level of the HA2539 is speci fi ed at -1 13 dBm in a 1 kHz bandwi dth. Since the HA2539 was incap able of driving 50 ohm loads, the HA5002 was used as an impedance matchi ng mechanism. The HA5002 is a 110 MHz buffer amplifier capable of driving 50 ohm load s. The maxi mum output cu rre nt for the HA5002 is 200 rnA, well ab ove the 20 rnA needed to dri ve l vol t into 50 ohms. The noise level for the HA5002 was ap proxi mated at -1 00 dBm in a l kHz bandwidth. Voltage regulators were used on each chan nel to isolate the amplifiers from power line noi se. Many bypas s capaci tors were co nnected from power pi ns to ground to decouple the ampl ifiers fro m one an other. The input of the amplifier co nsists of a 10 dB attenuator and a 10 MHz low-pas s fi l ter. The attenuator provides a constant 50 ohm load wi th a USWR of less than 1.5. The fi l ter has a sharp ro lloff after 10.7 MHz whi ch effectively attenuates hi gh-frequency 43 signal components which might be misinterpreted . The output of the HA5002 contains a series RC network. The resistor is 50 ohms which provides an in-band match for the mixers of the adjustable filter. The capacitor was a 5600 pF component which effectively blocked DC without affecting in-band performance. The filters which were designed to eliminate signals of fre quencies greater than 450 kHz in the low-frequency section of each channel are "composite filters ." These fil ters were used for several purposes in the thesis project. The design concept and design equations are presented in the local oscillator generation section. Experimental Results The preamplifier met all minimum specifications. However, the system was difficult to stabili ze against oscillations . The HA2539 chip was inclined to oscil late when the circuit •s physical layout was not perfect. It was necessary to connect bypass capaci tors directly to power pins. The 9.5 MHz speci fication for the full power bandwidth of the HA2539 was very misleading. This specification was derived and not measured. ;ull power bandwidth should mean maximum frequency at which a maximum output sine wave can be fed into a load with some specified mi nimal distortion. Harris Semiconductor derived the full power bandwidth using the slewrate of a square wave in a calculation to obtain their 9. 5 MHz for the full power bandwidth. The distortion of the waveform at 44 amp litudes greater than 0 dBm at 10 MHz was unac ceptabl e. The dis tortion was in the form of slewing. At 10 MHz with an output of 0 dBm the harmonics were ap proximately 40 dB be low the funda mental . Figure 19 contains data of harmonics at 5 MHz and 10 MHz. Figure 20 contains a Bo de plot of the low- frequency section of each chan n el . No tice the extremel y sharp cut off abo ve 400 kHz. The crosstal k between the low-frequency sections was very low. Figure 21 contains a Bo de plot of the high frequency sec tion of each channel . The crosstal k between the channels is asymmetrical , with val ues of -43 dB from left to right and -5 0 dB from right to left. 0 0.08 • AMP. AMP dBm dBm I I 5 10 15 10 -u20 30 F(MHz) F(MHz) Figure 19. Preampl ifier output for fundamenta ls at 5 MHz and 10 MHz. � (J'1 , _ - : li0fl1.1!llll. .I . !Uill]\_, i UUI I ill �- ! II 2 0 I Ill _ ll t+tffi1 l.lrlj n ll1 I l I I l ,. II - 3 0 �@l , j E ' I I (1) � I "0 -40 f II 11 I l.d j_ il l 0 I �f·- - . - 50 IiI' : ::J � ) . i 1-- -- l ...... 1 I I - I, 0.. -60 r- � <[ I - H_J_ I I I, ' ' 70 . . I I I I I JII I I ! i -80 �II : - I • . . I -�0 I I 1 , ,1 � � 1t i 1 I II,dk 1w I\-,I' . A l�-c �' -100 l I ' i. II l'�i.!�tliW.'� · W!',/ llil�/1 IOK lOOn1 K IM IOM FREQUE NCY Figure 20 . Bode plo t of low-frequency secti on . � (J) 0 -·0 - 20 _. v / \ - 3 0 E (l) v�--' -c / \ 4 - - 0 \ w 0 / ::J - 50 1- - _J a.. - 60 ., :2 r - 80 - 90 --!- ,..... -100 !OK lOOK IM tOM FREQUENCY Fi gure 21 . Bode plot of hi gh-frequency secti on. _p,. "-J V. LOCAL OSC ILLATOR GENERATION SYSTEM Introduction The final part of the HP3577A modification is the local oscillator generation sys tem . In the preceding secti ons dealing with the operation of the moving filter, it was assumed that the local oscilla tor driv1ng the filter board mixers would track the source frequency of the HP3577A with a 45 MHz offset . Thi s is not completely straightforward in this appl ication. Conceptually, it is desired to add 45 MHz to the sourc e frequency . Since the addition of frequenc ies can be accompl i shed by mixing, mixing the HP3577A source frequency wi th a 45 MHz signal wi ll generate the desired local osci