Frequently Asked Questions
Vector Network Analyzer
Application Note 1EZ38_3E
Subject to change
19 January 1998, Olaf Ostwald
2. What is the maximum IF bandwidth?
The maximum selectable measurement bandwidth is 26.5 kHz. It may be reduced from 10 kHz
to 1 Hz in half-decade steps. After PRESET the IF
bandwidth is 10 kHz so that a high measurement
speed is ensured. Even in the case of large bandwidths the dynamic range is more than 80 dB.
3. What about aging and temperature
The stability and reproducibility of measurements
is a key feature of all quality network analyzers. In
the development of the ZVR network analyzer
family particular emphasis was placed on these
characteristics and a number of measures
concerning concept, construction and electrical
characteristics were taken to reduce temperature
drift to a minimum. This also applies to the
ventilation used which ensures a uniform
temperature in all receiver channels particularly
for critical modules such as the front end. In
addition, scattering parameters are always
derived from the vector ratio of measurementchannel and reference-channel parameters, the
RF cabling in the analyzer test set being such that
the measurement and reference channels have
the same electrical length. Phase variations
caused by temperature drift are compensated for
by the calculation of vector ratios. Moreover, a
thin coaxial compensation line for the reference
channels is provided together with the balun
which is housed in SWR bridges. The coaxial
cable for the baluns and the compensation lines
in the two bridges are taken from the same cable.
This ist the reason why the SWR bridges for
ZVRE and ZVR can only be ordered in pairs.
Frequently asked
…on measurements with the
1. Can frequency range, the number
of test points or the measurement
bandwidth be varied without the
calibration being lost?
Yes. Vector Network Analyzers ZVR and ZVRL
allow to vary the measurement bandwidth
(IFBW) any time without the calibration becoming
invalid, as the measurements will be carried out at
the same frequency points at which the calibration
was performed. The CAL enhancement label
signals that the calibration is valid. By the way,
during calibration the bandwidth is automatically
reduced to 1 kHz when a larger bandwidth was
used before. This reduces the effect of noise
without affecting the calibration speed. After
termination of the calibration the original IF
bandwidth is automatically restored.
Furthermore, all instruments are subjected to preaging in the test shop to minimize the drift. Due to
these measures, it is sufficient to calibrate only
once as long as the test setup is not changed
whatever the application. A calibration is typically
valid for months. For instance, after normalization,
switching off and complete cooling down the
analyzer and switching the cold unit on again a
few minutes later typical values for analyzer
stability are <0.02 dB and <0.5° for transmission
measurements. For reflection measurements the
stability of the effective directivity is typically
>60 dB. With higher stability requirements, it is
advisable to give the instrument one hour to warm
up. After that time the typical difference to a
measurement the day before with the instrument
switched off overnight is <0.005 dB, and <0.1°. If
higher stability is required, the network analyzer
should be operated under constant ambient
conditions in an air-conditioned room and after a
By contrast, changing the frequency range or
the number of measurement points means in
most cases that the analyzer measures at other
than the calibrated frequencies. In this case
calibration data have to be interpolated. This
can be done by selecting the CAL INTERPOL
function. Measurements can then be carried out in
any subrange within the previously calibrated
frequency range and with any number of measurement points. Of course, calibration cannot be
extrapolated to ranges outside the previously
calibrated range. To identify calibration interpolation the enhancement label changes from CAL
to CAI.
29 May 1998
warm-up of at least two hours. For the most
stringent requirements it is necessary to carry out
a calibration immediately before each measurement. In all other cases, previously determined
calibration data sets can be used as long as the
test setup is not significantly changed.
6. Is it also possible to measure very
low impedances below 1 Ω?
Yes, and in this case a simple normalization
calibration is generally sufficient, where the DUT
is replaced by a short. Thanks to its excellent
basic characteristics (test port matching etc.) and
thermal stability, Network Analyzer ZVR is able to
measure impedances near 0 Ω with an
impressive accuracy. For instance, a 560 mΩ
mini-resistor was measured up to 100 MHz with
an uncertainty of typically less than 40 mΩ.
With the aid of more suitable calibration
techniques like TRM the measurement accuracy
can be further improved. For measurements at
high frequencies it is very important for the
electrical length of the DUT to be taken into
account to avoid phase shift of the reflection
coefficient which increases the impedance of the
4. How can PCBs be tested up to
around 2 GHz?
There are different solutions to this problem,
depending on whether the measurements are
carried out in 50 Ω or 75 Ω systems or whether a
high-impedance measurement is required. With
low-impedance measurements on a few interfaces only, PCBs may be fitted into test fixtures.
