Semiconductor switch

Abstract

Proposed is a semiconductor switch that is formed with a multiplicity of series-connected circuit modules. The circuit modules each comprise: a first transistor (M3) having a source terminal (source3), a control terminal (Gate3) and a drain terminal; and a second transistor (Q2) having a source terminal, a control terminal and a drain terminal. The source terminal of the second transistor (Q2) is connected to the control terminal (Gate3) of the first transistor (M3), and the drain terminal of the second transistor (Q2) is connected to the source terminal (source3) of the first transistor (M3). Also provided is a driver (V2) that is connected via connections to the control terminals of the second transistors (Q1, Q2, Q3, Q4, Q5, Q6) of the individual circuit modules. For a subset of the plurality of circuit modules, a device for reducing the control terminal voltage of the first transistor is incorporated in the relevant connection to the driver. This measure allows improved setting of the maximum short-circuit current, and a better-controlled procedure in the event of defects of the semiconductor switch according to the invention.

Description

[0001] Description

[0002] Haiblei ter schal ter

[0003] The invention relates to a semiconductor switch.

[0004] Advances in the development of semiconductor components have led to new switch concepts that can replace conventional, typically electromechanical low-voltage switches. These new concepts apply to circuit breakers or motor starters, for example, but can also, in principle, be used for switching higher currents, e.g., through circuit breakers. The term SSCB (Solid State Circuit Breaker) is also commonly used for circuit breakers and circuit breakers.

[0005] Of key importance here are the fast response times of the semiconductor components, which prevent damage from overload. This means that the switch switches off the current before the semiconductor components can be damaged. Nevertheless, given the greater sensitivity of semiconductor components compared to conventional switching elements, it can happen that the semiconductor-based switching mechanism becomes damaged and the switch no longer functions properly.

[0006] An efficient way to realize a semiconductor switch with comparatively moderate conduction losses is to design the semiconductor switch as a series connection of transistors (e.g. MOSFETs or IGBTs).

[0007] The invention aims to improve switches formed by series circuits of transistors.

[0008] This object is achieved by claim 1. Advantageous further developments are specified in the subclaims. The invention is based on a semiconductor switch with a plurality of circuit modules connected in series. The circuit modules are each formed with a first transistor with a source terminal, a control terminal and a drain terminal, wherein the source terminal is a source terminal or an emitter terminal, the control terminal is a gate terminal or a base terminal and the drain terminal is a drain terminal or a collector terminal. In addition, the circuit modules each comprise a second transistor with a source terminal, a control terminal and a drain terminal, wherein the source terminal is connected to the control terminal of the first transistor and the drain terminal is connected to the source terminal of the first transistor (M3).

[0009] The semiconductor switch is formed with a driver which is connected via connections to the control terminals of the second transistors of the individual circuit modules.

[0010] For the optimization described in more detail below, for a subset (e.g. for one circuit module or two circuit modules) of the plurality of circuit modules, a device (e.g. Zener diode) for reducing the control terminal voltage of the corresponding first transistor is introduced in the respective connection to the driver.

[0011] The first transistor can be a unipolar transistor, e.g. a MOSFET. But it can also be implemented using a bipolar transistor, e.g. an IGBT (insulated-gate bipolar transistor) with an anti-parallel protection diode (integrated or as an external freewheeling diode). In a unipolar transistor, the control terminal voltage would then be the gate-source voltage (hereinafter also referred to as the gate voltage), while in a bipolar transistor it would be the base-emitter voltage. In an IGBT (insulated-gate bipolar transistor), which has the insulated gate of a voltage-controlled component (e.g. MOSFET) combined with the bipolar output characteristic field of a bipolar transistor, it would then be the gate-emitter voltage.

[0012] The improvements sought through this measure concern the following points:

[0013] 1. Without the control terminal voltage-reducing device, current limitation would depend on the driving voltage, which could result in increased design complexity in the application. As illustrated below in an example, the targeted selection of circuit modules for the installation of a control terminal voltage-reducing device allows the countervoltage built up during a short circuit, and thus the value of the maximum short-circuit current, to be adjusted.

[0014] 2. The first individual transistors of the circuit modules of the semiconductor switch are loaded to different degrees. In the event of a rapid current increase as a result of a short circuit, the outer transistors (near the connection terminal) are not loaded very heavily because the current is quickly commutated to the relief network. The transistor supplied directly from the driver also remains in the linear or ohmic characteristic range and is not loaded very much. One of the transistors shifts its operating point into the pinch-off range and is therefore heavily loaded (depending on the type of transistor, the terms used here for a MOSFET transistor may differ, e.g. an IGBT is referred to as a saturation range and an active range; however, the principle applies generally and the problem arises regardless of the type of transistor).Which of the transistors in the circuit assumes this unfavorable state is not clear and depends on the specific conditions of the individual case. However, for easy repair in the sense of sustainability and resource conservation, it would be important to have a single transistor with a higher thermal stress so that this then also has a significantly worse MTBF (mean time between failures). This would ensure that if there is a defect in the semiconductor switch, this one power semiconductor is very likely to be defective. This would create a type of "sacrifice transistor" which could be prepared for quick and easy replacement (either just the transistor or the module containing the transistor) (e.g. particularly easily accessible, pluggable, marked for easier identification, etc.).This can be achieved by selectively reducing the control terminal voltage for a semiconductor switch formed with identical circuit modules.