llator . Unfortunately, the mixing process generates ma ny other frequency components besides the sum frequency . The mixer output mu st be fil tered to reduce these und esired frequency components to acceptable level s. The local oscillator shoul d meet the fol lowi ng specifications : The output signal should track the HP3577A source wi th a 45 MHz offset . The frequency bounds of the local oscillator should extend from 45.45 MHz to 55 MHz . Any spurious signal s should have power level s at least 40 dB below the power in the fu nda mental . The ri ppl e in the ampl itude should less than 5 dB. The nomi na l power output of the local oscill ator generation system is 26 dBm. The power of the output signal must be 26 dBm 48 49 due to power division in the fi l ter board . The power level at the local oscillator port of each mixer in the moving fil ter will be 7 dB less than the output of the loca l oscillator generation system . The rema inder of Section V contains the fol lowing : Local osci lla tor generation system design, experimental results, and the effects of undesirable spectral component s on system perform ance. In addition, composite fil tering , which played an important role in this thesi s, will be discussed . Des ign The block diagram of the local oscillator generatio n sys tem is shown in Figure 22. The local oscillator generation system mixes toget her the HP3577A source frequency and a synthesi zed 45 MHz signal . The mixer 1s output is fil tered to pass only the sum frequency . The remaining signal is then ampli fied to the proper level . The ampl ifier used for the power ampl ification to the 26 dBm level was a Mi ni -Ci rcuit ZHL-32A . The maximum output power of the ZHL-32A is 30 dBm. The frequency ra nge of thi s device is between 50 kHz and 130 MHz . The gai n of the ZHL-32A is 25 dBm. The mi xer used was again Mi ni -Ci rcuits 1s TFM-3 mi xer . The reasons for thi s selecti on are the same as explained in the mi xer MIXER B BANDPASS ,� SYNTHESIZER FILTER TO A F 45MHz t 45.5- FIG. = 1 .. 13 o 55MHz EXT IOMHz/N TO FROM SOURCE REAR REFERENCE HP3577A HP35T1A 500 kHz TO IOMHz Figure 22 . Local osci llator generati on. 01 0 51 subsection of Section III. The higher level local oscillator resul ts in lower harmonic content in the output. The TFM-3 mi xer 's output contains three undesi red spectra l components that must be el iminated . These include the difference frequency , the second harmonic , and the thi rd ha rmonic of the local oscil lator. The power level of the input signal driving the mixer will be +1 5 dBm for the 45 MHz loca l oscillator and -7 dBm for the HP3577A source. The fil ters neces sary to el imin ate the difference frequency and the second ha rmonic cf the local oscillator are not straight forwa rd . At the lower limi t of the system's operation (source frequency = 450 kHz) , the difference fr equency is only 900 kHz bel ow the desired sum frequency . In the 45 MH z region, 900 kHz is a very short frequency spa ce . Sta ndard fil ters could not el imi nate the difference frequency wi thout affecting the sum frequency . The ty pe of fil ter u for this operation is called a "composite fi lte r." The composite filters for the local oscillator generati on system were divided into two parts , cascaded high-pass and 1ow-pass fi l ters . An ampl ifier bu�fer wa s cascaded between the two fi l ters to provide isolation and in. site Fi l ters Composit e fi l ters were develo ped ma ny years ago to hel p el iminate crosstal k between channel s on mul ti p le carrier tel ephone lines . The ba sic concGpts needed to understa nd the operation of 52 composite fil ters are these: Two-part networks can be designed such that loading the network1S output wi th a particular impedance will result in that impedance bei ng measured at the input. This impedance is cal led the image impedance of the network . Many such networks with the same image impedance can be cascaded without affecting the transfer function of any individual two-part component . Imaged ma tc hed LC circuits containing series and parall nk circuits can be designed . Series LC circuits and paral lel LC circuits have special characteristics at resonance. In the case of a series LC circuit, a short circuit is obtai ned at the resonance frequency; for the para l lel, an open circuit. In the design of composite fil ters , image matched high and low-pass two ports are cascaded with image matched networks con taining LC tank circ uits. The resonance frequencies of these tank circuits are outside the bandpa