Aeroflex 2019 Data Sheet

T
oday’s wireless communications market, from cellular phones to wireless data, is
expanding at an incredible rate. Along with this growth comes an increasing need for test equipment that verifies the performance of these devices and systems. Signal genera­tors play a multifaceted role in the develop­ment of both receivers and transmitters. They are used for generating signals ranging from simple sinusoidal tones for LO substitution to
fully modulated signals for receiver testing. This article focuses on the importance of us­ing a signal generator with relatively high spectral purity for RF communications test­ing. The ideal signal generator would pro­vide perfect sinusoids at carrier and sideband frequencies, but in re­ality all signals have imperfections. The
foresight to take these flaws into account allows the engineer to se­lect the appropriate signal generator and re­duce development time.
WHAT IS SPECTRAL PURITY?
Spectral purity is the inherent frequency sta­bility of a signal. Stability is defined over a peri­od of time: short or long term. Long-term sta­bility, or drift, is usually defined as frequency changes over a period of time greater than one second. Short-term stability is defined as fre-
quency changes over less than one second. Current signal generator technology generally offers good long- and short-term stability. For wireless communications testing, short-term stability is of greater concern. This article dis­cusses key spectral purity components and the importance of spectral purity in testing wireless communications equipment. Implications of spectral purity are briefly covered for LO sub­stitution, phase noise measurements, receiver performance tests and radar applications.
Phase Noise
Perhaps the most common method for specifying the spectral purity of a signal gener­ator is its phase noise. In the time domain, phase noise is exhibited as a jitter in the zero crossings of a sine wave, as shown in Figure 1. For a high performance signal generator, the phase noise is not usually discernible in the time domain. In the frequency domain, the phase noise appears as noise sidebands on the
SIGNAL GENERATOR SPECTRAL PURITY CONSIDERATIONS IN RF COMMUNICATIONS TESTING
TUTORIAL
BRIAN CHENG
Hewlett-Packard Co.,
Microwave Instruments Division Santa Rosa, CA
“...in reality all signals have imperfections. The foresight to take these flaws into account allows the engineer to select the appropriate signal generator and reduce development time.”
Fig. 1 Time domain phase noise jitter.
PHASE NOISE
V (t)
TIME (t)
carrier, as shown in Figure 2. The US National Bureau of Standards de­fines single-sideband (SSB) phase noise +(f) as the ratio of the noise power in a 1 Hz bandwidth at a fre­quency f away from the carrier to the signal power of the carrier:
+(f) is expressed as decibels relative to the carrier per hertz (dBc/Hz). A 1 Hz bandwidth is used to allow the phase noise in other bandwidths to be easily calculated for comparison.
The SSB phase noise at a specified carrier frequency is often graphically represented on a log-log plot, as shown in Figure 3. Phase noise can be conveniently displayed for a wide range of frequency offsets by using a log scale on the frequency axis.
+(f) =
noise power in a 1 Hz
bandwidth at a frequency
f (Hz) away from the carrier
power level of the carrier
Spurious: Harmonics, Subharmonics and Nonharmonics
Spurious signals are frequency spikes that appear in the spectrum. These spectral components may be divided into three categories: har­monic, subharmonic and nonharmon­ic, as shown in Figure 4.
Harmonics are generated by de­vice nonlinearities in the signal gen­erator and are integer multiples of the carrier frequency. For example, a 100 MHz carrier frequency will have harmonics at 200 MHz, 300 MHz and so on. The amplitudes of the har­monics (relative to the amplitude of the carrier signal) are determined by the nonlinear characteristics of the components in the signal generator.
Subharmonics are generated when frequency multiplying to create the carrier frequency. The frequency be­ing multiplied may leak through the signal path and appear at the output. For example, a 500 MHz signal mul­tiplied by two to arrive at a 1 GHz carrier frequency might appear as a subharmonic.
Nonharmonics are frequency com­ponents that do not appear related to the carrier frequency. Although signal generator designers can determine the location of these spurious signals, they are unpredictable to the user. Today’s signal generators are able to suppress harmonics, subharmonics and nonharmonics to a level accept­able for most applications.
Residual FM
Residual FM is another method commonly used to specify the fre­quency stability of signal generators. Residual FM includes the effects of both spurious signals and phase noise. It is the integral of the SSB curve with limits set by the post-de­tection bandwidth. Common band­widths are 300 Hz to 3 kHz and 20 Hz to 15 kHz.
