Protection of equipment against surge voltage and switching noise
Writing this text led me to a feeling of not knowing many of the principles of operation, use (or even ignorance of the existence) of parallel protection against impulse overvoltages in the network, including those caused by lightning discharges.
Impulse noise in the network is quite common, they can occur during a thunderstorm, when powerful loads are turned on / off (since the network is an RLC circuit, it causes fluctuations that cause voltage surges) and many other factors. In low-current, including digital, circuits, this is even more relevant, since switching disturbances penetrate well enough through the power sources (Backward converters are the most protected - in them the transformer energy is transferred to the load when the primary winding is disconnected from the network).
In Europe, the installation of modules for protection against impulse overvoltages (hereinafter, for simplicity, call lightning protection or SPD) is de facto almost obligatory, although their networks are better than ours and thunderstorm areas are smaller.
The use of SPDs of the last 20 years has become particularly relevant, when scientists began to develop more and more variants of field MOSFET transistors, which are very afraid of exceeding the reverse voltage. And such transistors are used in almost all switching power supplies up to 1 kVA, as keys on the primary (network) side.
Another aspect of the use of SPDs is the provision of voltage limiting between the neutral conductor and the earth conductor. Overvoltage on the neutral conductor in the network can occur, for example, when switching the Automatic Transfer Switch with a divided neutral. During the switch, the neutral conductor will be "in the air" and there can be anything on it.
Overvoltage impulses in a network are characterized by a waveform and current amplitude. The shape of the current pulse is characterized by its rise and fall times — for European standards, these are pulses of 10/350 µs and 8/20 µs. In Russia, as it happens often lately, they adopted the standards of Europe and GOST R 51992-2002 appeared. The numbers in the designation of the pulse shape mean the following:
- the first is the time (in microseconds) of the rise of the current pulse from 10% to 90% of the maximum current value;
- the second - the time (in microseconds) of the current pulse to fall to 50% of the maximum current value;
Protection devices are divided into classes depending on the power of the pulse, which they can dissipate:
1) Class 0 (A) - external lightning protection (in this post we do not consider);
2) Class I (B) - overvoltage protection characterized by impulse currents with amplitude from 25 to 100 kA with a 10/350 µs waveform (protection in building's distribution boards);
3) Class II (C) - overvoltage protection characterized by impulse currents with amplitude from 10 to 40 kA with a waveform of 8/20 μs (protection in floor panels, electrical panels of premises, electrical equipment inputs);
3) Class III (D) - overvoltage protection, characterized by impulse currents with an amplitude of up to 10 kA with a wavelength of 8/20 μs (in most cases the protection is built into the equipment - if it is made in accordance with GOST);
The main two SPD devices are arresters and varistors of various designs.
')
Main characteristics of arresters:
1) Protection class (see above);
2) Nominal operating voltage - continuous, recommended by the manufacturer operating voltage of the arrester;
3) Maximum operating alternating voltage - the ultimate long-term voltage of the arrester, at which it is guaranteed not to work;
4) Maximum pulse discharge current (10/350) μs - the maximum value of the current amplitude with a waveform (10/350) μs, at which the discharger will not fail and will provide a voltage limit at a given level;
5) Rated impulse discharge current (8/20) µs - the nominal value of the current amplitude with a waveform (8/20) µs, at which the arrester will provide a voltage limit at a given level;
6) Voltage limiting - the maximum voltage on the electrodes of the discharger during its breakdown due to the occurrence of an overvoltage pulse;
7) Response time - arrester opening time (for almost all arresters - less than 100 ns);
8) (rarely indicated by manufacturers parameter) static voltage breakdown of the spark gap - static voltage (slowly varying in time) at which the gap will open. Measured by applying constant voltage. In most cases, it is 20-30% higher than the maximum operating alternating voltage reduced to a constant (alternating voltage multiplied by the root of 2);
Choosing a spark gap is quite a creative process with numerous “spitting at the ceiling” - after all, we don’t know in advance the value of the current that will occur in the network ...
