How it Works
HP
Loading...
A
Loading...
Loading...
an283
schematic
51 pgs11.4 Mb0

HP an283 schematic

f)
Application Note 283-3 Low Phase Noise Applications
of the HP 8662A and 8663A
Synthesized Signal Generators
Whp*
HEWLETT
PACKARD
Chapter 7 Using the HP 8662A/8663A at Microwave Frequencies With the HP 3048A Phase Noise Measurement System 26
Why Use the HP 8662A/8663A at Microwave Frequencies Effect of Multiplication on Signal Noise HP 8662A/8663A Phase Noise Performance at Microwave Frequencies Using the HP 8662A/8663A and the HP 11740A for Low-Noise Microwave Signal Generation Using the HP 3047A/11740A to Make Phase-Noise Measurements on Microwave Sources Measurement Techniques Data Interpretation HP 304 8A/11740A System Performance Measurements on Pulsed Sources Summary
Chapter 8 Voice Grade Receiver Testing With the HP 8662A/8663A 33
Receiver Test Basics: In-Channel and Out-of-Channel Testing Using the HP 8662A/8663A for Adjustment Channel Receiver Tests Using the HP 8662A/8663A for Spurious Attenuation Testing
Chapter 9 HP 8662A/8663A as an External LO with the HP 8901A/B Modulation Analyzer and HP 8902 Measuring Receiver 37
Measured Performance ■ Measurement Considerations and Procedure
Chapter 10 Using an HP 8662A/8663A With the HP 8505A RF Network Analyzer 39
Measurement Setup
Typical Operating Characteristics
Frequency Characteristics Range and Resolution Typical System Residual FM Output Characteristics Delay and Electrical Length Characteristics
Chapter 11 Using the HP 8662A/8663A as a Substitute LO With the HP 8672A
Microwave Synthesized Signal Generator 41
System Operation Hardware Modifications
System Performance Resolution Frequency Algorithm
Modulation
Chapter 12 Fast Frequency Switching With the HP 8662A/8663A 44
Standard HP-IB Frequency Control Fast Learn Frequency Switching Fast Frequency Switch Option H-50 Summary
Appendix A Calibration of Phase Noise of Three Unknown Sources .50
Appendix B 10 MHz Low-Noise Bandpass Amplifier 50
Appendix C
Low-Noise Amplifier 50 Appendix D
References 51
Hewlett-Packard Applications Notes Other References
Chapter 1
Introduction
^^M Range (MHz) ^^m Resolution (Hz)
^H Stability
^H OUTPUT
^^M Range (dBm) ^^H Resolution (dB) ^^H Accuracy (dB)
^H MODULATION
^^H External ^H SPECTRAL PURITY
TABLE 1.1.
HP 8662A/8663A Performance Comparison
FREQUENCY
Harmonics Spurious
(-dBc)
320
to 640 MHz band
(-dBc)
in
The stringent performance requirements of modern radar and communications systems call for high frequency signals with extremely good spectral purity. The Hewlett-Packard 8662A and 8663A Synthesized Signal Generators provide extremely good overall spectral purity by combining the low close in phase noise of a frequency synthesizer with the low spurious and noise floor typically found only in cavity-tuned generators. These
characteristics make the HP 8662A and HP 8663A excellent choices for many low noise applications, particularly as local oscillators in low noise systems, low noise RF signals when multiplied up in frequency, or as versatile signal simu­lators through their flexible modulation formats. The HP 8662A operates up to 1280 MHz and is well suited to radio receiver testing by providing simulta­neous AM and FM modulation. The HP 8663A covers another frequency
octave, up to 2560 MHz, for applications in the low
S-band
range and provides simultaneous AM, FM, phase and pulse modulation. This allows simulation of radar returns and transmitted communi­cations signals.
The HP 8662A and the HP 8663A share the same frequency synthesis circuitry and therefore yield the same spectral purity. Their performance differs primar-
ily in frequency range, output power level, and modulation format. Table 1.1 and Figure 1.1 illustrate the HP 8662A and HP 8663A performance similarities and differences.
HP 8662A
0.01 to 1280
0.1 to 0.2 5xi0-10/day
+13 to -14D
0.1
±1
AM,
FM
AM,
FM
<30 <90
HP 8663A
0.1 to 2560
0.1 to 0.4 5xl0-10/day
+16 to-130
0.1
±1
AM,
FM,
Phase, Phase,
Pulse Pulse
AM,
FM,
<30 <90
Figure 1.1.
Measured Residual SSB phase noise versus offset from carrier. Carrier frequency 159 MHz, 639 MHz and 2.56 GHz
This application note discusses phase noise in detail (Chapter 2) to provide an understanding of its implications for cer­tain critical applications such as
out-of­channel receiver testing, doppler radar, and local oscillator substitution.
In Chapter 3, key design aspects of the
HP 8662A and HP 8663A, and the resulting phase noise performance, are presented, followed in Chapter 4 by a discussion of the effects of external refer­ences on their performance. Chapters 5 and 6 present techniques of applying the excellent phase-noise performance of the HP 8662A/8663A to solve problems that commonly arise in the measurement of low phase noise. Chapter 7 extends these techniques to the microwave frequency range via HP 8662A/8663A-based sys­tems specifically intended to measure
low phase noise microwave signals.
HP 8662A and at
159
MHz
The effects of signal generator phase noise on receiver testing are discussed in Chap­ter 8. The next three chapters present meth­ods of applying the HP 8662A/8663A to enhance the performance of several other Hewlett-Packard instruments. Finally, Chap­ter 12 discusses the fast frequency switching capability of the HP 8662A/8663 A.
HP
8663A "
HP 8662A and HP 8663A at 639 MHz
v-
"HP 8663A at—T
2.56 GHz «■• "
vv-j?,
o
^tes XW*i
-,?".„,
*.
»\1i"
Chapter 2
The Phase Noise Density Spectrum and Its Implications
What is Phase Noise?
Every RF or microwave signal displays some frequency instability. A complete description of such instability is gener­ally broken into two components, long-
term and short-term. Long-term fre­quency stability, commonly known as frequency drift, describes the amount of variation in signal frequency that occurs over long time periods - hours, days, or even months. Short-term frequency sta­bility refers to the variations that occur over time periods of a few seconds or less.
This application note deals primar-
ily with short-term frequency stability.
DIFFERENT MEASURES OF SHORT­TERM FREQUENCY STABILITY
There are a number of methods for specifying short-term frequency stability. Three of these methods, fractional frequency deviation, residual FM, and single sideband (SSB) phase noise are discussed in this chapter.
Fractional frequency deviation uses a time domain measurement in which the frequency of the signal is repeatedly measured with a frequency counter, with the time period of each measurement held constant. This allows several calculations of the fractional frequency difference, y, over a time period, T. A special variance of these differences, called the Allan variance, can then be calculated. The square root of this variance is generally repeated for several different
time periods, or T, and versus T as an indication of the signal's short-term frequency stability. (See also NBS Technical Note 394, "Characterization of Frequency Stability", reference 9 in Appendix D.)
