Text preview for : 5988-5635EN.pdf part of HP 5988-5635EN HP Publikacje 5988-5635EN.pdf
Back to : 5988-5635EN.pdf | Home
Multiport & Balanced
Device Measurement
Application Note Series
Concepts in Balanced Device
Measurements
Application Note 1373-2
Introduction
The use of differential components such as surface
acoustic wave (SAW) filters and differential amplifiers
is becoming more common in the wireless industry
because they have greater performance than their sin-
gle-ended counterparts, such as the traditional single-
ended three-terminal duplexer filters used in mobile
handsets. Accurate measurement of these components
is challenging as most vector network analyzers have
single-ended RF ports that cannot directly measure dif-
ferential parameters. However, there are several alter-
native methods of obtaining the differential parameters
needed to characterize these devices. This article
describes the challenges designers face in measuring
the performance of differential components, and
describes each of the most widely used techniques.
While each technique produces a specific level of accu-
racy that depends on the characteristics of the device
to be tested, the "calculated mix-mode" method pro-
vides the most accurate device characterization and
has the fewest drawbacks.
Differential devices
Two types of differential components, one with a single-
ended output and the other with a differential output
are shown in figure 1. The differential port of the
devices consists of a pair of physical terminals.
Differential components are unique in that signals are
referenced not only to a common ground but to each
other as well. The signals referenced to each other are
called "differential mode" and the signals referenced to a
common ground are called "common mode." Differential
components can have both common mode and differen-
tial mode signals.
In most cases, the differential mode signals are out of
phase because their phase relative to each other is 180
degrees, which creates a virtual ground along the axis of Figure 1: Two types of differential components, showing single-ended and differential
symmetry of the device. At the virtual ground, the outputs.
potential at the operating frequency does not change
with time, regardless of the signal amplitude. Common-
mode signals are induced at the terminals of the device
with the same phase and amplitude relationship.
While a differential component has no performance
advantage over a single-ended component when used in
common mode, it exhibits significant benefits when used
in differential mode, because it will pass differential-
mode signals and reject common-mode signals. For
example, noise from a power supply will affect both ter-
minals of the device equally with the same phase rela-
tionship. The device will characterize this noise as a
common-mode signal on its terminals, and keep it from
passing through. Since no device is ideal, some of the
differential signal applied is converted into the common
mode and some of the common-mode signal is converted
into the differential mode as shown in figure 2. This is Figure 2: Mode conversion occurs when some of the differential signal is converted
into the common mode and some of the common-mode signal is converted
referred to as mode conversion, and is directly related to into the differential mode.
noise immunity of the device.
2
Measurement obstacles
While vector network analyzers are typically employed Another method uses a mathematical transform to con-
to measure an RF component, most are not designed for vert the single-ended data to differential parameters,
measuring differential parameters. Their RF ports are and is called the mixed-mode S-parameter technique. It
single-ended with common impedance values and cannot provides the common-mode and differential-mode
supply differential and common-mode signals to the parameters of the device, and is similar to single-ended
device. Single-ended devices have impedances of 50 to measurement except that instead of stimulating a single
75 ohms, while differential components have no stan- terminal of the DUT, pairs of terminals are considered
dard impedance values. To measure a four-terminal to be stimulated in either a differential (out-of-phase) or
(two-port differential) component requires 16 S-parame- a common (in-phase) mode. A physical differential/com-
ters. These single-ended S-parameters are not sufficient mon-mode stimulus to the device is not being provided.
to accurately characterize a differential component oper- A single stimulus signal is actually supplied to each of
ating in differential mode. As a result, measuring differ- the ports, and the response is measured. This single-
ential components accurately using a single-ended ended data can then be transformed to mixed mode. The
analyzer cannot be accomplished without applying some mixed-mode S-parameter technique essentially seeks to
type of hardware or software conversion to the single- determine (with a differential-mode stimulus on a differ-
ended data. Two common approaches use either a balun, ential port) the corresponding differential and common
or perform some type of mathematical transform. mode responses on all of the device ports. For a com-
mon-mode stimulus, it attempts to determine the differ-
The balun-based technique requires the balun to be ential-mode and common-mode responses. A
placed between the differential port of the device and mixed-mode S-parameter matrix can be organized in a
the single-ended port of the analyzer. The balun trans- way similar to the single-ended S-parameter matrix, in
forms the single-ended signal of the analyzer to a differ- which each column represents a different stimulus con-
ential signal that is applied to the device. There are a dition, and each row represents a different response
number of problems with this approach. While the meas- condition (figure 3). Unlike the single-ended example,
urement plane should be at the terminals of the device, the mixed-mode S-parameter matrix not only considers
this is difficult to realize because there are no standard the port but the mode of the signal at each port.
calibration standards for differential mode. As a result,
a calibration must be performed at the single-ended The naming convention for the mixed-mode S-parame-
input side of the balun. The analyzer is now measuring ters must include mode information as well as port
the performance of the differential component and information. Consequently, the first two subscripts in
balun as one device. The balun has finite return loss, the matrix describe the mode of the response and stimu-
insertion loss, amplitude balance, phase balance, and lus respectively, and the next two subscripts describe
bandwidth, and becomes a major limiting factor in meas- the ports of the response and stimulus. The mixed-mode
uring the component. A balun also will not pass com- S-parameter matrix fully describes the linear perform-
mon-mode signals, so none of the mode conversion ance of a differential two-port network.
parameters of the device can be measured. Finally, the
balun method will only provide information about the
differential mode of the component.
Figure 3: A mixed-mode S-parameter matrix can be organized in a similar way to the
single-ended S-parameter matrix, in which each column represents a
different stimulus condition, and each row represents a different response
condition.
3
Differential measurement techniques
The techniques employed to measure a differential
device in the following discussion are based on measure-
ments made of a three-terminal SAW filter in an LTCC
antenna switch module using an Agilent E8358A PNA
Series vector network analyzer. The results are shown in
figures 7 and 8.
The measured results are saved first into S-parameter
files (s2p) using a two-port analyzer. For a three-termi-
nal device this will require three two-port measurements
resulting in three files for a total of 12 S-parameters.
The next step is to import the files into ADS using the
ADS instrument server as shown in figure 4. The data is
read into Agilent's Advanced Design System (ADS)
datasets with names that were specified during the ADS Figure 4: The ADS instrument server is used to import S-parameter files
read process. A three-terminal device will have a total
of nine S-parameters, which will require the cancellation
of redundant terms in the 12 S-parameters. This cancel-
lation procedure is performed by mapping the three
datasets to one dataset as shown in figure 5. The result-
ing dataset is then used for the analysis. A three-port
analyzer would directly provide the nine S-parameters,
which could then be imported into ADS.
A simple extension of the mixed-mode concept can be
applied to devices having a combination of differential
and single-ended ports, as is the case with the three-ter-
minal SAW filter. The four-terminal matrix can be con-
verted to a three-terminal matrix by removing the port 1
differential mode stimulus and response as illustrated
by the shaded row and column in figure 3, page 3. In
this scenario, the differential and common modes on the
differential ports and one mode on the single-ended port Figure 5: Cancel redundant terms by mapping three datasets to one
must be considered.
The S-parameter matrix (figure 6) for such a device is
arranged with the stimulus conditions in the columns,
and the response conditions in the rows. The mode on
the single-ended port is referenced with an 'S' for single-
ended instead of a 'C' for common mode because only
one mode is available on this port. Two columns and
two rows describe each differential port, and one col-
umn and one row describe each single-ended port. In
this case the four parameters in the lower right corner
describe the four types of reflection that are possible on
a differential port. The single parameter in the upper
left describes the reflection on the single-ended port,
and the other four parameters describe the differential
and common mode transmission characteristics in the
forward and reverse directions. Figure 6: When the four-terminal matrix is converted to a three-terminal matrix, the
differential and common modes on the differential ports and one mode on
the single-ended port must be considered. The S-parameter matrix for this
configuration is arranged with the stimulus conditions in the columns, and
the response conditions in the rows.
4
Single-ended method Delta method
Measuring the differential device as a single-ended mul- With the delta method, the single-ended transmission
tiport device is the simplest method, but it is time-con- phase characteristics of the device are measured. The
suming because multiple two-port measurements are topology of most differential devices will constrain the
needed to fully characterize the device. In addition, it electrical length of the two terminals comprising the dif-
can produce misleading results because the single-ended ferential port to give a 180-degree phase shift between
data may not give a representative indication of the per- them. This parameter is directly related to how well the
formance of the device when it operates in one of its dif- device performs in differential mode. Figure 8 shows the
ferential modes. This occurs because the single-ended phase difference between the two terminals (S31 and
data does not provide accurate information of differen- S21) of the differential port measured single-ended as
tial performance. For example, S21 is the insertion loss
with the single-ended method. There should be 180
measurement from the antenna terminal (terminal 1) to degrees of phase difference between the two terminals.
the rx+ terminal (terminal 2). It is not the same as the The difference in phase shown in the figure is not exact-
insertion loss measurement from the antenna to the dif- ly 180 degrees, which results from the asymmetries of
ferential port. In figure 7, there is offset between the val- the device. This method also does not yield insight into
ues of S31 and S21, which represents the overall balance the full mixed-mode S-parameter matrix.
between the two terminals that make up the differential
port. This offset may be caused by an asymmetrical
device topology that will result in a decrease in differen-
tial mode performance. Ideally, S31 and S21 should have
the same amplitude characteristics.
Figure 7: Offset between the values of S31 and S21 represents the overall balance Figure 8: The phase difference between the two terminals (S31 and S21) of the
between the two terminals that make up the differential port. This image differential port measured single-ended as with the single-ended method.
also contains the comparison of the balun methods and mixed-mode
methods.
Physical balun method
As was discussed earlier, a balun may be used to convert
the single-ended port of the network analyzer to the dif-
ferential port of the device, which transforms the imped-
ance of the differential device to the impedance of the
network analyzer. In this case, the differential port
impedance is 100 ohms and the single-ended port
impedance of the analyzer is 50 ohms. This method will
provide some degree of accuracy about the differential
characteristics of the device but does not provide infor-
mation on common-mode performance. The accuracy of
this method is also highly dependent on the calibration
reference plane and the characteristics of the balun.
5
Mathematical "ideal balun" method Simulated mixed-mode method
The single-ended data may also be imported into a cir- A circuit simulator can also be used to measure mixed-
cuit simulator such as Agilent's ADS. This data can then mode S-parameters of the differential device as illustrat-
be transformed to differential data using a balun circuit ed in figure 10. A center-tapped balun is used to
component in the simulator (figure 9). As with the phys- perform the differential-mode conversion and also pro-
ical balun method, the common-mode performance of vides the mechanism for the common-mode terms. The
the device cannot be measured. The circuit component is common-mode conversion occurs at the center tap of the
an ideal balun, so the common-mode impedance is infi- balun where only common-mode signals will appear
nite (where the non-center tapped reflection coefficient because of the characteristics of the balun. These com-
equals +1). Any common-mode signals at the output of mon-mode signals are then terminated through a balun
the device will reflect from the balun and possibly back into a 25-ohm termination, which is the common-mode
to the output as an error signal, depending on the impedance of the SAW device. This configuration will
mixed-mode performance of the device. The same will be allow all the mixed-mode characteristics of the device to
true when using a physical balun, but the reflection be measured. It also provides the appropriate termina-
coefficient will differ depending on its characteristics. tions for differential- and common-mode signals so that
The mixed-mode performance of the device cannot be mode conversion terms do not cause errors like those
measured using a balun, so there is no way to determine produced by the balun method.
what the error result may be. The same is true for the
center-tapped balun (where the reflection coefficient
equals -1).
Figure 9: If the single-ended data is imported into Agilent's Advanced Design System Figure 10: A circuit simulator can also be used to measure mixed-mode parameters
(ADS) simulation tool, it can be transformed to differential data using a of the differential device.
balun circuit component.
Calculated mixed-mode S-parameters
method
Bocklemann and Eisenstad1 have analyzed a method to
convert the single-ended data to mixed-mode using
mathematical algorithms. These algorithms show the
relationship between nodal waves generated by a stan-
dard vector network analyzer and the associated com-
mon and differential waves that realize mixed-mode
S-parameters. This method is highly beneficial because
of the quick and simple method of conversion. It does
not require a circuit simulator and therefore can be per-
formed in real time using a compiled math function
library. For example, the mixed-mode S-parameters of a
differential device can be accurately measured in a man-
ufacturing environment in which differential measure-
ment speed and accuracy are of high concern.
6
Balun versus mixed-mode method
The results obtained from the mathematical balun and
mixed-mode measurement methods can be quite differ-
ent as shown in figure 11. The device under test used in
the example exhibited mode conversion, and in conjunc-
tion with the balun, induced an error in the measure-
ment. To continue this discussion, the differential
insertion loss (Sds21) of the three-terminal SAW filter is
measured with a balun at the output to perform the dif-
ferential-to-single-ended conversion to the network ana-
lyzer. The balun is again assumed to be ideal (lossless)
in this analysis.
The network analyzer supplies a signal to the single-
ended port of the device, and this signal is attenuated by
the device's insertion loss and shows up at the output of
the device as a differential signal. The signal is then con-
Figure 11: The error due to the mode conversion of the device can be calculated by
verted back to single-ended using the balun so it can be comparing the mathematical balun results to the mixed-mode results.
measured using the network analyzer. This signal is the
desired differential measurement result using the balun
method. A portion of the incident signal that was
applied to the input of the device is converted to com-
mon mode (Scs21) at the output, which is called mode
conversion. The common-mode signal encounters the
common-mode match (Scc11) of the balun, which has a
common-mode reflection coefficient of +1. A grounded
center-tapped balun would have a common-mode reflec-
tion coefficient of