When choosing a spark gap you can follow the following rules:
1) When installing protection in the input panels from an overhead power line or in areas where there are frequent thunderstorms, install arresters with a maximum discharge current (10/350) μs at least 35 kA;
2) Choose the maximum long-term voltage a little more than the estimated maximum mains voltage (otherwise there is a chance that at high mains voltage, the arrester will open and fail from overheating);
3) Choose arresters with the lowest possible voltage limits (in this case, the implementation of rules 1 and 2 is mandatory). Typically, the voltage limits of class I arresters are from 2.5 to 5 kV;
4) Install arresters specifically for this purpose between the N and PE conductors (manufacturers indicate that they are to be connected to N-PE conductors). In addition, these arresters are characterized by lower operating voltages, usually of the order of 250 V AC (there is no voltage at all between the neutral and the ground in normal mode) and a high discharge current from 50 kA to 100 kA and above.
5) Connect the arresters to the network with conductors with a cross section of at least 10 mm2 (even if the network conductors have a smaller cross-section) and as short as possible. For example, when a 2-meter-long conductor with a cross section of 4 mm2 of 40 kA appears, it will fall (in the ideal case without inductance, but it plays a big role here) around 350 V. If the arrester is connected with such a conductor, then at the network connection point the limiting voltage will be equal to the sum of the arrester limiting voltage and the voltage drop across the conductor at the pulse current (our 350 V). Thus, the protective properties are significantly deteriorated.
6) If possible, install dischargers in front of the input circuit breaker and always in front of the RCD (it is necessary to install a fuse with a gL characteristic for a current of 80-125 A in series with the discharger to ensure disconnection of the discharger from the network when it fails). Since no one will allow installing an arrester in front of the input automaton, it is desirable that the automaton be at a current of at least 80A with response characteristic D. This will reduce the likelihood of a false triggering of the automaton when the arrester trips. The installation of an SPD in front of the RCD is due to the low resistance of the RCD to impulse currents; moreover, when the N-PE spark gap is triggered, the RCD will falsely trip. Also, it is advisable to install SPDs in front of electricity meters (which, again, the power industry will not allow)
In the initial state, the varistor has a high internal resistance (from hundreds of ohms to tens and hundreds of MOhm). When the voltage at the contacts of the varistor reaches a certain level, it drastically reduces its resistance and begins to conduct significant current, while the voltage at the contacts of the varistor changes slightly. Like a discharger, a varistor can absorb the energy of an overvoltage pulse lasting up to hundreds of microseconds. But with long-term overvoltage, the varistor fails with a large amount of heat (explodes).
All varistors in the DIN rail version are equipped with thermal protection, designed to disconnect the varistor from the network when it is unacceptable overheating (at the same time, it is possible to determine from the local mechanical indication that the varistor has failed).
In the photo varistors with built-in thermal relay after exceeding the operating voltage of different values. With a significant overvoltage, such built-in thermal protection is practically not effective - the varistors explode so that the ears lay. However, the built-in thermal protection in varistor modules on a DIN rail is quite effective for any long-lasting overvoltages, and manages to disconnect the varistor from the network
A small video of naturalistic tests :) (supply of overvoltage to a varistor of 20 mm in diameter - 50 V excess)
The main characteristics of varistors:
1) Protection class (see above). Typically, varistors have a protection class II (C), III (D);
2) Nominal operating voltage - continuous, recommended by the manufacturer operating voltage of the varistor;
3) The maximum operating alternating voltage is the ultimate long-term voltage of the varistor at which it is guaranteed not to open;
4) Maximum pulse discharge current (8/20) μs - the maximum value of the current amplitude with a waveform (8/20) μs, at which the varistor will not fail and will provide a voltage limit at a given level;
5) Rated pulse discharge current (8/20) μs - the nominal value of the current amplitude with a waveform (8/20) μs, at which the varistor will provide a voltage limit at a given level;
6) The limiting voltage is the maximum voltage on the varistor when it is opened due to the appearance of an overvoltage pulse;
7) Response time - the opening time of the varistor (for almost all varistors - less than 25 ns);
8) (parameter rarely indicated by manufacturers) classification voltage of a varistor is a static voltage (slowly varying in time) at which the leakage current of a varistor reaches 1 mA. Measured by applying constant voltage. In most cases, it is 15-20% higher than the maximum operating alternating voltage reduced to a constant (alternating voltage multiplied by the root of 2);
9) (very rarely indicated by manufacturers parameter) the permissible error of the parameters of the varistor - for almost all varistors ± 10%. This error should be considered when choosing the maximum operating voltage of the varistor.