The inner conductor and the ground line of the
normally planar, in many cases microstrip or coplanar connectors of the PCB are contacted by
the test fixture and the coaxial connector of the
network analyzer is connected via an adapter. If
PCB interfaces cannot be contacted in this way,
eg because they are not located at the edge of
the board but somewhere inside, special test
adapters with suitable coaxial probes for contacting the testpoints may be used. For highimpedance measurements, special high-impedance RF probes are available. They comprise a
sensitive RF preamplifier and may be connected
to any point of the DUT. The advantage is that the
DUT is virtually unloaded, ie the voltages measured are the same as with no high-impedance
probe connected. The operating voltage for the
RF preamplifier can be supplied from the network
analyzer. To this end two connectors PROBE 1
and PROBE 2 are provided on the front panel of
ZVR to which commercial RF probes can be
7. How much faster is Network
Analyzer ZVR than other analyzers
in surface acoustic wave (SAW)
filter measurement?
The measurement speed of any network analyzer
is limited by the chosen measurement bandwidth
(IF bandwidth). The measurement bandwidth is
determined by the internal IF filter used for
filtering the test signal that has been downconverted in the receiver. Same as in most stateof-the-art network analyzers, this filter is digitally
realized in ZVR with the aid of a digital signal
processor (DSP). The narrower the selected IF
bandwidth, the lower the effect of noise and the
greater the dynamic range but also the longer the
measurement time. The latter corresponds to
approx. the reciprocal of the measurement
bandwidth irrespective of whether an analog or a
digital filter is used. The speed can be slightly
improved at the expense of the filter quality.
Conversely, extremely steep-edged IF filters with
high stop-band attenuation lengthen the
measurement time.
5. Can balanced-to-ground
measurements up to 100 MHz be
carried out with the ZVR?
This depends on the available test adapter and
particularly on the balun which is required as an
external component for transforming the applied
balanced signals into unbalanced signals. Thus
basically all network analyzers of the ZVR family
are able to carry out balanced measurements in
the full frequency range up to 4 GHz. The ultra
wideband balun used in the internal SWR bridges
of the passive test set comprises three pairs of
ferrite-core circuits and covers the frequency
range from 9 kHz to 4 GHz. Baluns for instance
for the frequency range 10 kHz to 125 MHz are
commercially available.
An advantage of Network Analyzer ZVR is that it
allows wider measurement bandwidths to be
used. The maximum IF bandwidth setting "Full" in
analyzers of the ZVR family is 26.5 kHz, that of
other network analyzers is 3 kHz. Thanks to the
wider bandwidth in conjunction with a slightly
more speed-optimized IF filter characteristic of the
ZVR, the measurement speed of the ZVR is
somewhat more than 10 times higher compared
to other analyzers.
ZVR also offers a special measurement mode,
the "fast mode" with reduced dynamic range and
accuracy. This mode allows less time for the
settling of the internal generator and receiver, and
29 May 1998
does not switch the internal IF amplifiers in front
of the A/D converters. This speeds up the
measurement again by a factor of 1.5
9. How can the advantages of the VXI
bus be utilized for the DC control
circuits in the system?
However, the minimum settling time of the DUT
must always be taken into account. For example,
measuring a narrowband crystal filter at the
highest measurement speed is not useful since
the filter is not able to settle within the brief period
of the stimulus signal which can be seen at times
as a tilting of the trace. This problem can be
solved with the aid of a special characteristic of
the ZVR, ie by linking two measurements. In one
measurement the test frequency is swept as
usual from low to high frequencies (forward
sweep) and in the second the sweep direction is
reversed (reverse sweep). The sum of the two
fast measurements is approximately the result
that would be obtained in a sufficiently slow
Network analyzers of the ZVR family are not able
to control system components via the VXI bus.
Conversely, the network analyzer cannot be
controlled via the VXI bus. For system application
of any kind the IEC/IEEE bus is recommended. Of
course, IEC/IEEE-bus-compatible DC voltage
sources may also be controlled via this bus.
For SAW filters these limitations are mostly not
observed. In this case the full measurement
speed of ZVR can be utilized which is
approximately 15 times greater in the fastest
mode than that of comparable network analyzers.
Network analyzers ZVR may comprise up to three
independent IEC/IEEE-bus interfaces, one
being standard equipment for universal
applications. The second, the so-called system
bus, serves for fast control of external generators
as may be required, for instance, as local
oscillators for measurements on frequencyconverting DUTs, with the ZVR acting as a
system controller. The third (optional) IEC/IEEEbus interface is used in conjunction with the
Computer Function Option ZVR-B15 for
controlling any test setups by the PC-compatible
computer in the ZVR.
8. What possibilities and references
are available for component tests?
10. What happens to the open path in
three-port operation?
