Menu
homeNavigation

Input and output filters for a frequency converter - purpose, principle of operation, connection, features. What is electromagnetic noise

In industry, a significant part of the consumption electrical energy fall on ventilation, pumping and compressor installations, conveyors and lifting mechanisms, electric drives of technological installations and machine tools. These mechanisms are most often driven by asynchronous motors. alternating current. To control operating modes induction motors, including to reduce their energy consumption, the world's largest manufacturers of electrical equipment offer specialized devices - frequency converters. Without a doubt, frequency converters (which are also called frequency converters, inverters or IF for short) are extremely useful devices that can greatly facilitate the starting and operation of asynchronous motors. In some cases, however, frequency converters can also have a negative effect on the connected motor.

Due to the design features of the frequency converter, its output voltage and current have a distorted, non-sinusoidal shape with a large number of harmonic components (noise). The uncontrolled rectifier of the frequency converter consumes a non-linear current that pollutes the power supply network with higher harmonics (5th, 7th, 11th harmonic, etc.). PWM - frequency converter inverter generates a wide range of higher harmonics with a frequency of 150 kHz-30 MHz. The supply of motor windings with such a distorted non-sinusoidal current leads to such negative consequences as thermal and electrical breakdown of the motor winding insulation, an increase in the aging rate of the insulation, an increase in the level of acoustic noise of a running motor, and bearing erosion. In addition, frequency converters can be a powerful source of interference in the electrical supply network, having a negative impact on other electrical equipment connected to this network. To mitigate the negative impact of harmonic distortion generated by the inverter during operation, on electrical network, the motor and the frequency converter itself use different filters.

The filters used in conjunction with frequency converters can be conditionally divided into input and output filters. Input filters are used to suppress the negative influence of the rectifier and PWM inverter, output filters are designed to combat interference generated by the PWM inverter and external sources of interference. Input filters include mains chokes and EMI filters (RF filters), output filters: dU / dt filters, motor chokes, sine filters, high-frequency common-mode noise filters.

Line chokes

The mains choke is a two-way buffer between the power supply network and the frequency converter and protects the network from higher harmonics of the 5th, 7th, 11th order with a frequency of 250Hz, 350Hz, 550Hz, etc. In addition, line chokes help protect the frequency converter from increased mains voltage and current surges during transients in the mains supply and load of the inverter, especially during a sharp surge in mains voltage, which happens, for example, when powerful asynchronous motors are turned off. Mains chokes with a specified voltage drop across the winding resistance of about 2% of the nominal value of the mains voltage are designed for use with frequency converters that do not regenerate the energy released when the motor is braked back into the power supply system. Chokes with a specified voltage drop on the windings of about 4% are designed to operate combinations of converters and autotransformers with the function of regenerating the braking energy of the engine into the power supply system.

  • if there is significant interference from other equipment in the power supply network;
  • when the asymmetry of the supply voltage between the phases is more than 1.8% of the nominal voltage value;
  • when connecting the frequency converter to a supply network with a very low impedance (for example, when the inverter is powered from a nearby transformer, the power of which is more than 6-10 times greater than the power of the inverter);
  • when connecting a large number of frequency converters to one power line;
  • when powered from a network to which other non-linear elements are connected that create significant distortions;
  • if there are capacitors (reactive power compensators) in the power supply circuit of the batteries, which increase the power factor of the network.

Advantages of using line chokes:

  • Protect the frequency converter from voltage surges in the network;
  • Protect the frequency converter from phase imbalances of the supply voltage;
  • Reduce the rate of rise of short-circuit currents in the output circuits of the frequency converter;
  • Increase the service life of the capacitor in the link direct current PC.

EMI filters

In relation to the supply network, the variable frequency drive (FC+motor) is a variable load. Together with the inductance of the power cables, this leads to high-frequency fluctuations in the mains current and voltage and, consequently, to electromagnetic radiation (EMR) from the power cables, which can adversely affect the operation of other electronic devices. EMI filters are necessary to ensure electromagnetic compatibility when installing the converter in places that are critical to the level of interference from the mains supply.