<T(T).
The whole process is
O(T)
is plotted
For this reason, the use of residual FM to specify the short-term stability of a signal generally provides the least amount of information of the methods listed. An additional disadvantage is that different post-detection bandwidths are specified in different measurement standards. For example, another common choice is 20 Hz to 15 kHz. As
a result, quite often comparisons of oscillator performance based on residual FM specifications cannot be made directly. However, for many communi­cations systems, residual FM is used because it matches the terms and condi­tions of the application.
Single sideband (SSB) phase noise mea-
sures short-term instabilities as low-level phase modulation of the signal carrier. Due to the random nature of the insta­bilities, the phase deviation must be rep­resented by a spectral density distribu­tion plot known as an SSB phase noise plot, see Figure 2.2.
Of all the methods commonly in use, SSB phase noise has the advantage of providing the most information about the short-term frequency stability of a signal. In addition, both fractional fre-
quency deviation and residual FM may be derived if the phase noise distribution of a signal is known. As a result, SSB phase noise has become the most widely used method of specifying short-term stability. For this reason, the majority of this application note is devoted to SSB phase noise to specify short-term fre­quency stability.
SSB PHASE NOISE DEFINITIONS
Due to phase noise, in the frequency domain a signal is not a discrete spectral line,
but "spreads out" over frequencies both above and below the nominal signal frequency in the form of modula­tion sidebands. Figure 2.1 illustrates the difference between ideal and real signals in the frequency domain. In some cases, phase-noise sidebands can actually be viewed and measured directly on a spec­trum analyzer. This has led to the common definition of phase noise in which the phase-noise level is repre­sented by a function <<f(f) called "script L". The U.S. National Bureau of Standards defines Jf(f) as the ratio of the power in one sideband, on a per-Hertz-
of-bandwidth spectral-density basis, to the total signal power, at an offset (mod­ulation) frequency f from the carrier. Jt({) is a normalized frequency-domain measure of phase-fluctuation sidebands expressed as dB relative to the carrier per Hz (dBc/Hz).
Power Density (One Phase Modu-
Jf(f) =
lation Sideband) dBc
Total Signal Power Hz
The second method of specifying short­term frequency stability is residual FM. This is a frequency-domain technique in which the signal of interest is examined using an FM discriminator followed by a filter. The bandwidth of the filter is set at some specified value, usually 300 Hz to 3 kHz, and the rms noise voltage at the filter output is proportional to the frequency deviation in Hz. In this
method, only the total short-term frequency instability occurring at rates that fall within the filter bandwidth is indicated. No information regarding the relative weighting or distribution of instability rates is conveyed.
Figure
2.1.
CW signal viewed in the frequency domain.
5
As mentioned, <Jf(f) can be measured directly on a spectrum analyzer if the following conditions are met:
1.
The spectrum analyzer noise floor is lower than the level of phase noise being measured. This means that the
phase noise of the spectrum analyzer's local oscillator must be lower than the level of the noise being measured. In addition, the dynamic range and selec­tivity of the analyzer must be sufficient to discern the measured phase noise.
2.
The signal's AM noise does not make a significant contribution to the noise measured. This can be determined by measuring the AM noise of the signal, or it can be deduced by understanding the nature of the source under test.
For more information on how to mea­sure phase noise directly on spectrum analyzers, refer to Hewlett-Packard Application Note 270-2, "Automated Noise Sideband Measurements Using the HP 8568A Spectrum Analyzer".
Another function frequently encountered in phase noise work is S,(f). S.(f) is the spectral density of the phase fluctuations in radians squared per Hz. The relation­ship between S^(f) and <Jf(f) is simply:
small relative to one radian. Close-in to the carrier this criterion may be violated. The plot of Jt({) resulting from the phase noise of a free running VCO (Figure 2.2) illustrates the erroneous results that can
occur if the rms phase deviation in a particular measurement exceeds a small angle. Approaching the carrier, «Jf(f) is increasingly in error, eventually exceed­ing the carrier amplitude and reaching a level of +45 dBc/Hz at a 1 Hz offset (45 dB more noise power at a 1 Hz offset in a 1 Hz bandwidth than the total power in the signal).
The —10 dB/decade line drawn on Figure 2.2 represents an rms phase deviation of approximately 0.2 radians integrated over any one decade of offset frequency. At approximately 0.2 radians, the power in the higher order sidebands of the phase modulation is still insignifi-
cant compared to the power in the first order sideband. This ensures that the simple calculation of Jf(f) from S^(f) is valid (the mean square phase fluctua­tions are small relative to one radian squared). Below this line the plot of Jf(f) is correct; above the line Jf(f) is invalid and S ,(f) is used to represents the noise of the signal. The data above the line
must be interpreted in radians squared per Hertz, not in dBc/Hz as of(f) is defined. In addition, the vertical scale must be adjusted by 3 dB since
S^(f)/2
is
actually graphed.
RESIDUAL AND ABSOLUTE PHASE NOISE
There are two measures of phase noise commonly used in specifying the short­term stability of signals - residual phase noise and absolute phase noise. Residual phase noise refers to that noise inherent in (added by) a signal processing device, independent of the noise of the reference oscillator driving it. Absolute phase noise is the total phase noise present at
the device output and is a function of both the reference-oscillator noise and the residual phase noise of the device. Absolute phase noise is the parameter generally considered.
Residual phase noise is used to help understand the additive noise generated
in frequency synthesizers. Although most synthesizers have internal reference oscil­lators,
many synthesizer users prefer to use external references of higher stability to improve the synthesizer performance or to synchronize a system of many instruments. In these cases, the residual noise specification conveys more informa­tion than the absolute noise specification, since it allows the user to calculate absolute noise performance from the
characteristics of his own reference oscil­lator. Chapter 4 discusses the effects of external references on the absolute noise of the HP 8662A and 8663A.
«*(f) = -y-
This relationship, however, only applies
if the mean-square phase deviations are
Figure 2.2.
Region of Validity of Jt(() = -|—
s (f
j
S ,(f) and «f(f) are discussed further in Chapter 5, where the two-source method of measuring phase noise is described.
Why is Phase Noise Important?
In recent years, advances in radar and communications technology have pushed system performance to levels previously unattainable. Design empha­sis on system sensitivity and selectivity
has resulted in dramatic improvements in those areas. However, as factors pre­viously limiting system performance have been dealt with, new limitations have emerged upon which attention is being focused. One of these limitations is phase noise. The ability to generate and measure low-phase-noise RF and microwave signals has become more important than ever before.
Because of extremely low SSB phase noise, the HP 8662A/8663A allow
users to meet these critical phase noise requirements with off the shelf equip­ment. To illustrate how low phase noise sources such as the HP 8662A/ 8663A can help achieve better system performance, three specific applications are presented.