The choice of varistors as well as arresters is fraught with difficulties associated with the unknown conditions of their work.
When choosing a varistor protection, you can use the following rules:
1) Varistors are installed as the second or third level of protection against impulse overvoltages;
2) When using varistor protection class II in conjunction with class I protection, it is necessary to take into account the different response speeds of varistors and arresters. Since the arresters are slower than the varistors, if the SPD is not matched, the varistors will take over most of the surge voltage and quickly fail. For matching I and II lightning protection classes, special matching chokes are used (ultrasound manufacturers have their assortment for such cases), or the cable length between the I / II class I / O SPDs must be at least 10 meters. The disadvantage of this solution is the necessity of cutting the chokes into the network or its elongation, which increases its inductive component. The only exception is the German manufacturer PhoenixContact , which has developed special class I arresters with the so-called “electronic ignition”, which are “matched” with varistor modules of the same manufacturer. These combinations of SPDs can be installed without additional coordination;
3) Choose the maximum continuous voltage a little more than the estimated maximum mains voltage (otherwise there is a chance that at high mains voltage, the varistor will open and fail from overheating). But here it is impossible to overdo it, since the voltage limiting of the varistor directly depends on the classification voltage (and consequently, on the maximum operating voltage). An example of an unsuccessful choice of the maximum operating voltage are IEC varistor modules with a maximum continuous voltage of 440 V. If they are installed in a network with a nominal voltage of 220 V, then its operation will be extremely inefficient. In addition, it should be borne in mind that varistors tend to "aging" (i.e., over time, with many actuators of a varistor, its classification voltage starts to decrease). Optimal for Russia will be the use of varistors with a long operating voltage from 320 to 350 V;
4) You need to choose with the lowest possible voltage limits (with the implementation of rules 1 - 3). Typically, voltage limiting class II varistors for mains voltage from 900 V to 2.5 kV;
5) Do not connect parallel to the varistors to increase the total power dissipation. Many manufacturers of protection of surge protective devices (especially class III (D)) sin by parallel connection of varistors. But, since 100% of the same varistors do not exist (even from the same batch they are different), one of the varistors will always be the weakest link and fail when an overvoltage pulse occurs. During subsequent pulses, the remaining varistors will fail, since they will no longer provide the required power dissipation (this is the same as connecting diodes in parallel to increase the total current - this is not possible)
6) Connect the varistors to the network with conductors with a cross section of at least 10 mm2 (even if the network conductors have a smaller cross section) and as short as possible (the reasoning is the same as for the arresters).
7) If possible, install varistors in front of the input circuit breaker and always in front of the RCD. Since no one will allow installing an arrester in front of the input automaton, it is desirable that the automaton be on a current of at least 50A with a response characteristic D (for class II varistors). This will reduce the likelihood of a false actuation of the machine when the varistor is triggered.
In addition, there are quite a few ready-made surge protectors on the market, including modules of one or two protection classes, as well as fuses for safety in case of failure of protective elements. In this case, the panel is fixed on the wall and connected to the existing electrical wiring in accordance with the recommendations of the manufacturer.
The cost of the SPD varies depending on the manufacturer at times. At one time (several years ago), I analyzed the market and selected a number of manufacturers of protection class II (some were not included in the list, due to the lack of module versions for the required long-term operating voltage of 320 V or 350 V).
As a quality note, I can only identify HAKEL modules (for example, PIIIMT 280 DS) - they have weak contact connections of the inserts and are made of combustible plastic, which is prohibited by GOST R 51992-2002. At the moment, HAKEL has updated a number of products - I can not say anything about it, because won't use hakel ever again
We will use the application of class III (D) SPDs and the protection of digital circuits of devices for later.
In conclusion, I can say that if after reading everything you have more questions than after reading the title, this is good, because the topic is interesting, and it is so vast that you can write more than one book.
Impulse noise in the network is quite common, they can occur during a thunderstorm, when powerful loads are turned on / off (since the network is an RLC circuit, it causes fluctuations that cause voltage surges) and many other factors. In low-current, including digital, circuits, this is even more relevant, since switching disturbances penetrate well enough through the power sources (Backward converters are the most protected - in them the transformer energy is transferred to the load when the primary winding is disconnected from the network).
In Europe, the installation of modules for protection against impulse overvoltages (hereinafter, for simplicity, call lightning protection or SPD) is de facto almost obligatory, although their networks are better than ours and thunderstorm areas are smaller.
