Agilent

Spectrum and Signal Analyzer

Measurements and Noise

Application Note 1303

Measuring Noise and Noise-like Digital

Communications Signals with Spectrum and Signal
Analyzers
 Table of Contents
3 Part I: Noise Measurements
3 Introduction
3 Simple noise—Baseband, Real, Gaussian
3 Bandpassed noise—I and Q
6 Measuring the power of noise with an envelope detector
7 Logarithmic processing
8 Measuring the power of noise with a log-envelope scale
8 Equivalent noise bandwidth
9 The noise marker
10 Spectrum analyzers and envelope detectors
12 Cautions when measuring noise with spectrum and Signal
analyzers

14 Part II: Measurements of Noise-like Signals

14 The noise-like nature of digital signals
14 Channel-power measurements
16 Adjacent-Channel Power (ACP)
16 Carrier power
17 Peak-detected noise and TDMA ACP measurements

18 Part III: Averaging and the Noisiness of Noise

Measurements
18 Variance and averaging
19 Averaging a number of computed results
19 Swept versus FFT analysis
19 Zero span
19 Averaging with an average detector
19 Measuring the power of noise with a power envelope scale
20 The standard deviation of measurement noise
21 Examples
22 The standard deviation of CW measurements

23 Part IV: Compensation for Instrumentation Noise

23 CW signals and log versus power detection
24 Power-detection measurements and noise subtraction
25 Log scale ideal for CW measurements

27 Bibliography

28 Glossary of Terms

2
Part I: Noise Measurements

Introduction The Gaussian PDF explains some of the characteristics of

Noise. It is the classical limitation of electronics. a noise signal seen on a baseband instrument such as an
In measurements, noise and distortion limit the dynamic oscilloscope. The baseband signal is a real signal; it has no
range of test results. imaginary components.

In this four-part paper, the characteristics of noise and its Bandpassed noise—I and Q
direct measurement are discussed in Part I. Part II contains In RF design work and when using spectrum analyzers, we
a discussion of the measurement of noise-like signals usually deal with signals within a passband, such as a com-
exemplified by digital CDMA and TDMA signals. Part III munications channel or the resolution bandwidth (RBW,
discusses using averaging techniques to reduce noise. Part the bandwidth of the final IF) of a spectrum analyzer. Noise
IV is about compensating for the noise in instrumentation in this bandwidth still has a Gaussian PDF, but few RF
while measuring CW (sinusoidal) and noise-like signals. instruments display PDF-related metrics.

Simple noise—Baseband, Real, Gaussian Instead, we deal with a signal’s magnitude and phase
Noise occurs due to the random motion of electrons. The (polar coordinates) or I/Q components. The latter are the
number of electrons involved is large, and their motions are in-phase (I) and quadrature (Q) parts of a signal, or the real
independent. Therefore, the variation in the rate of current and imaginary components of a rectangular-coordinate
flow takes on a bell-shaped curve known as the Gaussian representation of a signal. Basic (scalar) spectrum analyzers
Probability Density Function (PDF) in accordance with the measure only the magnitude of a signal. We are interested
central limit theorem from statistics. The Gaussian PDF is in the characteristics of the magnitude of a noise signal.
shown in Figure 1.

i
i
3
3
2
2

1 1

0 0 t
PDF (i)
–1
–1
–2
–2
–3
–3

Figure 1. The Gaussian PDF is maximum at zero current and falls off away from zero,
as shown (rotated 90 degrees) on the left. A typical noise waveform is shown on the right.

3
 We can consider the noise within a passband as being Spectrum analyzers respond to the magnitude of the signal
made of independent I and Q components, each with within their RBW passband. The magnitude, or envelope, of
Gaussian PDFs. Figure 2 shows samples of I and Q com- a signal represented by an I/Q pair is given by:
ponents of noise represented in the I/Q plane. The sig-
nal in the passband is actually given by the sum of the I venv = √ (vI 2 +vQ2 )
magnitude, vI , multiplied by a cosine wave (at the center
frequency of the passband) and the Q magnitude, vQ , mul- Graphically, the envelope is the length of the vector from
tiplied by a sine wave. But we can discuss just the I and Q the origin to the I/Q pair. It is instructive to draw circles
components without the complications of the sine/cosine of evenly spaced constant-amplitude envelopes on the
waves. samples of I/Q pairs as shown in Figure 3.

3 3

2 2

1 1

0 0

–1 –1

–2 –2

–3 –3
–3 –2 –1 0 1 2 3

–3 –2 –1 0 1 2 3

Figure 2. Bandpassed noise has a Gaussian PDF independently in both its I and Q components.

4
If one were to count the number of samples within each gram. This continuous function is the PDF of the envelope
annular ring in Figure 3, we would see that the area near of bandpassed noise. It is a Rayleigh distribution in the
zero volts does not have the highest count of samples. envelope voltage, v, that depends on the sigma of the sig-
Even though the density of samples is highest there, this nal; for v greater than or equal to 0
area is smaller than any of the other rings.
( ) (
PDF (v) = σv– 2 exp – —
1 v 2
2 (σ )
– )
The count within each ring constitutes a histogram of the
distribution of the envelope. If the width of the rings were The Rayleigh distribution is shown in Figure 4.
reduced and expressed as the count per unit of ring width,
the limit becomes a continuous function instead of a histo-

Q
3

0 I

3
3 2 1 0 1 2 3

Figure 3. Samples of I/Q pairs shown with evenly spaced constant-amplitude envelope circles

PDF(V)

0 V
0 1 2 3 4
σ σ σ σ
Figure 4. The PDF of the voltage of the envelope of a noise signal is a Rayleigh distribution.
The PDF is zero at zero volts, even though the PDFs of the individual I and Q components are
maximum at zero volts. It is maximum for v=sigma.

5
 Measuring the power of noise with an envelope The average envelope voltage is given by integrating the
detector product of the envelope voltage and the probability that
the envelope takes on that voltage. This probability is the
The power of the noise is the parameter we usually want
Rayleigh PDF, so:
to measure with a spectrum analyzer. The power is the
∞
heating value of the signal. Mathematically, it is the time- v– = ∫0 vPDF (v)dv = σ √ π–2
average of v2(t)/R, where R is the impedance and v(t) is the
voltage at time t.
The average power of the signal is given by an analogous
expression with v2 /R in place of the "v" part:
At first glance, we might like to find the average enve-
∞ 2
lope voltage and square it, then divide by R. But finding
the square of the average is not the same as finding the
p– = ∫0 ( v– )PDF
R
2 2σ
(v)dv = –R
average of the square. In fact, there is a consistent under-
We can compare the true power, from the average power
measurement of noise from squaring the average instead of
integral, with the voltage-envelope-detected estimate of
averaging the square; this under-measurement is 1.05 dB
v2 /R and find the ratio to be 1.05 dB, independent of s and
R.
–2
( )
10 lo g v p–/R = 10 lo g π–4 = –1.05 dB ( )
Thus, if we were to measure noise with a spectrum analyzer
using voltage-envelope detection (the linear scale) and
averaging, an additional 1.05 dB would need to be added to
the result to compensate for averaging voltage instead of
voltage-squared.