[0015] In another embodiment of the semiconductor switch according to the invention, the increased stress on the first transistors of the subset of switch modules with control terminal voltage-reducing devices is compensated for by using different first transistors than the other first transistors. Preferably, for specified operating conditions for the semiconductor switch (e.g., rated current), the first transistors of the subset are then designed for a service life or MBTF comparable to that of the other first transistors under the specified operating conditions.

[0016] In one embodiment of the semiconductor switch according to the invention, it has an even number of circuit modules, the first half of the circuit modules being connected in series one after the other with the same switching direction, and the second half of the circuit modules being connected in series one after the other with the opposite switching direction. Two of the circuit modules are then connected in series in opposite switching directions, and a device for reducing the control terminal voltage of the first transistor is introduced for one circuit module of the first half of circuit modules and for one circuit module of the second half of circuit modules in the respective connection to the driver, the position of the two circuit modules with respect tothe number of circuit modules between them and the circuit module with the corresponding switching direction, which is connected to a circuit module with the opposite switching direction, corresponds to one another. The semiconductor switch is then preferably designed to be mirror-symmetrical with respect to the line defined by the connection of two switch modules with different switching directions, with regard to the equipment of circuit modules with devices for reducing the control connection voltage.

[0017] There are a wide variety of design variants for the circuit modules. For example, as mentioned above, the first transistor can be a unipolar transistor, such as a MOSFET. However, implementation using a bipolar transistor, such as an IGBT (Insulated-Gate Bipolar Transistor) with an antiparallel protection diode (integrated or as an external freewheeling diode) is also possible.

[0018] According to one embodiment of the circuit module, the connection to the driver is formed by a conductor section into which a diode is introduced which blocks in the direction of the driver.

[0019] According to one embodiment of the circuit module, the control terminal and the source terminal of the second transistor are connected to one another, and a diode is arranged between the control terminal and the source terminal of the second transistor, which diode blocks in the direction of the control terminal.

[0020] According to one embodiment of the circuit module, the control terminal and the drain terminal of the second transistor are connected to one another, and a resistor (preferably an ohmic resistor) is arranged between the control terminal and the drain terminal of the second transistor.

[0021] According to one embodiment of the circuit module, a voltage limiter is provided in parallel with the first transistor. A capacitive resistor can also be provided in parallel with the first transistor.

[0022] According to a first alternative embodiment of the circuit module, the control terminal and the source terminal of the first transistor are connected to one another, and a diode is arranged between the control terminal and the source terminal of the first transistor, which blocks current in the direction of the control terminal. This diode is an optional feature, implemented, for example, by a Zener diode, which limits the maximum control terminal voltage.

[0023] For example, protective diodes and voltage limiters can also be provided selectively only for the circuit modules where the specific design of the semiconductor switch requires it. For example, the circuit module with the shortest distance to the driver, or for a symmetrically constructed semiconductor switch, the two circuit modules with the shortest distance to the driver, can be formed without a second transistor.

[0024] According to a second alternative embodiment of the circuit module, the circuit module is formed with a third transistor, wherein the third transistor is of the same type as the first transistor and the first and third transistors are each directly connected to one another via the source terminal (i.e., arranged practically in opposite directions). In this embodiment, the source terminal of the second transistor is connected to the control terminals of the first and third transistors, and the drain terminal of the second transistor is connected to the junction of the source terminals of the first and third transistors.

[0025] The invention is described in more detail below within the framework of an exemplary embodiment with reference to figures.

[0026] Fig. 1a and 1b: a circuit module for forming a semiconductor switch according to the invention, Fig. 2: the structure of a semiconductor switch formed with circuit modules according to Fig. 1a and Fig. 1b,

[0027] Fig. 3: the semiconductor switch according to Fig. 2, wherein according to the invention Zener diodes for selected circuit modules are incorporated in the connection to the driver,

[0028] Fig. 4a and 4b: diagrams showing the shift of the operating point and the reduction of the forward resistance for the circuit modules with the Zener diode inserted according to Fig. 3, and

[0029] Fig . 5a and 5b : Tables illustrating the effect of the Zener diode introduced according to Fig . 3 on the build-up of blocking voltages in circuit modules of a semiconductor switch according to Fig . 2 and Fig . 3 respectively in the event of a short circuit.