The use of SPDs of the last 20 years has become particularly relevant, when scientists began to develop more and more variants of field MOSFET transistors, which are very afraid of exceeding the reverse voltage. And such transistors are used in almost all switching power supplies up to 1 kVA, as keys on the primary (network) side.
Another aspect of the use of SPDs is the provision of voltage limiting between the neutral conductor and the earth conductor. Overvoltage on the neutral conductor in the network can occur, for example, when switching the Automatic Transfer Switch with a divided neutral. During the switch, the neutral conductor will be "in the air" and there can be anything on it.
Overvoltage impulse characteristics
Overvoltage impulses in a network are characterized by a waveform and current amplitude. The shape of the current pulse is characterized by its rise and fall times — for European standards, these are pulses of 10/350 µs and 8/20 µs. In Russia, as it happens often lately, they adopted the standards of Europe and GOST R 51992-2002 appeared. The numbers in the designation of the pulse shape mean the following:
- the first is the time (in microseconds) of the rise of the current pulse from 10% to 90% of the maximum current value;
- the second - the time (in microseconds) of the current pulse to fall to 50% of the maximum current value;
Protection devices are divided into classes depending on the power of the pulse, which they can dissipate:
1) Class 0 (A) - external lightning protection (in this post we do not consider);
2) Class I (B) - overvoltage protection characterized by impulse currents with amplitude from 25 to 100 kA with a 10/350 µs waveform (protection in building's distribution boards);
3) Class II (C) - overvoltage protection characterized by impulse currents with amplitude from 10 to 40 kA with a waveform of 8/20 μs (protection in floor panels, electrical panels of premises, electrical equipment inputs);
3) Class III (D) - overvoltage protection, characterized by impulse currents with an amplitude of up to 10 kA with a wavelength of 8/20 μs (in most cases the protection is built into the equipment - if it is made in accordance with GOST);
Surge Protection Devices
The main two SPD devices are arresters and varistors of various designs.
')
Discharger
A discharger is an electrical device open (air) or closed (filled with inert gases) of the type, containing in the simplest case two electrodes. When the voltage on the electrodes of the spark gap exceeds a certain value, it “breaks through”, thereby limiting the voltage on the electrodes at a certain level. During the breakdown of a spark gap, a considerable current flows from it (from hundreds of Amps to tens of kiloAmp) for a short time (to hundreds of microseconds). After removing the overvoltage impulse, if the power that the arrester is capable of dissipating has not been exceeded, it goes to the initial closed state until the next impulse.Main characteristics of arresters:
1) Protection class (see above);
2) Nominal operating voltage - continuous, recommended by the manufacturer operating voltage of the arrester;
3) Maximum operating alternating voltage - the ultimate long-term voltage of the arrester, at which it is guaranteed not to work;
4) Maximum pulse discharge current (10/350) μs - the maximum value of the current amplitude with a waveform (10/350) μs, at which the discharger will not fail and will provide a voltage limit at a given level;
5) Rated impulse discharge current (8/20) µs - the nominal value of the current amplitude with a waveform (8/20) µs, at which the arrester will provide a voltage limit at a given level;
6) Voltage limiting - the maximum voltage on the electrodes of the discharger during its breakdown due to the occurrence of an overvoltage pulse;
7) Response time - arrester opening time (for almost all arresters - less than 100 ns);
8) (rarely indicated by manufacturers parameter) static voltage breakdown of the spark gap - static voltage (slowly varying in time) at which the gap will open. Measured by applying constant voltage. In most cases, it is 20-30% higher than the maximum operating alternating voltage reduced to a constant (alternating voltage multiplied by the root of 2);
Choosing a spark gap is quite a creative process with numerous “spitting at the ceiling” - after all, we don’t know in advance the value of the current that will occur in the network ...
When choosing a spark gap you can follow the following rules:
1) When installing protection in the input panels from an overhead power line or in areas where there are frequent thunderstorms, install arresters with a maximum discharge current (10/350) μs at least 35 kA;
2) Choose the maximum long-term voltage a little more than the estimated maximum mains voltage (otherwise there is a chance that at high mains voltage, the arrester will open and fail from overheating);
3) Choose arresters with the lowest possible voltage limits (in this case, the implementation of rules 1 and 2 is mandatory). Typically, the voltage limits of class I arresters are from 2.5 to 5 kV;
4) Install arresters specifically for this purpose between the N and PE conductors (manufacturers indicate that they are to be connected to N-PE conductors). In addition, these arresters are characterized by lower operating voltages, usually of the order of 250 V AC (there is no voltage at all between the neutral and the ground in normal mode) and a high discharge current from 50 kA to 100 kA and above.
