Agilent

Pulsed Antenna Measurements

Using PNA Network Analyzers
White Paper
 Abstract
This paper presents advances in the instrumentation techniques that can be
used for the measurement and characterization of antennas that are to be
tested in a pulsed mode of operation. A digital filtering process is described
which allows accurate measurements under a wide range of pulse conditions
using a single receiver. A novel approach to achieving point-on-pulse
measurements using receiver time-gating at the IF frequency is described.
Measurements made using an Agilent E8360 PNA series Microwave Network
Analyzer are presented as a demonstration of a practical implementation of
these techniques.

Keywords: antenna measurements, gating, pulsed measurements,

measurement systems, receiver sensitivity, S-parameters.

1. Introduction
In some applications, it is preferable to make antenna measurements using
pulsed-RF signals. Examples are active array antennas, which may only
operate using pulsed RF signals, and antennas designed for use in pulsed
applications. Among the most common techniques used to make accurate
vector antenna pattern or S-parameter measurements under pulsed condi-
tions are the wide-bandwidth, or full pulse characterization approach, and
the narrow-bandwidth, or high PRF approach[1]. The selection of which
approach to use for a given measurement depends upon several factors,
including the characteristics of the measurement receiver, the pulse width
and pulse repetition frequency (PRF) of the signal being measured, and the
desired time resolution of the measurement.

To use the full pulse characterization approach successfully, the rise time of
the receiver must be sufficiently fast to capture the pulse being measured
without distortion. This technique provides good dynamic range, and the
ability to measure single or non-periodic pulses. However, since the rise time
of any receiver is limited, all receivers have a lower limit to the measurable
pulse width that they are able to characterize in the wide-bandwidth mode.

When the pulse width of the signal being measured is less than several times
the rise time of the receiver, the receiver will not reach a steady-state value
for an individual pulse. This complicates the measurement and calibration
significantly, and it is, therefore, preferable to use an alternative technique
to measure narrow pulses.

If a pulsed RF signal is repetitive with a constant PRF (as is frequently

the case), the frequency domain spectrum of the pulsed signal will consist
of a series of evenly spaced discrete tones, centered at the RF signal
frequency and spaced by the PRF[2]. If we define the stop bandwidth of
the receiver as the bandwidth around the receiver center frequency beyond
which the receiver has no significant sensitivity, then for pulsed signals
with a PRF higher than 1/2 the receiver stop bandwidth, the receiver will
be sensitive only to the central tone of the pulsed RF signal spectrum, and
will therefore measure the pulsed signal as though it were a CW signal.

2
Figure 1 illustrates this situation. In this figure, a segment of the constant
PRF pulse spectrum is shown in the bottom half of the diagram, while the
frequency response of an example receiver is shown in the top half. This
ability to convert a pulsed RF waveform into a representative CW signal
makes the measurement of pulsed signals possible with almost any receiver,
so long as the PRF is greater than 1/2 the stop bandwidth of the receiver.
This has been referred to as the high PRF technique for making pulse
measurements.

Pass bandwidth

Stop bandwidth

PRF

Figure 1. Relationship between receiver filter and Pulsed RF spectrum in standard high
PRF technique

Unfortunately, the simple high PRF technique described above has four
significant characteristics that may limit its usefulness in any given antenna
measurement. These are:

1. The lack of time discrimination (no point-on-pulse).

2. The requirement that the PRF must be greater than 1/2 the stop
bandwidth, which may force the PRF and/or the duty cycle of the pulse to
be too high.

3. The potential presence of other signals in the pulse spectrum that fall
within the receiver stop-bandwidth, degrading the measurement.
4. The reduction of dynamic range due to the loss of the energy in the
harmonics rejected by the receiver.

