Network analyzer
A network analyzer ( English Network Analyzer , short: NWA , VNA or NA ) is used in electronics , communications technology and especially in high-frequency technology to determine the scattering parameters (S-parameters), i.e. the wave size of the reflection and transmission at electrical gates as a function of Measure frequency . Network analyzers are used in the field of electronic circuit development and as test equipment in production. They are not to be confused with the protocol analyzers used in computer networks .
Applications of network analyzers range from determining the transmission properties of filters or amplifiers , for example , to measuring complex transmission paths . Due to the generality of the measurement, network analyzers in combination with appropriate antennas and software for evaluating the measurement data can also be used as simple synthetic aperture radar (SAR), for example in the field of materials science to find foreign inclusions in material samples.
As an essential property in a network analyzer, the measurement object (short MO or DUT , English Device Under Test by built-in network analyzer measurement generators (transmitters)), for example, an electronic assembly such as a filter, fed and recorded at the same time the signal changes occurring by the test receiver in the network analyzer. In this way, the measurement of the transmitted and received signals can be designed as a relative measurement and the measured values can be related to one another. With a network analyzer, the DUT is not left in its regular application environment and different types of measurements are made there by a measuring device that behaves as passively as possible, as is the case with a spectrum analyzer, for example , but the DUT is supplied by the network analyzer on its own and in its Measure properties. That is why network analyzers are among the most extensive measuring devices in the field of electrical measurement technology .
principle
In the simplest case, the network analyzer generates a sinusoidal test signal via its test generator . The frequency range spans several decades; the range of devices covers practically all technically used frequency ranges between 10 Hz and 100 GHz. The frequency range actually used is, however, mostly restricted to a significantly narrower frequency range by the structure of the respective analyzer, the type of test and the properties of the test object. In the case of a linear test object (DUT), the test signal at a certain frequency causes a likewise sinusoidal response at its output, which generally differs in amplitude and phase position from the test signal.
Scalar network analyzer
A scalar network analyzer (SNA) - these devices are rarely used today as measuring devices - only records the different amplitudes of the test signal and the response signal supplied by the DUT and is of simple design.
Vector network analyzer
A vector network analyzer (VNA) - usually and also in the following text, a network analyzer is only understood as such - records the amplitude and phase position as a complex variable and can thus also express the S-parameters in complex values. Strictly speaking, no vectors are measured , but the phasor , composed of an amount and an angle. Each type of network analyzer measures both the test signal it generates and the response signal modified by the DUT and relates these signals to each other. As a result, the absolute value measurement, which is complex for high accuracy requirements, can generally be reduced to the relative measurement that can be realized with less effort.
Vector network analyzers are characterized by a number of inherent advantages:
- A system error correction as part of the calibration is only possible with complex-valued signal processing.
- Processes related to system error correction, such as the computational compensation of the measurement object recording or the so-called computational or virtual embedding of the test object in a physically non-existent external coupling network ( English embedding and deembedding ) are principally only possible with complex signal processing.
- A time domain analysis is only clearly possible with complex-valued measurement data, since the representation as a function of time requires the transformation of the measurement data into the time domain.
- Usual images such as the representation in the Smith diagram are only unambiguous through complex measurement data.
Network analyzers can be constructed according to the homodyne principle or, more complexly, according to the heterodyne principle :
- With the homodyne principle, there is only a single oscillator in the measuring device , which supplies the test signal and serves as an oscillator source for the mixer in the receiver.
- More complex network analyzers are based on the heterodyne principle, where the oscillator for the test generator is separated from the oscillators in the individual receivers, the so-called LO oscillators, and allows greater variation in measurements.
construction
A network consists of two or more ports as a test port designated. In order to be able to measure a two-port , for example a filter, a cable or an amplifier stage, in the S parameters, two ports are necessary. They are labeled P1 and P2 in the diagram on the device under test (DUT) . Conventional network analyzers therefore have two, larger devices with four or more ports in order to be able to measure multiple ports without laborious repositioning. External changeover switches, so-called switching matrices, can be used to achieve an even higher number of goals.
Each test port on a conventional network analyzer can be operated both as a transmitter and as a receiver, the test ports are symmetrical and each have a dedicated test port circuit built into the NA. In order to reduce asymmetries in the necessary in the course of measurement may repositioning of the specimen and of the different properties of plugs and sockets NA also with special "genderless connectors" as the APC compound ( English Precision Connector equipped), in the drawing with A1 and A2 designated. In addition, N connectors with high-quality test cables are also used .
The measuring gate circuit installed in each gate of the NA basically consists of the following components, as shown in the block diagram:
- A directional coupler ( DC ) directly at the measuring gate, which separates the wave running out of the gate of the NA and the wave entering the gate. The incoming wave (reflected at the DUT) is decoupled and fed directly to a measuring receiver assigned to the gate, referred to as the RX test in the block diagram .
