Agilent AN 1315

Optimizing RF and Microwave

Spectrum Analyzer Dynamic
Range
Application Note
 Table of Contents

3 1. Introduction
3 What is dynamic range?
3 Why is dynamic range important?

4 2. Dynamic Range Interpretations

4 Measurement range
4 Display range
5 Mixer compression
6 Internal distortion
7 Noise
7 Sensitivity
7 Phase noise
7 Second- and third-order dynamic range
9 Summary

10 3. Making Harmonic or Intermodulation

Distortion Measurements
10 Measurement uncertainty
11 Optimizing measurements
11 Preamplifier
12 Atttenuator
12 External filter
13 RBW filter
13 Measuring other signals

14 4. Summary

15 5. References

2
1. Introduction

What is dynamic range? the presence of large signals and therefore is one
The dynamic range of a spectrum analyzer is tradi- of the most important performance figures for a
tionally defined as the ratio, in dB, of the largest spectrum analyzer. It is often misunderstood and
to the smallest signals simultaneously present at misinterpreted, since the display range, measure-
the input of the spectrum analyzer that allows ment range, noise floor, phase noise, and spurious
measurement of the smaller to a given degree of response of the instrument all play important roles
uncertainty. The signals of interest can either be in determining dynamic range. By understanding
harmonically or nonharmonically related. which dynamic range interpretation applies to a
specific measurement, you can make more accu-
rate, reliable, and repeatable spectrum analyzer
Why is dynamic range important?
measurements.
The dynamic range specification determines
whether or not low-level signals will be visible in

+30 dBm Maximum power level

0 dBm Mixer compression

–37.5 dBm Third-order distortion

CRT-Display
range Compression/noise
range= –45 dBm Second-order distortion
85 dB/120 dB
135 dB
Signal/third-order 0 dBm Phase noise
Measurement
distortion range=
range=
97 dB
181 dB
Signal/second-order Signal/phase noise
distortion range= –90 dBc/10 kHz offset
90 dB range=80 dB, 10 Hz RBW

–135 dBm = Displayed Average Noise Level (DANL) without preamp

–151 dBm = DANL with preamp

Figure 1. Dynamic range interpretations

3
2. Dynamic Range Interpretations

Figure 1 shows several different interpretations for Display range

dynamic range. In this chapter we will explain The display range is the calibrated amplitude range
each interpretation. of the display. If an analyzer display has 10 vertical
divisions, it is usually assumed that in the 10 dB/
Measurement range division logarithmic display mode the analyzer is
The measurement range is the difference between able to display signals 100 dB apart. However, the
the largest and smallest signals that can be meas- log amplifier often limits the range. For example, if
ured on a spectrum analyzer, allowing for different we have an 85 dB log amplifier and 10 divisions on
instrument settings. Usually, the maximum power our screen, we only have 8.5 calibrated divisions.
level that can be applied to the input of the analyzer An analyzer’s reference level setting positions the
without damaging the front-end hardware dictates display range anywhere within the measurement
the largest signal; this is +30 dBm (1 watt) for most range.
analyzers. The noise floor of the instrument deter-
mines the lower limit of the measurement range; In some spectrum analyzers, the narrow resolution
signals below the noise level are not visible on bandwidths are implemented with digital filters.
screen. Note that the lowest noise level cannot be The log amplifier in this case is bypassed and the
achieved with the same input attenuator setting limitation now comes from the digital-to-analog
used for a 1 watt measurement. converter plus any auto-ranging. A calibrated dis-
play range in this case can be as much as 120 dB.

Hint: optimize using reference level

If small signals need to be measured in the presence of
larger signals, the larger signals can often be moved
above the upper limit of the display by up to 10 dB by
changing the reference level, to place the small signals in
the calibrated display range. This will have little effect on
the accuracy of measuring the smaller signal; however,
mixer compression must be avoided.