ss of the filter. At resonance frequencies, the signal is either blocked or shorted , preventing the signal from reaching the output . The high or low-pass net works el iminate frequencies far away from the band pa ss. The LC circuits are used to el iminate particular frequencies. Since resonance is a narrow ba nd phenomenon, the rolloff in the resonance region is sharp. These tank circuits can be pl aced just outside the pas sband of interest to achieve rapid rol loffs . Since the response of the tank circuits is very narrow band , sev era l tank circui ts across the reject ba nd may be necessary . These extra networks prevent the total filter1s response from rising in the 53 rej ect band . The design equations for the composite fil ters are shown in Appendix C. Three composite fi l ters were designed and integra ted into the modification. The first, conta i ned in the preampl ifier, was desi gned to have an extremely sharp rol loff about 450 kHz . The results for this fi l ter are shown in Figure 20, page 46 . The second , is the high-pass fil ter in the local osci llator generati on system that pa sses signals above 45 MHz and rejects signal s be1ow . The third is a low-pass fil ter which filters out unwanted higher harmonics generated by the mixer i� the local osci llator genera tion system. Experimental Results The mixer's inp�t and output signal s are shown in Figure 23a . Note the ma ny spurious signal s in the mixer 's output. Thi s spectrum is then fed into the bandpass fi lter . The high and low pass filters contained in the local osci llator generation system are shown in Figure 24. The composite fil ter response for the high-pass low-pass confi guration is shown in Figure 25. The Bode pl ot in Figure 25 is the most important factor affecting the performa nce of the entire modification . Even though the rol l off at 45 MHz is sharp, it is not idea l . Therefore, the idea l goa l of el imi nating the difference frequency and not affect ing the sum frequency is not obta i ned for frequencies very close to 45 MHz . Figure 23b is the spectrum obtai ned from the local IF DE SIRED o-----·· - SIDE BA ND / A 13 / r dBm -13 // LO -5()- 2 / 50 d�m 5 ,: -60 l-7 __ j __ __ 5_lj 10 15 35 40 45t 50 1�55 85 90 95 f(MHz) f(MHz_lli�-) ' +15 A dBm l - 3 5 �---45-----L 90 f(MHz ) (a ) z Local illator Spectrum 23 . Local oscil lator generation operation 5 MHz . U1 -+::> 55 +2 A dBm -45 40 45 50 55 85 90 95 f(MHz) FILTERED OU TPUT (b) Loca l osci llator generation spectrum after band-pass fil ter. Figure 23 (conti nued ). 56 -5 A dBm -35 � -40 I 44 45 46 89 90 91 f( MHz) (c) Local oscil l ator generation spectrum after band-pass fi lter for 1 MH z source frequency . Fi gure (co ntinued). 450 KHz 10 IOMHz HIGH- PA SS FILl ER fc = 45MHz -7dBm 12 160 8 LOW , FILT ER PASS +15 55MHz 10 POWER 176 176 15db AMPL IFIER � � U1A- 108 I I ( r I 8 8 /� ! Figure 24 . Local osci llator generation system schematic . U1 '--1 0 - -·-----·· -·- -·--··-----· r-· ·0 :------ -20 -"'- . �· " � ·- --·-- � ___ -- - 30 ( ,...._... E (1) -o 40 1---- � \ w v- 0 - 50 \ r----. :::l ----- 1- AMPLITUDE i .....1 -60 a.. ::2: <( - 7 0 � ------··- -·--- y -80 \AJ +180 � ---- � - - �- PHASE - 90 l___ -180 i v r\ �' -100 ('\ � �- 35.0 41.5 48.0 54.5 61.0 67. 5 74.0 80.5 87.0 93.5 100.0 FREQU ENCY MHz Figure 25. Bode plot 45 .45 MHz to 55 MHz ba ss fi lter . CJ1 co 59 osci llator generation system when the HP3577A sourc e frequency is 5 MHz . Notice the excel lent spectra l purity. Figure 23c is the spectrum obtained when the source frequency is 1 MHz . Not only is the sum frequency attenuated , the difference freq uency is at a higher level . Thi s results in lower spec tral purity. Local Oscillator Specifications : Nomi nal Ampl i tude 26 dBm Frequency Ra nge 450 kHz to 10 MHz Spurious Signal s ( 1 .5 MHz to 10 MHz) 40 dB below fu ndamenta l Spurious Signals ( 450 KHz to 1.5 MHz ) 30 dB below fu ndamenta l Local oscillator fu ndamental frequency tracks HP3577A source wi th 45 MHz offset. The effect of the local oscilla tor 1s spectral purity en the performa nce of a mi xer must now be discussed . The rel ationship between local oscillator spectral puri ty and the resul ting mi xer output is compl ex . A mi xer is not a two part . Therefore, spurious signal s and harmonics contained in the local oscillator spec trum do not lin early affect the output. This phenomonon is best demonstrated wi th an exampl e. Suppose a square-wave local osci llator ( frequency Fl ) is used to mix with a sine wave (frequency F2) in the mixing process . A square wave is composed of ma ny odd harmonics in addition to the fundamental . However , when the mixer1s output is exami ned , only the sum and difference frequencies ( Fl ! F2) are present . Why don 1t the harmonics of the square wave produce an effec t? The rea son is because a diode mi xer , 60 a bi-pha se modulator , changes the phase of the IF signal 180 degrees each time the sign of the local oscillator changes . Si nce the operation of the local oscillator is dependent on sig n change alone, the harmonics of the square wave, which add together , have no effect. As a rule thumb , unl ess a local oscillator spurious component is a harmonic of the fundamental , a spu rious response wi ll be present in the mi xer 1s output. The ampl itude and frequency rel ati onshi p between the desi red mi xer output and the spurious response wi ll be the same as that of the actual local oscillator signal to the spurious signal in the local oscill ator. VI . EXPER IMENTAL RESULTS tion This sect ion di scusses the experimenta l resul ts for the entire sy stem. First , the results of the system are compared with the HP3577A alone. Second , accuracy , harmonic generat ion, lin eari ty, and cal ibra tion of the system are examined . Third , the ability of the sys tem to ma ke phase coherency mea surements is di sc ussed . Fourth, the sys tem dynamic range is discussed and limi tations expl ained. Fifth, crosstal k between channel s is pre sen ted . As a final exampl e, the frequency spectrum of a square wa ve is measured and the results compared to a spectrum anal yzer 's results ( Appendix B). Fi gures 3-6, pages 10-1 3, show the results of directly feeding a 4 MHz sine wa ve into the HP3577A . Figures 26-29 show the HP3577A display when the 4 MHz sine wave is fi rst proces sed by the modification. No tice the tremendou s reduction in the image response in Figure 27 as compared tJ Figu r e 4, Jage 11 . The level of the other spuri ous signal s fou nd in the low freq uency ra nge ha s al s o been greatl y lowered . The noise level s at the edges of the Fi gures 27-36 traces looks as if they rise. This is not the case. Th is phenomenon is due to cal ibration. When the system is cal i brated over the entire 450 kHz to 10 MHz range, the video trace is stored in the HP3577A 61 - 10 .--- 0 ;> - 10 -20 E CD '0 - 3 0 - ! 40 -i- _.J -50 /' 0... v .J.� :E , ..N":� � ... ·,/"' ,, -8 0 - j_ -90 3 .80 3.84 3.88 3.92 3.96 4.00 4.04 4.08 4.12 4.16 4.20 FREQUENCY MHz FisJure 26. d i lay a 4 �Hz sine wa ve �v ith system modification. 0') N 10 0 - - 10 - 20 -- E CD -o -30 -- � w 0 - 40 - :::> �-- _j - 50 :,� �Q.. <( dol.. __..... � -60 -- II"- .k � A¥ - 70 -80 ------·-·� - 90 - --'--- 3400 3463 3526 3589 3652 3716 3779 3842 3905 39 68 4032 FREQUENCY kHz Figure 27 . HP3577A display showing attenuation of image signal . Q) w 10 I f I . [ I l I I 0 L-- I - 10 -20 -1 E ro "0 - 30 � LLJ I 0 - 4 I- :::> 0 1- - _J - 50 .t' 0... � <( -60 � ;..,. ��� •� NITJ..NlV� - 70 11.1.. W.. --f------� 'Y t 11 --t----t--L---.J -so ·---T--r----f----f-.J-_j----t·· - 90 I I I --J I I j .50I .55 .60 .65 .7 0 .75 .80 .85 .90 .95 1.00 FREQUENCY MHz Figure 28 . HP3577A displ showi ng lowered spurious responses . 0"> .p.. 65 L(') I I Vl QJ ' Vl � ' c 0 0.. ' <;t Vl Cl.J I I I I S- \�"" I Vl I i ::I l 0 S- I ::I I 0.. I Vl I t<> N S- I I I :r: QJ 1 I ' I � 3 ·> 0 P- I II I i >- I 0 O"l l I z c I IJ.l ·..- l I 3 I :;:) 0 I N 0 £ I � I II IJ.l Vl i I 0:: i I u... ' I I I I I ' I I Vl I I ! l ...... I ! j I '"'0 I I I I I I I - � : I ;. I I I � 0'> ! N Cl.J I I 0 S- I � ::I 0 0 0 0 I O"l 0 0 0 0 0 0 C\.1 rn L(') "- I I "' \1) Q) Ll... wep '30n.L ildl/IJ\1 .------·-- --., HP3577A REF. IN SOURCE RA EXT. IOMHz OUT AH BH SYNTHESIZER -4d8m AL BL 45MHz MOD. +15d8m • -- SOURCE • _ 7d8m I 45MHz IN 1 I IN HP CONDITIONS -AMP. SOURCE = - 7 - A OB SWEEP TIME = 10 SEC. L_jTO A OR 8 -PRESS NORMALIZE 1 BUTTON TO CALIBRATE EITHER A OR B. Fi gure 30 . Cal i brati on system block di agram. 0'1 0'1 10 -�- -�------ 0 ;> -10 I E - 20 en "0 - 30 w 0 :::::> - 40 -1- _J Q_ - 50 ::;,! v/1 <( �� -60 .J.<."'J'" lj!�!.tt�.,.-.} �..,..,. \�� . ·, �PI � ���fll' - -70 � -- - 80 ---- -90 � 280 2.84 2.88 2.92 2.96 3.00 3.04 3.08 3.12 3.16 3.20 FREQUENCY MHz Fiqure 2.9 sine wa ve fundamental . 0) -...) --- 10 -- -- 0 [ -10 - E -20 CD "'0 -30 w 0 ------:::J - 4 0 ------r-- 1-- --r--- ...... J ------Q_ - 50 - - :i! 1' t--· vLJ ------70 -�-�-· ------r-- -- - -80 ----- 1- -90 '---··- ---- __.__ _ 5.70 5.74 5.78 5.82 586 5.90 594 5.98 6.02 6. 