- In order to be able to actively feed the gate from the NA, there is a power splitter behind the input directional coupler , which splits the test signal generated by the generator: One part is fed directly to the second measuring receiver, referred to in the block diagram as RX-Ref , the other part is sent to the Supply of the measurement object used.
- Furthermore, a test port has two permanently assigned receivers: A receiver ( RX test ) for measuring the signal coming externally at the input and a measuring receiver ( RX-Ref ) for measuring the generator signal .
- Depending on the scope of the equipment, the NA only has one switchable test generator, as shown in the picture, which can optionally be switched to one of the outputs. More complex NAs have several independent test generators. Each test generator can usually also be varied in its amplitude via a downstream adjustable attenuator .
More extensive network analyzers also offer, additional options, the individual compounds in the Messtorschaltung via outwardly guided calibration lines to separate. Normally, these attenuators are closed as a bridge connection; by opening the bridges, other connections can be made in the entrance area if necessary. For example, a direct signal feed to the input of the measuring receiver is possible, bypassing the directional coupler on the input side, or additional attenuators can be connected in front of the receivers.
The individual measuring receivers - four measuring receivers are required for a network analyzer with two symmetrical ports as shown in the block diagram - have a local oscillator independent of the test generator, which can be changed independently of the test generator and converts the received signals into the intermediate frequency position (IF) with the usual heterodyne principle . This IF is digitized by fast analog-digital converters . In the subsequent digital signal processing in a microprocessor , in combination with special hardware such as field programmable gate arrays (FPGAs) and I / Q demodulators built therein, complex-valued baseband signals are generated, which form the raw measured values. After running through the system error correction, the individual scattering parameters of the measurement object are calculated numerically from the raw measured values.
Measurement sequence
The measurement of the scattering parameters, each at a specific frequency, can be described as the following simplified sequence, as shown on the two-terminal pole shown on the right:
- S 11 is determined from the ratio of reflected to transmitted signal. It represents the entrance reflection at the first gate.
- S 21 is determined from the ratio of transmitted to transmitted signal. It represents the forward transmission from the first to the second gate of the measurement object.
In order to determine all four parameters of the two-terminal network, the test generator is switched to the other port with the switch SW1 . The sequence of both of the above points is repeated in a mirror-inverted manner in order to determine the two missing parameters S 22 (output reflection at gate 2) and S 12 (backward transmission from gate 2 to gate 1). The switching process is not required for network analyzers with their own test generator per port. Each of the four S-parameters has a complex value, i.e. it consists of the specification of an amount and an angle.
This run ( English sweep ) is performed automatically over the selected frequency range with a certain spectral step size. Practically all network analyzers have the option of saving tables of measured values on data carriers or transferring them via data connections such as LAN connections . A common data exchange format is the Touchstone file format, which can also be read in directly as a data record in many programs for circuit simulation and processed further.
Network analyzers with a built-in screen display the measured S-parameters as amplitude or phase response as a function of frequency or in a complex representation in a Smith chart. The representation in the Smith diagram is of interest for the input and output reflection (S 11 and S 22 ). It serves, for example, the proper impedance matching ( english matching ) for power adjustment to be determined.
Some network analyzers also offer further display options, such as the display of the group delay over the frequency. The group delay of the selected S-parameter is determined by deriving the phase response, the numerical calculation and display as a diagram is carried out by the computer built into the measuring device.
Calibration and system error correction
The individual components in the test port circuit of a network analyzer and the additional components used such as measuring cables are faulty. A distinction must be made between two basic types of errors:
- Random measurement errors that occur, for example, in the context of noise . These deviations can only be recorded statistically and cannot be minimized with the system error correction.
- Systematic measurement deviations such as the frequency and phase response of the measuring device and the cables are characterized by the fact that they are reproducible and unchangeable over time and have a strong effect on the measurement results, especially at higher frequencies. They can be largely compensated for by means of a numerical system error correction.
For system error correction it is necessary to first record the systematic measurement deviations in order to then be able to remove them from the measured raw measurement values. This measurement of the system errors takes place within the scope of the calibration of the network analyzer, usually immediately before the actual measurement and with a physical setup that comes as close as possible to the actual measurement setup. As part of the calibration, the components used later in the actual measurement, such as coaxial cables, plug connections or additional attenuators, should also be recorded. Nevertheless, minor systematic deviations can occur in the course of long-lasting measurement processes, for example as a result of a temperature change in the device. In these cases it may be necessary to repeat the calibration at certain intervals.
For calibration, instead of the measurement object, different calibration standards with known electrical properties are used one after the other and the measured values that occur are determined. Since the properties of the calibration standards are already known within the scope of a certain error, the deviations measured are a consequence of the systematic deviation. In the subsequent measurements, the raw measured values obtained are offset against the error coefficients determined during the calibration and the errors systematically caused by the network analyzer and the structure are compensated for.