4
Mixer compression The mixer compression specification describes the
The mixer compression level is the maximum total mixer input power level below which the ana-
power level that we can put into the analyzer with- lyzer compresses the displayed signal less than 1 dB.
out compromising the accuracy of the displayed Keep in mind that the input level at the mixer is
signal. When the signal level at the mixer is well the sum of the input power at all frequencies, even
below the compression point, the level of the if all the input signals cannot be seen on the dis-
desired mixing product (IF signal) is a linear func- play. There are three different ways to evaluate
tion of the input and little energy is diverted to compression: CW compression, two-tone compres-
distortion. As the mixer level increases, the trans- sion, and pulse compression. Each is a different
fer function becomes nonlinear because significant compression mechanism with its own compression
energy is lost to the distortion products. At this threshold [1].
point the mixer is considered to be in compression,
and the displayed signal level is below the actual Hint: limit the mixer level
signal level. For accurate measurements of high-level signals, it is
important to determine the input attenuation setting that
will prevent compression by limiting the power that
reaches the mixer. In Agilent ESA spectrum analyzers the
maximum mixer level can be set. The attenuator setting
will then change automatically to keep power levels
equal to or less than the selected level at the input mixer,
for any signal that is displayed on-screen. The default and
maximum setting is –10 dBm.

5
Internal Distortion for a 1 dB change in the fundamental power, the
Internal distortion is one of the factors that second-harmonic distortion product changes by 1
determines the dynamic range when measuring dBc (dB relative to the carrier or fundamental),
distortion products, such as harmonic distortion and the third-order product changes by 2 dBc.
from a single tone or the intermodulation distortion
from two or more tones. The internally generated Referring to the Agilent Technologies E4402B spec-
intermodulation and harmonic distortions are a trum analyzer’s specifications [4] for spurious
function of the input signal amplitude at the mixer. responses, the second-harmonic distortion for this
To understand these effects, it is helpful to look at analyzer is –75 dBc for a –30 dBm signal at the
how our input mixer behaves. mixer input and third-order intermodulation dis-
tortion (TOI) is –80 dBc for two –30 dBm signals
Most analyzers use diode mixers, which are non- at the mixer. We can create a graph (Figure 3) for
linear devices and behave according to the ideal a variety of mixer input levels since we know the
diode equation [1]. Using the Taylor Series expan- relationship between the fundamental and the
sion, it can be shown that in nonlinear devices a internally generated second- and third-order dis-
1 dB change in the fundamental signal power at tortion products. The slope is one for the plot of
the input results in a 2 dB change in second-order the internally generated second harmonic, and
(for example, second-harmonic) distortion and a two for the plot of the third-order intermodulation
3 dB change in third-order (for example, third- distortion. It appears from the graph that, if the
harmonic) distortion at the output (see Figure 2). mixer level is low enough, there is no need to be
We can equate the dynamic range to the differ- concerned about internally generated distortion.
ences between the fundamental tone or tones and This is true, but as our signal gets lower we need
the internally generated distortion. We find that to take into account the effects of noise.

1 dB

20 dB
21 dB

2 dB 3 dB

f0 2f0 3f0

Figure 2. Changes in the relative amplitudes of distortion

products with input power level

6
Noise signals that are close in frequency (<1 MHz apart).
Two types of noise contribute to dynamic range: Also called sideband noise, phase noise is caused
phase noise and sensitivity. Noise is a broadband by the instability of the local oscillator (LO). There
signal; therefore, as the resolution bandwidth is no such thing as a perfect oscillator; all are
(RBW) filter is widened, more random noise ener- phase-modulated to some extent by random noise,
gy is allowed to hit the detector. This increases the and any instability in the LO is translated to the
level of phase noise as well as the noise floor of the displayed signal through the mixer. The more sta-
analyzer. Therefore, noise specifications must be ble the LO, the further down the phase noise will
referenced to the RBW. be, assuming a sufficiently stable input signal. For
dynamic range considerations, phase noise is like
Sensitivity third-order intermodulation distortion in that it
The sensitivity of the analyzer, also called the matters when the test tones are close to each other.
Displayed Average Noise Level (DANL) or noise
floor, determines the smallest signal we can meas- Second- and third-order dynamic range
ure. The lower limit of DANL is theoretically kTB,1 To incorporate the effect of noise in the graph of
or –174 dBm, for a noise bandwidth of 1 Hz at internal distortion levels (Figure 3), we need to
room temperature. consider how signal-to-noise changes with respect
to changes in the fundamental level at the mixer
Phase noise input. For every dB that we increase the signal
While DANL is the key parameter when measuring level at the mixer, we gain 1 dB of signal-to-noise
two signals that are far apart in frequency, phase ratio.
noise is the key parameter when measuring two