06 6.10 FREQUENCY MHz Figure 32 . Exampl e of second harmonic (5.89 MHz). 01 CXl - I 0 .-----.. .----.---.----,...- - 0 1------+----+---t- ---l -I 0 E - 20 (l) "0 . -30 1------f------1 -+-·· f------+---- llJ 0 :::> 0 -- - --·-·-- -- r- -4 1------t-- ---+--- -- +------+ - -+ 1 ---+----+ -- _J 0.. 50 � <1: -t -60 �ts:J:J--�- ,.,��1.."-��-'-'-',J.-'---'-���+ �--+---+---+----+---/-1 -70 - -so - 90 �----�--�------�- --�-----�----�--�---- �----�--� 8. 65 8.69 873 877 8.81 8 85 8.89 893 8.97 9.01 9.05 FREQUENCY MHz Figure Exampl e of third harm onic (8 .84 t� Hz ). OJ <.o 10 --- ·--r--r------r---r-�-- 1 ·- 0 1 T -I 0 ; - I __ r---+--- '�--J------�- --- t-·-- -T · - E -·- 1 - 20 --l...------f--· J CD 1- -T/-r---+------+-· -· "0 - 30 f-·------t---+-- w 0 --� I ::J - 4 0 _l -1- _J Q_ - 50 z: <{ '\?! - GO .�N� Jill#r�,jj ��� i{�J � �f I �� - 7 0 �li��-�_: I__,.---1------·· - - 80 �-- - L- �- -- --t---il---r---+-- 1---' I � I I 90 �._ __j_ _ I _J_ I 8 07 8.08 8.09 9.00 9.01 9.0 2 9.03 9.04 9.05 9.06 9.07 FREQUENCY MHz Figure 34 . 9.015 MHz sine wd ve wi th 0 dB attenuation. -....J 0 10 -- 0 --t----- 10 -- -- ,______(_ E ------CD -20 - -o - 30 w 0 ::> - 40 -!-- -t--- _j 0.. - 5 0 2. � /1 l'll ··-- -7 0 fj - 80 - 90 -· 8.07 8. 08 8.09 9 .00 9.01 9.02 9.03 9.04 9.05 9.06 9.07 FREQUENCY MHz Figure 35 . 9.015 sine wa ve wi th 10 dB attenu ation. ...... J 72 1'- 0 (]) '\ f..O t\ I 0 I O'l 1:: 0 •r- I I I' l() +> I � r wsp '3anJ.. IIdii'J'1 73 memory . When the frequency band is changed , the trace modification is ma inta i ned as before. , it is necessary to recalibrate the system over each bandwidth to el iminate this probl em. Several points on system performa nce can now be discussed : The reader might have wondered why graphs of different frequency regions are separa ted . The rea son for examining the spectrum in discrete bands is due to the limited number of data po ints the HP3577A ta over a frequency sweep and the limited resolution bandwidth of the HP3577A . The unit can take up to 401 data points across a frequency sweep, no matter the size of the frequency range bei ng swept. If the system user is not careful , spectral compo nents can fall into the region between data poi nts and mi ssed . For good accuracy , 500 KHz regions must be exami ned separately. Spurious freq uency components seen around the spectral compo nents may tern generated . e unit genera ted spurs arise from the that, since th� mov ing fi l ter ras some fi nite band�idth, the spectral component is sti ll introduced into the HP3577A during the times the receiver is not exactly tuned to that frequency . So \v h ile the signal is in the tern bandwidth, but not exactly at the source frequency, some spurious signal s may be generated . These spurious frequency components will have power level s at least 40 dB low the fundamental. 74 Accuracy , Linearity, and Cal i bration The modificati on ha s a system loss of 15 dB from input to output as wel l as high and low-frequency rol loff . These problems , which cause inaccuracy , can be cal i brated out of the system's performa nce . The ca l ibration proced ure is quite simpl e and is shown on Figure 30. The sourc e frequency from the HP3577A is spl it and one component is fed directly into the preampl ifier. Since the sourc e frequency wi ll always stay at the system 's tuned frequency , it can be used as a referenc e. Due to the way it wa s necessary to set the system up, the actual ampl i tude of any input signal wi ll be 7 dB bel ow the display value. Figures 31 -33 show a 3 MHz sine wave and the harmonics generated . The actual level of the input signal is -10 dBm. The system shows the correc t -3.8 dBm level ( allowi ng for the 7 dB offset and l dB accuracy) . The power level of the harmonics is at least 40 dB bel ow the fundamental . Figures 34-36 show a 9 MHz sine wave at different ampl itudes . Figure 34 is at a level of 0 dBm (-7 dBm ) . The signal is attenu ated 10 dB in Figure 35 and 40 dB in Figure 36 . Notice the linearity over the different ampl itudes . Pha se Coherency Mea surements A very important part of the system performa nce is pha se coherency mea surements. Figure 37 shows the spectra of a spl it 10 0 --1--·- r -1 0 - r E 2 0 en "0 -30 w 0 :J - 4 0 -1-- ...J \.11,.' ,. -� �oh 1�1Jri'lf. /y;lt - 9 0 ---- 8. 07 8.08 8.09 9. 00 9.01 9.02 9.03 9.04 9.05 9.06 9.07 FREQUENCY MHz Fig ure 37 . 