Any change in the measuring frequency range, such as higher or lower frequencies, changing the measuring lines, modifications to the connections, changes to the attenuators and the like, make a recalibration necessary. Depending on the method and number of measuring ports, several measurements must be carried out with the corresponding standards for calibration. A calibration process can therefore take several minutes. The time also depends on the measurement configuration of the NA. The more measuring points are within the desired frequency range and the longer the waiting time (settling time) is for a measuring point, the longer the calibration takes. In addition, there are also automatic calibration devices that combine the different standards required for calibration in a compact housing. During the calibration process, in which the standards are automatically switched on or over by the network analyzer, manual rewiring is no longer necessary.
There are a number of different calibration methods that differ according to the possibilities of the network analyzer, the effort and accuracy requirements. They are usually named after the first letters of the calibration standards used:
- OSL or MSO: Open-Short-Load or Match-Short-Open
- SOLT: Short-Open-Load-Through
- TAN: Through Attenuation Network
- TRL: Through-Reflect-Line
When calibrating a two-port device, there are generally twelve possible systematic error variables; with the 12-term error model, all are recorded during calibration. In relation to the first goal to the second goal, the six error sizes are:
- Parasitic reflection right on the first gate
- Attenuation and phase shift of the transmission line at the first port
- Mismatch between the test generator and the target (reflection measurement)
- Attenuation and phase shift of the transmission line at the second port
- Mismatch of the measurement object at the receiver input (transmission measurement)
- Crosstalk from first gate to second gate
Since a two-goal is usually measured from both sides, these six error sizes occur again mirror-symmetrically from the second goal to the first goal, which results in a total of twelve systematic error sizes.
Calibration standards
Some of the common calibration standards are described below.
Open
In an Open (German Open ) is the measurement line defined open, that is connected to "nothing". An open end of the line causes total reflection of the transmitted signal. If you look at the complex data of a reflection measurement at a port of the network analyzer in the Smith diagram, the open defines the point infinite on the X-axis.
Short
In a short (German short-circuit ) is the measurement line defined by the line shielding (ground), d. H. shorted. A short-circuited line end also causes total reflection of the transmitted signal, but the phase of the signal is rotated by 180 ° compared to the open . In the Smith chart, the short defines the point zero on the X-axis.
match
In the match calibration (German Restated ) the measuring line is terminated with the characteristic impedance. A value of 50 Ω is usual. If the measuring port is closed with its characteristic impedance, no signal reflections occur, in the Smith diagram the match defines point one on the X axis, i.e. the center of the diagram. This point is often referred to as system impedance in connection with network analyzers .
Through
When Through measurement (German Continuous ) two gates are connected to each other via a defined link. Since the Through standard, unlike the previous calibration standards, has two ports, it is counted among the two port standards.
Reflect
The Reflect standard represents a more general form of short or open , in which the exact properties do not have to be known. It can therefore only be used for calibration procedures that have at least one self-calibration standard, which is a standard that is not fully known. Suitable calibration methods for Reflect are, for example, TRL or TRM .
Attenuation
The attenuation standard is the same as the Reflect a Selbstkalibrierstandard is, is contrary to this but a two-port, which is connected between the two gates to be calibrated. The attenuation standard should have an insertion loss that is as constant as possible, the exact value of which does not have to be known. Furthermore, it must have reciprocal behavior, that is, it must not have any direction-dependent properties.
Line
The Line standard is similar to the through a non-reactive as possible two-port whose impedance must be known. The electrical length must be known exactly if this line standard replaces a through . If TRL is fully programmed, the length of L only needs to be known to ± 90 °.
Frequency converting measurements
Using appropriate network analyzers, additional software with special calibration procedures such as Without Thru , as well as two further calibration standards, a comb generator and a power sensor, frequency-converting measurements are also possible. In addition to vector harmonic measurements, calibrated vector intermodulation and mixer measurements can also be carried out. The vector information allows a description of non-linear effects and their localization. In addition, such an NWA can be used as a precise sampling oscilloscope, as mismatches to the measurement object, in contrast to oscilloscopes, are eliminated by the system error correction of the network analyzer.
Manufacturer overview
Well-known manufacturers of professional network analyzers:
| Manufacturer | Current devices | Former devices |
|---|---|---|
| Rohde & Schwarz | ZNA, ZNB | ZVA, ZVB |
| Keysight (formerly Agilent, HP) | PNA, ENA | |
| Anritsu |
For some years now, entry-level devices have also been commercially available for less than a hundred euros, mainly from amateur radio . Although these offer significantly worse properties than professional devices, they are often sufficient for private users.
literature
- Michael Hiebel: Basics of vector network analysis . 3. Edition. Rohde & Schwarz Publication, 2006, ISBN 978-3-939837-05-3 ( online ).
- Joachim Müller: practical introduction to vector network analysis . beam-Verlag, Marburg 2011, ISBN 978-3-88976-159-0 .
Web links
- Calibration of network analyzers (PDF; 203 kB)