Design and scope of dU/dt filters

The dU/dt filter is an L-shaped filter low frequencies consisting of chokes and capacitors. The inductance ratings of the chokes and capacitors are selected in such a way as to ensure the suppression of frequencies above the switching frequency of the power switches of the FC inverter. The inductance of the filter inductor winding dU / dt is in the range from several tens to several hundreds of μH, the capacitance of the filter capacitors dU / dt is usually in the range of several tens of nF. By using a dU/dt filter, it is possible to reduce the peak voltage and the ratio of dU/dt pulses at the motor terminals to approximately 500 V/µs, thereby protecting the motor winding from electrical breakdown.

  • Frequency controlled drive with frequent regenerative braking;
  • Drive with a motor that is not designed for frequency converter operation and does not comply with IEC 600034-25;
  • Drive with old motor (low insulation class), or with motor general purpose not complying with the requirements of IEC 600034-17;
  • Drive with short motor cable (less than 15 meters);
  • Variable frequency drive, the motor of which is installed in an aggressive environment or operates at high temperatures;

Since the dU / dt filter has relatively low inductance and capacitance values, the voltage wave on the motor windings still has the form of bipolar rectangular pulses instead of a sine wave. But the current flowing through the motor windings already has the shape of an almost regular sinusoid. dU/dt filters can be used at switching frequencies below nominal value, but avoid using them at a switching frequency higher than the nominal value, as this will cause the filter to overheat. dU/dt filters are sometimes called motor chokes. Most motor chokes are designed without capacitors, and the coil windings have a higher inductance.

Design and scope of sinus filters

The design of sine filters (sine filters) is similar to the design of dU / dt filters, with the only difference that they have larger chokes and capacitors that form an LC filter with a resonance frequency of less than 50% of the switching frequency (carrier frequency of the PWM inverter). This provides more effective smoothing and suppression of high frequencies and a sinusoidal form of phase voltages and motor currents. The value of the inductance of the sine filter is in the range from hundreds of μH to tens of mH, the capacitance of the sine-wave filter capacitors is from units of μF to hundreds of μF. Therefore, the dimensions of sine filters are large and comparable to the dimensions of the frequency converter to which this filter is connected.

When using sine filters, there is no need to use special motors with reinforced insulation certified for operation with frequency converters. It also reduces acoustic noise from the motor and bearing currents in the motor. The heating of the motor windings caused by the presence of high frequency currents is reduced. Sine filters allow longer motor cables to be used in applications where the motor is installed far from the frequency converter. At the same time, the sine filter eliminates impulse reflections in the motor cable, thereby reducing losses in the frequency converter itself.

  • When it is required to eliminate acoustic noise from the motor during switching;
  • When starting old engines with worn insulation;
  • In case of operation with frequent regenerative braking and with motors that do not comply with the requirements of IEC 60034-17;
  • When the engine is installed in an aggressive external environment or works at high temperatures;
  • When connecting motors with shielded or unshielded cables from 150 to 300 meters long. The use of motor cables longer than 300 meters depends on the specific application.
  • If necessary, increase the engine maintenance interval;
  • When stepping up the voltage or in other cases when the frequency converter is powered by a transformer;
  • With general purpose motors using 690V.

Sine filters can be used with a switching frequency higher than the nominal value, but they cannot be used with a switching frequency lower than the nominal value (for this filter model) by more than 20%. Therefore, in the settings of the frequency converter, it is necessary to limit the minimum possible switching frequency in accordance with the rating data of the filter. In addition, when using a sine filter, it is not recommended to increase the frequency of the inverter output voltage above 70 Hz. In some case, it is necessary to enter the capacitance and inductance values ​​of the sine filter into the inverter.

During operation, a sine filter can release a large amount of thermal energy (from tens of W to several kW), so it is recommended to install them in well-ventilated places. Also, the operation of a sine filter may be accompanied by the presence of acoustic noise. At the rated load of the drive, a voltage of about 30 V will drop on the sine filter. This must be taken into account when choosing an electric motor. The voltage drop can be partly compensated by reducing the field weakening point in the frequency converter settings, and up to this point the correct voltage will be applied to the motor, but at rated speed the voltage will be reduced.

dU/dt chokes, motor chokes and sine filters must be connected to the frequency converter output with the shortest possible shielded cable. Maximum recommended cable length between frequency converter and output filter:

  • 2 meters with drive power up to 7.5 kW;
  • 5-10 meters with drive power from 7.5 to 90 kW;
  • 10-15 meters with drive power above 90 kW.