*.;.-■*:■«
.,1s *» <•-.*/.
-W!?>;
wspe.
LOCAL-OSCILLATOR APPLICATIONS
Phase noise can be a major limiting factor in high performance frequency­conversion applications dealing with
signals that span a wide dynamic range. The first down conversion in a high-performance superheterodyne receiver serves as a good example for illustration. Suppose that two signals (Figure 2.3a) are present at the input of such a receiver. These signals are to be mixed with a local oscillator signal down to an intermediate frequency (IF) where highly selective IF filters can
separate one of the signals for amplifi-
cation, detection, and baseband pro­cessing. If the desired signal is the
larger signal, there should be no diffi­culty in recovering it, if the receiver is> correctly designed.
signal can degrade a receiver's useful dynamic range as well as its selectivity. To achieve the best performance from a given receiver design, its local-oscillator phase noise must be minimized. This is where the HP 8662A/8663A can help. First the HP 8662A/8663A can provide a low-phase-noise signal to serve as the reference when measuring the phase noise of the local-oscillator signal under test. This measurement is described in detail in Chapters 5 and 6. Second, the HP 8662A/8663A can provide the local-
oscillator signal typical output power, 0.1 Hz frequency resolution, 420/510 microsecond frequency switching speed, and full HP-IB programmability, the HP 8662A/ 8663A can serve in almost any demand­ing local-oscillator application.
itself.
With +16 dBm
DOPPLER RADAR APPLICATIONS
Doppler radars determine the velocity of a target by measuring the small doppler shifts in frequency that the return echoes have undergone. Return echoes of tar-
gets approaching the radar (closing tar­gets) are shifted higher in frequency than the transmitted carrier, while return echoes of targets moving away from the radar (opening targets) are shifted lower in frequency. Unfortunately, the return signal includes much more than just the
target echo. In the case of an airborne radar, the return echo also includes a large "clutter" signal which is basically the unavoidable frequency-shifted echo from the ground. Figure 2.4 shows the typical return frequency spectrum of an
airborne pulsed-doppler radar. In some situations, the ratio of main-beam clutter to target signal may be as high as 80 dB. This makes it difficult to separate the target signal from the main-beam clutter. The problem is greatly aggravated when the received spectrum has frequency
instabilities—high phase noise—caused by either the transmitter oscillator or the receiver LO. Such phase noise on the clutter signal can partially or totally mask the target signal, depending on the relative level of the target signal and its frequency separation from the clutter sig­nal.
Recovering the target signal is most
difficult when the target is moving slowly and is close to the ground because then the ratio of clutter level to target level is high and the frequency separation between the two is low.
Figure 2.3.
Effect of L.O. phase noise in mixer application.
A problem may arise, however, if the desired signal is the smaller of the two, because any phase noise on the local­oscillator signal is translated directly to the mixer products. Figures 2.3b and c show this effect. Notice that the trans­lated noise in the mixer output com­pletely masks the smaller signal. Even though the receiver's IF filtering may be sufficient to remove the larger signal's mixing product, the smaller signal's mix­ing product is no longer recoverable due to the translated local-oscillator noise.
This effect is particularly noticeable in receivers of high selectivity and wide dynamic range.
The key point here is that the phase­noise level of the local-oscillator signal often determines the receiver's performance. A noisy local-oscillator
This effect is similar to that in the local-
oscillator application described in the preceding section. A small signal, the target echo, must be discerned in the
Figure 2.4.
Typical return spectrum for airborne doppler radar.
7
• MM
■.-'
"iv***""!
presence of the much larger clutter
signal that is very close in frequency. Again, the system performance is limited by phase noise. In this case, it is the phase-noise level of either the transmit­ter oscillator or the receiver local oscilla­tor that is limiting.
The HP 8662A/8663A can improve the radar's performance by serving as a low­phase-noise source for phase-noise mea-
surement or signal substitution. Since most radars operate at microwave fre­quencies, it is usually necessary to multi­ply the generator's outputs to the micro­wave frequency range. This multiplication is discussed in Chapter 7.
OUT-OF-CHANNEL RECEIVER TESTING
Modern communications receivers have excellent selectivity and spurious rejec-
tion characteristics. These are called the out-of-channel characteristics and require very high quality test signals for verifica­tion. Typically, two signal generators are used for testing the out-of-channel char­acteristics of a receiver. One generator is tuned in channel, the other is tuned out of channel, typically one channel spacing away.
Due to the masking effect described for
\„^
local oscillator applications, the phase noise and AM noise of the out-of-channel generator may limit the selectivity that can be measured. As a result, the measured selectivity may be much worse than the actual receiver selectivity. The limiting
level of phase noise on the out-of-channel generator is determined by the level of performance of the receiver that is being measured. More selective receivers require lower phase noise on the out-of-channel generator. Out-of-channel receiver testing and the phase noise requirements of the out-of-channel generator are described in more detail in Chapter 8.
Chapter 3
The HP 8662A/8663A: Designed for Low Phase Noise
The HP 8662A and HP 8663A Synthe­sized Signal Generators offer a superior combination of spectral purity, frequency resolution, and frequency switching speed in programmable RF signal gene-
rators.
To understand how these prod­ucts achieve such performance, it is nec­essary to examine their basic operation.
Theory of Operation
Figures 3.1 and 3.2 show the basic block
Figure 3.1.
HP 8662A block diagram.
Phaee-Locked Loop Section
I ' High Frequency Loop*
erence section synthesizes many differ­ent frequencies from a high stability 10 MHz quartz oscillator. The phase­locked loop section uses these reference-
section signals to synthesize output fre­quencies of 320 to 640 MHz in 0.1 Hz steps.
The output section modulates and amplifies the output signal from the phase-locked loop section and translates its frequency to the desired output fre­quency. This frequency translation is done by doubling, dividing, or mixing.
Reference
Sum
Loop
310 to 620 MHz
* 320 to 640 MHz
signals are used as a basis for synthesiz­ing the final output signal.
All of the reference section signals are directly synthesized; i.e., they are derived by multiplying, mixing, and
dividing from an internal high stability
10 MHz reference oscillator. As a result, the long-term frequency stability of the HP 8662A/8663A is derived directly from the internal reference and is speci­fied to be less than 5X10~10 per day after a 10-day warmup. As an example of how stable this is, when the HP 8662A/ 8663A is set for an output frequency of 500 MHz, the frequency will drift no
more than a quarter of a hertz per day after the specified warmup!
The frequency accuracy of the HP 8662A/8663A is directly related to the frequency accuracy of the internal reference oscillator. The reference fre­quency can be mechanically adjusted over a range of about 20 Hz to allow close calibration against a standard. The frequency accuracy of the output is dependent on: 1) how closely the inter­nal reference oscillator is adjusted to match an accepted standard and 2) how far the reference oscillator drifts over time (the primary drift component is crystal aging, specified to be less than 5X10-10/day). For most applica­tions,
the stability of the internal refer-
ence is adequate.