Second-order distortion (500 MHz to 1.5 GHz)

Distortion products relative to mixer level (dB)

Third-order intermodulation distortion

–40

–50

–60

–70

–80

–90

–100
–70 –60 –50 –40 –30 –20 –10 0
Mixer level (dBm)

Figure 3. Internally generated second-harmonic and third-order

intermodulation distortion levels as a function of mixer input level for
the Agilent E4402B

1. Where k=Boltzmann’s constant, T=absolute temperature in degrees

Kelvin, and B=bandwidth in Hertz.

7
Therefore, the DANL curve is a straight line having Hint: determine maximum achievable second- or third-
a slope of –1 (see Figure 4). The horizontal line at order dynamic range
–80 dB in Figure 4 represents phase noise, rather Typically, the maximum achievable dynamic range is
than DANL, for a 10 kHz offset and 10 Hz RBW. specified, and refers to the maximum difference in dB
Phase noise is the limiting factor when making between two harmonically related signals without inter-
measurements close to the carrier (<1 MHz away); ference from the analyzer’s internally generated distor-
for measurements that are far away it can be tion products. From Figure 4, we see that the maximum
ignored. Noise affects our ability to minimize inter- dynamic range is achieved when the mixer level corre-
nally generated distortion products by limiting how sponds to the intersection of the noise and distortion
small a signal we can measure. components. The maximum dynamic range is calculated
with the following equations and can be used to deter-
Second- or third-order dynamic range is limited by mine if your spectrum analyzer has enough dynamic
three factors: distortion performance of the input range to make a particular measurement:
mixer, DANL (sensitivity), and phase noise of the
local oscillator. It is important that all of these Equation 1 Maximum third-order dynamic range =
factors are taken into account and optimized for (2/3)(DANL–TOI)
a particular measurement. Figure 4 shows the
achievable dynamic range as a function of mixer Equation 2 Maximum second-order dynamic range =
level for second- and third-order distortion meas- (1/2) (DANL–SHI)
urements. For maximum distortion-free dynamic
range, the power at the mixer should be adjusted Where: TOI = mixer level – (1/2) (level of
so that the DANL and the internally generated distortion products in dBc)
distortion are equal. SHI = mixer level – level of distortion
products in dBc.

DANL (10 MHz–1 GHz, 10 Hz RBW)

DANL (1–2 GHz, 10 Hz RBW)
DANL (2–3 GHz, 10 Hz RBW)
Second-harmonic distortion (500 MHz to 1.5 GHz)
Third-order intermodulation distortion
Second-harmonic distortion (10 MHz to 500 MHz)

–40
DANL and distortion products
relative to mixer level (dB)

–50
Phase noise
–60 (10 kHz offset, 10 Hz RBW)
–70

–80

–90

–100
–70 –60 –50 –40 –30 –20 –10 0
Mixer level (dBm)

Figure 4. Sensitivity, phase noise, third-order intermodulation,

and second-harmonic distortion versus mixer level for the
Agilent E4402B

8
Summary
A popular use of a spectrum analyzer is to meas-
ure harmonic or intermodulation distortion prod-
ucts. Measurement range, display range, and mixer
compression do not take into account all of the
restrictions encountered when making these types
of measurements. Therefore, “dynamic range” is
frequently understood to mean either second- or
third-order dynamic range. This is the interpreta-
tion of dynamic range used in Section 3, which
discusses techniques for making better distortion
measurements by improving and understanding
the effects of dynamic range.

9
3. Making Harmonic or Intermodulation Distortion Measurements

Measurement uncertainty If we want no more than ±1dB of uncertainty in

Theoretical dynamic range calculations do not our measurement, we must be sure that the inter-
account for measurement uncertainties. If internally nally generated distortion product is at least 19 dB
generated components occur at the same frequen- below the distortion product that we wish to meas-
cies as the signals that we want measure, as is the ure. If we add this 19 dB guardband to the curves,
case in distortion measurements, our accuracy will our dynamic range is reduced by 9.5 dB for second
be degraded. Depending on the phase relationship order and 12.7 dB for third order, as seen in Figure
of the external and internal distortion products, 5. Notice that we do not lose the full 19 dB that we
the measured amplitude can fall anywhere between added for a guardband.
the sum and the difference of the two. Since we
have no way of knowing what the phase relation- Not only does the instrument’s internal distortion
ship is, we can only determine the measurement play a factor; low distortion-to-noise ratio can con-
uncertainty: tribute to the uncertainty as well. If the distortion
components to be measured are at or very close
Equation 3 Uncertainty (dB) = 20 log (1 ± 10d/20) (within a few dB) to the noise level of the spec-
trum analyzer, the displayed signal level (which is
where d = difference in dB between internal actually the level of signal plus noise) is greater
and external distortion products (a negative than the actual signal level. A correction factor can
number) be applied to compensate for this measurement
error [3]. It is common practice to make sure that
the signal is at least 5 dB above the displayed noise
floor, giving an error in the displayed signal level
of less than 0.5 dB.
DANL (2–3 GHz, 10 Hz RBW)
Third-order intermodulation distortion
Second-harmonic distortion
Corrected DANL (2–3 GHz, 10 Hz RBW)
Corrected third-order distortion
Corrected second-order distortion
DANL and distortion relative to mixer level (dB)

–40

–50

–60

–70

–80

–90

–100
–70 –60 –50 –40 –30 –20 –10 0 10
Mixer level (dBm)

Figure 5. Dynamic range graph corrected for amplitude

uncertainty

10
Optimizing measurements One of the drawbacks of a preamplifier is that it
Preamplifier must be calibrated into the system so its effects
A preamplifier is useful when measuring low-level can be accounted for in measurement results. In
signals. There are two important factors to consider the Agilent ESA-E series spectrum analyzers, there
when choosing a preamplifier: gain and noise fig- is an optional built-in preamplifier that will lower
ure. The noise figure (NF) of the preamplifier must the noise floor of the system without compromis-
be lower than the NF of the spectrum analyzer ing dynamic range. It is internal to the system
(NFSA= DANL + 172 dBm). If the NF plus gain of and already calibrated for easy use. Figure 6 shows
the preamplifier is less than the NF of the spec- the noise floor without using the internal pream-
trum analyzer, then the preamplifier will reduce plifier, and Figure 7 shows the approximately 15
the analyzer’s DANL (after correcting for the pre- dB improvement in noise floor when the internal
amplifier gain) by an amount nearly equal to the preamplifier is turned on.
preamplifier gain. If there is too much gain, the
input attenuator can be used to optimize the power
level at the mixer.

Figure 6. DANL of an Agilent ESA-E series spectrum ana- Figure 7. DANL of an Agilent ESA-E series spectrum ana-
lyzer without preamplification lyzer with preamplfication

11
Attenuator Attenuator test
As discussed earlier, when measuring distortion While making distortion measurements it is easy to see
products it may be necessary to adjust the power if the distortion products displayed on the analyzer have
level at the mixer. If the power level at the mixer is been generated internally. The input attenuator attenuates
low enough, we can be guaranteed that the distor- the incoming signal and the analyzer amplifies it again by
tion products we are seeing are from our Device- the same amount after the signal has gone through the
Under-Test (DUT). However, if the power is too low, mixer. Therefore, if the input attenuation is changed and
the analyzer’s noise floor covers up the distortion the distortion products remain the same, the distortion is
products. part of the incoming signal. If the amplitudes of the dis-
tortion products change, all or part of the distortion has
There must be enough power at the mixer for the been internally generated.
DUT’s distortion to be above the DANL, but not
so much that the analyzer’s distortion masks that External filter
of the DUT. The Agilent ESA spectrum analyzers Another way to limit the power reaching the mixer
allow you to change the attenuator in 5 dB incre- for reduced internally generated distortion prod-
ments to better approach the optimum balance for ucts is to attenuate or reject signals that are not of
a specific measurement. interest. This can be done with an external filter,
which acts as a preselector. For example, if we
want to test an oscillator for harmonics, we can
use a high-pass filter to reject the fundamental but
not the harmonic of interest. For closely spaced
signals (for example, in an intermodulation test),
a high-Q filter may be required.