Design and scope of high-frequency common mode filters

The high frequency common mode filter is a differential transformer with a ferrite core, the "windings" of which are the phase wires of the motor cable. The high-pass filter reduces high-frequency common-mode currents associated with electric discharges in the motor bearing and also reduces high-frequency emissions from the motor cable, for example, in cases where unshielded cables are used. The ferrite rings of the high-frequency common-mode filter are oval-shaped for ease of installation. Through the hole in the ring, all three phase wires of the motor cable are passed, connected to the output terminals U, V and W of the frequency converter. It is important to pass all three phases of the motor cable through the ring, otherwise it will saturate. It is equally important not to pass the protective earth wire PE, any other earth wires or neutral conductors through the ring. Otherwise, the ring will lose its properties. In some applications, it may be necessary to assemble a package of several rings to prevent their saturation.

Ferrite rings can be installed on the motor cable at the frequency converter output terminals (terminals U, V, W) or in the motor connection box. Installing high-frequency filter ferrite rings on the terminal side of the frequency converter reduces both the load on the motor bearings and the high-frequency electromagnetic interference from the motor cable. When installed directly in the motor junction box, the common mode filter only reduces the load on the bearings and does not affect the EMI from the motor cable. The required number of rings depends on their geometric dimensions, the length of the motor cable and the operating voltage of the frequency converter.

During normal operation, the temperature of the rings does not exceed 70 °C. Ring temperatures above 70 °C indicate saturation. In this case, you need to install additional rings. If the rings continue to saturate, this means that the motor cable is too long, there are too many parallel cables, or a high capacitance cable is being used. Also, do not use a sector-shaped cable as a motor cable. Only cables with round cores should be used. If the temperature environment above 45 - 55 °C, the derating of the filter becomes very significant.

When using several parallel cables, the total length of these cables must be taken into account when choosing the number of ferrite rings. For example, two cables of 50 m each are equivalent to one cable of 100 m. If many parallel motors are used, a separate set of rings must be installed on each of them. Ferrite rings can vibrate when exposed to an alternating magnetic field. This vibration can wear the insulation material of the ring or cable through gradual mechanical abrasion. Therefore, the ferrite rings and the cable should be firmly fixed with plastic cable ties (clamps).

Frequency converters, like many other electronic converters powered by AC mains with a frequency of 50 Hz, due to their device alone, distort the shape of the consumed current: the current does not depend linearly on voltage, since the rectifier at the input of the device is usually normal, i.e. uncontrollable. So are the output current and voltage of the frequency converter - they also differ in a distorted shape, the presence of many harmonics due to the operation of the PWM inverter.

As a result, in the process of regularly supplying the motor stator with such a distorted current, its insulation ages faster, the bearings deteriorate, the motor noise increases, and the likelihood of thermal and electrical breakdowns of the windings increases. And for the network that feeds, this state of affairs is always fraught with interference that can harm other equipment powered by the same network.

To get rid of the problems described above, additional input and output filters are installed to frequency converters and motors, which save both the supply network itself and the motor fed by this frequency converter from harmful factors.

Input filters are designed to suppress interference generated by the rectifier and PWM inverter of the frequency converter, thus protecting the network, and output filters protect the motor itself from interference generated by the PWM inverter of the frequency converter. The input filters are chokes and EMI filters, and the output filters are common mode filters, motor chokes, sine filters and dU/dt filters.

The inductor, which is connected between the network and the frequency converter, is, it serves as a kind of buffer. The mains choke does not allow higher harmonics (250, 350, 550 Hz and beyond) from the frequency converter to the network, while protecting the converter itself from power surges in the network, from current surges during transients in the frequency converter, etc.

The voltage drop across such a choke is about 2%, which is optimal for normal operation throttle in combination with a frequency converter without the function of regeneration of electricity at the moment of braking the motor.