Figure 3.2.
HP 8663A block diagram.
diagrams for the HP 8662A and HP 8663A, respectively. The HP 8662A and HP 8663A block diagrams are fun­damentally the same. The major differ­ences are attributable to an extended fre­quency range and the addition of pulse
o
and phase modulation in the HP 8663A. In general, the block diagram can be divided into three main sections: the erence section, the phase-locked loop section, and the output section. The ref-
ref-
THE REFERENCE SECTION
The main function of the reference sec-
tion is to provide a synthesized octave band of frequencies from 320 to 640 MHz in 20 MHz steps. The refer­ence section also generates frequencies of 10-, 20, 120, and 520 MHz for use as local-oscillator signals in the phase-
locked loop and output sections. Both the short-term and long-term frequency stability of the signals from the refer­ence section are critical, since these
If greater stability is required, provision has been made in the HP 8662A/8663A to substitute an external 5 or 10 MHz reference for the internal reference. A cesium or rubidium standard used as an external reference can provide frequency accuracies on the order of one part in
1X1011.
also provide improved phase noise at
some offsets compared to the internal
reference. The use of external references with the HP 8662A/8663A is discussed in Chapter 4.
The short-term frequency stability or phase noise of the reference oscillator affects the phase noise on the HP 8662A/8663A output signal. Although the internal reference has very low inherent phase noise, as its frequency is multiplied up to produce the higher frequency reference section signals, the phase noise also increases at
a rate of 6 dB/octave. To reduce this
effect, monolithic crystal filters in the
reference multiplier chain at 40 and
Such an atomic standard may
9
160 MHz filter the noise sidebands at offsets greater than about 4 kHz. The resulting phase noise of the reference section output at 500 MHz is typically
-110 dBc (dB relative to the carrier) at a
10 Hz offset decreasing to a noise floor of about -148 dBc at offsets greater than 10 kHz.
The mechanical configuration of the crystal filters is critical, since any small
mechanical vibrations in the filter trans­late directly into microphonic spurious sidebands on the signal. The most common source of instrument vibration is the cooling fan which causes spurious signals at about 53 Hz offsets with 60 Hz power lines. This spurious mechanism is minimized in the HP 8662A/8663A by a special shock mounting arrangement which mechanically isolates the crystal filters from instrument vibration and by dynamically balancing each fan before
installation in the instrument.
THE PHASE-LOCKED LOOP SECTION
The phase-locked loop section consists of seven phase-locked loops that provide the frequency programmability, frequency modulation, and fine frequency resolution of the HP 8662A/
8663A without compromising the excel-
lent frequency stability and spectral pur­ity of the reference section. Using an indirect-synthesis technique (i.e., synthe­sis using phase-locked loops as con­trasted with direct synthesis by mixing, multiplying, or dividing as is done in the reference section), the phase-locked loop section takes the 320 to 640 MHz in 20 MHz steps from the reference section and synthesizes an output of 320 to 640 MHz in 0.1 Hz steps.
The phase-locked loop section is divided into two areas, the high-frequency loops and the low-frequency loops. The two high-frequency loops are nearly identical with specially designed, low-noise voltage-controlled oscillators (VCOs).
The low-frequency loops consist of five phase-locked loops; three that provide the HP 8662A/8663A's 0.1 Hz frequency resolution and two which generate frequency modulation and sum the resulting FM signal with the final output signal.
High-Frequency Loops
The first of the two high-frequency loops,
the reference sum loop, tunes
over a 310 to 620 MHz frequency range.
This loop sums the reference section's main output of 320 to 640 MHz with 10 or 20 MHz also from the reference sec­tion. The reference sum loop's primary function is to filter out spurious signals on the reference section output beyond
the loop bandwidth and to improve the resolution from 20 MHz steps to 10 MHz steps.
The loop provides 60 dB of spectral filtering, thereby reducing the spurious level from —40 dBc to
-100 dBc. Such filtering is an advantage of indirect synthesis, since the bandwidth of the phase-locked loop can be set so that the loop VCO will only track the loop reference signal within the bandwidth of the loop. Reference signal sidebands falling outside the loop band
width are therefore rejected by the loop.
Figure 3.3.
320 to 640 MHz switched reactance oscillator.
The second high-frequency loop is the output sum loop. This loop sums the 310 to 620 MHz output of the reference sum loop with a 10 to 20 MHz signal from the low-frequency loops. This 10
to 20 MHz signal has a resolution of
0.1 Hz and is frequency modulated when FM is enabled. The resulting output from the output sum loop is 320 to 640 MHz in 0.1 Hz steps. In the HP 8662A, this signal is sent to the output section for translation to the final output frequency and amplitude modulation. In the HP 8663A, this signal is sent to the phase modulator (if phase modulation option 002 is included) and then to the output section for translation to the final HP 8663A frequency, amplitude, pulse, and BPSK modulation.
The reference sum loop and the output sum loop are nearly identical, since they both contain identical, specially designed low-noise VCO's. These VCOs employ a
switched-reactance resonator of novel design (Figure 3.3). The resonator con­sists of an array of five inductors switched in a binary sequence to provide 32 frequency steps. Thus, for continuous frequency coverage of 320 to 640 MHz, the varactor has to tune over only
10 MHz spans. Compared to a conven­tional VCO with a varactor covering the entire 320 to 640 MHz frequency range, this switched scheme results in greatly reduced oscillator tuning sensitivity. Therefore, any noise on the VCO tuning line causes very little phase noise as compared with a conventional VCO. In addition, the design of the oscillator
yields very high signal levels (±10 volts peak),
high Q (150 to 250), fast switch-
ing, and precise pretuning.
These properties of the VCOs result in
excellent phase noise performance com­bined with fast frequency switching. The actual phase noise of the VCO is shown in Figure 3.4. The noise at offsets beyond about 100 kHz is particularly important since this noise will not be reduced by the action of the phase-locked loop as will the noise closer in.
Several important considerations were taken into account in the design of the loops that phase-lock these VCOs. Using the reference sum loop as an example, to get the lowest possible overall phase noise, the loop bandwidth was selected to minimize the noise con-
tributions of both the VCO and the
ref­erence section. The special efforts made to lower the noise in the reference sec­tion allow a relatively wide loop bandwidth (250 to 450 kHz).
10
•y;' v ■
, -.if. n
if
^
o
■■
\i'^-?'^r;W
«t90
^>
:
"V:
iff
4
■'
m
-1201
Figure 3.4.
Typical phase noise reactance oscillator.