9.013 MHz sine wa ves fed into ea ch cha of system . -.1 U'1 76 9 MHz sine wave fed into both channel s of the modification. Notice the ampl i tude of the signal is identical on both channel s, as it should be . Figure 38 is the ratio of the pha se of the two signal s. The phase ratio is constant, confirming the system's ability to ma ke phase mea surements . System Dynamic Range Figure 39 is the graph that outl i nes the dynamic range in ampl i tude and frequency for the system. The frequency rol l off on the low end is caused by the dec reasing system loca l oscilla tor level . The frequency rolloff on the hi gh end is caused by the band limiting 10 MHz filters . The maximum ampl i tude input is determined by the mixer 's ma ximum input level and the preampl ifier's slew rate . The no ise level of the system is due to the direct feed through of the synthesi zed 45 MHz signal . The 45 MHz feeds through the input mixer , pa sses the crystal filter , and is mixed to the current receiver frequency. Figure 40 shows the ampl itude and phase of the system across the 450 kHz to 10 MHz band before cal ibration. Fi gure 41 shows the cal i bra ted sys tem response. Figure 42 is the normal ized crosstalk between channel s. The crosstalk is less than -40 dB over the frequency change. A mea surement ta ken wi th the low frequency section of a channel is shown in Figure 43. Figure 43 is the HP3577A display with a -25 dBm sine wave at 300 kHz . A sine wave at 1 MHz wa s also 225 180 - 135 90 (f) r n,� w w �'t/� Ar"i 45 �1\M lA,�rM� �r/' �M.)I�J� F1 \J,�� I�,X Lv�,�.-vNir �. a: v I I � I� ' (9 11 r w 0 0 w (f) - '1 - 90 1----- -1 35 - I -180 lI - 225 8.07 8.08 8.09 9.00 9.01 9.02 9.03 9.04 9.05 9.06 9.07 FREQUENCY MHz Figure 38 . Pha se relatio n between two 9.013 MHz sine wa ves. '-I '-I 0 - r· I j I I I 10 - - -20 _,... f-. v - 30 ---�----- "' E """ ro "0 -40 - - ___, l.J.j 0 - 50 ------::::J ....., 1- I .....J - 60 ---- - I 0.. � 100 j I 0.40 I 3.32J 4. j78 6. J24 7.7 J 0 9.J16 10.62 12.j 08 131.54 1 5.00 FREQUENCY MHz Figure 39 . ic range high-frequency section. -..j co -- r------.------· 10 .-- ------0 ------>-- -- t--- r--- - 10 - -- � - ·- 20 +- v E ------� m - 3 0 --- c- -o -· 0 l.JJ +180 0 -40 :::> f- - ° ------90 � 0 -�--- � _j r--· � r- �------r-- + Q_ ::?. ------0 0 � - 180 - 80 -- ··- - � �- f-- -- � -----'------90 -- '--- 040 86 3.32 4.78 6.24 7.70 9. 16 I 0.62 12.08 13.54 15.00 I. FREQUENCY MHz Figure 40. Amf)l itude and f)hase response of high-frequency section. '--.1 \.0 I 0 .------.-- -.------�---� -�----r---·---r----···-r----- 0 �--�----�--��--+----+----�--�----�----�--� - ---·-+ --- - 10 ·------+ --- +---- 1� ------+-----l------1 -20 E m -o 30 - w (J :J - 4 0 �---r----r----+-----�---+-----�- +180 1-·· __j - 50 r----r----r-----�-----4----4---�----�-- --1----J----_j +SO Q_ :2: <( -60 r----r----r----+�--+----+----�--��--�----�--� 0 70 ----+- ---+-- --1- I -+ - 90 I I I I I I I _ _ 80 f--- 180 -90 0.40 1.86 3.32 4.78 6.24 7.70 9. 16 0.62 12.08 1 3 . 54 15.00 FREQUENCY MHz Figure 41 . Normal ized itude and phase response of high-frequency secti on. co 0 10 0 1------f-.------ 10 ' -·- -20 f------t----r----Jf---+---t---l--- E m "D - 3 � 0 w 0 :::J 40 1- _j - 50 / 1\ l----1- --r---- a.. -� ::2: � , �v � <( � -6 0 ��l/ 1 1 v -7 0 - 80 - 90 0 50 1.45 2.40 3.35 4.30 5.25 6. 20 7. 15 8. 10 9.05 10.00 FREQUENCY MHz Figure 42 . rmal ized crosstal k channel s. OJ 10 0 ·0 -20 E co -o - 30 w 0 -4 :::> 0 1- _J 50 Q_ � <( -60 - 7 0 -eo -90 0 50 100 150 200 250 300 350 400 450 500 FREQUENCY kHz Figure HP3577A di spl ay 300 kHz sine wave in low-frequency cha co "' 83 fed into the law frequency ; no spurious signal s were displayed in the 5 Hz to 450 kHz ra nge. lusians The system worked as expected . The unit specifi cations are shown again for clarity. Freq uenty Ra nge 5 Hz to 450 kHz and 450 kHz to 10 MHz Low Frequency Section Dy namic Ra nge 0 to -90 dBm Hi gh Frequency Section Dynami c Ra nge -6 to -65 dBm Low Frequency Ha rmonics -40 dBc Hi gh Frequency Ha rmonics -40 dBc Hi gh Frequency Section Spurious Signal s: 450 to 1500 kHz >-30 dBc 1 . 5 to 1 0 �1 Hz >-40 dBc Accuracy : Low Frequency Section 0.001 dB ma gnitude and 1 degree phase Hi gh Frequency Section 1 dB ma gni tude and 5 degrees phase Cl'o ssta 1 k: Low Frequency Section >-60 dB High Frequency Section >-40 dB REFERENC ES ERENCES 1. Microwave Sol i d-Sta te Devices, Samu el Y. Liao , Prentice-Hall. Engl ewood Cl iffs , New Jersey 07632 . pp. 306-307 , 1985. 2. Operating Manual Model 3577A Network Analyzer , Hewl ett-Packard Company , P. 0. Box 69, Marysvi lle, Wa shington, 98270 U. S. A. , Microfiche Part No . 03577-90050 , Section 5, p. 8, November 1983. 3. Single-Wafer Four-Pole Monolithic Crystal Filters ( Brochure ) , Piezo Technol ogy , Inc. , P. 0. Box 7859, Orlando , Florida 3 54-7859, pp. 1-4. 