SPECTRAL PURITY CONSIDERATIONS IN RF RECEIVER DESIGN
A spectrally pure signal generator provides high value to those design­ing and verifying analog and digital communications devices. As an exam­ple, a simple communications receiv­er, shown in Figure 5, is used to il­lustrate the effects of phase noise and spurious signals on practical applica­tions and measurements. Three ma­jor applications discussed here are LO substitution, phase noise mea­surements and receiver performance tests. All of these applications require the use of a signal generator with suf­ficient spectral purity.
LO Substitution
In receiver development, as well as transmitter development, a spec­trally clean LO is required for upcon­version and downconversion of sig­nals. A signal generator is often used to substitute an onboard LO for test­ing and system troubleshooting. Looking at the downconversion in the receiver, the importance of spectral purity for LO substitution is readily apparent. Suppose that two signals are present at the input of the receiv­er, as shown in Figure 6. These sig­nals are mixed with an LO signal down to an intermediate frequency (IF) where highly selective IF filters separate one of the signals for ampli­fication, detection and baseband pro­cessing. If the desired signal is the larger signal, there is no difficulty in recovering it.
On the other hand, a problem might arise if the desired signal is the smaller of the two because any phase noise on the LO signal is translated directly to the mixer products. Notice that the translated noise in the mixer output completely masks the smaller signal. Even though the receiver’s IF filtering might be sufficient to re-
TUTORIAL
Fig. 2 A frequency carrier (a) without
and (b) with phase noise sidebands.
Fig. 3 A typical phase noise plot.
Fig. 4 Harmonic, subharmonic
and nonharmonic signals.
Fig. 5 A simple communications receiver.
AMPLITUDE
f
0
(a)
(b)
FREQUENCY
AMPLITUDE
f
FREQUENCY
+(f)
FREQUENCY OFFSET FROM CARRIER (f)
SUB-
HARMONICS
AMPLITUDE
0.5 f
0
CW OUTPUT
PHASE NOISE
f
0
FREQUENCY
HARMONIC
SPUR
NON-
HARMONIC
SPUR
2 f
0
RF
MIXERLOFILTER
0
PRESELECTOR
AMLIFIER
A/D TO DSP
move the larger signal’s mixing prod­uct, the smaller signal’s mixing prod­uct is no longer recoverable due to the translated LO noise.
Phase Noise Measurements
Eventually, the signal generator that is substituting as the LO must be replaced by the actual LO. The phase noise of this onboard oscillator must be measured to ensure a quality sig­nal. In this case, a low phase noise signal generator can be used to make the measurement.
Many methods exist to measure phase noise. One of the most sensi­tive measurement techniques is the two-source phase detector technique. Here, the signal under test is down­converted to 0 Hz and examined on a low frequency spectrum analyzer. A low noise LO is required as the phase detector reference. The basic mea­surement setup for measuring phase noise using the two-source technique is shown in Figure 7.
The noise measured by this two­source technique represents the com­bined noise of both the source under test and the reference source. This lev­el is the upper limit for the phase noise of either device. Therefore, if the phase noise of the reference is better than the source under test, the phase noise of the source under test can be determined.
Receiver Performance Tests
After the design of the receiver is complete, various tests must be per­formed to confirm design parameters. The primary goal of most receiver
tests is to measure the receiver’s ability to maintain a certain sensitivity level in the presence of un­wanted signals.
Receiver perfor­mance verification tests may be divided into in-channel and out-of-channel tests. Common in­channel tests in­clude sensitivity and
co-channel rejection. Common out­of-channel tests are spurious and in­termodulation rejection, and adja­cent-channel selectivity. All of these tests, except for sensitivity, require a modulated or unmodulated interfer­ing signal with allowable uncertain­ties, phase noise and spurious content as defined in the communications standard. Figure 8 shows the test setup for co-channel or out-of-chan­nel rejection measurements.
For analog receivers, sensitivity is defined as the minimum power level at which the receiver can successfully detect and demodulate the incoming signal. For digital receivers, sensitivity is defined as the median level of the received signal that produces a speci­fied bit error rate when the signal is modulated with a pseudorandom bi­nary sequence of data. The important specification of the signal generator for sensitivity tests is power level ac­curacy (rather than spectral purity).
Co-channel rejection is the ability of the receiver to maintain sensitivity in the presence of an in-channel in­terfering signal. Frequently, this co­channel interfering signal will be a continuous-wave (CW) signal, as shown in Figure 9. The specific com­munications standard defining this test will set phase noise and spurious signal requirements for the CW tone.