Suitable adapters play a key role in the testing of
components not provided with common coaxial
connectors. The range of test adapters is as wide
as the variety of components. R&S supports the
user in many ways in the selection of suitable
adapters for components tests and optimum
calibration and measurement procedures.
Thus tailor-made adapters can be fabricated for
the components to be tested and suitable
references or calibration standards can be
implemented for the customer. Many application
problems can be elegantly solved with the aid of
the highly suitable R&S calibration procedure TNA
which comes as standard for all four-channel
network analyzers ZVR. In addition, the new
Virtual Embedding Networks ZVR-K9 for which a
patent is pending, allows component tests to be
carried out without the use of matching networks
whose production, maintenance and variety often
prove to be problematical. The matching networks
are replaced by a sophisticated numerical
technique as part of the system error correction
which has to be carried out in any case (see
questions on virtual transformation networks,
page 10).
When the 3-Port Adapter Option ZVR-B8 is used,
PORT1 of the network analyzer is extended to
two ports, ie PORT1 and PORT3 by means of an
electronic FET switch (SPDT). This allows seven
of the altogether nine S-parameters of an arbitrary
three-port DUT to be directly measured.
Controlled by the network analyzer the measurement path from PORT1 to PORT2 or the path
from PORT3 to PORT2 are alternately switched
on. The unused port of the open path is
terminated with a low-reflection 50 Ω resistor
integrated in the 3-Port Adapter.
The following Application Notes are available for
further information on three-port measurements:
3-Port Measurements (1EZ26_0E) and Multiport
Measurements using Vector Network Analyzer
ZVR (1EZ37_0E).
11. Can also 4-port measurements be
This can be done in several ways. A direct
solution is by using the optional 4-Port Adapter
ZVR-B14. This adapter comprises two electronic
switches and thus extends the two ports PORT1
and PORT2 of the network analyzer to a total of
four ports PORT1 to PORT4. These ports are
then activated in pairs depending on the network
analyzer channel currently active so that different
transmissions and all reflections of a 4-port DUT
can be determined.
29 May 1998
…regarding calibration:
The 4-port adapter is available in two different
models, ie .02 and .03. The first model has two
single pole double throw switches (2 x SPDT) and
is thus especially suitable for DUTs with two
inputs and two outputs such as directional
couplers or double pole double throw switches
(DPDT). The following S-parameters of the DUT
can be determined: S11, S22, S33, S44, S21,
S12, S32, S23, S41, S14, S43 and S34.
12. Is full system error correction also
valid in the case of 4-port
The mathematical routines implemented in the
analyzers of the ZVR family for system error
correction apply to 1-port and 2-port measurements. They are based on different, partly
simplified models (normalization, One Path Two
Port, TOSM, TOM, TNA, etc.) through to the full
model (TOM-X), which detects and numerically
eliminates any coupling between the two
measurement channels and the reference
The second model on the contrary has a single
pole triple throw switch (SP3T) and is more
suitable for DUTs with one input and three outputs
or vice versa such as filter banks or power
dividers. It allows the measurement of the
following 4-port S-parameters: S11, S22, S33,
S44, S21, S12, S31, S13, S41 and S14.
Application Note Multiport Measurements using
Vector Network Analyzer ZVR (1EZ37_0E)
provides further detailed information.
Because of the enormous amount of calculations
involved, full models are not implemented at
present for 3-port and 4-port measurements.
Simplified models are used instead. For instance,
in the case of 3-port measurements, a full 2-port
calibration is performed for each of the two
available paths, PORT1 to PORT2 and PORT3 to
An alternative solution is to operate two 3-Port
Adapters ZVR-B8 together at the optional
MULTIPORT ADAPTER socket on the rear panel
using a simple Y cable, PORT1 and PORT2 of
the network analyzer are then extended to two
ports each. The second 3-Port Adapter is
configured so that it does not switch
synchronously to the even or odd channels as
usual, but is in the setting PORT1 for channels
CH1 and CH2 and setting PORT3 for channels
CH3 and CH4. The 4-port DUT is now connected
to the four available ports. Depending on the
selected channels CH1 to CH4, the 4-port
S-parameters S31, S13, S32, S23, S41, S14,
S42 and S24 can be measured like with model
.02 of the 4-port adapter in addition to
transmission S-parameters S21 and S12 and all
reflection coefficients S11 to S44.
If 4-port measurements are performed with the
aid of two 3-Port Adapters ZVR-B8 or one 4-Port
Adapter ZVR-B14, up to four separate 2-port
calibrations can be carried out. If 4-port DUTs are
measured with the aid of the modified Extra Input
4-Port Option ZVR-B26, full 2-port calibration is
only possible for PORT1 and PORT2 and a
simple normalization calibration is carried out for
the paths to PORT3 and to PORT4.