So, line chokes are installed between the network and the frequency converter under the following conditions: in the presence of interference in the network (for various reasons); when the phases are skewed; when powered by a relatively powerful (up to 10 times) transformer; if several frequency converters are fed from one source; if capacitors of the KRM installation are connected to the network.

The line choke provides:

    protection of the frequency converter against mains voltage surges and phase imbalance;

    circuit protection against high short circuit currents in the motor;

    extending the service life of the frequency converter.

To eliminate radiation, to ensure electromagnetic compatibility with radiation-sensitive devices, an EMI filter is just needed.

The three-phase EMI filter is designed to suppress interference in the range from 150 kHz to 30 MHz according to the Faraday cage principle. The EMI filter is connected as close as possible to the input of the frequency converter in order to provide surrounding devices with reliable protection against all interference generated by the PWM inverter. Sometimes the EMI filter is already built into the frequency converter.

The so-called dU/dt filter is a three-phase L-shaped low-pass filter, consisting of chains of inductors and capacitors. Such a filter is also called a motor choke, and often it may not have any capacitors at all, while the inductances will be significant. The filter parameters are such that all interference at frequencies higher than the switching frequency of the keys of the PWM inverter of the frequency converter is suppressed.

If the filter contains , then the capacitance of each of them is in the range of several tens of nanofarads, and - up to several hundred microhenries. As a result, this filter reduces the peak voltage and pulses at the terminals of a three-phase motor to 500 V / µs, which saves the stator windings from breakdown.

So, if the drive experiences frequent regenerative braking, is not initially suitable for operation with a frequency converter, has a low insulation class or a short motor cable, is installed in an aggressive environment or is used at a voltage of 690 volts, a dU / dt filter between the frequency converter and the motor is recommended. install.

Even though the voltage supplied to the motor by the frequency converter may be in the form of bipolar square waves rather than a pure sine wave, the dU/dt filter (with its small capacitance and inductance) acts on the current in such a way that it makes it in the windings engine almost exactly. It is important to understand that if you use a dU / dt filter at a frequency higher than its nominal value, then the filter will experience overheating, that is, it will bring unnecessary losses.

A sine filter is similar to a motor choke or dU/dt filter, but the difference is that the capacitances and inductances are large here, such that the cutoff frequency is less than half the switching frequency of the PWM inverter switches. Thus, a better smoothing of high-frequency interference is achieved, and the voltage shape on the motor windings and the current shape in them are much closer to the ideal sinusoidal.

Capacitances in a sine filter are measured in tens and hundreds of microfarads, and coil inductances in units and tens of millihenries. The sine filter is therefore characterized by a large size compared to the dimensions of a conventional frequency converter.

The use of a sine filter makes it possible to use, together with a frequency converter, even a motor that was originally (according to the specification) not intended for operation with a frequency converter due to poor insulation. In this case, there will be no increased noise, no rapid wear of the bearings, no overheating of the windings by high-frequency currents.

It is possible to safely use a long cable connecting the motor to the frequency converter when they are far apart, while avoiding impulse reflections in the cable that could lead to losses in the form of heat in the frequency converter.

    noise needs to be reduced. if the motor has poor insulation;

    experiences frequent regenerative braking;

    works in an aggressive environment; connected with a cable longer than 150 meters;

    should work for a long time without maintenance;

    during engine operation, the voltage rises step by step;

    the rated operating voltage of the motor is 690 volts.

At the same time, it should be remembered that a sinus filter cannot be used with a frequency below its passport rating (the maximum allowable downward frequency deviation is 20%), so in the settings of the frequency converter, you must first set the frequency limit from below. A frequency above 70 Hz must be used with great care, and in the settings of the converter, if possible, pre-set the capacitance and inductance of the connected sinus filter.

Remember that the filter itself can make noise and emit a noticeable amount of body, because even at rated load about 30 volts fall on it, so the filter should be installed under proper cooling conditions.

All chokes and filters must be connected in series with the motor with a screened cable as short as possible. So, for a 7.5 kW engine maximum length shielded cable must not exceed 2 meters.