The Agilent PNA series of microwave network analyzers incorporate

many features that make them attractive for use as receivers in antenna
measurement applications. These include integrated microwave LO and RF
synthesizers, fast synthesized sweeping, flexible microwave architecture, and
the connectivity of the Windows® operating system. However, the maximum
receiver bandwidth of 40 kHz, along with other architectural considerations,
limit the PNA to minimum pulse widths of approximately 30 µs when
using the full pulse characterization technique. For narrower pulses, the
narrow bandwidth high PRF approach may be used, but the characteristic
limitations of this technique listed above may unacceptably limit the
performance in some pulse applications. In this paper, we describe
techniques that have been implemented in the Agilent PNA with Pulsed-RF
Measurement Capability (Option H08) that reduce or eliminate these
limitations and enhance the pulsed measurement capability in narrow
bandwidth receivers.

3
 2. Time Discrimination in High PRF Measurements:
Time Gating
In many applications, it is desirable to determine the response of the device
being tested at some particular time during the pulse. If the response of
interest occurs over a fraction of the pulse period, but significant signal
strength is present for a longer period of time, then simple high PRF meas-
urements will not provide the desired results, since the signal measured will
represent the average response over the duration of the pulse. Fortunately,
this limitation may be eliminated through the use of time gating. A time gate
is a fast switch that is inserted in the signal path between the device under
test and the bandwidth-limiting receiver. By properly synchronizing the
opening and closing of the switch with the pulsed RF signal, only the signal
of interest is allowed to charge the band-limiting filter, and therefore only
the desired portion of the pulse is measured.

Time gating may either be performed at the signal frequency or at the

intermediate frequency (IF). In IF gating, the gate switch is inserted between
the broadband mixer and the bandwidth-limiting filter. Since the receiving
mixer is a broadband linear device, the pulse is preserved through the
mixing process (although for short-duration pulses, care must be taken that
the pulse harmonics, which may “wrap” around DC, do not degrade the
measurement). Since IF gate switches operate at a single frequency, it is
usually possible to obtain better switching speed, on to off ratio, and full
frequency coverage for less cost using IF gating. Also, the mixer provides
significant isolation between the gate switch and the device being measured,
minimizing any errors that may otherwise occur due to the differing
impedance of the switch in its on and off state. A low-noise IF amplifier with
adequate bandwidth to preserve the pulse shape within the desired time
resolution may be inserted between the mixer and the gate switch. This
further isolates the gate and can improve the sensitivity of the system.

Using the Agilent PNA with Option H08 and H11, gating may either be
performed using either internal (provided) IF gate switches or external
(customer supplied) RF or IF gate switches. The internal gate switches have
a minimum time resolution of less than 50 ns, with an on-to-off isolation of
greater than 90 dB. For narrower time resolutions in distributed antenna
measurement systems with external mixers, such as the system shown in
Figure 7, it may be preferable to use an external gate switch, located close to
the device under test, in order to reduce the degradation in time resolution
that may be caused by multiple reflections in the long IF signal path.

4
Examples of time-gated pulse measurements made with the PNA are shown in
Figure 2 and Figure 3 below. Both figures show transmission measurements
made on a microwave switch, which is switched on 0.7 µs after the start of
an 5 µs RF pulse, and switched off 3 µs later. The receive channel gate
resolution was set to 100 ns. The PRF was 32 kHz. In Figure 2, the switch is
measured at a single frequency (10 GHz), while the delay of the receive gate
is swept from 0 to 5 µs. This type of measurement is called a pulse profile
measurement, and provides useful information about the dynamic perform-
ance of the device being measured in both magnitude and phase. Note the
variation in the transmission of the switch in both magnitude and phase as
it turns on, reaching a stable value by 3 µs. Note also the dynamic range of
the measurement, with the measured noise level prior to the switch turning
on at approximately –85 dB (relative to a through).

dB

1 µs 2 µs 3 µs

4.5 µs

deg

Figure 2. Pulse profile measurement example

In Figure 3, the frequency response of the switch from 2 to 19.9 GHz is

measured for delays of 1 µs, 2 µs, 3 µs, and 4.5 µs, again with a time
resolution of 100 ns. This is called a point-on-pulse measurement.

dB

1 µs 2,3 µs

4.5 µs

deg

Figure 3. Point-in-pulse measurement example

5
 3. Using Digital Filtering to Allow Lower PRFs in
Narrowband Pulsed measurements.
For accurate high PRF pulsed measurements, it is important that only the
center tone of the pulse response pass through the filter. If the PRF of the
measurement is sufficiently high, the rejection of the receiver analog filter
will be sufficient to accomplish this. In some cases, however, it is desirable
to make measurements at lower PRFs.

In Figure 1 all of the harmonics of the pulse spectrum are eliminated by the
stopband rejection of the bandpass filter. While this approach is effective, all
that is actually required is that the receiver filter reject all of the discrete
tones (other than the center frequency) present in the pulsed RF signal. The
frequency response of an FIR filter with a rectangular window[3] is shown
in Figure 4, juxtaposed with the pulsed spectrum. Note that this response
has evenly occurring nulls, spaced by 1/(N*T) Hz, where N is the number of
filter taps and T is the sample time. By aligning the nulls in the filter
response with the tones in the pulse spectrum, the desired result is
achieved. This is accomplished by collecting the number of samples N,
spaced by the sample time T, such that the equation PRF=K/(NT)1 is satisfied
for some integer K (in Figure 4, K=2). Using this approach, it is in principle
possible to measure down to an arbitrarily low periodic PRF by selecting an
appropriate number of taps N, although in practice the dynamic range losses
due to pulse desensitization limit the lowest practical PRF to approximately
100 Hz.

Figure 4. FIR filter response and pulse spectrum

1. In the actual implementation used in the PNA, a slightly non-rectangular window is used, and the null equation
is modified to PRF=K/((N-6)*T).

6
4. Identifying and Filtering Other Interfering Signals
Up to this point, the only tones assumed to be present in the pulse spectrum
measured by the receiver are those centered at the stimulus frequency and
spaced by the PRF, and the only receiver sensitivity has been assumed to be
given by the digital filter response centered at the stimulus frequency. In
practice, however, there are several other potential signals and receiver
responses that could interfere with measurement accuracy. These include
source harmonics, gate switch video feed-through, and receiver effects such
as sensitivity to the first LO image frequency. For CW signals, these sources
are insignificant due to the inherent frequency selectivity of the PNA. For
pulsed signals, however, each source of interference causes a corresponding
series of interfering frequencies in the frequency domain, centered at the
source interference frequency and spaced by the PRF. If the sources of
interfering signals and receiver sensitivities are known, their effects can be
reduced to acceptable levels through careful selection of the digital filter
characteristics and pulse repetition frequencies. For example, in the
Agilent PNA with pulsed-RF measurement capability (Option H08), a filter
constructing algorithm is used to reduce the potential interference from
eight different combinations of pulse spectral energy and receiver
sensitivities to negligible levels.

7
 5. Dynamic Range Considerations
Since only the central tone in the pulse spectrum is allowed to pass through
the filter, The high-PRF pulse technique results in a signal loss of
20*log10 (duty cycle) relative to a non-pulsed signal measured by the same
receiver. In this equation, the duty cycle is given by the ratio of either the
gate width or the pulse width (whichever is smaller) to the pulse period
when gating is being used, or by the pulse width to the pulse period when no
gating is used.

Because of this inherent loss, pulsed measurements and pulse measurement

systems using the narrow-band technique should be designed to obtain the
best possible sensitivity and dynamic range. Some of the factors to consider
are the gate resolution setting and PRF of the measurement, the gain of the
IF amplifiers, and the analog filter bandwidth of the system.