13. Are calibration data lost when the
instrument is switched off?
Application Note Multiport Measurements
(1EZ37_0E) mentioned above provides further
detailed information.
No. After a calibration, the calibration data are
stored automatically on the internal hard disk and
activated when the instrument is switched on
again. In addition, the user may store any
calibration data on the hard disk or on floppies to
be able to call them up later to suit the test setup
used. The only limitation on the number of
calibration records that can be stored is the
memory size of the disks.
Another way of carrying out 4-port measurements
is to use the Extra Inputs 4-Port Option ZVR-B26.
The two optional inputs INPUT b1 and INPUT b2
are used as PORT3 and PORT4 supplementing
the test ports PORT1 and PORT2. Fast
switchover between the four ports is ensured by
two additional electronic switches integrated in the
test set. This configuration permits the 4-port
S-parameters S11, S22, S21, S12, S31, S32,
S41 and S42 to be measured without the need for
3-port or 4-port adapters. Application Notes
4-Port Measurements (1EZ25_0E) and Multiport
Measurements (1EZ37_0E) provide further
information on this solution.
29 May 1998
of the SHORT is displayed that is obtained by
transformation of its short plane via a 15.1 mm
line that is connected ahead. If another SHORT
with a different electrical length between its short
plane and the reference plane is measured
instead of the calibration short as DUT, the
analyzer indicates its reflection coefficient
correctly as expected from a correctly measuring
network analyzer. With a long length between the
short plane and the reference plane, the open
circuit point is passed. This is exactly the case for
the frequency at which the electrical length of the
standard is a quarter of the wavelength. If the
length of the standard is 0 however, the expected
short circuit point is obtained in the Smith chart
over the complete span.
14. Can a DUT be used for calibration?
A DUT may be used instead of a calibration
standard for normalizing the test setup. The
characteristics of all subsequently measured
DUTs are then compared with this golden device
which is used as a reference. This procedure
corresponds to a normalization calibration, eg
transmission normalization with a throughconnection selected in the CAL menu of the
network analyzer. This procedure is ideal for the
determination of the identity of two DUTs.
However, due to the non-specified measurement
accuracy it should not be used for accurate
quantitative measurements of the differences of
two DUTs.
Since such a short of the length 0 is normally not
available (an exception is the calibration standard
SHORT (F) of Calibration Kit ZCAN), a circular
arc rather than a point is usually obtained whose
length is determined by the product of the
frequency range (SPAN) and the electrical length
of the calibration standard (LENGTH).
15. Have any measurements been
made with this calibration?
Measurements with a golden device being used
as a reference have been carried out with the
ZVR frequently and without any problems. For
such measurements, the reference DUT, eg a
bandpass filter, is connected first and then the
filter to be measured or adjusted is connected to
the same terminals. This allows the characteristics of the two DUTs to be compared. An
interesting variation is the use of two synchronously switched 3-Port Adapters ZVR-B8 or a
4-Port Adapter ZVR-B14. They permit the
reference device and the DUT to be measured
simultaneously which considerably simplifies a
comparison of temperature and drift effects.
The position of the reference plane can be
mathematically shifted after calibration with the
aid of operating function OFFSET. With the length
of the standard selected as offset, ie 15.1 mm as
used in the example (ELECTRICAL LENGTH),
the network analyzer indicates the measured
phase in the short plane of the standard. This
phase is accurately 180° and the measured curve
lies at the short circuit point of the Smith chart.
Similar conditions apply to OPENs but a further
effect occurs: the so-called fringing capacitance
of the open line end. In contrast to SHORTs
which can be produced practically ideally (apart
from the electrical length to be taken into
account), OPENs have a significant stray
capacitance of the open line end amounting to
several tenths of femtofarad and depending on
the frequency. The fringing capacitance of the
calibration standards used for ZVR is accurately
known and is taken into account during calibration
so that no additional measurement uncertainties
have to be coped with. With an OPEN used as
the DUT, the fringing capacitance causes a phase
shift which in turn results in the trace not being
coincident with the open point after mathematical
correction of electrical length (OFFSET). This
does not mean, however that the network
analyzer performs an inaccurate measurement
but is a correct representation of the real
capacitance of the OPEN. If an ideal OPEN could
be made whose reflection coefficient were exactly
+1 over the complete span, ie with the magnitude
1 and the phase 0°, the analyzer (provided that
such a standard is connected) would measure a
point as data trace at the open circuit point (at the
right edge of the unity circle of the Smith chart).
16. Why are there no points in the
Smith chart on reconnecting the
OPEN or SHORT after calibration?