Spurious immunity is a measure of the ability of the receiver to receive a modulated input signal in the pres­ence of unwanted input signals at fre­quencies other than those specified for adjacent- and alternate-channel tests. The specific communications standard defines the spurious signal frequency location and tolerable phase noise level.
Intermodulation rejection is a measure of the capability of the re­ceiver to receive a wanted modulated signal without exceeding a given degradation due to the presence of two or more unwanted signals with a
TUTORIAL
Fig. 6 Phase noise effects at the mixer;
the (a) RF input, (b) LO and (c) mixer output spectra.
Fig. 7 Basic measurement setup for the two-source
phase detector technique.
Fig. 8 The test setup for co-channel or out-of-channel rejection measurements.
Fig. 9 Co-channel rejection.
(a)
AMPLITUDE
FREQUENCY
f
f
1
2
AMPLITUDE
f
LO
(b)
FREQUENCY
(c)
AMPLITUDE
f1−fLOf2−f FREQUENCY
LO
SOURCE
UNDER
DOUBLE-
BALANCED
MIXER
LOW NOISE REFERENCE
SOURCE
TEST
OSCILLOSCOPE
LOWPASS
FILTER
LOW NOISE
AMPLIFIER
IN-CHANNEL SIGNAL
(MODULATED SIGNAL)
OUT-OF-CHANNEL SIGNAL
(CW OR MODULATED SIGNAL)
LOW
FREQUENCY
SPECTRUM ANALYZER
AUDIO
SIGNAL
OR
DATA
CO-CHANNEL
INTERFERER
LEVEL (dB)
DESIRED
SIGNAL
FREQUENCY
IF REJECTION
CURVE
SSB PHASE
NOISE
specific frequency relationship to the wanted signal frequency. Typically, two out-of-channel CW tones are placed so that their third-order inter­modulation distortion product falls on top of the desired signal, as shown in Figure 10. Intermodulation rejection measures how well the receiver re­jects this unwanted distortion.
Adjacent-channel selectivity mea­sures a communications receiver’s ability to process a desired signal while rejecting a strong signal in an adjacent channel. Alternate-channel selectivity is a similar test where the interfering signal is spaced two RF channels away from the passband of the receiver. These tests are very im­portant for both analog and digital units where channel spacings are nar-
row and many signals may be encoun­tered in a small geographical area.
PHASE NOISE REQUIREMENTS FOR ADJACENT-CHANNEL SELECTIVITY
For many receivers, the SSB phase noise of the signal generator used to produce the interfering signal is a critical spectral characteristic. If the phase noise energy inside the pass­band of the IF filter is excessive, the receiver might appear to fail the test. This case is shown in Figure 11.
The required signal generator SSB phase noise may be calculated using
where Φn= signal generator SSB phase
noise (dBc/Hz) at the channel spacing offset
Be= receiver noise-equivalent
bandwidth (Hz)
Pac= adjacent- or alternate-channel
selectivity specification (dB)
P
mar
= test margin (dB)
Since Beand Pacare fixed by the specifications or design, the test mar­gin determines the power that the signal generator phase noise is al­lowed to contribute to the IF pass­band of the receiver. A large test mar­gin increases confidence that the re­ceiver operates properly in the presence of signal-to-noise degrada­tion due to fading in the channel or imperfections in receiver compo­nents. For a system using a new tech­nology or new operating frequencies, a large test margin should be used to compensate for uncertainties.
Φ
n
e
ac mar
B
P P=
 
 
101log
For a receiver with a noise-equiva­lent bandwidth of 14 kHz, Pacat the adjacent channel of 70 dB, margin of 10 dB and channel spacing of 25 kHz, the required SSB phase noise is –121 dBc/Hz at 25 kHz offset. This condi­tion is typical for an analog FM re­ceiver. Unlike the FM receiver in this example, most digital communica­tions receivers have adjacent-channel selectivity values less than 15 dB. For a GSM receiver with a noise-equiva­lent bandwidth of 200 kHz, a Pacat the adjacent channel of 9 dB, margin of 10 dB and channel spacing of 200 kHz, the required SSB phase noise is –72 dBc/Hz at 200 kHz offset. The required SSB phase noise is driven primarily by Pac.
Table 1 lists the values of adja­cent- and alternate-channel selectivi­ty for various communications sys­tems as well as the required signal generator SSB phase noise. A 10 dB test margin was used. Clearly, for ad­jacent- and alternate-channel selec­tivity testing on many digital RF com­munications formats, the signal gen­erator SSB phase noise is not as important as for analog FM systems.