5) Connect the arresters to the network with conductors with a cross section of at least 10 mm2 (even if the network conductors have a smaller cross-section) and as short as possible. For example, when a 2-meter-long conductor with a cross section of 4 mm2 of 40 kA appears, it will fall (in the ideal case without inductance, but it plays a big role here) around 350 V. If the arrester is connected with such a conductor, then at the network connection point the limiting voltage will be equal to the sum of the arrester limiting voltage and the voltage drop across the conductor at the pulse current (our 350 V). Thus, the protective properties are significantly deteriorated.
6) If possible, install dischargers in front of the input circuit breaker and always in front of the RCD (it is necessary to install a fuse with a gL characteristic for a current of 80-125 A in series with the discharger to ensure disconnection of the discharger from the network when it fails). Since no one will allow installing an arrester in front of the input automaton, it is desirable that the automaton be at a current of at least 80A with response characteristic D. This will reduce the likelihood of a false triggering of the automaton when the arrester trips. The installation of an SPD in front of the RCD is due to the low resistance of the RCD to impulse currents; moreover, when the N-PE spark gap is triggered, the RCD will falsely trip. Also, it is advisable to install SPDs in front of electricity meters (which, again, the power industry will not allow)
Varistor
Varistor - a semiconductor device with a "cool" symmetric current-voltage characteristic.In the initial state, the varistor has a high internal resistance (from hundreds of ohms to tens and hundreds of MOhm). When the voltage at the contacts of the varistor reaches a certain level, it drastically reduces its resistance and begins to conduct significant current, while the voltage at the contacts of the varistor changes slightly. Like a discharger, a varistor can absorb the energy of an overvoltage pulse lasting up to hundreds of microseconds. But with long-term overvoltage, the varistor fails with a large amount of heat (explodes).
All varistors in the DIN rail version are equipped with thermal protection, designed to disconnect the varistor from the network when it is unacceptable overheating (at the same time, it is possible to determine from the local mechanical indication that the varistor has failed).
In the photo varistors with built-in thermal relay after exceeding the operating voltage of different values. With a significant overvoltage, such built-in thermal protection is practically not effective - the varistors explode so that the ears lay. However, the built-in thermal protection in varistor modules on a DIN rail is quite effective for any long-lasting overvoltages, and manages to disconnect the varistor from the network
A small video of naturalistic tests :) (supply of overvoltage to a varistor of 20 mm in diameter - 50 V excess)
The main characteristics of varistors:
1) Protection class (see above). Typically, varistors have a protection class II (C), III (D);
2) Nominal operating voltage - continuous, recommended by the manufacturer operating voltage of the varistor;
3) The maximum operating alternating voltage is the ultimate long-term voltage of the varistor at which it is guaranteed not to open;
4) Maximum pulse discharge current (8/20) μs - the maximum value of the current amplitude with a waveform (8/20) μs, at which the varistor will not fail and will provide a voltage limit at a given level;
5) Rated pulse discharge current (8/20) μs - the nominal value of the current amplitude with a waveform (8/20) μs, at which the varistor will provide a voltage limit at a given level;
6) The limiting voltage is the maximum voltage on the varistor when it is opened due to the appearance of an overvoltage pulse;
7) Response time - the opening time of the varistor (for almost all varistors - less than 25 ns);
8) (parameter rarely indicated by manufacturers) classification voltage of a varistor is a static voltage (slowly varying in time) at which the leakage current of a varistor reaches 1 mA. Measured by applying constant voltage. In most cases, it is 15-20% higher than the maximum operating alternating voltage reduced to a constant (alternating voltage multiplied by the root of 2);
9) (very rarely indicated by manufacturers parameter) the permissible error of the parameters of the varistor - for almost all varistors ± 10%. This error should be considered when choosing the maximum operating voltage of the varistor.
The choice of varistors as well as arresters is fraught with difficulties associated with the unknown conditions of their work.
When choosing a varistor protection, you can use the following rules:
1) Varistors are installed as the second or third level of protection against impulse overvoltages;
2) When using varistor protection class II in conjunction with class I protection, it is necessary to take into account the different response speeds of varistors and arresters. Since the arresters are slower than the varistors, if the SPD is not matched, the varistors will take over most of the surge voltage and quickly fail. For matching I and II lightning protection classes, special matching chokes are used (ultrasound manufacturers have their assortment for such cases), or the cable length between the I / II class I / O SPDs must be at least 10 meters. The disadvantage of this solution is the necessity of cutting the chokes into the network or its elongation, which increases its inductive component. The only exception is the German manufacturer PhoenixContact , which has developed special class I arresters with the so-called “electronic ignition”, which are “matched” with varistor modules of the same manufacturer. These combinations of SPDs can be installed without additional coordination;
3) Choose the maximum continuous voltage a little more than the estimated maximum mains voltage (otherwise there is a chance that at high mains voltage, the varistor will open and fail from overheating). But here it is impossible to overdo it, since the voltage limiting of the varistor directly depends on the classification voltage (and consequently, on the maximum operating voltage). An example of an unsuccessful choice of the maximum operating voltage are IEC varistor modules with a maximum continuous voltage of 440 V. If they are installed in a network with a nominal voltage of 220 V, then its operation will be extremely inefficient. In addition, it should be borne in mind that varistors tend to "aging" (i.e., over time, with many actuators of a varistor, its classification voltage starts to decrease). Optimal for Russia will be the use of varistors with a long operating voltage from 320 to 350 V;
4) You need to choose with the lowest possible voltage limits (with the implementation of rules 1 - 3). Typically, voltage limiting class II varistors for mains voltage from 900 V to 2.5 kV;
5) Do not connect parallel to the varistors to increase the total power dissipation. Many manufacturers of protection of surge protective devices (especially class III (D)) sin by parallel connection of varistors. But, since 100% of the same varistors do not exist (even from the same batch they are different), one of the varistors will always be the weakest link and fail when an overvoltage pulse occurs. During subsequent pulses, the remaining varistors will fail, since they will no longer provide the required power dissipation (this is the same as connecting diodes in parallel to increase the total current - this is not possible)
6) Connect the varistors to the network with conductors with a cross section of at least 10 mm2 (even if the network conductors have a smaller cross section) and as short as possible (the reasoning is the same as for the arresters).
7) If possible, install varistors in front of the input circuit breaker and always in front of the RCD. Since no one will allow installing an arrester in front of the input automaton, it is desirable that the automaton be on a current of at least 50A with a response characteristic D (for class II varistors). This will reduce the likelihood of a false actuation of the machine when the varistor is triggered.
Overview of the manufacturers of SPDs
The leading manufacturers specializing in SPDs for low-voltage networks are: Phoenix Contact ; Dehn ; OBO Bettermann ; CITEL ; Hakel . Also, many manufacturers of low-voltage equipment, in the products there are modules of SPDs (ABB, Schneider Electric, etc.). In addition, China successfully copies the SPDs of world manufacturers (since the Varistor is a fairly simple device, Chinese manufacturers produce fairly high-quality products — for example, TYCOTIU modules).In addition, there are quite a few ready-made surge protectors on the market, including modules of one or two protection classes, as well as fuses for safety in case of failure of protective elements. In this case, the panel is fixed on the wall and connected to the existing electrical wiring in accordance with the recommendations of the manufacturer.
The cost of the SPD varies depending on the manufacturer at times. At one time (several years ago), I analyzed the market and selected a number of manufacturers of protection class II (some were not included in the list, due to the lack of module versions for the required long-term operating voltage of 320 V or 350 V).
As a quality note, I can only identify HAKEL modules (for example, PIIIMT 280 DS) - they have weak contact connections of the inserts and are made of combustible plastic, which is prohibited by GOST R 51992-2002. At the moment, HAKEL has updated a number of products - I can not say anything about it, because won't use hakel ever again
We will use the application of class III (D) SPDs and the protection of digital circuits of devices for later.
In conclusion, I can say that if after reading everything you have more questions than after reading the title, this is good, because the topic is interesting, and it is so vast that you can write more than one book.
Source: https://habr.com/ru/post/188972/
All Articles