For measurements with SHORTs the calibrated
network analyzer does not display a short point
(r = -1) in the Smith chart as might be expected. A
trace is shown instead which, at low frequencies,
starts at the short circuit point and, at high
frequencies, runs along the unity circle of the
Smith chart and passes, starting from 180°,
phase values up to 35° at 4 GHz, for instance and
thus comes even close to the open circuit point
(r = +1).
This behaviour is fully correct and can be
explained by the electrical length of the SHORT.
For reasons of mechanical design the short plane
is generally not in the reference plane, which is
defined by the outer edge of the outer conductor,
but 15.1 mm behind the reference plane as is the
case for the standards of Calibration Kit ZV-Z21.
Since the network analyzer measures and
displays the reflection coefficient referred to the
reference plane, the correct reflection coefficient
29 May 1998
18. Can a 50 Ω line be used for
17. How often should a calibration be
repeated in production applications?
Yes. A known 50 Ω line is one of the best
standards for verifying the measurement
accuracy of a network analyzer. For coaxial
systems so-called air lines are used. Air lines are
high-precision coaxial lines without dielectric
supports. The space between the inner and outer
conductor is filled with air over the whole length of
the line. This allows an exact description of the
field distribution and consequently precise
assessment of the characteristic impedance. A
typical 50 Ω air line with N connectors has an
outer conductor of nominally 7 mm diameter and
a 3.0396 mm inner conductor, production
tolerances being approx. 2 µm. The accuracy of
the characteristic impedance is influenced by the
line dimensions and the deviations of the coaxial
tubes from the ideal cylindrical shape. The
accuracy also depends on the eccentricity of the
tubes and the field penetration, which is
frequency-dependent because of the skin effect
and thus changes the effective line diameter, and
finally on the roughness of the conductor
surfaces. In practice the characteristic impedance
can be specified to a maximum accuracy of 0.1 Ω
which corresponds to a return loss of between
0 dB and 60 dB.
The answer to this question depends on the
stability of the network analyzer (see question 3
concerning aging and temperature drift) but also
on the stability of the appropriate user-specific
test setup and on the required accuracy of the
particular measurement. For many applications
involving measurements carried out either directly
or via short test cables (eg ZV-Z11) at the test
ports of the network analyzer and moderate
accuracy requirement, the network analyzer need
not be calibrated at all. In this uncalibrated state
(CAL enhancement label off) measurements are
automatically corrected via factory calibration in
which a TOM calibration (for ZVR) between
PORT1 and the end of a Test Cable ZV-Z11
connected to PORT2 is performed. Thanks to the
excellent stability of the analyzer, this factory
calibration is fully sufficient for a great number of
If the user has a different and more elaborate test
setup with, for instance, adapters or long cables,
which also comprises frequency- or even
temperature- or time-dependent components, a
calibration of the test setup is indispensable.
For these cases Vector Network Analyzer ZVR
offers a great variety of partly worldwide new and
exclusive calibration methods that offer distinct
advantages (see: "Die ZVR-Familie“ in ElektronikPraxis No. 3/96 of 9.2.96, pp 116-119).
To verify an important system characteristic of the
network analyzer, eg test port matching (source
match), scattering parameter S11 is selected and
the air line connected to PORT1 of the network
analyzer. A short-circuit at the other end of the
line causes a total reflection. A perfect network
analyzer should now yield a measured broadband
return loss of exactly 0 dB, the low loss of the air
line being neglected. Because of the finite source
match of a real network analyzer, the whole of the
signal reflected by the short circuit is not received.
Part of the reflected signal, eg 10%, is reflected
back by the port to the shorted end via the air line.
Here it is reflected again and superimposed at the
test port onto the direct signal. Neglecting multireflections, two signals are superimposed in the
receiver of the analyzer, one approaching the
correct magnitude of 1, the other with a magnitude of approx. 0.1 because of the assumed 10%
reflection at the test port. Superposition of the two
signals is not scalar but vectorial and because of
the length of the air line (typically 300 mm) the
phase of the second signal varies against that of
the first as a function of frequency. The two
extreme values are obtained through the addition
and subtraction of the signals depending on the
frequency and this is shown as a ripple of ±10% in
the linear representation of the trace. Therefore,
the ripple of the trace corresponds to an important
As long as the external test setup is stable in its
characteristics, a calibration carried out with the
ZVR can be used for months for all
measurements tolerating an inaccuracy of a few
tenths of dB or a few degrees although the
operating temperature of the instrument has an
important influence. Let us assume the instrument
is calibrated when completely warmed up and is
then switched off overnight. The cold instrument
directly after power-up will show a deviation of
typically 0.05 dB and 2° which continuously
decreases to typically 0.005 dB and 0.1° as the
instrument is allowed to warm up for something
like an hour to attain thermal equilibrium. For
measurements for which this stability is
acceptable, a recalibration is only required when
some changes have taken place with the test
setup. In all other cases stored calibration data
may be used for many months.