12
RBW filter The DANL is also specified for a particular RBW
Since noise is a function of the RBW filter, use and must be adjusted if the measurement RBW is
caution when interpreting phase noise or DANL different. The same calculation can be used to
specifications. determine the corrected DANL:

Phase noise specifications are normalized to a 1 Hz DANL correction (dB) =

RBW filter even if the analyzer does not have a 10 log (measurement RBW/specified RBW)
1 Hz RBW filter. In order to determine the actual
achievable phase noise level, it is necessary to con- The minimum RBW filter is clearly an important
vert from the specified 1 Hz RBW to the actual factor in determining whether or not the analyzer
RBW filter used for the measurement. The correc- has enough dynamic range for certain measurements.
tion factor is determined as follows:
Measuring other signals
Phase noise level correction (dB) = 10 log (actual RBW /1 Hz) When comparing small signals (other than second-
or third-order distortion) to large signals, the dis-
For example, the phase noise specification for the tortion curves that we have been discussing do not
Agilent E4411B at a 10 kHz offset is –90 dBc/Hz, apply. In this situation, the distortion products
but the minimum RBW setting is 10 Hz. The correc- created by the mixer do not affect the measure-
tion factor for the 10 Hz RBW is: ments. For example, if we are measuring the fourth
harmonic, the internally generated distortion is
10 log (10 Hz/1Hz) = 10 dB small enough that it can be neglected. For these
measurements, phase noise and DANL will be the
Taking this correction factor into account, the limiting factors, as long as the total power at the
actual achievable phase noise is –80 dBc in the mixer is below mixer compression and there is
analyzer’s minimum RBW of 10 Hz. enough display range.

13
4. Summary

Dynamic range limits the range of signal ampli-

tudes we can measure with a spectrum analyzer.
There are several interpretations of dynamic range,
including measurement range, display range, mixer
compression, internal distortion, and noise; the
applicable interpretation depends on the type of
measurement. Second- and third-order dynamic
range can be optimized for a particular measure-
ment by controlling various parameters, such
as mixer input power and DANL. For maximum
distortion-free dynamic range, the power at the
mixer should be adjusted so that the DANL and
the internally generated distortion are equal.

14
5. References

Pub. number
1. Spectrum Analysis Basics,
Agilent Application Note 150 5952-0292
2. Effective Spectrum Analysis
Testing for Consumer
Electronics Production Lines,
Agilent Application Note 130 5966-0367E
3. Spectrum Analyzer
Measurements and Noise,
Agilent Application Note 1303 5966-4008E
4. Agilent ESA-E Series
Spectrum Analyzers,
Data Sheet 5968-3386E

15
By internet, phone, or fax, get assistance with all your
test and measurement needs.
Online Assistance
www.agilent.com/find/assist
Phone or Fax
United States:
(tel) 1 800 452 4844
Canada:
(tel) 1 877 894 4414
(fax) (905) 206 4120
Europe:
(tel) (31 20) 547 2323
(fax) (31 20) 547 2390
Japan:
(tel) (81) 426 56 7832
(fax) (81) 426 56 7840
Latin America:
(tel) (305) 269 7500
(fax) (305) 269 7599
Australia:
(tel) 1 800 629 485
(fax) (61 3) 9272 0749
New Zealand:
(tel) 0 800 738 378
(fax) (64 4) 495 8950
Asia Pacific:
(tel) (852) 3197 7777
(fax) (852) 2506 9284
Product specifications and descriptions in this
document subject to change without notice.
Copyright © 1999, 2000 Agilent Technologies
Printed in U.S.A. 5/00
5968-4545E