As noted previously, the sensitivity of the measurement will decrease as the

gate width becomes smaller. Therefore, in order to optimize the measure-
ment performance, the widest gate width that will provide the required time
resolution for the measurement should be used. One approach to determin-
ing the optimal gate width setting is to perform a pulse profile measurement
(described in section 2), in order to identify the time dependency of the
device being measured. This will inform the selection of both the maximum
permissible gate width and the optimal delay for the parameter in question.
Similarly, to optimize measurement sensitivity, the PRF should be set as high
as possible.

In CW measurements, IF amplification is often selected to optimize the

trade-of between increased sensitivity (more gain) and higher compression
levels (less gain). In pulsed measurements, if the compression occurs after
the bandwidth-limiting filter, it may be beneficial to use higher IF gain, since
the filter will reduce the peak signal level for narrow pulse widths.

Other factors that may effect the dynamic range for an antenna measure-
ment include the power from the source; losses prior to the receive mixer;
mixer conversion loss and noise figure; IF amplifier gain, compression, and
noise figure; and noise in the LO distribution system. Although a thorough
discussion on the proper design of an antenna measurement system is
beyond the scope of this paper, some of these factors will be considered
briefly in the following sections, especially as they pertain to pulse
measurements.

8
 6. Pulsed Antenna Measurement System
Figure 5 shows a block diagram for the pulsed measurement system used
to perform the measurements shown in this paper. This is one of several
possible system architectures that can be constructed using the PNA as a
receiver. The system shown employs fundamental mixing, using the Agilent
85320 mixers. The Agilent 85309 LO-IF Distribution Test Set provides
isolated LO signals at the proper power levels to the two mixers, as well as
IF amplification. The RF and LO signals are provided by the PNA (with
Option H11). An additional gain block of 12 dB or more in the LO path is
used to set the LO power at the 85309 input to at least 0 dBm. This ensures
that the LO signal is compressed, which reduces AM noise on the LO. RF
pulse modulation is performed with an external pulse modulator. IF gating
is performed by the internal IF gates provided as part of the PNA Option
H11. One or more Agilent 8110 (or equivalent) pulse generators provide the
pulse control signals to the internal IF gates, the RF pulse modulator, and
the test device. One of these is designated the master pulse generator, and it
generates the PRF, which is phase locked to the 10 MHz time base provided
by the PNA. This ensures the synchronization required by the digital filter.
All other pulse generators are synchronized to the master using their
external trigger inputs. The IF frequency is approximately 8.33 MHz,
although the precise value is set as required by the digital filter constructor
of the PNA Option H08. The IF amplification of approximately 23 dB provided
by the 85309 is perhaps 10 dB more than optimal for non-pulsed systems,
but the extra gain was found to increase the sensitivity for gate widths of
1 µs or less. This effect is discussed in the next section. Some benefits of this
system architecture are that
1. No additional microwave sources are required.
2. Since The PNA controls the frequencies in a standard way (frequency
offset mode), and since the signal being measured by the PNA is a CW
signal, all normal features and performance of the PNA are available for
use, once the pulse set-up is established by the H08 software.
3. Fundamental mixing provides excellent sensitivity and dynamic range for
the measurements.

Reference Test antenna

Source antenna antenna

85320A
85320B test mixer
Reference
mi xer
LO
in
85309
12 dB gain
IF signals block
RF pulse
modulator

RF out
PNA with option H 11 and H08 Up to 4
independent
IF gating signals
10 MHz
Time Base

External
RF pul se control Trigger
OUT/IN

Pulse Generators

Figure 5. PNA based pulsed antenna system block diagram used for measurements
9
 7. Measurements
The results shown below were obtained on the system described in the
previous section.

Predicted and Measured Sensitivity, Dynamic Range, and

Compression vs. Gate Resolution for 10 kHz PRF (5 GHz)
150

100

50

dB(m)
0

–50

–100

–150
100 10 1 0.1 0.01
Gate Resolution (µs)

Predicted Sensitivity Measured Sensitivity

Predicted Dynamic Range Measured Dynamic Range
Predicted .01 dB Compression Actual .01 dB Compression

Figure 6.