If higher accuracy is required, the network
analyzer should be operated under constant
ambient conditions in an air-conditioned room and
be given at least two hours to warm up. Only for
the most stringent accuracy requirements will it be
necessary to carry out a calibration immediately
before each measurement.
29 May 1998
system characteristic of the analyzer, ie the
source match to be verified.
an open-circuit O and a matched load M. Whether
AutoKal can be used for planar connectors or not
depends on whether the standards required for a
TOM calibration are available in planar form.
Basically, AutoKal can also be linked to the TNA
technique for fundamental calibration. AutoKal
can then be used reliably for nearly all planar
applications. This is not yet possible in the ZVR
but can be implemented, if required.
This method may be used in the same way for
verifying the so-called effective source match
which is obtained after system error calibration
and correction. Values of 1% corresponding to
40 dB are typical for analyzers of the ZVR family.
19. How can a calibration be
performed in 3-port operation?
21. Can the case capacitance C 0 of a
one-port resonator be eliminated
in a TOM-X calibration?
The 3-Port Adapter option ZVR-B8 is controlled
by the analyzer via a rear-panel connector. The
path from PORT1 to PORT2 is throughconnected for the two odd channels CH1 and
CH3 and the path from PORT3 to PORT2 for the
two even channels CH2 and CH4. Three-port
calibration is then performed as two
independent 2-port calibrations for each of the
two paths. Any calibration method, eg TOM, may
be used for the measurements between PORT1
and PORT2 with the channel CH1 active and
between PORT3 and PORT2 with channel CH2
Yes. To do this, the calibration standards should
be fitted into the empty case of the resonator, eg
into the case of a crystal. What is essential is
good reproducibility, ie the case capacitance for
calibration and DUT measurements should be the
same. If this criterion is met to a sufficient degree,
the calibration method TOM-X, which is particularly suitable for eliminating X = crosstalk by
calculation, takes the case capacitance as a
source for the crosstalk. If the capacitance is
constant in the calibration and DUT measurements and is not DUT-dependent, the case
capacitance C0 is eliminated particular
effectively due to the full calibration model used
in the TOM-X calibration.
With the 3-port DUT connected and parameter
S21 displayed in channels CH1 and CH2 in the
DUAL CHANNEL SPLIT mode, S-parameter S21
of the 3-port DUT is measured in CH1 and
S-parameter S23 in CH2 as the 3-Port Adapter is
controlled as described above.
…regarding virtual
transformation networks:
20. Can AutoKal be used with a test
fixture in a planar environment?
Yes. Generally, calibrations are performed to
improve the measurement accuracy of the whole
network analyzer system, in other words the ZVR
including cables, adapters and all other components of the test setup. The calibration method
also determines the reference planes for the
vector measurements of complex S-parameters
of DUTs, irrespective of whether calibration is
carried out in a coaxial or planar environment.
The essential criterion is that the contacts to the
calibration standards and later on to the DUT are
reliable, reproducible and of good quality and that
the calibration standards feature the required
characteristics. In this respect there is no
difference between AutoKal and the other
calibration methods. The difference and an
advantage of the AutoKal technique is that a
complete 2-port calibration can be performed with
only one calibration standard, namely a throughconnection T. AutoKal needs a single fundamental calibration prior to its first use to determine
the characteristics of the transfer standards within
the AutoKal unit. TOM is the recommended
method for this fundamental calibration which
requires, in addition to the through-connection T,
22. What does "measuring in a
customer-specific environment"
The Virtual Embedding Networks Option ZVR-K9
allows any customized circuit environment, eg a
matching network or PCB to which the DUT will
be fitted later on, to be virtually integrated into
the test fixture of an automatic test system.
The function of a whole module, which may only
be theoretically known or does not exist at all, can
be tested in this way and yet only one critical
component needs to be measured. Thus an
overall function check can be made at an early
development stage and the suitability of a
component for the environment for which it is
intended can be guaranteed. Should a component
have to be changed on the PCB, the interplay with
the DUT can be tested immediately and the
number of rejects in production can thus be
considerably reduced.
29 May 1998
in planar structures using microstrip lines, for
instance. In this case the same rules should be
observed when dimensioning the through
standard T as are used for calculating the lines
around the DUT in the real circuit. Thus the
characteristic impedance obtained per
definition is ideal. Dispersion will not be
disturbing, on the contrary it is desirable that the
through standard shows the same dispersion as
the environment to which the DUT has to be
matched with low reflection.