Common mode filters are designed to suppress high frequency noise. This filter they are a differential transformer on a ferrite ring (more precisely, on an oval), the windings of which are directly three-phase wires connecting the motor to the frequency converter.

This filter is used to reduce common mode currents generated by discharges in the motor bearings. As a consequence, the common mode filter reduces possible electromagnetic emissions from the motor cable, especially if the cable is not shielded. The three-phase wires pass through the core window, while the protective earth wire remains outside.

The core is fixed on the cable with a clamp to protect the ferrite from the damaging effects of vibration on the ferrite (during engine operation, the ferrite core vibrates). The filter is best installed on the cable from the terminal side of the frequency converter. If the core during operation heats up to more than 70 ° C, then this indicates saturation of the ferrite, which means you need to add cores or shorten the cable. It is better to equip several parallel three-phase cables each with its own core.

Chapter 3

Digital IF overview

Since the 1980s, one of the most significant changes in spectrum analysis has been the use of digital technology to replace instrument clusters that were previously exclusively analog. With the advent of high-performance ADCs, new spectrum analyzers are able to digitize the incoming signal much faster than instruments created just a couple of years before. The biggest improvements have come in the IF section of the spectrum analyzers. Digital IF 1 has produced a strong improvement effect in speed, accuracy, and ability to measure complex signals through the use of advanced technologies digital processing signals.

Digital filters
A partial digital implementation of the IF circuits takes place in the Agilent ESA-E Series analyzers. While resolution bandwidths of 1 kHz and wider can usually be achieved with traditional analog LC and chip filters, the narrowest resolution bandwidths (1 Hz to 300 Hz) are realized digitally. As shown in Fig. 3-1, the linear analog signal is down-converted to 8.5 kHz IF and then passed through a bandpass filter only 1 kHz wide. This IF signal is amplified, then measured at 11.3 kHz and digitized.


Figure 3-1. Digital implementation of resolution filters 1, 2, 10, 30, 100 and 300 Hz in ESA-E series devices

Being already in the digitized state, the signal is passed through the Fast Fourier Transform algorithm. To convert a valid signal, the analyzer must be in a fixed state (no sweep). That is, the transformation must be performed on the time domain signal. Therefore, instead of a continuous sweep in the digital resolution bandwidth mode, the ESA-E series analyzers implement stepped increments of 900 Hz. This stepped tuning can be observed on the display, which is updated in 900 Hz increments while digital processing is being performed.
As we will see shortly, other spectrum analyzers, such as the PSA series, use an all-digital IF and all of their resolution filters are digital. A key advantage of the digital processing performed by these analyzers is the bandwidth selectivity of approximately 4:1. This selectivity is available on the narrowest filters - the ones we need to separate the closest signals.

In Chapter 2, we performed selectivity calculations for two signals separated by 4 kHz using a 3 kHz analog filter. Let's repeat this calculation for the case of digital filtering. A good model for the selectivity of a digital filter would be a near-Gaussian model:

Where H(Δ f) is the filter cutoff level, dB;
Δ f – frequency detuning from the center, Hz;

α is the selectivity control parameter. For an ideal Gaussian filter α=2. The swept resolution filters used in Agilent analyzers are based on a near-Gaussian model with α=2.12, which provides a selectivity of 4.1:1.

Substituting the values ​​from our example into this equation, we get:


At 4kHz offset, the 3kHz digital filter drops to -24.1dB, compared to the analog filter, which was only -14.8dB. Due to its superior selectivity, the digital filter can distinguish much more closely spaced signals.

Fully digital IF
For the first time, Agilent's PSA Series Spectrum Analyzers have combined several digital technologies to create a fully digital block PC. A pure digital IF provides a whole host of benefits to the user. The combination of FFT analysis for narrow and swept analysis for wide spans optimizes the sweep to provide the fastest measurements. Architecturally, the ADC has moved closer to the input port, made possible by improvements in A/D converters and more. digital equipment. Let's start by looking at the block diagram of the PSA series all-digital IF analyzer shown in Fig. 3-2.