4. RF/ IF Signal Processing Handbook, Mini- Circ uits, Inc., P. 0. Box 166, Brooklyn , New York 11235, pp. 33, 17-29, and 164- 165, 1985-86. 5. 36 Frequently Asked Questions About ecting Mixers, Appl ica- tion Note 4, Mini ircuits Co., P. 0. Box 166, Brooklyn , New York 11235, pp. 1 5, 1985-86. 6. Practical Design and Evaluation of High Frequency Circ uits, Appl ications Note 7, Hewl ett-Packard Company , P. 0. Box 69, Marysville, Wa shington , 98270 U. S. A. , pp. 1-10. 7. Mixer Distortion Meas urements , Appl ication Bul letin 7, Hewlett-Packard Company , P. 0. Box 69, Marysvi lle, Washington, 98270 U. S. A., P. 0. Box 166, Brooklyn, New York 11235, pp. 1-6. 8. Introduction to Modern Network Synthesis, Van Val kenberg , John Wi ley and Sons , New York, Chapter 16, p. 447 , 1960. 9. Transmi ssion Networks and Wa ve Filters , T. E. Shea , M. S. , D. Van Nostrand Compa ny , Inc., New York, pp. 20-36, 1929. 10. Network Analysis and Synthesis, Lou is Wei nberg , McGraw-Hill Book Company, Inc., New York, Cha pters ll-12, 1962. 11. Microwave Sol id-State Devices , Samuel Y. Liao , Prentice-Hall, Engl ewood Cl iffs , New Jersey 07632 , pp. 306-307 , 1985. 12. Referenc e Data for Radio Engineers , Fourth Edition , H. P. Westman, ed itor, International Tel ephone and Tel egraph Corporation , pp. 166-169, 1956 . 85 APPENDIXES APPENDIX A HP3577A CONVERSION SYSTEM SCHEMATICS 450kHz v v v TO J O MHz 31 2 1 2 1 2 FROM PREAMP 2IMIXFR II I I 3 I 3 J 1 3 437 1 4371 4371 - 4 .. 1 1 �� c - 1- 1- .89uH � .89uH .89uH IOpF f 16pF 16pF 5pF 3 T ! I Hz ! 1�4 2 TO INPUT _ OU OM co co HP3577A � LPF U TA 108 4 + 18d8m : TO MIXER I .>CHANNEL 2 4 4545 TO NOTE: 55MHz FROM +26dBm CRYSTA L LOCAL 1 4371 - 45MHz OSCILLATOR 25d8 FILTER • GENERATION SYSTEM � Figure A- 1 . Fi l ter board schemati c (1 channel }. -12 .001 I � & I 1 2 5 1 0. 51"1 HA .. � . 00 f" 8 ' 4 :?6 ·-l , 1 s �- \tVv'v- o0 2 _;;;--'VvVv----j r----o 5 -�- �-� to z �A: / 50kHz 0 :-ra�-- 4 AT���-10 IOl IO��lMHz?[J: -- !A A 6 !OMHz LL LL q 51U NOTES: l.l 81 PA SS CAPAC ! TORS IN qF 2. l!IZVOLT SUPPL IES REGU ---·��------LA TED FFlOM :t 15 VOLTS 3 lEACH CHANNEL HAS SEPAR AT E REGULATORS 4) COMPOSITE FILTER COME'ON ENTS: • C pF H:qH TO INPUT 32 32 HP357lA '---,-----<) 450 kHz to IOMHz .001 J� 4480 r···o -12 -12 J 00 Fi A-2 . Preamplifier schemati c (l channel ). \..0 450KHz TO IOMHz HIGH- Pt�ss FILTER d8m fc =45 MHz _ 7 128 160 _ __ L=_ TFM }- 56 135 97 1· 4 I I I 1------.---l t-----.----; 1------.---i MIXER 30 3 - 160 88 300 l I 30 I -10 .l 30 J � � LOW - PASS FILTER fc" 55MHz 176 176 TO POWER I AM P LIFIER 0 UTA -108 88 �,r C pF 1 I = nH Fi gure A-3 . Local osci llator generation system schematic. 91 + . E � u 0 r- ..0 >, 0 l() � (\j > u � w (\j + (l) 1... :::::; l() o; .,... l() I '-'- - I 0 0 l() z > C) u I w l()- I + I APPENDIX 8 THEORETICAL AND ACTUAL MEASUREMENTS OF THE HARMONICS OF A SQUARE WAVE � -4 -4 - 1 2 �- 1 3.5 - 1 5 - 17. 9 A A - 1 9 I. dBm -20.9 dBm - 2 4 � -23 '-0 J w - -43 I 45 I -45 L .. 3 5 7 9 23 456 7 89 HARMON IC NUMBER HARMONIC NUMBER Theoretical Actual Figure B-1 . Theoreti cal vs . actual spectral components of square wave . 10 ··- - 0 f--· p 10 -· - E -20 c:D "0 - -3 0 ·-· w 0 :::> 40 - 1- - _j 0... - 50 . � � - go - r· 0.80 0.84 0.88 0.92 0.96 1.00 1.04 I .08 1.12 1.16 1.20 FREQUENCY MHz Fiaure B-2 . Fundamental of square wave meas ured by conversion system . \.0 � 0 0 ·0 -· E -20 ro -o - 3 0 --- w 0 :::> -40 - I--- _j (L - 5 0 :2: i - 70 I -· 80 90 1.80 1.84 1.88 1.9 2 1.96 2.00 2.04 2.08 2.1 2 2.1 6 2.20 FREQUENCY MHz Fi gure B-3. Second harmonic of square wa ve . \..0 U1 10 �·-- - r- 0 0 -· -- - r· ·-�- - t> E -20 ··- CD "0 --- - 30 l.il 0 ::::> - 40 I- - _J 0.. - 50 I � 70 - -80 -· - 9 0 - 2.80 2.84 2.88 2 .92 2 .96 3 .00 3.04 308 3. 12 3.16 3.20 FREQUENCY MHz figure B-4 . ird harmonic of square wave. \.0 O'l --· .,, . '' 10 I 0 -- --·- -·��- -10 �- ,,. ------E 20 £D "'0 -30 - w 0 ::::> - 4 0 ---· --· -I- ..J Q.. - 5 0 � <1 A. � ·" -60 � .J'J.u\...... � - 70 - -80 ·-· - 90 3.80 3.84 3,88 3.92 3.96 4.00 4.04 4.08 4.12 4.16 4.20 FREQUENCY MHz Figure B-5. Fourth c of square wa ve. 1.0 "'-J 0 ------.-- 0 -10 -- ;:> E -20 1- -- al "0 - 3 0 -!------1-- w 0 � 40 ·-- ....J - CL 50 - ::E -80 1- ---. - - 9 0 - - 4.80 4.84 4.88 4.92 4.96 5.00 5.04 5.08 5 12 5. 16 5.20 FREQUENCY MHz Fi B-6 . Fi harmonic of square wave. <..o 00 99 0 ""! t! \0 I I \0 l \0 I I �Irl . I C\J (LJ � > I r1j I I I \0 3: I j I � (JJ 00 .._ Q r1j I :::::l I ( \0 c:r I I I ! til <;j :.;... l i l Q N 0 .. ' I I \0 I :;! u < < ...... I' ! >- ' I 0 l I I _1 0 u I : z I \0 w -, I I �� I \0 a-' l ' I � I 01 w r I l() 0:: X \ lL ...... (/") I 41 i I C\J l I i ! ' ()) i i I . Ii I L() ;--... I I I I I I cc I I H--f-I 1 l ! 00 I l l I ! aJ I i 1 i I I i.(l I I I I ...... I I ' <;j00 l ..t... I � u) I 0 I I I 00 0 0 0 0 0 0 0 J 0 0 0 l(j N 'T l() I r<> \.0 ClJ G) ' I I wgp '30nll ldv.JV' 10 r-----, ------r---- r--- --· ·,-· ---- 0 +---··-+-----+ +---+----+-----! -10 -+-----+----+------4 -- ---+---+- E -20 1---+-- m "0 +--t ·�- -+---··1 ------t+---+-----+-----+ w - 30 0 � -40 I I--+ -+ --ttr' --+--- +---+--+-----+ _J a.. � --:70: 1-bt:liM�>'J::];��dd=��--+----+-----1 ------. -+ -----t-----+- - 8 0 1------+ +- -- ·+-----+----+---+-----i----l ___ -90 �- �.____._ �---'--- 6. 80 6.84 6.88 6.92 6.96 7.00 7. 04 7. 08 7. 12 7.16 7.20 FREQUENCY MHz Fi gure B • Seven th harmon ic of square wave . 0 0 10 r----.-----�----r---- ·----�----;r-----·----�----�--� 0 0 -I --+------+----+·------+-----1----+---·+----+---+ E -20 Q) "'CJ - 3 0 -- �--+---·-- -�- - w 0 :::> - 4 0 f- - _J - 5 0... 0 ------� �� -+ --4- -� �----if-----r--+-----i - 70 - ·- - - ' -80 - 90 --- ��------� -- 8.80 8.84 888 8.92 8.96 9.00 9.04 9.08 9.12 9.16 9.20 FREQUENCY MHz �igure B-9. Ninth harmoni c of square wave. 0 AP PENDIX C DESIGN DATA SHEETS FO R COMPOS! FILTERS 166 CHAPTER 6 Low-pass filter design type and hall section impedance characlerisfics Constant-k Series m-derived :t Zn = Zn ;Z R .. R [ 1 - �;!1 - m2) J vi - W:./w.2 _.., Shunt m•derived t ... ' z R R'/Zrl z .., Notation>: Z in ohms, a in neoers, and ;1 in rodions m = w, = 2-.:f. angular cutoff frequenc:y R = nominal terminating resistance = vt,fc1 w., = 2Jrf., = angular frequenc:y of pea it VZn Z .. " attenuation Figure C-1 . Composite fi l ter low-pass design equations. SOURCE : Reference Data for Radi o Engi neers, fourth edi ti on, H. P. Westman, editor, International Te lephone and Te l egraph Corporation, pp . 166-1 67. 103 104 FllTERS !MAGE-PA RAMETER DESIGN 167 de•ign formulas full-section half-section hall-section allenuo!ion a ond phase d characteristics series arm •hunt arm • "'J ! a . I When 0 � w � w< a=O 0 : w- ��... Vvhen w< < w < ro �i = 1T t' .. Q m ... c, = me. me. c, For comlant-k type .�' Zu Za = P When w,., < w < co. For m•derived type Curves drown for m = 0.6 o: = cosh-1 1 [ R' = Z·n z�, = Zti .....ri .... w.,,:, Zz(.. hunt�"') = cosh-• l - 2 ---:�-::-----""""7" [ = Z t (•h•lnt�m) Zz'-r•r�·'") Fi gure C-1 (continued). 105 lbti CHAPTER 6 High-pass filter design type and half section impedance characteristics Conslanl-k �---- . � 'J- 1 w} II I = Zn R or R --+ Z rl = ---c=== \):------U.'e2 Zrk z1Tk 1-- w' Series m-derived - R 1 - =-��11 m21 [ v: J --,_.-= V 1 - :.;.;c� 'vf Shunt m-derived R "' z _ �::�' ; t T'! - 1 - Wn2i�-:_ , z R = fi"/Zrt .J Z-n = Zrt Notations: Z in ohms, a m nepers, and B In radians We = 27rfr = anqular cutoA' frequency R = nominal terminating resistance = v'L,jC, Wro = 2rrf00 = angular frequency of peak ottenuaf10n figure C-2. High-pass design equations for composite filters. SOURCE : Reference Data for Radio Engi neers, fourth edition, H. P. Westman , editor , Internati onal Telephone and Telegraph Corporati on, pp . 168-1 69. 106 f!LTERS IMAGE-PARAMETER DESIGU Jo9 design formulas -- full-section _h_a_u=-•• e ciion hall-section attenuation a and phase iJ charflcteristict series orm shunt arm t When 0 < w w 0 '---w�c ------=., B = _,. '..rJc R I,V hen w, < w < a=O .. ;= w-: t:y! fJ ·11' . c. m � - When W� < W = cosh-1 (J,t" [" ,. W;:.. When 0 < w < w., 5 0 and Fo, constan!-k type R1 ZaZu = P w� < w < Fo, m-d erived type n: = 0 onj Curve-s drawn for ,...., ;:::::; C.6 R' = lnz�, Figure C-2 (conti nued ), VITA The author- was born in Dickson, Tennessee� on May 2, 1960. He attended high school at Father Ryan in Nashville, Tennessee. He obtained a Bachelor 's degree in Electrical Engineering from The University of Tennessee, Knoxville in December of lS82. The author was employed with Boeing Aerospace Company from January 1983 to January 1985. The position 's responsibilities in cluded design and development of electronic and microwave sy stems and components. In January 1985, the author returned to The University of Tennessee , Knoxville to work on his Master 's degree in Elec trical Engineeri ng. The major courses of study for the Master 's degree were electronics and networks. The author received his M. S. in Electrical Engineering in March 1987 . He then went to wor k for a company called Seal Tech in Birmingham, Alabama. 107