23. Where do the data for the virtual
transformation network come
The virtual transformation network can be
determined directly by measuring
S-parameters, for instance when an existing
matching transformer should be replaced later on
by a virtual transformer with the aid of the option
ZVR-K9. Another possibility is to synthesize a
network with the aid of a simulation program
and to transfer the measured S-parameters to the
network analyzer, ie to implement the virtual
Care should also be taken that the impedance of
the A standard matches the characteristic
impedance of the T standard as accurately as
required, but this depends on how the A standard
has been realized. For instance, the A standard
may be made up of two concentrated resistors,
which unlike the line have almost no dispersion.
The standard may also be implemented in the
form of a line like the T standard, with a tapered
absorbing rubber used for attenuation.
The new calibration technique TNA (R&S patent)
allows the implementation and use of virtual
networks also in planar circuit environments
where conventional calibration techniques often
fail. The latter require calibration standards of
accurate and known characteristics, which in
practice often cannot be realized with sufficient
quality. The calibration technique TNA, on the
contrary, places low demands on the characteristics of the calibration standards. It permits
almost any unknown calibration standards to be
used and an accurate calibration to be performed
at the reference plane of the DUT within the test
fixture and at the coaxial interfaces of the DUT.
Thus, the networks between these interfaces can
also be determined and considered in the
calculation or, the other way round, the network
can be calculated as a difference and eliminated.
25. Are TNA standards traceable?
This question concerns the T (Through) and A
(Attenuator) standards since apart from reflection
symmetry no special demands are placed on the
N (Network) standard which is realized in most
cases by open test terminals, ie by simply
disconnecting the DUT. The decisive characteristics of the T and A standards have been
discussed in the previous section on the
characteristic impedance. Whether the standards
can be traced to national or international
standards depends on the type of line of the
calibration standards. As far as coaxial calibration
standards are concerned, the answer is clearly
24. How can it be guaranteed that TNA
standards have the desired
characteristic impedance?
Compared with other calibration techniques, the
TNA technique is the least demanding with regard
to having to know the characteristics of the
calibration standards. However, a few characteristics of the standards have to be known or
represented as exactly as possible. This is for one
the electrical length of the through-connection T
(Through) which determines the reference plane
for phase measurements in reflection and transmission. In addition, the input impedance at both
ends of the A standard (Attenuator) and the
characteristic impedance of the through-connection T should be of the same value as both
together determine the reference characteristic
impedance. If the impedances are different or if
one of them or both do not comply with the
desired reference impedance, measurement
errors will occur resulting in a reduced effective
directivity of the test setup.
In the case of planar structures, standards are
often only indirectly traceable. Theoretically,
even the most exotic structure can be traced back
to a primary standard. If required, this is checked
individually and performed with the aid of the R&S
calibration laboratory.
26. Can a dummy be used so that
calibration can be performed
within and outside the housing?
When performing a calibration for measurements
on housed components, the connectors both on
the inside and outside of the housing may be
used as reference planes. If a calibration is to be
performed at the connectors inside, empty
housings (dummies) can be modified - by
using calibration substrates, for instance - such
that they can be employed as calibration
standards. If the housings are well matched and
have a low loss, these standards can also be
In coaxial test setups these problems can be
easily mastered, but here the TNA technique will
hardly ever be necessary. TNA is of more interest
29 May 1998
used for calibration on the connectors on the
outside of the housing. The TNA calibration
technique (R&S patent) is especially suitable as it
comprises only the easy-to-realize calibration
standards T = Through, N = Network and
A = Attenuator. For T a direct through-connection
may be used instead of the DUT, N can be
realized with any unknown network featuring
reflection symmetry and is best obtained by
simply disconnecting the DUT, ie by leaving the
terminals open (for calibration outside the
housing) or by an empty housing (for calibration
within the housing). Standard A is characterized
by matching at both ends, any unknown crosstalk
between the test ports being acceptable. For
many applications, the mentioned standards can
be set up in the form of the housing or be fitted
into the empty DUT case. This has the additional
advantage that in automatic test stations both the
calibration standards and the DUTs can be
connected automatically to the test fixture.
27. Does embedding/ deembedding
reduce the measurement speed?
No. The data required for virtual transformation
networks are first mathematically linked to the
calibration data of the network analyzer via a
modification of the system error matrices. During
the measurement the modified matrices are used
instead of the original system error matrices.
Therefore, no additional calculations are
required during the measurement beyond the
usual, high-speed calculations for system error
correction. The measurement speed is as high
as with conventional system- error-corrected
measurements without the use of virtual
Olaf Ostwald, 1ES3
Rohde & Schwarz
19 January 1998
28. Can the virtual transformation
network be determined by
measurements only or can it also
be synthesized?