A direct consequence
of
HP 8662A/8663A switched
of
wide bandwidth
is faster frequency switching. As a result,
the reference sum loop can switch
about 50 microseconds. This larly significant considering the overall phase noise also shown
of
the reference sum loop,
in
Figure 3.4. The reference phase-locked loop filters the close-in noise
of
ing absolute phase noise
the VCO,
HP 8662A/8663A
This combination fast frequency switching achieve
in
poration
synthesizer design. The incor-
of
to
provide the result-
of
as
shown (Figure 3.4).
of
both low noise and
is
these characteristics distin­guish the HP 8662A/8663A from other signal generators, noise applications
for
example,
for
doppler radar, and in fast switching applications jam communications systems. The fast switching capability 8663A
is
discussed
Low-Frequency Loops
Careful design
of
the HP 8662A/
in
Chapter 12.
in
the low-frequency loops optimizes the tradeoffs between resolution, switching speed, and phase noise
of
the 10
to
20 MHz signal from
these loops. Fractional-N techniques sim-
Frequency Range Heterodyne Band Divide-by-4 Band
Divide-by-2 Band Fundamental Band 1st Doubled Band 2nd Doubled Band
Table 3.1.
HP 8662A/8663A frequency bands.
s^:
^:y
V
Absolute Phase Noise of 320 to 640 MHz
'Switched Reactance Oscillator at 500 MHz
Offset From Carrier (Hz)
similar HP Synthesizers (Models 3325A, 3326A and 3335A) are used Loop"
the N Loop, technique achieves 1 MHz resolution while minimizing the multiplication phase noise number. The Fractional N Loop uses
_
corrected fractional-N technique
is
particu-
in
achieve 0.1 Hz overall resolution with relatively low spurious content. This
the
loop overall frequency switching speed
HP 8662A/8663A. of about 400 microseconds.
difficult
to
The overall phase noise
20 MHz low frequency loop
-145 dBc
in
for
low-
anti-
THE OUTPUT SECTION
The output section translates
to 640 MHz signal from locked loop section HP 8662A/8663A output frequency by doubling, dividing, modulates discussed section. This process produces distinct frequency bands covering HP 8662A
ranges,
HP 8662A
0.01
to
1280 MHz
0.01
to
120
MHz 120 to 160 MHz 160 to 320 MHz
320 to 640 MHz 640 to 1280 MHz (not applicable)
'>;>0'^$§i
HP 8662A/8663A
~
Absolute Phase Noise
to
those used
and the "Fractional N Loop".
an
by
is
the primary determinant
at a
the
in the
and
as
shown
HP 8663A
0.1
to
2560 MHz
0.1
to
120 MHz 120 tO 160 MHZ 160 to 320 MHz 320 to 640 MHz 640 to 1280 MHz 1280
to
2560 MHz
in
lower-frequency
in
both the
uncorrected fractional-N
using a low divide
It
has a settling time
of
the 10
is
10 kHz offset.
the
to the
signal
desired
or
mixing,
as
previously
high-frequency loop
the
8663A frequency
in
Table 3.1.
I Offset H from H Carrier
H
10 Hz
II
100 Hz
U 1 kHz I
10 kHz
■
100 kHz
*HP 8663A only,
"N
In
of
a
to
of
the
of
to
about
the 320
phase-
and
Heter­odyne
0.01
120 MHz
-113
-126
-133
-137.
-134
The ways
in
which these bands derived determine the short-term stability characteristics and the maximum available peak FM deviation band. For example, since frequency doubling results
in a
6 dB increase phase noise (for offsets greater than 1 kHz), the phase noise
of
HP 8662A/8663A output
bands should higher than that
be
about 6 and 12 dB
in
the main band. Likewise, the phase noise by-2 and divide-by-4 bands should about 6 and 12 dB lower. The phase noise
in
about the same as
the heterodyne band should
in
the main band,
except that some noise cancellation
a
the
occurs close
lation noise Similarly, deviation
number, increased in the heterodyne band, same
to
of
in
the carrier due
correlated reference section
the down conversion process.
in
divide bands, maximum FM
is
reduced
in
the multiply bands
by
by
the divide
the multiply number, and
it
as in
the fundamental band.
The actual residual phase noise over the entire frequency range HP 8662A and 8663A
is
shown Table 3.2. For each divide-by-2 multiply-by-2 from
frequency, increases
the
by
the
main band
phase noise decreases
6 dB, respectively. Note how closely the actual correlates with the expected values. This close correla­tion results from careful design parts
of
the output section. Areas particular concern included designing the AGC loop
noise conversion fully controlled levels
for
minimum AM-to-PM
and
obtaining care-
at the the heterodyne band mixer. resulting broadband noise floor HP 8662A/8663A
is
less than -148 dBc
at offsets greater than 1 MHz.
Table 3.2
Typical HP 8662A/8663A residual SSB phase noise.
Carrier Frequency
Main-
+4
120 to
160 MHz
-119
-129
-138
-147
-145
-H2-
160 to
320 MHz
-113
-124
-133
-142
-142
band
320 to
640 MHz
-107
-119
-128
-136
-136
X2
640 to
1280 MHz
-101
-in
-122
-130
-130
are
of
each
the
in
the doubled
in
the divide-
to
cancel-
it is
remains
of
the
in or
in all
inputs
The
of
X4
1280 to
2560 MHz*
-95
-106.
-116
-124
-124
in
be
the
of
the
'
•
be
or
to
11
m#n,
er
4
Improving Frequent^ Stability With External References
',V
"^°1Si'
' •
<*
1^
Jv <?img&ffif - -SS'
A synthesizer source are derived from a single fixed-frequency
reference oscillator, where short-term stability translated
examines ence oscillator affects output frequency
8663A. shows bility
of the HP internal reference output signal. a specific case an external reference the close-in short-term stability as
the HP 8662A/8663A. This specific case then expanded arbitrary external reference ity parameters
Why
Use an
The internal reference 8663A absolute phase noise
frequency stability
8663A apply only with this internal erence. Often, however, erence accepts level
of 1 V^
reference 50 ohms.)
often desirable components common reference. in
the
system reference, stability altered. Since erence does alter these frequency stability parameters, ence
can be
Reference Effects
is
in
The
how the
long-term stability
defined
which
all
output frequencies
to the
how the
of the
output. This chapter
stability
the
of the HP
first part
of the
long
and
8662A/8663A's
are
The
translated
chapter then describes
of
using a cesium beam
to
to
discuss
of the HP
External Reference?
is a 10 MHz
is
used.
any
external
±0.1
at a
level
For
example,
of the
is
the
long-
of the HP
the use of an
used
in the HP
crystal oscillator.
and
of the HP
(The HP
5 MHz
V or any 10 MHz
of 0.5 to 0.7 V_
to
operate
system from If
another reference
chosen
and
8662A/8663A will
an
to
improve them.
on
as a
signal
the
long-
reference
of the
stability
short-term sta-
improve both
of the the
on the
8662A/8663A.
long-term
an
external
8662A/8663A
in a
all the
as the
short-term
external
external refer-
refer-
of the 8662A/ chapter
own
to the
as
well
effect
stabil-
8662A/
8662A/
standard
rms
system
a
common
is
of an
ref­ ref-
into
it is
ref-
Long-Term
and
as
is
The
at a
be
Stability
Frequency stability degree
to produces a specified period
of frequency stability includes cepts dental modulation, ations
which
the
of
random noise, residual
of the
can be
the
same frequency throughout
of
and any
output frequency.
defined
oscillating source
time. This definition
as the
the
con-
and
other fluctu-
inci-
synthesizers, in fractional parts week, month, ity usually results from aging components oscillating source.