Figure 3-2. Block diagram of a fully digital IF in the PSA series

Here all 160 resolution bands are implemented digitally. Although there are analog circuits before the ADC, starting with several stages of down conversion and ending with a pair of single-pole pre-filters (one LC filter and one on-chip filter). The pre-filter helps prevent third-order distortion from entering the downstream circuit, just like in an analog IF implementation. In addition, it makes it possible to expand the dynamic range by automatic switching measuring ranges. The signal from the output of the single-pole pre-filter is routed to an auto-switching detector and an anti-aliasing filter.
As with any FFT-based IF architecture, a smoothing filter is needed to eliminate aliasing (the contribution of out-of-band signals to the ADC data sample). This filter is multi-pole, so it has a significant group delay. Even a very sharply rising RF burst transferred down to the IF will experience a delay of more than three ADC cycles (30 MHz) when passing through the anti-aliasing filter. The delay gives time to recognize a large incoming signal before it causes the ADC to overload. The logic circuit driving the autoranging detector will reduce the gain in front of the ADC before the signal gets there, thus preventing clipping. If the envelope of the signal remains low for a long time, the auto-tuning circuit will increase the gain, reducing the effective noise at the input. The digital gain after the ADC is also changed to match the analog gain before the ADC. The result is a floating point ADC with a very wide dynamic range when autotuning is enabled in sweep mode.


Figure 3-3. Auto-tuning keeps the ADC noise close to the carrier and below the LO noise floor or resolution filter characteristics

On Fig. Figure 3-3 shows the PSA Series Analyzer's sweep behavior. The single-pole pre-filter allows the gain to be increased while the analyzer is tuned away from the carrier frequency. As you get closer to the carrier, the gain decreases and the ADC quantization noise increases. The noise level will depend on the level of the signal and its frequency offset from the carrier, so it will look like stepped phase noise. But the phase noise is different from this auto-tuning noise. Phase noise in spectrum analyzers cannot be avoided. However, reducing the pre-filtering width helps to reduce the lock-in noise at most frequency offsets from the carrier. Since the pre-filter bandwidth is about 2.5 times the resolution bandwidth, reducing the resolution bandwidth reduces the lock-in noise.

Dedicated Signal Processing IC
Let's return to the digital IF block diagram (Fig. 3-2). After the ADC gain has been set to match the analog gain and corrected by the digital gain, the ASIC begins processing the sample. First, the 30 MHz IF samples are split into I and Q pairs in half steps (15 million pairs per second). The I and Q pairs are then high-frequency boosted by a single-stage digital filter whose gain and phase are roughly the opposite of those of an analog single-pole pre-filter. Then the I and Q pairs are filtered by a low-pass filter with a linear phase response and an almost perfect Gaussian frequency response. Gaussian filters have always been the most suitable for swept frequency analysis due to the optimal compromise between the behavior in frequency domain(shape factor) and in the time domain (response to fast sweep). With the reduced signal bandwidth, the I and Q pairs can now be decimated and sent to the processor for FFT processing or demodulation. Even though the FFT can be performed for up to 10 MHz span segment of the anti-aliasing filter band, even in the narrower 1 kHz interval, with a narrow resolution bandwidth of 1 Hz, the FFT would require 20 million data points. Using data decimation for narrower intervals significantly reduces the number of data points required for the FFT, which greatly speeds up calculations.
For frequency swept analysis, the filtered I and Q pairs are converted to amplitude and phase pairs. In traditional swept analysis, the amplitude signal is filtered over the video bandwidth and sampled by the display's detector circuit. The choice of display mode "log/linear" and scaling "dB/unit" are made in the processor, so that the result is displayed in any of the scales without remeasuring.

Additional video processing options
Usually, a video bandpass filter smooths out the logarithm of the signal amplitude, but it has a lot of additional features. It can convert the logarithm of the amplitude to the voltage envelope before filtering, and translate back before the display is detected, for consistent readings.
Line-to-line voltage amplitude filtering is desirable for observing the envelopes of pulsed radio signals with zero frequency sweep. A signal with a logarithmic amplitude can also be converted to power (amplitude squared) before filtering and then back. Power filtering allows the analyzer to give the same average response to noisy signals (digital communication signals) as it does to CW signals of the same RMS voltage. Nowadays, it is increasingly necessary to measure full power per channel or across the entire frequency range. With these measurements, a dot on the display can show the average power over the time that the local oscillator passes through this dot. The video band filter can be reconfigured to collect data for logarithm, voltage, or power averaging.

Frequency count
Frequency sweep spectrum analyzers usually have a frequency counter. It counts the number of zero crossings in the IF signal and offsets this count by known offsets from the local oscillator in the rest of the conversion chain. If the count goes 1 second, you can get a frequency resolution of 1 Hz.
Thanks to digital oscillator synthesis and all-digital resolution bandwidth implementation, the inherent frequency accuracy of the PSA series analyzers is quite high (0.1% of span). In addition, the PSA has a frequency counter that tracks not only zero crossings but also phase changes. Thus, it can resolve frequencies in the tens of millihertz in 0.1 seconds. With this design, the ability to resolve frequency variations is no longer limited by the spectrum analyzer, but rather by the noisiness of the signal under investigation.

Other benefits of a fully digital IF
We've already covered a number of features of the PSA series: logarithm/voltage/power filtering, high-resolution frequency reading, logarithmic/linear scale switching of stored data, excellent shape factors, data dot average detector mode, 160 different resolution bandwidths, and , of course, frequency swept or FFT processing mode. When analyzing a spectrum, filtering on resolution filters introduces an error into the amplitude and phase measurements, which are functions of the sweep speed. At a certain fixed level of such errors, the resolution filters of a purely digital IF with a linear phase allow more high speeds frequency sweep than analog filters. The digital implementation also provides some compensation in frequency and amplitude acquisitions, thus allowing sweep speeds twice as fast as older analyzers, and performs well even at quadruple the sweep speed.
Implemented in digital form logarithmic amplification is highly accurate. Typical errors that are typical for the analyzer as a whole are much smaller than the measurement errors with which the manufacturer evaluates the reliability of the logarithm. At the analyzer's input mixer, log confidence is specified as ±0.07 dB for any level down to -20 dBm. Log gain range per low levels does not limit the reliability of the logarithm, as it would be with an analog IF; the range is limited only by noise of the order of -155 dBm at the input mixer. Due to single-tone compression in subsequent chains at higher powers, the confidence parameter degrades to ±0.13 dB for signal levels down to -10 dBm at the input mixer. By comparison, an analog logarithmic amplifier typically has tolerances of the order of ±1 dB.
Other IF-related accuracies also experienced improvement. The IF pre-filter is analog and must be configured like any analog filter, so it is subject to tuning errors. But it is still better than other analog filters. While only one stage needs to be made for it, it can be made much more stable than the 4- and 5-stage filters used in analog IF analyzers. As a result, gain drops between resolution filters can be kept within ±0.03 dB, ten times better than pure analog designs.
The accuracy of the IF bandwidth is determined by the limitations of the settings in the digital part of the filter and the calibration uncertainty in the analog pre-filter. Again, the pre-filter is very stable, and introduces only 20% of the error that would be present in an analog implementation of a resolution bandwidth consisting of five such steps. As a result, most resolution bandwidths fit within 2 percent of their advertised width, as opposed to 10-20 percent for analog IF analyzers.
The most important aspect of bandwidth accuracy is minimizing the error in channel power measurements and similar measurements. The noise bandwidth of the resolution filters is even better than the 2 percent tolerance for setup processes, and noise markers and channel power measurements are corrected to ±0.5%. Thus, bandwidth errors contribute only ±0.022 dB to noise amplitude density and channel power measurement errors. And, finally, with the complete absence of analog gain stages depending on the reference level, there is no “IF gain” error at all. The sum of all these improvements is such that a pure digital IF provides a significant improvement in spectral analysis accuracy. It is also possible to change the analyzer settings without any significant impact on the accuracy of the measurement. We will talk about this in more detail in the next chapter.

1 Strictly speaking, once a signal is digitized, it is no longer at the intermediate frequency, or IF. From now on, the signal is represented by digital values. However, we use the term "digital IF" to describe those digital processes that have replaced the analog IF section of traditional spectrum analyzers.)