6
Logarithmic processing between markings is the probability that the log
Spectrum Analyzers are most commonly used in their of the envelope voltage will be within that 1 dB interval.
logarithmic (log) display mode, in which the vertical axis is Figure 6 represents the continuous PDF of a logged signal
calibrated in decibels. Let us look again at our PDF for the which we predict from the areas in Figure 5.
voltage envelope of a noise signal, but let’s mark the x-axis
with points equally spaced on a decibel scale, in this case
with 1 dB spacing. See Figure 5. The area under the curve

PDF (V)

0 V
0 1 2 3 4

Figure 5. The PDF of the voltage envelope of noise is graphed. 1 dB spaced marks on the x-axis
shows how the probability density would be different on a log scale. Where the decibel markings
are dense, the probability that the noise will fall between adjacent marks is reduced.

PDF (V)

X
20 15 10 5 0 5 dB 10

Figure 6. The PDF of logged noise is about 30 dB wide and tilted toward the high end.

7
 Measuring the power of noise with a Equivalent noise bandwidth
log-envelope scale Before discussing the measurement of noise with a spec-
When a spectrum analyzer is in a log (dB) displaymode, trum analyzer noise marker, it is necessary to understand
averaging of the results can occur in numerous ways. the RBW filter of a spectrum analyzer.
Multiple traces can be averaged, the envelope can be aver-
aged by the action of the video filter, or the noise marker The ideal RBW has a flat passband and infinite attenuation
(more on this below) averages results across the x-axis. outside that passband. But it must also have good time
Some recently introduced analyzers also have a detector domain performance so that it behaves well when signals
that averages the signal amplitude for the duration of a sweep through the passband. Most spectrum analyzers use
measurement cell. four-pole synchronously tuned filters for their RBW filters.
We can plot the power gain (the square of the voltage gain)
When we express the average power of the noise in deci- of the RBW filter versus frequency as shown in Figure 7.
bels, we compute a logarithm of that average power. When The response of the filter to noise of flat power spectral
we average the output of the log scale of a spectrum density will be the same as the response of a rectangular
analyzer, we compute the average of the log. The log of filter with the same maximum gain and the same areaunder
the average is not equal to the average of the log. If we go their curves. The width of such a rectangular filter is the
through the same kinds of computations that we did com- equivalent noise bandwidth of the RBW filter. The noise
paring average voltage envelopes with average power density at the input to the RBW filter is given by the output
envelopes, we find that log processing causes an under- power divided by the equivalent noise bandwidth.
response to noise of 2.51 dB, rather than 1.05 dB.1

The log amplification acts as a compressor for large noise

peaks; a peak of ten times the average level is only 10 dB
higher. Instantaneous near-zero envelopes, on the other
hand, contain no power but are expanded toward negative
infinity decibels. The combination of these two aspects of
the logarithmic curve causes noise power to measure lower
than the true noise power.

1. Most authors on this subject artificially state that this factor is due to
1.05 dB from envelope detection and another 1.45 dB from logarithmic
amplification, reasoning that the signal is first voltage-envelope
detected, then logarithmically amplified. But if we were to measure the
voltage-squared envelope (in other words, the power envelope, which
would cause zero error instead of 1.05 dB) and then log it, we would
still find a 2.51 dB under-response. Therefore, there is no real point in
separating the 2.51 dB into two pieces.

8
The ratio of the equivalent noise bandwidth to the –3 dB 1. Under-response due to voltage envelope detection (add
bandwidth (An RBW is usually identified by its –3 dB BW) 1.05 dB) or log-scale response (add 2.51 dB).
is given by the following table:
2. Over-response due to the ratio of the equivalent noise
Filter type Application NBW/–3 dB BW bandwidth to the –3 dB bandwidth (subtract 0.52 dB).

4-pole sync Most SAs analog 1.128 (0.52 dB) 3. Normalization to a 1 Hz bandwidth (subtract 10 times
5-pole sync Some SAs analog 1.111 (0.46 dB) the log of the RBW, where the RBW is given in units
of Hz).
Typical FFT FFT-based SAs 1.056 (0.24 dB)

Most spectrum analyzers include a noise marker that

accounts for the above factors. To reduce the variance of
The noise marker the result, the Agilent 8590 and 8560 families of spectrum
As discussed above, the measured level at the out put of a analyzers compute the average of 32 trace points cen-
spectrum analyzer must be manipulated in order to repre- tered around the marker location. The Agilent ESA family,
sent the input spectral noise density we wish to measure. which allows you to select the number of points in a trace,
This manipulation involves three factors, which may be compute the average over one half of a division centered
added in decibel units: at the marker location. For an accurate measurement, you
must be sure not to place the marker too close to a discrete
spectral component.

The final result of these computations is a measure of the

noise density, the noise in a theoretical ideal 1 Hz band-
width. The units are typically dBm/Hz.

Power gain
1

0.5

0 Frequency
2 1 0 1 2

Figure 7. The power gain versus frequency of an RBW filter can be modeled by a rectangular filter
with the same area and peak level, and a width of the “equivalent noise bandwidth.”

9
 Spectrum analyzers and envelope detectors

display detector
envelope log amp peak
detector S&H A/D
Vin

RBW VBW
sample
LO
resets
processor and display
sweep
generator

Figure A. Simplified spectrum analyzer block diagram

Notice that there is a second set of detectors in the block
A simplified block diagram of a spectrum analyzer is shown diagram: the peak/pit/sample hardware of what is normally
in Figure A. called the detector mode of a spectrum analyzer. These
display detectors are not relevant to this discussion, and
The envelope detector/logarithmic amplifier block is shown should not be confused with the envelope detector.
configured as they are used in the Agilent 8560 E-Series
spectrum analyzers. Although the order of these two cir- The salient features of the envelope detector
cuits can be reversed, the important concept to recognize are two:
is that an IF signal goes into this block and a baseband
signal (referred to as the “video” signal because it was 1. The output voltage is proportional to the input
used to deflect the electron beam in the original analog voltage envelope.
spectrum analyzers) comes out.
2. The bandwidth for following envelope variations
is large compared to the widest RBW.

(a) rms
Vin
x p average
R R 2

Figure B. Detectors: a) half-wave, b) full-wave

implemented as a “product detector,” c) peak.
Practical implementations usually have their gain
(b) rms terms implemented elsewhere, and implement
Vin xp average buffering after the filters that remove the residual
2 2 IF carrier and harmonics. The peak detector must
limiter be cleared; leakage through a resistor or a switch
with appropriate timing are possible clearing
mechanisms.
peak
(c) rms
Vin x1
2

10
Figure B shows envelope detectors and their associated A peak detector may act like an envelope detector in the
waveforms in (a) and (b). Notice that the gain required to limit as its resistive load dominates and the capacitive load
make the average output voltage equal to the r.m.s. voltage is minimized. But practically, the nonideal voltage drop
of a sinusoidal input is different for the different topologies. across the diodes and the heavy required resistive load
make this topology unsuitable for envelope detection. All
Some authors on this topic have stated that “an envelope spectrum analyzers use envelope detectors, some are just
detector is a peak detector.” After all, an idealized detector misnamed.
that responds to the peak of each cycle of IF energy inde-
pendently makes an easy conceptual model of ideal behav-
ior. But real peak detectors do not reset on each IF cycle.
Figure B, part c, shows a typical peak detector with its gain
calibration factor. It is called a peak detector because its
response is proportional to the peak voltage of the signal.
If the signal is CW, a peak detector and an envelope detec-
tor act identically. But if the signal has variations in its
envelope, the envelope detector with the shown LPF (low
pass filter) will follow those variations with the linear, time-
domain characteristics of the filter; the peak detector will
follow nonlinearly, subject to its maximum negative-going
limit, as demonstrated in Figure C. The nonlinearity will
make for unpredictable behavior for signals with noise-like
Figure C. An envelope detector will follow the envelope of the shown
statistical variations. signal, albeit with the delay and filtering action of the LPF used to
remove the carrier harmonics. A peak detector is subject to negative
slew limits, as demonstrated by the dashed line it will follow across
a response pit. This drawing is done for the case in which the
logarithmic amplification precedes the envelope detection, opposite
to Figure A; in this case, the pits of the envelope are especially sharp.

11
 Cautions when measuring noise with If the indicated power with the 1 kHz RBW is –20 dBm at
spectrum and signal analyzers the input mixer (i.e., after the input attenuator), then the
mixer is seeing about +11 dBm. Most spectrum analyzers
There are three ways in which noise measurements can
are specified for –10 dBm CW signals at their input mixer;
look perfectly reasonable on the screen of a spectrum ana-
the level below which mixer compression is specified to be
lyzer, yet be significantly in error.
under 1 dB for CW signals is usually 5 dB or more above
this –10 dBm. The mixer behavior with Gaussian noise is
Caution 1, input mixer level. A noise-like signal of very high
not guaranteed, especially because its peak-to-average
amplitude can overdrive the front end of a spectrum ana-
ratio is much higher than that of CW signals.
lyzer while the displayed signal is within the normal display
range. This problem is possible whenever the bandwidth of
Keeping the mixer power below –10 dBm is a good practice
the noise-like signal is much wider than the RBW. The
that is unlikely to allow significant mixer nonlinearity. Thus,
power within the RBW will be lower than the total power
caution #1 is: Keep the total power at the input mixer at or
by about ten times the log of the ratio of the signal band-
below –10 dBm.
width to the RBW. For example, an IS-95 CDMA signal with
a 1.23 MHz bandwidth is 31 dB larger than the power in a 1
kHz RBW.

output [dB]
ideal log amp

clipping log amp

–10 dB
≈

input [dB]
noise response minus
ideal response average response
+2.0 to noise
error
+1.0 –10 dB
+10 dB

average noise
level re: bottom clipping
≈

average noise
–10 –5
level re: top clipping
[dB]
average response
to noise error –0.5 dB
clipping log amp
–1.0 dB
ideal log amp noise response minus
ideal response

Figure D. In its center, this graph shows three curves: the ideal log amp behavior, that of a log amp that clips
at its maximum and minimum extremes, and the average response to noise subject to that clipping. The lower
right plot shows, on expanded scales, the error in average noise response due to clipping at the positive
extreme. The average level should be kept 7 dB below the clipping level for an error below 0.1 dB. The upper
left plot shows, with an expanded vertical scale, the corresponding error for clipping against the bottom of the
scale. The average level must be kept 14 dB above the clipping level for an error below 0.1 dB.

12
Caution 2, overdriving the log amp. Often, the level dis- Caution 3, underdriving the log amp. The opposite of the
played has been heavily averaged using trace averaging or overdriven log amp problem is the underdriven log amp
a video bandwidth (VBW) much smaller than the RBW. In problem. With a clipping model for the log amp, the results
such a case, instantaneous noise peaks are well above the in the upper left corner of Figure D were obtained. Caution
displayed average level. If the level is high enough that the #3 is: Keep the displayed average log level at
log amp has significant errors for these peak levels, the least 14 dB above the minimum calibrated level
average result will be in error. Figure D shows the error due of the log amp.
to overdriving the log amp in the lower right corner, based
on a model that has the log amp clipping at the top
of its range. Typically, log amps are still close to ideal for a
few dB above their specified top, making the error model
conservative. But it is possible for a log amp to switch
from log mode to linear (voltage) behavior at high levels,
in which case larger (and of opposite sign) errors to those
computed by the model are possible. Therefore, caution #2
is: Keep the displayed average log level at least 7 dB below
the maximum calibrated level of the log amp.

13
 Part II: Measurements of Noise-like Signals

In Part I, we discussed the characteristics of noise and its A typical example is IS-95 CDMA. Performing spectrum
measurement. In this part, we will discuss three different analysis, such as the adjacent-channel power ratio (ACPR)
measurements of digitally modulated signals, after showing test, is usually done using the 30 kHz RBW to observe the
why they are very much like noise. signal. This bandwidth is only one-fortieth of the symbol
clock rate (1.23 Msymbols/s), so the signal in the RBW is
The noise-like nature of digital signals the sum of the impulse responses to about forty pseudo-
Digitally modulated signals can be created by clocking a random digital bits. A Gaussian PDF is an excellent approxi-
Digital-to-Analog Converter (DAC) with the symbols (a mation to the PDF of this signal.
group of bits simultaneously transmitted), passing the DAC
output through apre-modulation filter (to reduce the trans- Channel-power measurements
mitted bandwidth), and then modulating the carrier with Most modern spectrum analyzers allow the measurement
the filtered signal. See Figure 8. The resulting signal is obvi- of the power within a frequency range, called the channel
ously not noise-like if the digital signal is a simple pattern. bandwidth. The displayed result comes from the computa-
It also does not have a noise-like distribution if the band- tion:
width of observation is wide enough for the discrete nature n2
of the DAC outputs to significantly affect the distribution of Pch = – ( )( )
Bs 1
– ∑ 10 (p i /10 )
Bn N i n 1
=
amplitudes.
Pch is the power in the channel, Bs is the specified
But, under many circumstances, especially test conditions, bandwidth (also known as the channel bandwidth), Bn
the digital signal bits are random. And, as exemplified by is the equivalent noise bandwidth of the RBW used, N
the channel power measurements discussed below, the is the number of data points in the summation, pi is the
observation bandwidth is narrow. If the digital update sample of the power in measurement cell i in dB units (if
period (the reciprocal of the symbol rate) is less than one- pi is in dBm, Pch is in milliwatts). n1 and n2 are the end-
fifth the duration of the majority of the impulse response of points for the index i within the channel bandwidth, thus
the resolution bandwidth filter, the signal within the RBW N=(n2 – n1) + 1.
is approximately Gaussian according to the central limit
theorem.

modulated
carrier
DAC filter
digital word
symbol clock ≈
Figure 8. A simplified model for the generation of digital communications signals.

14
The computation works well for CW signals, such as from must be used. Spectrum analyzers have lower effective
sinusoidal modulation. The computation is a power-sum- video bandwidths in sample detection than they do in peak
ming computation. Because the computation changes the detection mode, because of the limitations of the sample-
input data points to a power scale before summing, there and-hold circuit that precedes the A/D converter. Examples
is no need to compensate for the difference between the include the Agilent 8560E-Series spectrum analyzer family
log of the average and the average of the log as explained with 450 kHz effective sample-mode video bandwidth, and
in Part I, even if the signal has a noise-like PDF (probability a substantially wider bandwidth (over 2 MHz) in the Agilent
density function). But, if the signal starts with noise-like ESA-E Series spectrum analyzer family.
statistics and is averaged in decibel form (typically with a
VBW filter on the log scale) before the power summation, Figure 9 shows the experimentally determined relationship
some 2.51 dB under-response, as explained in Part I, between the VBW:RBW ratio and the under-response of
will be incurred. If we are certain that the signal is of noise- the partially averaged logarithmically processed noise sig-
like statistics, and we fully average the signal before per- nal.
forming the summation, we can add 2.51 dB to the result
and have an accurate measurement. Furthermore, the aver- However, the Agilent PSA is an exception to the relation-
aging reduces the variance of the result. ship illustrated by Figure 9. The Agilent PSA allows us to
directly average the signal on a power scale. Therefore, if
But if we don’t know the statistics of the signal, the best we are not certain that our signal is of noise-like statistics,
measurement technique is to do no averaging before power we are no longer prohibited from averaging before power
summation. Using a VBW ≥ 3RBW is required for insignifi- summation. The measurement may be taken by either using
cant averaging, and is thus recommended. But the band- VBW filtering on a power scale, or using the average detec-
width of the video signal is not as obvious as it appears. tor on a power scale.
In order to not peak-bias the measurement, the detector

0 0.3 1 3 10 30 ∞
≈
≈

0
0.045 dB RBW/VBW ratio
0.35 dB
–1.0

–2.0
power summation
≈

–2.5
error
1,000,000 point simulation
experiment

Figure 9. For VBW ≥ 3 RBW, the averaging effect of the VBW filter does not significantly affect
power-detection accuracy.

15
 Adjacent-Channel Power (ACP) Carrier power
There are many standards for the measurement of ACP Burst carriers, such as those used in TDMA mobile sta-
with a spectrum analyzer. The issues involved in most tions, are measured differently than continuous carriers.
ACP measurements are covered in detail in an article in The power of the transmitter during the time it is on is
Microwaves & RF, May, 1992, "Make Adjacent-Channel called the "carrier power."
Power Measurements." A survey of other standards is
available in "Adjacent Channel Power Measurements in the Carrier power is measured with the spectrum analyzer
Digital Wireless Era" in Microwave Journal, July, 1994. in zero span. In this mode, the LO of the analyzer does
not sweep, thus the span swept is zero. The display then
For digitally modulated signals, ACP and channel-power shows amplitude normally on the y axis, and time on the x
measurements are similar, except ACP is easier. ACP is axis. If we set the RBW large compared to the bandwidth
usually the ratio of the power in the main channel to the of the burst signal, then all of the display points include all
power in an adjacent channel. If the modulation is digital, of the power in the channel. The carrier power is computed
the main channel will have noise-like statistics. Whether simply by averaging the power of all the display points that
the signals in the adjacent channel are due to broadband represent the times when the burst is on. Depending on the
noise, phase noise, or intermodulation of noise-like signals modulation type, this is often considered to be any point
in the main channel, the adjacent channel will have noise- within 20 dB of the highest registered amplitude. (A trig-
like statistics. A spurious signal in the adjacent channel ger and gated spectrum analysis may be used if the carrier
is most likely modulated to appear noise-like, too, but a power is to be measured over a specified portion of a burst-
CW-like tone is a possibility. RF signal.)

If the main and adjacent channels are both noise-like, Using a wide RBW for the carrier-power measurement
then their ratio will be accurately measured regardless of means that the signal will not have noise-like statistics. It
whether their true power or log-averaged power (or any will not have CW-like statistics, either, so it is still wise to
partially averaged result between these extremes) is mea- set the VBW as wide as possible. But let’s consider some
sured. Thus, unless discrete CW tones are found in the examples to see if the sample-mode bandwidths
signals, ACP is not subject to the cautions regarding VBW of spectrum analyzers are a problem.
and other averaging noted in the section on channel power
above. For PDC, NADC and TETRA, the symbol rates are under
25 kb/s, so a VBW set to maximum will work well. It will
But some ACP standards call for the measurement of abso- also work well for PHS and GSM, with symbol rates of 380
lute power, rather than a power ratio. In such cases, the and 270 kb/s. For IS-95 CDMA, with a modulation rate of
cautions about VBW and other averaging do apply. 1.23 MHz, we could anticipate a problem with the 450 kHz
effective video bandwidth discussed in the section on chan-
nel power above. Experimentally, an instrument with 450
kHz BW experienced a 0.6 dB error with an OQPSK (mobile)
burst signal.

16
Peak-detected noise and TDMA ACP Tau (t) is the observation period, usually given by either the
measurements length of an RF burst, or by the spectrum analyzer sweep
time divided by the number of cells in a sweep. BWi is the
TDMA (time-division multiple access, or burst-RF) sys-
impulse bandwidth of the RBW filter.
tems are usually measured with peak detectors, in order
that the burst "off" events are not shown on the screen
For the four-pole synchronously tuned filters used in most
of the spectrum analyzer, potentially distracting the user.
spectrum analyzers, BWi is nominally 1.62 times the –3 dB
Examples include ACP measurements for PDC (Personal
bandwidth. For ideal linear-phase Gaussian filters, which
Digital Cellular) by two different methods, PHS (Personal
is an excellent model for digitally implemented swept ana-
Handiphone System) and NADC (North American Dual-
lyzers, BWi is 1.499 times the –3 dB bandwidth. In either
mode Cellular). Noise is also often peak detected in the
case, VBW filtering can substantially reduce the impulse
measurement of rotating media, such as hard disk drives
bandwidth.
and VCRs.

The peak of noise will exceed its power average by an Note that vpk is a "power average" result; the average of
amount that increases (on average) with the length of time the log of the ratio will be different.
over which the peak is observed. A combination of analysis,
approximation and experimentation leads to this equation for The graph in Figure E shows a comparison of this equation
vpk, the ratio of the average power of peak measurements with some experimental results. The fit of the experimental
to the average power of sampled measurements: results would be even better if 10.7 dB were used in place
of 10 dB in the equation above, even though analysis does
vpk = [10 dB] log 10 [loge (2  τ BW i +e)] not support such a change.

12

10

8
Peak:
average
ratio, dB 6

0
0.01 0.1 1 10 100 1000 104

t C RBW
Figure E. The peak-detected response to noise increases with the observation time.

17
 Part III: Averaging and the Noisiness of Noise Measurements

The results of measuring noise-like signals are, not surpris- If we were to measure the standard deviation of logged
ingly, noisy. Reducing this noisiness is accomplished by envelope noise, we would find that s is 5.57 dB. Thus, the
three types of averaging: s of a channel-power measurement that averaged log data
• increasing the averaging within each measurement over, for example, 100 measurement cells would be 0.56
cell of a spectrum analyzer by reducing the VBW, or dB (5.6/√(100)). But averaging log data not only causes
using an average detector with a longer sweeptime. the aforementioned 2.51 dB under-response, it also has a
• increasing the averaging within a computed result like higher than desired variance. Those not-rare-enough nega-
channel power by increasing the number of tive spikes of envelope, such as –30 dB, add significantly to
measurement cells contributing to the result. the variance of the log average even though they represent
• averaging a number of computed results. very little power. The variance of a power measurement
made by averaging power is lower than that made by aver-
Variance and averaging aging the log of power by a factor of 1.64.
The variance of a result is defined as the square of
its standard deviation; therefore it is symbolically s2. The Thus, the s of a channel-power measurement is lower than
variance is inversely proportional to the number of indepen- that of a log-averaged measurement by a factor of the
dent results averaged, thus when N results are combined, square root of this 1.64:
the variance of the final result is s2 /N.

The variance of a channel-power result computed from

independent measurement cells is likewise s2 /N where s is
the variance of a single measurement cell. But this s2 is a σ noise = 4.35 dB/√ N [power averaging]
very interesting parameter. σ noise = 5.57 dB/√ N [log processing]

18
Averaging a number of computed results Zero span is sometimes used for other noise and noise-like
If we average individual channel-power measurements to get measurements where the noise bandwidth is much greater
a lower-variance final estimate, we do not have to convert than the RBW, such as in the measurement of power spec-
dB-format answers to absolute power to get the advantages tral density. For example, some companies specify IS-95
of avoiding log averaging. The individual measurements, CDMA ACPR measurements that are spot-frequency power
being the results of many measurement cells summed spectral density specifications; zero span can be used to
together, no longer have a distribution like the "logged speed this kind of measurement.
Rayleigh" but rather look Gaussian. Also, their distribution
is sufficiently narrow that the log (dB) scale is linear Averaging with an average detector
enough to be a good approximation of the power scale. With an averaging detector the amplitude of the signal
Thus, we can dB-average our intermediate results. envelope is averaged during the time and frequency interval
of a measurement cell. An improvement over using sample
Swept versus FFT analysis detection for summation, the average detector changes the
In the above discussion, we have assumed that the vari- summation over a range of cells into integration over the
ance reduced by a factor of N was of independent results. time interval representing a range of frequencies. The inte-
This independence is typically the case in swept-spectrum gration thereby captures all power information, not just that
analyzers, due to the time required to sweep from one mea- sampled by the sample detector.
surement cell to the next under typical conditions of span,
RBW and sweep time. FFT analyzers will usually have many The primary application of average detection may be seen
fewer independent points in a measurement across a chan- in the channel power and ACP measurements, discussed in
nel bandwidth, reducing, but not eliminating, their theoreti- Part II.
cal speed advantage for true noise
signals. Measuring the power of noise with a power
envelope scale
For digital communications signals, FFT analyzers have The averaging detector is valuable in making integrated
an even greater speed advantage than their throughput power measurements. The averaging scale, when autocou-
predicts. Consider a constant-envelope modulation, such pled, is determined by such parameters as the marker func-
as used in GSM cellular phones. The constant-envelope tion, detection mode and display scale. We have discussed
modulation means that the measured power will be con- circumstances that may require the use of the log-envelope
stant when that power is measured over a bandwidth wide and voltage envelope scales, now we may consider the
enough to include all the power. FFT analysis made in a power scale.
wide span will allow channel power measurements with
very low variance. When making a power measurement, we must remember
that traditional swept spectrum analyzers average the log of
But swept analysis will typically be performed with an the envelope when the display is in log mode. As previously
RBW much narrower than the symbol rate. In this case, mentioned, the log of the average is not equal to the aver-
the spectrum looks noise-like, and channel power measure- age of the log. Therefore, when making power measure-
ments will have a higher variance that is not influenced by ments, it is important to average the power of the signal,
the constant amplitude nature of the modulation. or equivalently, to report the root of the mean of the square
(r.m.s.) number of the signal. With the Agilent PSA ana-
Zero span lyzer, an "Avg/VBW Type" key allows for manual selection,
A zero-span measurement of carrier power is made with as well as automatic selection, of the averaging scale (log
a wide RBW, so the independence of data points is deter- scale, voltage scale, or power scale). The averaging scale
mined by the symbol rate of the digital modulation. Data and display scale may be completely independent of each
points spaced by a time greater than the symbol rate will other.
be almost ompletely independent.

19
 The standard deviation of measurement noise The left region applies whenever the integration time is
Figure 10 summarizes the standard deviation of the mea- short compared to the rate of change of the noise envelope.
surement of noise. The figure represents the standard As discussed above, without VBW filtering, the s is 5.6 dB.
deviation of the measurement of a noise-like signal using When video filtering is applied, the standard deviation is
a spectrum analyzer in zero span, averaging the results improved by a factor. That factor is the square root of the
across the entire screen width, using the log scale. tINT is ratio of the two noise bandwidths: that of the video band-
the integration time (sweep time). The curve is also useful width, to that of the detected envelope of the noise. The
for swept spectrum measurements, such as channel-power detected envelope of the noise has half the noise band-
measurements. There are three regions to the curve. width of the undetected noise. For the four-pole synchro-
nously tuned filters typical of most spectrum analyzers, the
detected envelope has a noise bandwidth of (1/2) x 1.128
times the RBW. The noise bandwidth of a single-pole VBW
filter is π/2 times its bandwidth. Gathering terms together
yields the equation:
σ = (9.3 dB ) √ VBW/RBW

σ left asymptote: for VBW >1/3 RBW: 5.6 dB

for VBW ≤ 1/3 RBW: 9.3 dB VBW
RBW
5.6 dB VBW = ∞

5.2 dB right asymptote:

1.0 dB
VBW = 0.03 . RBW center curve: t . RBW [left asymptote]
INT
Ncells

N=400
N=600

0.1 dB
N=600,VBW=0.03 . RBW

≈ Average detector, any N

. RBW
≈

tINT
1.0 10 100 1k 10k

Figure 10. Noise measurement standard deviation for log-response (see text for power-response) spectrum analysis
depends on the product of the sweep time and RBW, the ratio of the VBW to RBW, and the number of display cells.

20
The middle region applies whenever the envelope of the will prevent power detection and lead to about a 2.3 dB
noise can move significantly during the integration time, under-response (see Figure 9) for which we must manually
but not so rapidly that individual sample points become correct. The sweep time will be 84 s. With the channel
uncorrelated. In this case, the integration behaves as a taking up one-fourth of the span, the sweep time within
noise filter with frequency response of sin (π tINT ) and an the channel is 21 s, so that is the integration time for our
equivalent noise bandwidth of 1/(2 tINT ). The total noise x-axis. Even though the graph is meant for zerospan analy-
should then be 5.6 dB times the square root of the ratio of sis, if the noise level is flat in our channel, the analysis is
the noise bandwidth of the integration process to the noise the same for swept as zerospan. t INT RBW= 6300; if the
bandwidth of the detected envelope, giving center of Figure 10 applied, sigma would be 0.066 dB.
Checking the right asymptote, Ncells is 150, so the asymp-
5.2 dB /√ t INT RBW tote computes to be 0.083 dB. This is our predicted standard
deviation. If the noise in the adjacent channel is not flat, the
In the right region, the sweep time of the spectrum ana- averaging will effectively extend over many fewer samples
lyzer is so long that individual measurement cells, mea- and less time, giving a higher standard deviation.
sured with the sample detector, are independent of each
other. Information about the signal between these samples In a third example, let’s measure W-CDMA channel power
is lost, increasing the sigma of the result. In this case, the in a 3.84 MHz width. We’ll set the span to be the same
standard deviation is reduced from that of the left-side 3.84 MHz width. Let’s use RBW=100 kHz, and set the
case (the sigma of an individual sample) by the square sweep time long (600 ms) with a 600-point analyzer, using
root of the number of measurement cells in a sweep. But the average detector on a power scale. Assume that the
in an analyzer using a detector that averages continuously spectrum is approximately flat. We are making a measure-
across a measurement cell, no information is lost, so the ment that is equivalent to a 600 ms integration time with
center curve extends across the right side of the graph an unlimited number of analyzer points, because the aver-
indefinitely. age detector integrates continuously within the buckets. So
we need only use the formula from the center of the graph;
The noise measurement sigma graph should be multiplied the cell-count-limited asymptote on the right does not
by a factor of about 0.8 if the noise power is filtered and apply. t INT is 600 ms, so the center formula gives sigma =
averaged, instead of the log power being so processed. 0.021 dB. But we are power-scale averaging, not log aver-
(Sigma goes as the square root of the variance, which aging, so the sigma is 20% lower, 0.017 dB.
improves by the cited 1.64 factor.) Because channel-power
and ACP measurements are power-scale summations, this Alternatively, we could think of example 3 as 600 individual
factor applies. However, when dealing with VBW-filtered one-measurement-cell readings that are then summed
measurements, this factor may or may not be valid. Most together. Each measurement cell would have an integration
spectrum analyzers average VBW-filtered measurements time of 1 ms. The center formula would give sigma = 0.52
on a log scale in which case the multiplication factor would dB on a log scale, or 0.412 dB for power averaging. The
not apply. In comparison, the Agilent PSA allows VBW- standard deviation of the sum of the power in the 600 cells
filtering on a power scale, making the multiplication factor would be lower than that of one cell by the square root of
applicable for such measurements. 600, giving the same 0.017 dB result for the entire channel
power measurement.
Examples
Let’s use the curve in Figure 10 for three examples.
In the measurement of IS-95 CDMA ACPR, we can power-
average a 400-point zero-span trace for a frame (20.2 ms)
in the specified 30 kHz bandwidth. Power averaging can be
accomplished in all analyzers by selecting VBW >>RBW. For
these conditions, we find t INT RBW = 606, and we approach
the right-side asymptote of or 0.28 dB. But we are power
averaging, so we multiply by 0.8 to get sigma = 0.22 dB.

In a second example, we are measuring noise in an adja-

cent channel in which the noise spectrum is flat. Let’s use
a 600-point analyzer with a span of 100 kHz and a channel
BW of 25 kHz, giving 150 points in our channel. Let’s use
an RBW of 300 Hz and a VBW = 10 Hz; this narrow VBW
21
 The standard deviation of CW measurements We can expand the log in a Taylor series and
CW signals have a variance due to added noise within the generalize this formula as:
resolution bandwidth. That noise can be decomposed into S/N+3.01
–
two components: one component is in phase with the CW σCW = 8.69 · 10 20

signal, and one component is in quadrature.

In this equation, the units of the signal-to-noise ratio, S/N,
Let’s make the assumption that the signal to noise ratio is and of the result, are decibels. VBW filtering, trace averag-
large. Then the quadrature noise does not change the mea- ing, noise marker averaging or the average detector can all
sured result for the CW signal. But the in-phase component reduce the sigma.
adds to or subtracts from the signal voltage vector.
In this equation, the units of the signal-to-noise ratio, S/N,
The interfering noise vector is Gaussian in both its in-phase and the result, σCW, are decibels.
and quadrature components. The power of the noise vec-
tor is the sum of the variances of the two components. The sigma can be reduced by filtering, such as VBW filter-
Therefore, the variance of the in-phase component is ing, or averaging. The bandwidth of the envelope modula-
half of the power of the noise signal. Let’s use a numeric tion noise represented by sigma is approximately one-half
example. of the noise bandwidth of the RBW filter. The noise
bandwidth of a VBW filter is π/2 times its bandwidth.
Let the noise power be 20 dB below the signal power. Then Therefore, if the VBW is under about one-third of the RBW,
the variance of the noise is 1% of the signal power. The the sigma is improved by the square root of the noise band-
in-phase variance is 0.5% of the signal power. Expressed in widths:
voltage, the in-phase noise is 0.0707 times the CW signal.
With this Gaussian noise of 0.0707 times as large a signal
riding on the apparent CW voltage vector length, its sigma π
– · VBW

√
becomes 20*log(1 + 0.0707) in decibels, or 0.59 dB. 2 –
S/N+3.01
σCW = 1 · 8.69 · 10 20

– · NBWRBW
2

For time averaging, such as that which occurs with the

average detector, the noise bandwidth of the averaging
process is 1/(2xtINT ), where tINT is the integration time.
Therefore, if the integration time is longer than about 1/
RBW, sigma is:

S/N+3.01
1 –
σCW =
√ · 8.69 · 10 20
1
2 · t INT · –· NBWRBW
2

22
Part IV: Compensation for Instrumentation Noise

In Parts I, II and III, we discussed the measurement of a CW signal with equal power within the RBW by 2.5
noise and noise-like signals respectively. In this part, we’ll dB. But to the first order, the noise doesn’t even affect
discuss measuring CW and noise-like signals in the pres- the S+N measurement! See "Log Scale Ideal for CW
ence of instrumentation noise. We’ll see why averaging Measurements" later in this section.
the output of a logarithmic amplifier is optimum for CW
measurements, and we’ll review compensation formulas for Figure 11 demonstrates the improvement in CW measure-
removing known noise levels from noise-plus-signal ment accuracy when using log averaging versus power
measurements. averaging.

CW signals and log versus power detection To compensate S+N measurements on a log scale
When measuring a single CW tone in the presence of for higher-order effects and very high noise levels, use this
noise, and using power detection, the level measured is equation where all terms are in dB units:
equal to the sum of the power of the CW tone and the powercw = powers+n –10.42x10 –0.333(deltaSN)
power of the noise within the RBW filter. Thus, we could
improve the accuracy of a measurement by measuring powerS+N is the observed power of the signal with noise,
the CW tone first (let’s call this the "S+N" or signal-plus- deltaSN is the decibel difference between the S+N and
noise), then disconnect the signal to make the "N" measure- N-only measurements. With this compensation, noise-
ment. The difference between the two, with both measure- induced errors are under 0.25 dB even for signals as small
ments in power units (for example, milliwatts, not dBm) as 9 dB below the interfering noise. Of course, in such a
would be the signal power. situation, the repeatability becomes a more important
concern than the average error. But excellent results can
But measuring with a log scale and video filtering or be obtained with adequate averaging. And the process of
video averaging results in unexpectedly good results. As averaging and compensating, when done on a log scale,
described in Part I, the noise will be measured lower than converges on the result much faster than when done in a
power-detecting environment.

23
 2.54 dB
0.63 dB

a.) b.) c.)

2.51 dB

Figure 11. Log averaging improves the measurement of CW signals when their amplitude is near
that of the noise. (a) shows a noise-free signal. (b) shows an averaged trace with power-scale
averaging and noise power 1 dB below signal power; the noise-induced error is 2.5 dB. (c) shows
the effect with log-scale averaging—the noise falls 2.5 dB and the noise-induced error falls to
only 0.6 dB.

Power-detection measurements and noise The power equation also applies when the signal and the
subtraction noise have different statistics (CW and Gaussian respec-
If the signal to be measured has the same statistical tively) but power detection is used. The power equation
distribution as the instrumentation noise— in other words, would never apply if the signal and the noise were cor-
if the signal is noise-like—then the sum of the signal and related, either in-phase adding or subtracting. But that will
instrumentation noise will be a simple power sum: never be the case with noise.

power S+N = power S + power N [mW] Therefore, simply enough, we can subtract the mea-
sured noise power from any power-detected result to get
Note that the units of all variables must be power units improved accuracy. Results of interest are the channel-pow-
such as milliwatts and not log units like dBm, nor voltage er, ACP, and carrier-power measurements described in Part
units like mV. Note also that this equation applies even if II. The equation would be:
powerS and powerN are measured with log averaging. power S = power S+N – power N [mW]

Care should be exercised that the measurement setups for

and are as similar as possible.

24
Log scale ideal for CW measurements The average response to the signal plus the quadrature
If one were to design a scale (such as power, voltage, log noise component is the response to a signal of magnitude
power, or an arbitrary polynomial) to have response to √ 1+ x2
signal-plus-noise that is independent of small amounts of
noise, one could end up designing the log scale. The average response to the signal plus in-phase noise
will be lower than the response to a signal without noise
Consider a signal having unity amplitude and arbitrary if the chosen scale is compressive. For example, let x be
phase, as in Figure F. Consider noise with an amplitude ±0.1 and the scale be logarithmic. The response for x =
much less than unity, r.m.s., with random phase. Let us +0.1 is log (1.1); for x = –0.1, log (0.9). The mean of these
break the noise into components that are in-phase and two is 0.0022, also expressible as log (0.9950). The mean
quadrature to the signal. Both of these components will response to the quadrature components is log(√2(1+(0.1)2)),
have Gaussian PDFs, but for this simplified explanation, we or log (1.0050). Thus, the log scale has an average deviation
can consider them to have values of ±x, where x << 1. for in-phase noise that is equal and opposite to the devia-
tion for quadrature noise. To first order, the log scale is
noise-immune. Thus, an analyzer that averages (for exam-
ple, by video filtering) the response of a log amp to the sum
of a CW signal and a noise signal has no first-order depen-
dence on the noise signal.
Q

+jx
+x

–x –jx
I

Figure F. Noise components can be projected into in-phase and

quadrature parts with respect to a signal of unity amplitude and
arbitrary phase.

25
 Figure G shows the average error due to noise addition for
signals measured on the log scale and, for comparison, for
signals measured on a power scale.

Error 5
[dB]
4

power summation
2

1 log scale

0 S/N ratio [dB]

2 0 2 4 6 8 10

Figure G. CW signals measured on a logarithmic scale show very little effect due to the addition of
noise signals.

26
Bibliography

1. Nutting, Larry. Cellular and PCS TDMA Transmitter 4. Ballo, David and Gorin, Joe. Adjacent Channel Power
Testing with a Spectrum Analyzer. Agilent Wireless Measurements in the Digital Wireless Era, Microwave
Symposium, February, 1992. Journal, July 1994, pp 74-89.

2. Gorin, Joe. Make Adjacent Channel Power 5. Peterson, Blake. Spectrum Analysis Basics. Agilent
Measurements, Microwaves & RF, May 1992, Application Note 150, literature part number 5952-0292,
pp 137-143. November 1, 1989.

3. Cutler, Robert. Power Measurements on Digitally 6. Moulthrop, Andrew A. and Muha, Michael S. Accurate
Modulated Signals. Hewlett-Packard Wireless Measurement of Signals Close to the Noise Floor on a
Communications Symposium, 1994. Spectrum Analyzer, IEEE Transactions on Microwave
Theory and Techniques, November 1991, pp. 1182-1885.

27
 Glossary of Terms

ACP: See Adjacent Channel Power. DAC: Digital to Analog Converter.

ACPR: Adjacent Channel Power Ratio. See Adjacent- Digital: Signals that can take on only a prescribed list of
Channel Power; ACPR is always a ratio, whereas ACP may values, such as 0 and 1.
be an absolute power.
Display detector: That circuit in a spectrum analyzer that
Adjacent Channel Power: The power from a modulated converts a continuous-time signal into sampled data points
communications channel that leaks into an adjacent chan- for displaying. The bandwidth of the continuous-time signal
nel. This leakage is usually specified as a ratio to the power often exceeds the sample rate of the display, so display
in the main channel, but is sometimes an absolute power. detectors implement rules, such as peak detection, for
sampling.
Averaging: A mathematical process to reduce the variation
in a measurement by summing the data points from mul- Envelope Detector: The circuit that derives an instanta-
tiple measurements and dividing by the number of points neous estimate of the magnitude (in volts) of the IF (inter-
summed. mediate frequency) signal. The magnitude is often called
the envelope.
Burst: A signal that has been turned on and off. Typically,
the on time is long enough for many communications bits Equivalent Noise Bandwidth: The width of an ideal filter
to be transmitted, and the on/off cycle time is short enough with the same average gain to a white noise signal as the
that the associated delay is not distracting to telephone described filter. The ideal filter has the same gain as the
users. maximum gain of the described filter across the equivalent
noise bandwidth, and zero gain outside that bandwidth.
Carrier Power: The average power in a burst carrier during
the time it is on. Gaussian and Gaussian PDF: A bell-shaped PDF which is
typical of complex random processes. It is characterized by
CDMA: Code Division Multiple Access or a communica- its mean (center) and sigma (width).
tions standard (such as cdmaOne®) that uses CDMA.
In CDMA modulation, data bits are xored with a code I and Q: In-phase and Quadrature parts of a complex signal.
sequence, increasing their bandwidth. But multiple users I and Q, like x and y, are rectangular coordinates; alterna-
can share a carrier when they use different codes, and a tively, a complex signal can be described by its magnitude
receiver can separate them using those codes. and phase, also known as polar coordinates.

Channel Bandwidth: The bandwidth over which power is Linear scale: The vertical display of a spectrum analyzer in
measured. This is usually the bandwidth in which almost all which the y axis is linearly proportional to the voltage enve-
of the power of a signal is contained. lope of the signal.

Channel Power: The power contained within a channel

bandwidth.

Clipping: Limiting a signal such that it never exceeds some

threshold.

CW: Carrier Wave or Continuous Wave. A sinusoidal signal

without modulation.

28
NADC: North American Dual mode (or Digital) Cellular. Rayleigh: A well-known PDF which is zero at x=0 and
A communications system standard, designed for North approaches zero as x approaches infinity.
American use, characterized by TDMA digital modulation.
RBW filter: The resolution bandwidth filter of a spectrum
Near-noise Correction: The action of subtracting the mea- analyzer. This is the filter whose selectivity determines
sured amount of instrumentation noise power from the the analyzer’s ability to resolve (indicate separately) closely
total system noise power to calculate that part from the spaced signals.
device under test.
Reference Bandwidth: See Specified Bandwidth.
Noise Bandwidth: See Equivalent Noise Bandwidth.
RF: Radio Frequency. Frequencies that are used for radio
Noise Density: The amount of noise within a defined band- communications.
width, usually normalized to 1 Hz.
Sigma: The symbol and name for standard deviation.
Noise Marker: A feature of spectrum analyzers that allows
the user to read out the results in one region of a trace Sinc: A mathematical function. Sinc(x) = (sin(x))/x.
based on the assumption that the signal is noise-like. The
marker reads out the noise density that would cause the Specified Bandwidth: The channel bandwidth
indicated level. specified in a standard measurement technique.

OQPSK: Offset Quadrature-Phase Shift Keying. A digital Standard Deviation: A measure of the width of the distri-
modulation technique in which symbols (two bits) are bution of a random variable.
represented by one of four phases. In OQPSK, the I and Q
Symbol: A combination of bits (often two) that are trans-
transitions are offset by half a symbol period.
mitted simultaneously.
PDC: Personal Digital Cellular (originally called Japanese
Symbol Rate: The rate at which symbols are
Digital Cellular). A cellular radio standard much like NADC,
transmitted.
originally designed for use in Japan.
Synchronously Tuned Filter: The filter alignment most com-
PDF: See Probability Density Function.
monly used in analog spectrum analyzers. A sync-tuned
Peak Detect: Measure the highest response within an filter has all its poles in the same place. It has an excellent
observation period. tradeoff between selectivity and time-domain performance
(delay and step-response settling).
PHS: Personal Handy-Phone. A communications standard
for cordless phones. TDMA: Time Division Multiple Access. A method of shar-
ing a communications carrier by assigning separate time
Power Detection: A measurement technique in which the slots to individual users. A channel is defined by a carrier
response is proportional to the power in the signal, or pro- frequency and time slot.
portional to the square of the voltage.
TETRA: Trans-European Trunked Radio. A communications
Power Spectral Density: The power within each unit of system standard.
frequency, usually normalized to 1 Hz.
Variance: A measure of the width of a distribution, equal to
Probability Density Function: A mathematical function that the square of the standard deviation.
describes the probability that a variable can take on any
particular x-axis value. The PDF is a continuous version of VBW Filter: The Video Bandwidth filter, a low-pass filter
a histogram. that smoothes the output of the detected IF signal, or the
log of that detected signal.
Q: See I and Q.
Zero Span: A mode of a spectrum analyzer in which the
local oscillator does not sweep. Thus, the display represents
amplitude versus time, instead of amplitude versus frequen-
cy. This is sometimes called fixed-tuned mode.

29
 www.agilent.com

For more information on Agilent

Remove all doubt Technologies’ products, applications
or services, please contact your local
www.agilent.com/find/emailupdates Agilent office. The complete list is
Our repair and calibration services
Get the latest information on the available at:
will get your equipment back to
products and applications you select. www.agilent.com/find/contactus
you, performing like new, when
promised. You will get full value out
of your Agilent equipment through- Americas
out its lifetime. Your equipment Canada (877) 894-4414
will be serviced by Agilent-trained Latin America 305 269 7500
technicians using the latest factory United States (800) 829-4444
calibration procedures, automated
Asia Pacific
repair diagnostics and genuine parts.
Australia 1 800 629 485
You will always have the utmost
China 800 810 0189
confidence in your measurements.
Hong Kong 800 938 693
For information regarding self main- India 1 800 112 929
tenance of this product, please Japan 0120 (421) 345
contact your Agilent office. Korea 080 769 0800
Malaysia 1 800 888 848
Agilent offers a wide range of additional Singapore 1 800 375 8100
expert test and measurement services Taiwan 0800 047 866
for your equipment, including initial Thailand 1 800 226 008
start-up assistance, onsite education
and training, as well as design, system Europe & Middle East
integration, and project management. Austria 01 36027 71571
Belgium 32 (0) 2 404 93 40
For more information on repair and Denmark 45 70 13 15 15
calibration services, go to: Finland 358 (0) 10 855 2100
France 0825 010 700*
www.agilent.com/find/removealldoubt
*0.125 €/minute
Germany 07031 464 6333
Ireland 1890 924 204
Israel 972-3-9288-504/544
Italy 39 02 92 60 8484
Netherlands 31 (0) 20 547 2111
Spain 34 (91) 631 3300
Sweden 0200-88 22 55
Switzerland 0800 80 53 53
United Kingdom 44 (0) 118 9276201
Other European Countries:
www.agilent.com/find/contactus
Revised: July 2, 2009

Product specifications and descriptions

in this document subject to change
without notice.

© Agilent Technologies, Inc. 2009

Printed in USA, September 16, 2009
5966-4008E