For
the HP ship between the reference
of
the Because process, racy
of the
of
the or external.
The internal reference 8663A
oscillator with specified long-term stabil­ity
of 5 X 10-10 per day warmup. function rate,
temperature effects, age effects. These parameters translated output frequency.
If
an
external reference
HP 8662A/8663A long-term stability
be either degraded
long-term stability
crystal oscillators A secondary standard such ium oscillator the order mary frequency standards such cesium beams have even less frequency drift—specifying stability 5 parts
in 10"12 for the
beam tube.
it is
commonly expressed
of a
cycle
per day,
or
year. Long-term stabil-
and
materials used
8662A/8663A,
the
long-term stability
and the
output frequency
of the
nature
the
frequency drift
output signal
reference, whether
is an
oven-controlled crystal
The
frequency accuracy
of
time base calibration, aging
to the HP
or
for
is 1 X
has
of 1 X 10~n per
long-term stability
the
long-term stability
is
simple.
of the
and
is
equal
it is
in the HP
after a 10-day
and
8662A/8663A
is
used,
improved. Typical
room temperature
10"6 per
as a
month. Pri-
on the
life
of the
of the
in the
relation-
of
synthesis
accu-
to
that
internal
8662A/
is a
line volt-
are
directly
the
can
month.
rubid-
on
as
order
of
cesium
Offset from Signal
f
1
Hz
10 Hz
100 Hz
1
kHz
10 kHz-
Effect
of the
Reference
on
Short-
Term Stability
A common measure frequency stability (SSB) phase noise; discussion implications. of phase noise residual noise synthesizer; that limit synthesizer. signal residual noise.
Absolute noise present lute noise includes of
the
with different references.
To examine ence oscillator translates the absolute noise
8663A, consider HP 8662A/8663A absolute
SSB phase noise (Figure 4.1). Note that the absolute noise with erence only than about 2 kHz. than 2 kHz, same reference typical phase noise
4.1. translated at a carrier frequency
plotted
ical phase noise 8663A
of
and
absolute. Residual phase
is the
phase noise inherent
on the
for
This phase noise
noise performance
The
can
never
or
total noise
reference used,
how the
is
greater than
offsets from
the
as the
absolute noise.
in the HP
to the
on the
in
Figure
of
short-term
is
single-sideband
see
phase noise
In a
are
at the
same graph with
Chapter
synthesizer,
usually specified—
is, it is a
noise
on the
be
better than
is the
device output. Abso-
the
noise contribution
and
noise
of the HP
the
plot
the
the
For
offsets greater
residual noise
8662A/8663A
as
shown
at 10 MHz is
equivalent phase noise
of 500 MHz and is
of the HP
4.1.
2 for a
and its
two
theoretical
of the
output
the
total phase
will change
on the
to or
affects
8662A/
of
typical
and
the
internal
residual noise
carrier less
is the
The
in
8662A/
types
in the
refer-
residual
ref-
internal
has
Table
the
typ-
Phase Noise Ratio
^(f)
-90 dBc
-120 dBc
-140 dBc
-157 dBc
-160 dBc
Long-term stability, often called fre­quency drift, refers output frequency over a period usually greater than
12
to the
a few
change
of
seconds.
in
time
For
Table 4.1.
HP 8662A/8663A internal reference oscillator phase noise.
The graph shows that the absolute phase noise of the HP 8662A/8663A closely follows the translated noise of the refer­ence to about 2 kHz offset from the car­rier. Beyond 2 kHz offset, the noise on the reference oscillator remains flat,
while the absolute noise of the HP 8662A/8663A continues to drop until it reaches the residual noise level. For offsets greater than about 2 kHz, the typical phase noise of the reference oscil­lator is actually greater than the typical absolute noise of the HP 8662A/8663A.
would be about -124 dBc at a 100 kHz offset. The filters, however, effect sub­stantial noise reduction, with about 35 dB of noise attenuation, to reduce the broadband noise floor to about —160 dBc. In addition to the noise reduc-
tion effected by the crystal filters, the bandwidths of the phase-locked loops were carefully chosen to minimize broadband noise. However, most of the noise reduction is due to the filtering. For more information on the design of the HP 8662A/8663A and the reference section, see Chapter 3.
HP 8662A/8663A Stability Using a
Cesium-Beam Reference
An excellent external reference source for improving the long-term stability of the HP 8662A/8663A is a cesium beam frequency standard. To see how the noise of a cesium standard affects the short-term stability or absolute noise of
the HP 8662A/8663A, and to expand that to the general effect of using an external reference, this section examines the measured absolute noise performance of the HP 8662A/8663A with the Hewlett-Packard Model 5061A Cesium Beam Frequency Standard (with high stability Option 004 for improved phase noise) as an external reference.
A good insight into the expected noise performance of the HP 8662A/8663A with the cesium-beam standard as an external reference can be gained by com­paring the specified single-sideband phase noise of the HP 5061A to that of the HP 8662A/8663A 10 MHz internal reference. Figure 4.2 plots these noise characteristics, with the noise of the
5 MHz HP 5061A converted up to the
equivalent noise at 10 MHz.
Figure 4.1.
Comparison of HP 8662A/8663A noise vs. noise of
internal reference.
The reference section of the HP 8662A/ 8663A was designed to ensure that this high reference noise at offsets greater than 2 kHz would not contribute to the absolute noise of the output signal; that
is,
the reference section includes filters to improve the broadband noise perform­ance over the noise of the internal refer­ence.
In the reference section, the
10 MHz reference signal is directly mul­tiplied up to 640 MHz for use in other parts of the HP 8662A/8663A.
Were nothing else done to this 640 MHz signal, the broadband noise would be translated to the output frequency. How­ever, to improve the broadband noise, monolithic crystal filters were added in the reference multiplier chain at 40 and 160 MHz. The 40 MHz filter has a band­width of about 6 kHz; the 160 MHz filter a bandwidth of about 18 kHz. With no filtering, the noise floor on the multiplied-up reference signal (640 MHz)
In summary, due to the design and fil­tering of the reference section, the noise
of the reference oscillator primarily affects the close-in absolute phase noise of the HP 8662A/8663A. Up to about 2 kHz, the dominant noise mechanism is that of the multiplied-up reference sec­tion. Beyond 2 kHz, the crystal filters in the reference multiplier chain filter the reference oscillator noise and the broad­band noise floor reaches the HP 8662A/ 8663A residual noise level. Absolute
noise can be improved by using a lower­noise reference. Again, by the definition
of residual noise, no external reference,
no matter how low in noise, could reduce the absolute noise of the HP 8662A/8663A to anything less than the residual noise. If the noise of the external reference is actually lower than the residual noise of the HP 8662A/ 8663A, the HP 8662A/8663A's residual
noise would dominate.
The phase noise of the HP 8662A/ 8663A internal reference is graphed with a dashed line for offsets from the carrier less than 1 Hz because the phase noise is actually specified only for offsets greater than 1 Hz. Phase noise information at offets greater than 1 Hz is normally
suf­ficient for those applications where a crystal would be used. However, the time domain stability (fractional­frequency deviation) for averaging times from tau equal to 10~3 to 102 seconds is specified for the HP 8662A/8663A refer­ence oscillator. These time-domain repre­sentations of short-term stability were translated to equivalent frequency­domain representations for offsets less
than 1 Hz by algebraic calculations
accepted by the U.S. National Bureau of Standards (NBS). For more information
on how to perform these translations, see NBS Technical Note 679, "Frequency Domain Stability Measurements: A Tutorial Introduction."
Figure 4.2 shows that the phase noise of the HP 5061A Cesium Beam is greater than that of the HP 8662A/ 8663A reference oscillator for offsets from the carrier greater than approxi­mately 2 Hz. Since the bandwidth of the first crystal filter in the HP 8662A/
8663A reference section at 40 MHz is
13
J . ."wr-.■"■<-••■■•.
i. ••- * ■:
•*.*,■;•"
> ■
'©&
wasKa
-20
-40
Iv.
•
1
*Ve' *
Internal Oscillator
Figure
4.2.
Noise comparison HP 5061A cesium beam.
approximately this higher noise would about
4 kHz
fore,
the HP 8662A/8663A with as
an
external reference the absolute noise with erence
at because loop bandwidths 8663A, this higher reference noise eventually attenuated until noise
is
of
internal reference oscillator
6 kHz,
attenuation
not
from
the
absolute noise
carrier. There-
of the
the HP
is
the
offsets greater than
of the
filtering
and
in the HP
dominant.
start until
5061A
higher than
internal
2 Hz. But effect 8662A/
the
residual
femnre
Re
10
100
Offset From Carrier
vs.
standard reaches filtering
of
ues that HP 8662/8663A continues offset from though broadband noise floor.
ref-
of
is
To show tages external reference, Figure the measured absolute noise
fc =
10 MHz
5061A
Opt.
^
(Hz)
in the HP
to
attenuate
the
absolute phase noise
the
of
using a cesium beam
the
the
carrier increases, even
reference
the
advantages
10K
100K
its
noise floor. Here
8662A/8663A contin-
reference noise
of the
to
decrease
has
reached
and
its
disadvan-
as an
4.4
compares
of the
the
so
HP 8662A/8663A with
crystal reference,
the
absolute noise with the cesium frequency standard, typical residual phase noise HP 8662A/8663A. Figure
the noise than oscillator 2 Hz. is translated of references
of the
the
As
the HP
cesium standard
noise
of the
for
offsets less than about
internal crystal
expected, this same relationship
to the
absolute phase noise
8662A/8663A when these
are
used.
The phase noise (less than HP 8662A/8663A of
the HP
Standard
5061A Option
as an iting greater than at
0.01 Hz
is
improved with
external reference, exhib-
10 dB of
offset, with improvement increasing the carrier decreases.
For offsets greater than lute phase noise 8663A with Cesium Standard
greater than
of the HP
the HP
5061A Option
as a
the
absolute noise with HP 8662A/8663A internal oscillator, predicted. standard continues
as
noise with
The
noise with
to be
the
internal crystal until
HP 8662A/8663A crystal filters
ficiently attenuate reference-noise floor residual noise. Figure reduction occurs carrier
of
at
around
the
160
tent with filter approximately
the
to
at an
25
kHz. This
fact that
MHz has a
18 kHz.
4.4
the
its own
internal
and the
of the
4.2
shows that
is
lower
very close-in
1 Hz
offset)
of the
use
004
Cesium
improvement
the
amount
as
2 Hz, the
of
offset from
abso-
8662A/
004
reference
is
the
as
the
cesium
higher than
the
the
can suf-
cesium's higher
less than
the
shows that this
offset from
is
consis-
the
second crystal
bandwidth
of
two
Figure
4.3
shows
the noise results; the
HP
the
8662A/8663A with HP 5061A Option is shown 100 kHz. between
for
offsets from
To
examine
the
noise
absolute phase
absolute phase noise
004
Cesium Standard
the
of the the resultant absolute noise HP 8662A/8663A, noise
of the
to
the
equivalent noise
also plotted.
nal reference, close for offsets less than phase noise very closely follows of
the
reference used. Between
and 1
kHz, the
the
HP
8662A/8663A generally follows the noise curve except that
cesium
the
is
smoothed
offsets greater than
14
the
specified phase
cesium standard converted
at 500 MHz is
As in the
case
to the
10 Hz) the
of the HP
8662A/8663A
the
noise spectrum
absolute phase noise
of the
cesium reference,
noise "plateau"
out by
1 kHz, the
the
0.1 Hz to relationship
reference
and
of the
of the
inter-
carrier (here
absolute
10 Hz
of the
filtering.
For
cesium
of
of
Figure
4.3.
Effect
of
cesium beam frequency standard
HP 8662A/8663A absolute noise.
on
jr->3
.Ktw >--,-( T^..
■"jf^y*
In summary, Figure
of
an HP
5061A Option Beam optimizes noise (less than 8663A.
For
some applications, this very close-in phase noise if offsets from
100
kHz are of
many types
of
4.4
shows that
004
the
very close-in phase
1 Hz) of the HP
is
the
critical. However,
carrier from
more concern,
receiver testing,
use
Cesium
8662A/
1 Hz to
as in
use of the HP 8662A/8663A internal crystal refer­ence provides better performance.
Effect
of an
Arbitrary Reference
Expanding of
any phase noise to
the 8663A output frequency, whether
noise or lower than that
oscillator.
rier, noise than noise will also until
reduce
the
results
external reference,
of the
absolute noise
of the
external reference
At
if the
the HP
greater offsets from external reference the
internal reference, this be
seen
8662A/8663A filtering
the
reference noise
to the
the
reference
of the HP
of the
internal crystal
as
absolute noise,
to
general case
close-in
is
translated
8662A/
the
is
higher.
the
has
car-
higher
can
less than the residual noise. This should normally occur
at an
However,
if the
reference noise extremely high, this might occur higher offset from tion
of the
frequency response
offset around
the
carrier
20 to 30 kHz.
is
at a
as a
func-
of the
crystal filters.
For
the
lowest phase noise
from
the
absolute noise
carrier, a combination
of the
offsets less than 1
HP 5061A
Option
Caslium Beam
~
'
Hz and the
004
at all
offsets
of the
cesium standard
absolute
*
■*
<1
BW
Lock
Reference
Oscillator
or other
Crystal
at
Hz
Box
I
in
o.01
Figure
4.4.
HP 8662A/8663A absolute noise comparison.
noise
of the other crystal reference, than 1
Hz mal solution solution
The "lock
would
is
technically feasible.
is
shown
box" is
at
be
optimal. This opti-
in
Figure
basically just
nal phase-locked loop with
internal oscillator,
standard acting and
the
crystal oscillator
as the
reference oscillator
or
offsets greater
4.5.
an
the
cesium
as the
voltage­controlled oscillator (VCO). Figure shows
the
lock
box in
simple block-
diagram form.
The phase-locked loop locks VCO
to the
cesium standard
HPS061A
K34-59991A
Ext.
Ref.
Input
HP 8662A/
8663A
the
in
less than
Offset from Carrier
some
One
exter-
4.6
crystal
100
1K 10K 100K
(Hz)
1
Hz
bandwidth. Within
of
the
loop,
But
the
loop
and the
to the
box" is
the equal outside
the
VCO is erence. loop, ence,
translated
This "lock
noise
no
noise
output.
the
bandwidth
at the
output
to the
noise
the
bandwidth
longer tracks
of the VCO
of
on the ref-
of the
the
refer-
will
commercially available as Hewlett-Packard Model 5061A K34-59991A, with a bandwidth approximately connected
0.16 Hz. It can be
to the HP
8662A/8663A
of
directly
external-frequency-control input.
This arrangement yields very close-in phase noise HP 5061A Option Frequency Standard,
of
the HP
ence oscillator
100 of
the HP
8662A/8663A internal refer-
at
kHz, the low
8662A/8663A
the
excellent
of the
004
Cesium Beam
the low
offsets from 1
phase noise
Hz to
broadband noise floor
and the
out­standing long-term frequency stability the cesium beam of
the
cesium beam tube.
±3 X 10~12 for the
life
be
of
Figure
4.5.
Using
two
phase noise.
references
for
optimal
HP
8662A/8663A
Figure
4.6.
Narrowband phase-lock loop system.
for
two-reference
Reference
Oscillator
(HP 5061A
Opt.
004
Cesium)
Phase
Detector
Low Pass
Filter
VCO
(Crystal
Osc.)
.'""
%
.-
-»
15
>T;SS5 Phase Noise Measurenierit
.,-f^>W
Common Measurement Methods
There are many methods of measuring SSB phase noise, each of which has its advantages. Here is a summary of the most common methods currently in use:
1.
Heterodyne frequency measurement technique. This is a time-domain technique in which the signal under test is down converted to an intermediate frequency and the fractional frequency deviation is measured using a computer­controlled, high-resolution frequency counter. a{r) is then calculated (see Chapter 2), and the computer trans­forms the time domain information to equivalent values of SSB phase noise. This method is particularly useful for phase noise measurements at offsets less than 100 Hz.
2.
Direct measurement with a spectrum analyzer. This is the frequency-domain technique discussed briefly in Chapter 2. This method is limited by the spectrum analyzer's dynamic range, selectivity, and LO phase noise. For more information, see Hewlett-Packard Application Note 270-2, "Automated Noise Sideband Measurements Using the HP 8568A Spectrum Analyzer."
3.
Measurement with a frequency
discriminator. In this frequency-domain
method, the signal under test is fed into
a frequency discriminator and the output
of the discriminator is monitored on a low-frequency spectrum analyzer. The best performance is obtained with a delay line/mixer combination as discriminator. Due to the inherent rela­tionship between frequency modulation and S if), the noise floor of this kind of
system rises rapidly for small offsets. The resulting higher noise floor limits the usefulness of this method for these small carrier offsets. Reference HP Product Note 11729C-2, "Phase Noise Characterization of Microwave Oscil-
lators Frequency Discriminator Method."
4.
The two-source technique. In this phase detector method, the signal under test is down converted to 0 Hz and examined on a low-frequency spectrum analyzer. A low-noise local oscillator (LO) is required as the phase detector reference. This is the most sensitive method of phase noise measurement. For this reason, and because the HP 8662A/ 8663A is ideally suited as the low-noise LO,
the phase detector method is
explored in detail in this chapter and the following two chapters. Also see HP Application Note 246-2, "Measuring Phase Noise with the HP 35 85A Spectrum Analyzer."
The Two-Source Technique
Basic Theory
The basic measurement setup used for
measuring phase noise with the two-
source technique is shown in Figure 5.1.
In this method, the signal of the source under test is down converted to 0 Hz or dc by mixing with a reference signal of the same frequency in a double-balanced mixer. The reference signal is set in phase quadrature (90 degrees out of phase) with the signal under test. When this condition of phase quadrature is met, the mixer acts as a phase detector, and the output of the mixer is propor­tional to the fluctuating phase difference between the inputs. Hence the SSB phase noise characteristics may be deter-
mined by examining the mixer output signal on a low frequency spectrum ana­lyzer. The frequency of the noise dis­played by the analyzer is equal to the offset from the carrier.
Source
Under
Test
Figure 5.1.
Basic two-source phase noise measurement setup.
The relationship between the noise mea­sured on the analyzer and Jf(f) (Chapter 2) is derived from
v
A0rms =
Ad>
rms
= rms
rms
where phase noise, V measured on spectrum analyzer, and K.
= phase detector constant which is
^bpeak- The level of the beat note
n
Krf
phase deviation of
= noise level
bpeak
(Vbrms-where V
= -^=
brms
)
produced in the calibration is described below. This assumes a sinusoidal beat note and a linearly operating mixer.
v?
ms
vf
ms
brms
vf
4 (V
2
)2
ms
brms
)2
(Vbpeak)
2 (V
(in a 1 Hz bandwidth)
m
of(f) =
(in a 1 Hz bandwidth)
=
(in a 1 Hz bandwidth)
_
S,(f)
2
This relationship reveals how to calibrate the measurement to obtain eJf(f). First the
reference source is offset by a small amount such as 10 kHz to produce a beat note from the mixer that can be measured on the spectrum analyzer (V
).
This beat note can be consid-
brms
ered as representing the carrier of the signal under test. This carrier reference level is noted, then the reference source is reset to the frequency of the source under test and adjusted for phase quad­rature. Quadrature is indicated by zero volts dc as monitored on the oscillo-
scope. The noise displayed on the spec­trum analyzer corresponds to phase noise and the spectrum analyzer's fre­quency scale corresponds to the carrier
offset frequency. To make an SSB phase noise measurement, the level of the noise on the spectrum analyzer is mea­sured referenced to the carrier level noted above (V
). The actual SSB
brms
16
Loading...
+ 35 hidden pages
Buy Points How it Works FAQ Contact Us Questions and Suggestions Privacy Policy Cookie Policy Terms of Use