Figure 6 shows the measured sensitivity, compression, and dynamic range at

5 GHz as a function of gate resolution for a PRF of 10 kHz. Sensitivity is
measured as equivalent noise power in a 10 Hz bandwidth at the test mixer
input. The gain compression is the power at the mixer input that results in
0.01 dB degradation in linearity. Notice that the sensitivity degrades less
than the 20*log10 (duty cycle) pulse desensitization. This is because the
excess average noise contributed by the IF amplifiers is also attenuated by
the pulse desensitization, although to a lesser degree than the signal.

Receiver Sensitivity in 10 Hz BW for

1 s Gate Resolution and Various PRFs
0
Sensitivity (dBm)

–20

–40 CW
100 kHz
–60
10 kHz
–80
1 kHz
–100
300 Hz
–120
100 Hz
–140

–160
0 5 10 15 20
Frequency (GHz)

Figure 7.

The net result is approximately 15 dB less loss in dynamic range than

predicted by duty cycle desensitization alone. Figure 7 shows sensitivity as a
function of frequency for a 1 µs gate resolution and PRFs of 100 kHz, 10 kHz,
1 KHz, 300 Hz and 100 Hz. Here again we see less degradation than predicted
by the signal desensitization of 20*log10 (duty cycle).

10
8. Summary
In this paper, we have described the narrow bandwidth high-PRF technique
for making measurements using pulsed RF signals. We have discussed several
techniques that may be used to overcome the traditional shortcomings of
this approach, including the use of time gating to provide time resolution
and the use of digital filtering, both to allow low PRF measurements, and to
eliminate potential interfering signals. We have described the factors that
contribute to dynamic range loss. We presented a system block diagram
based on the Agilent PNA series network analyzer, which we have used to
make several measurements demonstrating these concepts.

9. Conclusions
The high PRF technique for making pulsed measurements, which has always
been valuable due to its ability to measure arbitrarily narrow pulses, has
never the less been hampered by the lack of point on pulse capability and
the inability to measure low PRF pulses. The Agilent PNA Series Vector
Network Analyzer (with options H11 and H08) has successfully addressed
these limitations for a wide range of applications through the use of time
gating and the careful application of digital filtering. Although dynamic
range and sensitivity degradation are still realities for low duty cycle
measurements, this paper has presented methods of reducing this effect as
well, by using careful planning in the measurement design, and by taking
advantage of IF amplification prior to the gate switch. Together, these
techniques provide a viable approach to pulsed antenna measurements for a
broad range of applications.

10. References
[1] Swanstrom, J., and Shoulders, R., Pulsed Antenna Measurements With
the HP 8530A Microwave Receiver, 16th Proceedings of the Antenna
Measurement Techniques Association (AMTA-1994), Long Beach, CA,
pp. 416-420

[2] Ferrel G. Stremler, Introduction to Communication Systems (Reading,

Massachusetts: Addison-Wesley Publishing Company, 1982) pp. 44-50

[3] Alan V. Oppenheim and Ronald W. Schafer, Digital Signal Processing

(Englewood Cliffs, New Jersey: Prentice Hall, INC, 1975) pp. 239-250

11
Web Resources
For additional PNA Series product information visit our web site:
www.agilent.com/find/pna

Agilent Email Updates

www.agilent.com/find/emailupdates
Get the latest information on the products and applications you select.

Agilent T&M Software and Connectivity

Agilent’s Test and Measurement software and
connectivity products, solutions and developer
network allows you to take time out of connecting
your instruments to your computer with tools
based on PC standards, so you can focus on
your tasks, not on your connections. Visit
www.agilent.com/find/connectivity
for more information.

By internet, phone, or fax, get assistance with all

your test & measurement needs

Online Assistance:
www.agilent.com/find/assist

Product specifications and descriptions in this

document subject to change
without notice.

© Agilent Technologies, Inc. 2004

Printed in USA January 6, 2004
5989-0221EN