Yes, the virtual transformation network may be
designed and simulated, for instance, with the
aid of a common simulation program like
SuperCompact. The simulation calculation may
be carried out with the network analyzer using the
optional Computer Function ZVR-B15. The
S-parameters of the virtual transformation
network are transferred to the network analyzer in
the form of an FLP file (S-parameter data set for
multiports), combined with the system error
correction matrices so that modified matrices are
obtained which are then used in the subsequent
DUT measurement.
29 May 1998
Further Application Notes
Order designation
H.-G. Krekels: Automatic Calibration of Vector
Network Analyzer ZVR, Appl. Note 1EZ30_1E.
Vector Network Analyzers (test sets included) *
O. Ostwald: 4-Port Measurements with Vector
Network Analyzer ZVR, Appl. Note 1EZ25_1E.
T. Bednorz: Measurement Uncertainties for
Vector Network Analysis, Appl. Note 1EZ29_1E.
Ordering Information
O. Ostwald: 3-Port Measurements with Vector
Network Analyzer ZVR, Appl. Note 1EZ26_1E.
3-channel, unidirectional,
50 Ω, passive
3-channel, bidirectional,
50 Ω, passive
3-channel, bidirectional,
50 Ω, active
4-channel, bidirectional,
50 Ω, passive
4-channel, bidirectional,
50 Ω, active
3-channel, bidirectional,
50 Ω, active
4-channel, bidirectional,
50 Ω, active
P. Kraus: Measurements on FrequencyConverting DUTs using Vector Network Analyzer
ZVR, Appl. Note 1EZ32_1E.
J. Ganzert: Accessing Measurement Data and
Controlling the Vector Network Analyzer via DDE,
Appl. Note 1EZ33_1E.
Order No.
9 kHz to 4 GHz
9 kHz to 4 GHz
300 kHz to 4 GHz
9 kHz to 4 GHz
300 kHz to 4 GHz
20 kHz to 8 GHz
20 kHz to 8 GHz
Alternative Test Sets *
75 Ω SWR Bridge for ZVRL (instead of 50 Ω) 1)
J. Ganzert: File Transfer between Analyzers FSE
or ZVR and PC using MS-DOS Interlink, Appl.
Note 1EZ34_1E.
75 Ω, passive
9 kHz to 4 GHz
75 Ω SWR Bridge Pairs for ZVRE and ZVR (instead of 50 Ω) 1)
75 Ω, passive
75 Ω, active
O. Ostwald: Group and Phase Delay Measurements with Vector Network Analyzer ZVR,
Appl. Note 1EZ35_1E.
9 kHz to 4 GHz
300 kHz to 4 GHz
Time Domain
Mixer Measurements 2)
Reference Channel Ports
Power Calibration 3)
3-Port Adapter
Virtual Embedding
Networks 4)
4-Port Adapter (2xSPDT)
4-Port Adapter (SP3T)
0 to 8 GHz
same as analyzer
same as analyzer
same as analyzer
same as analyzer
0 to 4 GHz
same as analyzer
0 to 4 GHz
0 to 4 GHz
Controller (German) 5)
Controller (English) 5)
Ethernet BNC for ZVR-B15
Ethernet AUI for ZVR-B15
IEC/IEEE-Bus Interface for
Generator Step Attenuator
Generator Step Attenuator
PORT 2 6)
Receiver Step Attenuator
Receiver Step Attenuator
External Measurements,
50 Ω
same as analyzer
same as analyzer
same as analyzer
same as analyzer
10 Hz to 4 GHz
20 kHz to 8 GHz
O. Ostwald: Multiport Measurements using
Vector Network Analyzer, Appl. Note 1EZ37_1E.
[10] O. Ostwald: Frequently Asked Questions about
Vector Network Analyzer ZVR, Appl. Note
[11] A. Gleißner: Internal Data Transfer between
Windows 3.1 / Excel and Vector Network
Analyzer ZVR, Appl. Note 1EZ39_1E.
[12] A. Gleißner: Power Calibration of Vector Network
Analyzer ZVR, Appl. Note 1EZ41_2E
[13] O. Ostwald: Pulsed Measurements on GSM
Amplifier SMD ICs with Vector Analyzer ZVR,
Appl. Note 1EZ42_1E.
[14] O. Ostwald: Zeitbereichsmessungen mit dem
Netzwerkanalysator ZVR, Appl. Note 1EZ44_1D.
To be ordered together with the analyzer.
Harmonics measurements included.
Power meter and sensor required.
Only for ZVR or ZVC with ZVR-B15.
DOS, Windows 3.11, keyboard and mouse included.
For ZVR or ZVC only.
Step attenuators required.
* Note:
Active test sets, in contrast to passive test sets, comprise internal bias ne tworks,
eg to supply DUTs.
29 May 1998
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF