GB2542789A - Fault protection for voltage source converters - Google Patents

Abstract

This application describes methods and apparatus for fault protection of voltage source converters (VSCs), especially VSCs including a plurality of cells arranged as a chain-link circuit such as may be used for HVDC. An apparatus 200 for use in a voltage source converter has first and second cell terminals 104a, 104b, an energy storage element (such as a capacitor) 101 and a plurality of switches 102a, 102b configured such that first and second cell terminals can be selectively connected in a path that includes the energy storage element 101 or a path that bypasses the energy storage element. To provide the fault protection the apparatus also includes one or more surge protection devices 201 in parallel with at least one of said plurality of switches and said energy storage element. Each surge protection device 201 comprises a gas discharge tube in series with a metal oxide varistor.

Description

Fault Protection for Voltage Source Converters

This application relates to methods and apparatus for fault protection of voltage source converters and in particular to fault protection for modules, especially chain-link cells, of a voltage source converter. HVDC (high-voltage direct current) electrical power transmission uses direct current for the transmission of electrical power. This is an alternative to alternating current electrical power transmission which is more common. There are a number of benefits to using HVDC electrical power transmission.

In order to use HVDC electrical power transmission, it is typically necessary to convert alternating current (AC) to direct current (DC) and back again. Historically this has involved a six pulse bridge type topology based on elements such thyristors which, can be turned on at a desired point in the power cycle and remain conducting as long as they are forward biased. Such a converter is known as a line-commutated converter (LCC).

Recent developments in the power electronics field have led to an increased use of voltages-source converters (VSCs) for AC-DC and DC-AC conversion. VSCs make use of switching elements, typically insulated gate bipolar transistors (IGBTs) connected with respective anti-parallel diodes, that can be controllably turned on and off. Such converters are sometimes referred as self-commutated converters. For each AC phase the VSC has a phase limb with two converter arms connecting an AC terminal to high and low DC terminals respectively. The switching elements of each of the converter arms of a phase limb are switched in a sequence to provide conversion.

Various designs of VSC are known. Some types of VSC, such as the modular multilevel converter (MMC) or alternate arm converter (AAC) for example, make use of a series connection of a plurality of cells, each cell having an energy storage element such as a capacitor that can be selectively connected in series between the terminals of the cell or bypassed. The series connection of such cells is sometimes referred to as a chain-link circuit or chain-link converter or simply a chain-link and the cells are often referred to as sub-modules (with a plurality of cells forming a valve module).

The cells or sub-modules of the chain-link may be controlled to connect or bypass their respective energy storage element at different times so as to vary over the time the voltage presented across the chain-link so as to provide voltage wave-shaping for a converter arm of the VSC. For example in the MMC type converter, each converter arm comprises a chain-link having a relatively large number of sub-modules and the chain-links of each arm are switched together in sequence to synthesise a stepped waveform that approximates to a sine wave and which contain low level of harmonic distortion.

Figure 1 illustrates a typical cell or sub-module 100 of a chain-link. The cell 100 has a capacitance 101 (which may be formed by one of more capacitors) connected with a switch arrangement which comprises switches 102a and 102b, each switch comprising at least one IGBT having an anti-parallel diode 103. In the example shown in figure 1 there are two IGBTs connected in a half bridge arrangement, with switch 102a being connected in series with the capacitor 101 between the cell terminals 104a and 104b and switch 102b being directly connected between the cell terminals 104a and 104b. It will be appreciated however that cells based on a full bridge arrangement are also known and are used in some chain-links. It will also be appreciated that each switch may, in practice, comprise multiple IGBTs in series.

For DC fault protection each cell or submodule is typically further provided with bypass protection comprising a fast acting thyristor 105 and an electromechanical bypass switch 106 which is open in normal operation.

Initially in the event of a DC fault the electromechanical switch 106 will be open and the thyristor 105 non-conducting. Thus the fault current will initially flow between the cell terminals 104a and 104b via the anti-parallel diode of switch 102b, as illustrated by path 107-1. This diode will not typically be designed to permanently withstand a relatively high current such as may occur during a DC fault. Once the fault is detected by the control system, which may for instance be a short-circuit gate drive protector, the thyristor is activated and the fault current will then flow via the thyristor 105 as illustrated by path 107-2. In some circumstances classed as highly critical the electromechanical switch will also be commanded to close. As the electromechanical switch may take some time to close the fast acting thyristor 105 will carry the fault current whilst the electromechanical switch is closing. Once the electromechanical switch is closed the fault current will flow via path 107-3.

The thyristor 105 may be chosen to withstand a sustained fault current and/or may be designed to fail to short-circuit.

Such a bypass protection system does provide reasonable protection for some DC faults, but does require the presence of an electromechanical switch which is typically a relatively bulky, heavy and mechanically complex component that thus adds to the size and cost of a submodule or cell.

Also in some cases the fault current can flow between the cell terminals in the opposite direction to that illustrated in figure 1. For example, some disturbances on the AC side or a failure of control hardware may result in fault current in the opposite direction. When such a fault develops the current may flow via switch 102b, if conducting, or otherwise via the diode of switch 102a and the capacitor. When the fault is detected the IGBTs 102a and 102b will be blocked and a firing signal may be sent to thyristor 105. However, as the thyristor 105 is connected in the opposite direction to the current path in this instance, it will not conduct. Thus to provide a fault bypass the electromechanical switch 106 may be commanded to close but it is not until the electromechanical switch is finally closed that the fault current is diverted away from the sub-module semiconductor devices. By that time it is possible that the cell capacitor may have reached a voltage level that could destroy the semiconductor devices, the capacitor bank or both. Additionally whilst the electromechanical switch may be closed by a suitable command it may typically need to be physically reset to be reopened and thus may need to remain closed until the VSC is de-energised and access to the submodule is possible.

Embodiments of the invention relate to methods and apparatus for fault protection in a VSC that at least mitigate at least some of the above mentioned issues.

Thus according to an aspect of the present invention there is provided an apparatus for use in a voltage source converter comprising: first and second cell terminals; an energy storage element; a plurality of switches configured such that first and second cell terminals can be selectively connected in a path that includes the energy storage element or a path that bypasses the energy storage element; and one or more surge protection devices comprising a gas discharge tube in series with a metal oxide varistor. wherein the one or more surge protection devices are connected in parallel with at least one of said plurality of switches and said energy storage element.

The apparatus of embodiments of the invention, which may be a cell or sub-module of a VSC, for example a cell of a chain-link of a VSC, thus includes a surge protection device (SPD) arranged across at least some of the sensitive components of the cell. In use the SPD may be normally non-conductive but will become conductive in an overvoltage situation and thus act to divert a fault away from the sensitive components and/or reduce the overvoltage. The use of an SPD comprising a gas discharge tube in series with a metal oxide varistor provides advantageous characteristics compared to a standard surge arrestor as will be explained in more detail below.

The apparatus may further comprise a bypass semiconductor switching element, such as a thyristor, for providing a bypass path between the first and second cell terminals for fault current in a first direction. Thus the cell may comprise a bypass thyristor as described previously in addition to the SPD. As will be explained later however in some embodiments the previously described electromechanical switch need not be present. The surge protection device provides protection for faults with fault currents in a second direction, opposite to the first direction.

The one or more surge protection devices may be configured to fail to short-circuit.

This provides a safe failure mode.

The one or more SPDs may be connected across the components in various different embodiments. One of the one or more surge protection devices may be connected in parallel with the energy storage element. Additionally or alternatively one of the one or more surge protection devices may be connected across the first and second cell terminals. One of the one or more surge protection devices may additionally or alternatively be connected in parallel with one of said plurality of switches. In some embodiments there may be a respective surge protection device in parallel with each switch.

As mentioned in some embodiments a first surge protection device may be connected in parallel with the energy storage device, e.g. the capacitor of a cell or sub-module. In some embodiments the first surge protection device may form part of a discharge circuit for preventing overvoltage of the energy storage device. In such embodiments the apparatus may further comprise a discharge switching element in parallel with the first surge protection device and a discharge switch controller. If the voltage of the energy storage element becomes too high the first surge protection device may start conducting to discharge the energy storage element. In the event that the first surge protection device becomes conducting, the discharge switch controller may configured to control turn-on of the discharge switching element so as to extinguish conduction via the first surge protection device. This provides for a controlled discharge of the energy storage element. A discharge resistor may be in series with the first surge protection device and may, for example be connected in series with the parallel connection of the first surge protection device and the discharge switching element.

The first surge protection device may be configured to start conducting at a first voltage level. The first voltage level may be an overvoltage threshold. The discharge switch controller may configured to turn-on of the discharge switching element at a second, lower, voltage level. The second voltage level may a nominal voltage of the cell or the energy storage element of the cell. In this way the combination of the first surge protection device and discharge switching element provide overvoltage protection with a defined voltage level for triggering the surge protection device and also a defined voltage level for stopping discharge of the energy storage element.

In some embodiments the discharge switching element may comprise a switching element that can be controllably turned both on and off. The discharge switching element may for example comprise a gate-turn-off thyristor. In such embodiments the discharge switch controller may be configured to turn off the gate-turn-off thyristor a predetermined time after turning it on. The predetermined time may be relatively short, of the order of a few milliseconds or tens of milliseconds and may be just long enough for the SPD conduction to extinguish.

In other embodiments the discharge switching element may comprise a turn-on element only, such a thyristor, and the apparatus may comprise a capacitor in series with the discharge switching element. The capacitor in series with the discharge switching element may be rated so as to be able to charge to a voltage substantially equal to a nominal voltage of the energy storage device within a predefined period, e.g. of the order of a few milliseconds or tens of milliseconds.

In some embodiments there may be a single surge protection device for a cell but in some embodiments each cell apparatus may comprise a plurality of said surge protection devices.

In some embodiments at least one of the surge protection devices may comprise a triggerable gas discharge tube, i.e. a gas discharge tube that can be triggered to become conducting at a voltage lower than its nominal voltage rating - where the nominal voltage rating applies to the voltage across the gas discharge tube required to start conduction in the absence of suitable trigger signal. The controller may be configured to trigger the triggerable gas discharge tube in response to detection of a fault condition. In some embodiments the controller may comprise an over-current detector for detecting an over-current and may for instance be associated with the gating control electronics of an IGBT.

The one or more surge protection devices may each be located in a sealed housing. The housing may be a rupture proof housing such as a lightning strike rupture proof enclosure.

Each of the plurality of switches may comprise at least one insulated gate bipolar transistor with an anti-parallel diode. Where the surge protection device is connected in parallel with such a switch the surge protection device is thus also in parallel with the semiconductor switch and the anti-parallel diode.

The plurality of switches may comprise a first switch connected between said first and second cell terminals and a second switch connected in series with energy storage element between the first and second cell terminals. In other words the cell apparatus may comprise a half-bridge switch arrangement. In some embodiments however the plurality of switches may be formed in a full bridge arrangement.

In some embodiments the apparatus may be a multi-phase cell or sub-module of a VSC. In some embodiments the apparatus may have at least one additional cell terminal, e.g. at least a third cell terminal. The plurality of switches may be arranged such that the energy storage element may be selectively connected or bypassed between a selected pair of cell terminals. In such embodiments surge protection devices may be connected between at least some pairs of cell terminals, and possibly between each pair of cell terminals. Additionally or alternatively surge protection devices may be connected in parallel with one or more of the plurality of switches and/or the energy storage element.

The apparatus may be a cell forming part of a chain-link or chain-link converter of a VSC. Aspects also therefore relate to a chain-link for a VSC comprising a plurality of series connected cells wherein at least some of the cells comprise apparatus as described above.

The cell may form part of a VSC, for example such as a modular multilevel converter (MMC) or alternate arm converter (AAC) or other type of VSC. Aspects also therefore relate to a VSC comprising a plurality of switching cells wherein at least some of the cells comprise apparatus as described above.

The apparatus is for use in a VSC, in particular for a VSC for high voltage direct current (HVDC) power distribution.

The invention will now be described by way of example only with reference to the accompanying figures, of which:

Figure 1 illustrates a sub-module or cell of a chain-link with a conventional bypass protection system;

Figure 2 illustrates a sub-module or cell having fault protection according to an embodiment;

Figure 3a illustrates one example of a suitable surge protection device and figure 3b illustrates the voltage and current characteristics of such a device;

Figure 4 illustrates a further embodiment of a sub-module or cell having fault protection;

Figure 5 illustrates an embodiment of a sub-module or cell having multiple surge protection devices;

Figure 6 illustrates an additional embodiment of a sub-module or cell having multiple surge protection devices;

Figure 7 shows a further embodiment of a multi-phase sub-module or cell having surge protection devices; and

Figures 8a and 8b illustrate an embodiment for overvoltage protection and voltage clamping of a capacitor of a sub-module or cell.

Embodiments of the present invention relate to fault protection apparatus and methods for cells or sub-modules of a voltage source converter (VSC) and to an apparatus, e.g. a cell, for use in a VSC that include fault protection. Embodiments of the present invention use at least one surge protection device placed across at least some of the sensitive components of the cell, where the surge protection device comprises a Gas Discharge Tube (GDT) in series with a Metal Oxide Varistor (MOV).

Figure 2 illustrates a cell or sub-module apparatus 200 suitable for use in a chain-link of a VSC according to an embodiment of the present invention in which similar components to those described previously with reference to figure 1 are identified by the same reference numerals.

Figure 2 illustrates that the cell 200 includes a surge protection device 201. The surge protection device acts so that at voltage levels below a certain protection voltage level, sometimes referred to as a let-through voltage, the surge protection device is effectively open-circuit and thus does not provide a substantially conductive path. If, however, the voltage increases above the let-through voltage level the surge protection device effectively provides a conductive, i.e. relatively low impedance, path and thus provides a short circuit path to divert current away from the protected component(s) and/or reduce any overvoltage.

Surge arrestors using a Metal Oxide Varistor (MOV) are known for overvoltage protection for a range of applications. MOVs comprise a bulk semiconductor material, typically compressed zinc oxide powder, that presents a high resistive value when a voltage below the voltage rating of the MOV is applied, but a very low resistive value when the applied voltage rises above a certain protective level, i.e. the let-through voltage. Thus if the voltage level rises above the protective level the MOV effectively short circuits to provide a divert current path. The MOV recovers to a high resistive value by its own means when the applied voltage subsequently reduces below the protective level.

Ideally, if used to provide overvoltage protection for components of a cell of VSC, the transition between the MOV being high resistance, in the voltage range up to its rated voltage, and being low resistance at voltages above the protective level should be as abrupt as possible. For most practical MOVs however there is a transition zone between the low resistance and high resistance cases. This existence of a relatively significant transition zone in the voltage-resistance properties of an MOV can impact on the use of MOV surge arrestors.

For example consider placing an MOV surge arrestor in parallel with one of the IGBTs of the cell to provide overvoltage protection for the IGBT. The MOV should be arranged to ensure that it becomes conductive at a voltage that is lower than the breakdown voltage of the IGBT. However for a standard MOV surge arrestor, in order to ensure that the surge arrestor is sufficiently conductive at a desired protective voltage level, it is typically the case that the rated voltage level may not be high enough for the normal operating voltage range of the IGBT. Thus the IGBT would need to be under utilised in normal operation to ensure the MOV does not erroneously become conductive. For at least some standard MOV surge arrestors the transition zone may extend from a first voltage threshold, say X, to a second voltage threshold which may typically be about 1.7 times the threshold, i.e. 1.7X. In other words the MOV may start to become conductive at the first threshold of X but only be fully conductive at the second voltage threshold of 1.7X. As the MOV thus only becomes fully conducting at the voltage of 1,7X the IGBT should be rated for a voltage of 1,7X, but in use to prevent unwanted conduction the normal operating voltage should be kept below X.

Additionally MOVs exhibit leakage currents even in the nominally non-conductive state. This may not be a concern for some applications but if connected across an IGBT of VSC the leakage current could affect voltage balancing across IGBTs when in an off-state. The leakage current also contributes to power losses. MOVs also have a finite life expectancy and tend to degrade when exposed to voltage transients. MOVs are also prone to failure under sustained high current stress. The failure modes of MOVs may be such that the MOV fails to open-circuit, i.e. to a nonconducting state. If an MOV surge arrestor connected across, i.e. in parallel with, a sensitive component fails to open circuit during an overvoltage event the divert path provided by the MOV will disappear and the current/voltage will again be applied to the component it was desired to protect. Alternatively an MOV may fail to a state that provides a resistive current path. This would allow a fault current to be maintained that can result in significant overheating. This can be a significant danger for HVDC applications.

Embodiments of the present invention thus use a surge protection device that is not simply just an MOV surge arrestor. The surge protection device (SPD) used in embodiments of the present invention comprises a Gas Discharge Tube in series with a Metal Oxide Varistor. This ensures superior electrical characteristics, e.g. an enhanced trigger profile, compared to a conventional MOV surge arrestor in that the protection voltage level can be made closer to the operating voltage level without compromise of the operating voltage level and the transition between non-conducting and conducting is more abrupt. There is also no leakage current when below the protection voltage level. The SPD used in embodiments of the present invention also exhibits a failure mode to a short circuit, whereby if the SPD fails, it will permanently short circuit. This is a safer failure mode then a failure to open circuit for overvoltage protection; as a standard MOV arrestor would behave.

Figure 3a illustrates one example of a suitable surge protection device 201. The SPD has a gas discharge tube (GDT) 301 electrically connected in series with at least one Metal Oxide Varistor (MOV) 302.

The GDT 301 typically comprises two electrodes 303a and 303b spaced apart and separated by a gas 304 at a certain pressure. The electrodes and gas are housed in suitable housing 305 such as a ceramic enclosure. To ensure that the GDT 301 itself fails to short circuit the housing 305 may be provided with be a short circuit arrangement 306. Various short circuit arrangements for GDTs are known, for example including conducting elements biased to provide a low resistance bypass path, i.e. a short-circuit, but normally separated by at least one spacer element. The spaced element may be designed to melt at a given temperature such that the bypass path is established. Thus in the event of high temperatures due to, or indicative of, likely failure the short circuit arrangement is activated to result in the GDT being connected in a permanently short-circuit and relatively low resistance state.

The GDT is in series with an MOV 302. The MOV may comprise first and second electrodes 306a and 306b separated by a varistor material 307, for example Zinc Oxide and contained in a suitable housing 308. In various embodiments the MOV may have an integral fail to short circuit arrangement. For instance the MOV may be packaged using the so-called hockey puck packaging arrangement and/or may be arranged with electrodes that will start to arc if the MOV starts to fail, with a material arranged to be melted by the arcing resulting in the electrodes being connected in a low resistance path.

The GDT 301 and MOV 302 may be contained in a rupture proof housing 308.

This combination of a GDT and varistor in series is advantageous. Figure 3b illustrates the principles of how the SPD may act in terms of voltage and current in an overvoltage situation. Figure 3b illustrates plots of both voltage and current against time. It will be appreciated that these are notional plots to explain the principles and do not necessarily reflect accurate waveforms. Consider that the voltage across the SPD rises due to a fault leading to an over-voltage, as illustrated by line 310. Were there only an MOV on its own, i.e. without a series connected GDT, then the voltage would increase until the MOV was fully conductive and then the voltage would be clamped at a certain clamping voltage Vc. As the MOV will be at least partly conductive before this time however the current through the MOV would increase even at lower voltages, i.e. before the clamping voltage is reached at a time tc. The dot-dash lines of figure 3b illustrate the response of an MOV on its own.

With the GDT 301 connected in series with the MOV 302 however, the GDT is effectively open circuit until a spark-over voltage Vs is reached. Thus for the SPD including the GDT the voltage increases until the spark-over voltage is reached and no current flows until then. Once the spark-over voltage is reached gas discharge occurs. As will be appreciated by one skilled in the art once the gas discharge starts it can be maintained at a lower voltage in an arcing or glow regime. If there were only a GDT, i.e. there was no MOV connected in series, then the voltage may then drop to an arc voltage VA as illustrated by the dashed line in the voltage plot of figure 3b. This arc voltage may be below the voltage level of concern. However the series connection of the GDT and MOV provides the response indicated by the solid lines in the figure 3b. Thus until the spark-over voltage Vs is reached no significant current flows through the SPD. Once the spark over voltage is reached the GDT becomes conducting. At this point the voltage is still well above the clamping voltage of the MOV and so the MOV also becomes conductive. In this instance however the clamping voltage is determined by the MOV (together with the arc voltage of the GDT). Thus the clamping voltage can be controlled whilst providing a much more abrupt transition between being conducting and non-conducting.

It will be appreciated that the GDT does require a relatively high voltage spike in order to become conductive, i.e. the overvoltage must reach the spark-over voltage to trigger the GDT. For this reason GDTs are sometimes not considered suitable for overvoltage protection of some components. However in embodiments of the invention where the SPD is used for protection of power modules of voltage source converters, it has been appreciated that the capacitance within the VSC means that this voltage spike may not be experienced by the IGBTs of the power modules, or any power spike that is experienced is of too short a duration to result in any damage.

The GDT and MOV will thus be selected to provide an overall response that provides a desired protection level and packaged in a way that provide fail to short circuit operation. High voltage GDTs and MOV are known and series connections of GDTs and MOV have been proposed for other applications, such as lightening strike surge protection, however such devices would not previously have been considered for use with power modules of a VSC.

Referring back to figure 2, in this example the SPD 201 is connected across, i.e. in parallel with, the energy storage element of the cell 200, i.e. capacitance 101. The cell also has a bypass semiconductor switching element, e.g. thyristor 105, for providing a bypass path between the first and second cell terminals 104a and 104b for fault current in a first direction, i.e. a fault current such as illustrated with respect to figure 1. In the event of a fault such as illustrated with respect to figure 1 above that results in a fault current in the first direction, the bypass thyristor 105 may be operated as described to provide a bypass path and thus divert current away from the cell.

The surge protection device 201 provides protection form overvoltage resulting from a fault that leads to a fault current in a second direction, opposite to the first direction. In the event of such a fault, if the switch 102b is conductive the current may be initially be conducted via this switch as illustrated by path 202-1. In the event of the fault being detected the switches 102a and 102b are blocked and thus conduction may be via the diode 103 corresponding with switch 102a. With switch 102b off (and also switch 102a), the current path would thus be via the capacitance 101 as illustrated by path 202-2. However if the DC capacitor voltage increases to the protection level, i.e. the let-through voltage, of the SPD 201, the SPD will become conductive and thus discharge the DC capacitor (via path 202-3) and protect the IGBT from overvoltage.

The SPD 201 should therefore be able to absorb the energy stored in the capacitance 101.

If the fault current in such a situation is relatively very high, or is sustained for a relatively long time, it is possible that the SPD may fail. However as described previously the SPD is configured with a failure mode that results in a short circuit. Thus even if the SPD 201 fails the sub-module will be left in a safe state, and can be bypassed.

Since the surge protection device 201 and the thyristor 105 protect the sub-module from fault currents in both directions, the electromechanical switch 105 could be omitted thus removing a bulky and costly component from each cell of a chain-link of a VSC.

In the embodiment illustrated in figure 2 it will be noted however that if a fault does arise with a fault current in the second direction, the semiconductors of switch 102b are forced to support a sustained fault current until desaturation is detected in the gateboards and the IGBT is turned off.

Figure 4 illustrates an apparatus 400, i.e. a cell or sub-module, for use in a VSC according to another embodiment of the invention, where similar components are identified by the same reference numerals. In this embodiment the surge protection device 201 is connected across, i.e. between, the first and second cells terminals 104a and 104b. Thus the SPD is connected across all of the cell components. In such an embodiment, in the event of a fault leading to a fault current in the second direction, the SPD will automatically become conductive to provide a bypass path 202-3 for current in the second direction in the event of an overvoltage. This means that no sustained fault current will be experienced by the switches 102a and 102b, although the diode 103 corresponding to switch 102a may conduct an initial fault transient via path 202-2.

In the embodiment of figure 4 the voltage of capacitance 101 is not discharged in the event of a fault but is instead clamped.

It will be appreciated that in the event of a fault current that flows in the first direction, i.e. in the direction illustrated in figure 1, such a fault may be dealt with as described with reference to figure 1. That is the thyristor 105 may be triggered to become conductive to carry the fault current, thus bypassing the sensitive components of the cell. The SPD itself is bipolar, i.e. will conduct in both directions, and thus would act to become conductive and provide a bypass path for current flow in this direction if the let through voltage were reached. However activating the thyristor 105 will tend to collapse any voltage difference across the SPD 201 and thus the fault current will be carried by the thyristor. Were the thyristor to fail however and an over-voltage were to develop across the terminals of the cell the SPD would become conductive to provide a bypass path.

In theory the thyristor 105 could thus be omitted for fault handling reasons as an SPD located across the cell terminals can provide bipolar over-voltage protection. In practice however the thyristor 105 may be included to operate in such a fault mode to limit the current that would be carried by the diode associated with switch 102b.

Figures 5 and 6 illustrate further possible embodiments with more than one surge protection device. In the embodiment illustrated in figure 5 there is a first SPD 201a in parallel with switch 102b, i.e. in parallel with the IGBT, and an additional SPD 201b, i.e. a second SPD, in parallel with switch 102a. For the half-bridge cell illustrated this means that the first SPD 201a is also effectively connected across the cell terminals 104a and 104b, although it will be appreciated that this would not necessarily be the case for a full-bridge cell. Thus each switch of the cell, at least each switch in a path between the cell terminals, may have a SPD as described above connected in parallel with the switch. Both switches are thus protected by SPDs and in the event of a fault the fault current will be diverted away from semiconductor devices. If both SPD become conductive this will also effectively short circuit the capacitance 101 and allow the capacitance to be discharged.

Figure 6 illustrates a further embodiment of a cell apparatus 600 which has SPD 201 in parallel with the capacitance 101 as described above with reference to figure 2 and also an additional SPD 201a connected across the cell terminals 104a and 104b. The SPD 201a across the cell terminals will divert a fault current in the second direction away from the cell whilst the SPD 201a will prevent overvoltage of the capacitance 101.

Using more than one SPD within a cell may allow the individual SPDs to be designed with let-through voltage levels appropriate for the component being protected by that SPD. Thus the various SPDs may be designed to have different voltages ratings, i.e. different let-through voltages.

The embodiments of the present invention thus use one or more SPDs with improved triggering profiles to provide fault protection. The SPD will recover to its normal resistance state once the overvoltage ceases and the use of SPDs can prevent cascaded failure from occurring. The provision of the SPD may allow the bulky electromechanical switch to be omitted with a cost and size saving.

The SPDs described in the embodiments above are passive components in that they do not need any external control signal and act autonomously in response to any surge in voltage within the cell or submodule in use.

In some embodiments however the gas discharge tube for at least one surge protection device may be a triggerable gas discharge tube (GDT), where the discharge can be controllably triggered at a voltage lower than the normal let-through voltage in response to a suitable control signal. In other words the GDT can be triggered so as to lower the let-through voltage needed to initiate discharge.

Various types of triggerable GDT are known, for instance referring back to figure 3a the GDT 301 may comprise a third electrode 309 that in response to a suitable trigger signal Gt acts to reduce the spark-over voltage. For instance the electrode 309 may be an electrode that at least partly surrounds the gas and acts to at least partly ionise the gas when triggered. A suitable trigger signal can thus trigger the spark gap and allow gas discharge to start and large currents to flow below the nominal spark over voltage.

This could be useful, for instance, if there is a high surge current that does result in a voltage above the normal let-through voltage, for example as may occur with a surge current through the IGBT of switch 102b. In such a situation it would be desired to trigger the SPD to provide protection to divert the current away from the IGBT. In some embodiments therefore a controller, such as gating electronics of the IGBT, may detect an over-current and generate a trigger control signal for the IGBT. Referring back to figure 5 a controller 501 may therefore detect an over-current and generate a trigger control signal for SPD 201a, which includes a triggerable GDT. It will be appreciated however that any of the SPDs described in any of the embodiments discussed above may be implemented with a triggerable GDT.

The embodiments described above have illustrated cells or sub-modules with switches in a half-bridge configuration. The use of SPDs as described can also be useful for other cells however such as cells with full-bridge switch arrangements. In a full-bridge cell each of the first and second cell terminals is typically connected to both ends of the cell capacitor by separate switches such that, when no bypassed, the capacitor can be connected between the cell terminals in either of two orientations to present voltages of different polarity. An SPD as described may be connected between the cell terminals of a full-bridge cell and/or an SPD may be connected in parallel with a capacitor. At least some of the switches may additionally or alternatively be connected in parallel with an SPD.

In addition in some embodiments the cell or sub-module may be a multi-phase cell or sub-module, i.e. the cell or sub-module may have multiple switching arms in parallel with the energy storage element, each switching arm of the cell having an upper switch and a lower switch and a cell terminal connected between the upper and lower switches. Figure 7 for example illustrates an example of a three phase cell 700. This cell has first and second cell terminals 104a and 104b and also a third cell terminal 104c. SPDs 201 may be connected between each pair of cell terminals, e.g. (1) between terminals 104a and 104b, (2) between terminals 104a and 104c, and (3) between terminals 104b and 104c. Additionally or alternatively such a cell could have SPDs 201 connected across individual switches and/or across the energy storage device, e.g. capacitor 101.

The discussion above has focussed on the issues of fault conditions leading to voltage fault. The use of SPDs as described can also be advantageous in handling occasional hard switching events for switching elements of a VSC. Typically the switching elements of the cells of a VSC are arranged to switch at point in the power cycle where the switches are conducting no significant current. In some instances however it may be necessary to turn a switch off at a point at which it is conducting significant current. This is known as hard switching. This could be for instance due to some abnormal operating conditions. The use of an SPD such as described with reference to figure 3 could be used for switching elements of a VSC to help protect the switching elements in instances of hard switching. The SPD could be a passive SPD, i.e. one that simply reacts to an over-voltage that develops as the switch turns off whilst carrying significant current or may be a triggerable SPD, i.e. an SPD with a triggerable GDT with a suitable trigger control signal being generated on detection of a hard switching event.

As described above in relation to figure 2 an SPD 201 may be connected in parallel with the capacitor 101 of the cell to provide protection for a fault current in the second direction. Additionally or alternatively, in some embodiments, an SPD connected to the cell capacitor can be used to prevent an over-voltage of the capacitor and provide a rapid discharge function.

Typically the capacitor of the cell or sub-module may have a maximum voltage rating, say around 2kV for example. The cell may thus comprise a rapid discharge circuit for discharging the capacitor in the event of an overvoltage. Such a rapid discharge circuit may typically comprise a semiconductor switch such as an IGBT in series with a discharge resistor. If an overvoltage is detected the semiconductor switch can be turned on to discharge the capacitor via the discharge resistor.

In some embodiments an SPD such as described above can be used instead of such a rapid discharge circuit. Figure 8a illustrates an example where an SPD is arranged to provide overvoltage protection and discharge for a capacitor 101 of a cell or sub module. AN SPD 201, which is of the type described above with respect to figure 3a, is connected in parallel with the capacitor 201 and in series with a discharge resistor 801.

The SPD 201 is configured so that the let through voltage of the SPD is rated to desired overvoltage protection level for the capacitor. For example the spark-over voltage of the GDT of the SPD may be designed to be around 2.2kV. As described previously the SPD may activate autonomously and thus transitions from open circuit to short circuit automatically is an overvoltage develops on the capacitor 201. The automatic operation of the SPD avoids the need for break-over diodes or the like such as are needed for the conventional rapid discharge circuit.

In the embodiment of Figure 8a the SPD is arranged in parallel with a semiconductor switch element 802, which may conveniently be a gate turn off (GTO) thyristor. The combination of the SPD and semiconductor switch element 802 allows the circuit to be triggered at a pre-determined voltage (defined by the SPD) and also be turned-off at a predetermined voltage.

In response to an overvoltage the SPD will be triggered and will start conducting. Once the nominal voltage is reached the semiconductor switch element 802 is turned on.

This commutates the current through the semiconductor switch. The voltage across the SPD 201 thus collapses and the discharge of the GDT of the SPD ceases. The SPD thus returns to being open circuit. The semiconductor switching element need only be turned on temporarily to extinguish the SPD. The semiconductor switch may be turned off within a short period of conducting, say a few milliseconds or tens of milliseconds. The capacitor 201 can absorb any resulting voltage transient.

This arrangement thus provide a rapid discharge of the capacitor and protection against overvoltage that has a defined trigger voltage and which provides a controlled discharge of the capacitor to a desired voltage level.

The semiconductor switching element 802 thus may be controlled by suitable control circuitry 803, e.g. a gate board of a GTO thyristor. The control circuitry may be confirmed to perform current sensing on the SPD and to trigger the semiconductor switching element 802 when the nominal voltage is reached. The semiconductor switch may then be commutated off a predefined period later, say 5ms or so, at unity gain preferably. Once the semiconductor switching element is off the voltage of the capacitor 201 will be at or very near the nominal voltage and thus below the trigger voltage of the SPD. Normal use of the cell or sub-module can thus continue.

To perform unity gain turn off at 3kA, approximately 3mF on-board capacitance may be required to absorb the stored charge within the GTO thyristor during reverse bias of the gate to source terminals at turn-off. The on-board capacitance may be of the surface mount ceramic type which have high density in a small footprint, and thus the gate board may be of comparable size to that of the conventional IGBT. To turn off with surge currents, a Gate-Commutated-Thyristor (GCT) style segmented gate structure may be used, to ensure reverse gate current capability equals that of the expected surge current. Additionally or alternatively a capacitor may be arranged in series with the GTO thyristor to reduce the peak surge current to the point where the GTO can be safely turned off.

In some embodiments a capacitor 804 may be connected in series with the semiconductor switching element 802, as illustrated in figure 8b, the capacitor being rated such that the semiconductor switching element may be of the turn-on only type, e.g. a thyristor. The presence of the capacitor 804 avoids the need to control turn off of the semiconductor switching element 802 by driving the current in the thyristor to zero. This may simplify the gate circuitry of the semiconductor switch compared with the embodiment of figure 8a.

In the embodiment of figure 8b if an overvoltage triggers the SPD 201 and the voltage reaches the nominal value the semiconductor switching element 802, e.g. thyristor, is turned on. As described above the voltage of the SPD will collapse and conduction via the SPD extinguishes. In addition the capacitor 804 will charge up. The capacitance of the capacitor 804 is arranged such that the capacitor 804 charges relatively quickly, say a few milliseconds, to the same voltage level as the main DC capacitor 101. At this point the capacitor 804 is balanced with the main DC capacitor 101 and the thyristor 802 will thus naturally be commutated off.

In one example a capacitance of approximately 2mF for capacitor 804 may be sufficient for capacitor 804 to charge to approximately 1.4kV, resulting in turn off the thyristor within about 1ms. Such a value of capacitance could be achieved in a small space with high density ceramic surface mount capacitors. These capacitors may be arranged in a series-parallel array on a double sided PCB to obtain the desired capacitance and voltage rating.

This combination of an SPD and semiconductor switching element, in particular a thyristor or GTO thyristor, offer various advantages over the conventional rapid discharge circuit using an IGBT and discharge resistor. In the conventional circuit the IGBT carries significant sustained current during the discharge and thus must have a relatively high rating for continuous current, e.g. of the order of 1000A. The IGBT is thus a costly and bulky component. In the embodiment of figure 8a the SPD 201 carries the main discharge current and the semiconductor switch is provided to commutate the current away from the SPD for the relatively short time taken for the SPD conduction to extinguish, which may take of the order of 5ms or so. The semiconductor switch thus need only be rated to handle a relatively large surge current for a short period of time, say 10ms or so, and thus can tolerate a much lower continuous current rating. A suitable 100A GTO thyristor will typically satisfactorily tolerate a current of up to 3kA for a period of the order 10ms or so and such a thyristor is significantly less costly and bulky that the normally required IGBT. The SPD is also a relatively low cost and small component.

As noted above the SPD may be a solely passive device with no control logic or power supply required, thus avoiding the requirement for break-over diodes (BODs) or the like which are relatively expensive. The arrangement illustrated in figure 8a only needs triggering of the GTO when current is detected in the SPD, which can be achieved with a simple pulse transformer. A suitable SPD 201 may be able to tolerate currents in excess of 10kA and, as mentioned, a GTO thyristor may safely handle a surge current in excess of 3kA for the short period required. This means that the ohmic value of the discharge resistor may be lower than would normally be required with the conventional rapid discharge circuitry, say of the order of 1 ohm rather than of the order of 2 ohms as would typically be required. This has the advantage of reducing discharge time and possibly lowering the cost of the resistor. Although the resistor needs to handle a higher current, as it does so for a short time, thus energy dissipation requirements will remain the same as for the conventional circuit. However the thickness of the resistor arrangement may possibly be reduced due to lower ohmic value.

In addition the SPD, when conducting, will have a relatively high residual voltage. Thus the absorbs relatively a lot of energy during discharge of the capacitor. For example with a 2 ohm discharge resistor the SPD may absorb of the order of two-thirds of the total energy and during SPD conduction the circuit behaves like a potential divider between the SPD and the discharge resistor. This has potential cost/size benefits by reducing the impulse energy rating of the resistor.

The use of an SPD in parallel with a semiconductor switching element to provide controlled discharge of the main energy storage capacitor of a cell or sub-module of a VSC represents a novel aspect of the invention.

As mentioned the cells may be assembled in use to form a chain-link circuit for a VSC for use in HVDC. Embodiments also thus relate to a chain-link for a VSC comprising a plurality of series connected cells where at least some, and possibly all, of the cells have SPDs as described above. The cells may form part of a VSC and thus embodiments also relate to VSCs comprising cells such as described above. Various VSC designs use chain-links, for example an MMC type VSC or an AAC, and thus the cell apparatus may be a sub-module of an MMC or AAC VSC.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended claims. The word “comprising” does not exclude the presence of elements or steps other than those listed in a claim, “a” or “an” does not exclude a plurality, and a single feature or other unit may fulfil the functions of several units recited in the claims. Any reference signs in the claims shall not be construed so as to limit their scope.

Claims (15)

1. An apparatus for use in a voltage source converter comprising: first and second cell terminals; an energy storage element; a plurality of switches configured such that first and second cell terminals can be selectively connected in a path that includes the energy storage element or a path that bypasses the energy storage element; and one or more surge protection devices comprising a gas discharge tube in series with a metal oxide varistor. wherein the one or more surge protection devices are connected in parallel with at least one of said plurality of switches and said energy storage element.

2. An apparatus as claimed in claim 1 further comprising a bypass semiconductor switching element for providing a bypass path between the first and second cell terminals for fault current in a first direction.

3. An apparatus as claimed in claim 1 or claim 2 wherein the one or more surge protection devices are configured to fail to short-circuit.

4. An apparatus as claimed in any preceding claim wherein a first surge protection device of the one or more surge protection devices is connected in parallel with the energy storage element.

5. An apparatus as claimed in claim 4 further comprising a discharge switching element in parallel with the first surge protection device and a discharge switch controller, wherein, in the event that the first surge protection device becomes conducting, the discharge switch controller is configured to control turn-on of the discharge switching element so as to extinguish conduction via the first surge protection device.

6. An apparatus as claimed in claim 5 wherein the first surge protection device is configured to start conducting at a first voltage level and the discharge switch controller is configured to turn-on of the discharge switching element at a second, lower, voltage level.

7. An apparatus as claimed in claim 5 or claim 6 wherein the discharge switching element comprises a gate-turn-off thyristor and the discharge switch controller is configured to turn off the gate-turn-off thyristor a predetermined time after turning it on.

8. An apparatus as claimed in claim 5 or claim 6 wherein the discharge switching element comprises a thyristor and the apparatus comprises a capacitor in series with the discharge switching element.

9. An apparatus as claimed in any of claims 5 to 8 further comprising a discharge resistor in series with the first surge protection device.

10. An apparatus as claimed in any preceding claim wherein one of the one or more surge protection devices is connected across the first and second cell terminals.

11. An apparatus a claimed in any preceding claim wherein one of the one or more surge protection devices is connected in parallel with one of said plurality of switches.

12. An apparatus as claimed in any preceding claim wherein at least one of said one or more surge protection devices comprises a triggerable gas discharge tube that can be triggered to become conducting at a voltage lower than its nominal voltage rating.

13. An apparatus as claimed in claim 8 further comprising a controller configured to trigger the triggerable gas discharge tube in response to detection of a fault condition.

14. An apparatus as claimed in claim 13 wherein said controller comprises an overcurrent detector for detecting an over-current.

15. A voltage source converter comprising at least one phase limb, the phase limb comprising a plurality of cells, wherein at least some of the plurality of cells comprise an apparatus as claimed in any preceding claim.