For selectivity tests, the spectral shape of the signal is the characteris­tic of primary importance. The digital modulation formats used by GSM, CDMA, North American Digital Cel­lular (NADC) and personal digital cellular (PDC) characteristically leak a small amount of power into the ad­jacent channels. Figures 12, 13 and 14 show amplitude vs. frequency for the selectivity values specified previ­ously. The impact of the spectral shape on the adjacent and alternate channels of the receiver is evident. To properly test a digital radio receiver, the adjacent-channel power of a sig­nal generator must be below the re-
TUTORIAL
Fig. 10 Intermodulation rejection.
Fig. 11 Phase noise in adjacent-channel selectivity.
TABLE I
MAXIMUM TOLERABLE SSB PHASE NOISE
Analog FM GSM NADC PDC
Channel spacing (kHz) 25 200 30 25 Approximate receiver noise
14 200 35 33
bandwidth (kHz) Adjacent-channel selectivity (dB) 70 9 13 1 Maximum SSB phase noise –121 –72 –68 –56
at offset (dBc/Hz) at 25 kHz at 200 kHz at 30 kHz at 25 kHz Alternate-channel selectivity (dB) 41 42 42 Maximum SSB phase noise
–104 –97 –97
at offset (dBc/Hz) at 400 kHz at 60 kHz at 50 kHz
Fig. 12 A GSM adjacent- and alternate­channel selectivity spectrum.
OUT-OF-CHANNEL
CW TONES
THIRD-ORDER
INTERMODULATION
DISTORTION
DESIRED
SIGNAL
LEVEL (dB)
FREQUENCY
CHANNEL
SPACING
LEVEL (dB)
FREQUENCY
IF REJECTION
CURVE
44
76
9 dB
85
AMPLITUDE (dBm)
f
OFFSET FROM NOMINAL CENTER
41 dB
+200 +400
C
FREQUENCY (kHz)
quired system specification plus the desired test margin.
RADAR
Radar applications have tradition­ally required spectrally clean signal generators. Doppler radars deter­mine the velocity of a target by mea­suring the small Doppler shifts in fre­quency undergone by the return echoes. Return echoes of targets ap­proaching the radar are shifted high­er in frequency than the transmitted carrier, while return echoes of targets moving away from the radar are shift­ed lower in frequency. Unfortunately, the return signal includes much more than just the target echo. In the case of airborne radar, the return echo also includes a large clutter signal that is basically unavoidable frequen­cy-shifted echoes from the ground.
Figure 15 shows the typical re­turn frequency spectrum of airborne
pulsed-Doppler radar. In some situa­tions, the ratio of main-beam clutter to target signal might be as high as 80 dB. This problem is aggravated when the received spectrum has frequency instabilities, specifically 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.
CONCLUSION
As the wireless communications revolution moves forward and the frequency spectrum becomes in­creasingly crowded, the bandwidth requirements for signals become tighter and tighter. Systems must be designed such that only the desired signal is detected in the presence of adjacent-channel signals and other
channel interference. More stringent tests on communications devices must be passed. At the same time, test equipment must also meet these strict requirements. A spectrally pure signal generator complements the other test equipment on a develop­ment engineer’s bench and is highly valued for applications such as LO substitution and receiver testing.
Reference
1. “Testing and Troubleshooting Digital RF Communications Receiver Designs,” Hewlett-Packard Application Note 1314 (Literature # 5968-3579E).
Brian Cheng received his BSEE from the University of California at Berkeley in May
1998. He works at Hewlett-Packard Co. in the Microwave Instruments Division as an applications engineer where he supports the company’s RF and microwave sources.
TUTORIAL
Fig. 14 A PDC adjacent-
and alternate-channel selectivity spectrum.
Fig. 15 An airborne pulsed-Doppler
radar’s typical return frequency spectrum.
65
Fig. 13 An NADC adjacent- and
alternate-channel selectivity spectrum.
94
107
AMPLITUDE (dBm)
OFFSET FROM NOMINAL CENTER
13 dB
f
C
FREQUENCY (kHz)
42 dB
100
AMPLITUDE (dBm)
+60+30
58
1 dB
99
f
+25
C
OFFSET FROM NOMINAL CENTER
FREQUENCY (kHz)
42 dB
+50
OPENING
TARGET
SIDELOBE
CLUTTER
AMPLITUDE
MAIN BEAM
CLUTTER
ATTITUDE
LINE
f
Transmitter carrier
FREQUENCY
LO NOISE
CLOSING
TARGET
NOISE
Loading...