[0030] The invention is based on a semiconductor switch which is formed with circuit modules connected in series. Fig. 1a and Fig. 1b show an embodiment of such a circuit module. The module comprises a MOSFET M3 with a source connection source!, a gate connection Gate3 and a drain connection and a PNP bipolar transistor Q2 with an emitter connection, a base connection and a collector connection, the emitter connection being connected to the gate connection Gate3 of the MOSFET M3 and the collector connection being connected to the source connection source! of the MOSFET M3. The gate connection Gate3 and the source connection sourceS of the MOSFET M3 are connected to one another, and between these connections a diode D5 (preferably a Zener diode) is arranged which blocks in the direction of the gate connection Gate3.The base terminal and emitter terminal of the PNP bipolar transistor Q2 are connected together, and a diode D4 is arranged in the connection, which blocks conduction toward the base terminal. The base terminal of the PNP bipolar transistor Q2 is also connected to its collector terminal, with a resistor R5 looped into the connection.

[0031] In addition, a connection is provided from the base terminal of the PNP bipolar transistor Q2 to a driver V2. This can be seen in Fig. 2, which shows a semiconductor switch formed from series-connected modules with a driver V2. The driver V2 is connected to the PNP bipolar transistors Q1 - Q6 of six series-connected circuit modules. A diode D1, D6, D7, D12, D14, and D16 is inserted between the driver V2 and the base terminals of the PNP bipolar transistors Q1 - Q6, which blocks the direction of the driver V2. A fuse SI - S6 is also provided between the driver V2 and the diodes D1, D6, D7, D12, D14, and D16.In this switch geometry, the semiconductor switch is formed from an even number of modules, with the first half of the circuit modules connected in series with the same conduction direction and the second half of the circuit modules connected in series with the opposite conduction direction. In Fig. 1a and Fig. 1b, the semiconductor circuit module is shown twice, with different conduction directions. The semiconductor switch in Fig. 2 is therefore composed of half modules according to Fig. 1a and half modules according to Fig. 1b. The Zener diodes D5 and D18 in Fig. 1a and 1b are optional. In the middle of the circuit in Fig. 2, two modules with opposite conduction directions are connected in series. The driver V2 is connected to the junction of the two circuit modules connected in opposite conduction directions (source) and is connected to ground via this connection and the resistor R11.A resistor R2 or R6 is provided between this connection and the base terminals of the PNP bipolar transistors of the two circuit modules. In addition, a resistor RI is introduced between the positive pole of the driver V2 and the connections of the circuit modules in a forward direction. In addition, a capacitor C1 - C6 and a voltage limiter U1 - U6 (e.g. in the form of a suppressor diode or a varistor) are connected in parallel to the circuit modules. These voltage limiters of the circuit modules U1-U6 are optional components. Also shown is a voltage V1 and a load R3 fed from this voltage.

[0032] The following two points are relevant to the semiconductor switch design shown in Fig. 2:

[0033] Firstly, it is a circuit for alternating current flow (AC or DC with both current flow directions). For unipolar current flow, only half of the circuit shown would be required (i.e., only the modules with transistors in a forward direction).

[0034] On the other hand, the driver-related modules can, in principle, be designed without the transistors Q5 or Q6 for switching on. The elements D13, D14, D17, and R17, or D15, D16, D18, and R10, can then also be omitted, so that the modules then consist only of the MOSFETs M1 and M4. However, to ensure that all series-connected components are switched off as simultaneously as possible, or to symmetrically switch off the switching behavior, it is advantageous to construct these two modules in a similar manner to the others.

[0035] For the sake of simplicity, the operation is described below for a switch with unipolar current flow, which has only half of the circuit modules shown in Fig. 2 (i.e. only the modules with transistors of a forward direction).

[0036] With a suitable choice of driver voltage, the circuit shown here exhibits intrinsic short-circuit current limitation, since a voltage drop is generated across the series-connected components as a function of the load current. A Zener diode D19 and D20, respectively, is inserted into the gate path of MOSFETs M2 and M5 in the series circuit. These diodes specifically reduce the gate voltage for these MOSFETs and operate them close to the pinch-off region.

[0037] If the driver V2 has an output voltage of 15V and the MOSFETs M2 and M5 have Zener diodes D19 and D20 connected in series at a voltage of 8V, all MOSFETs except for the MOSFETs M2 and M5 are operated at their nominal gate voltage of 15V and thus have the lowest possible on-resistance RDSon. This is illustrated in Figures 4a and 4b. Figure 4a shows the current flowing through a typical MOSFET as a function of the voltage between drain and source V DS The relationship between current and voltage is initially essentially linear (which is equivalent to an essentially constant resistance) until the MOSFET reaches saturation and the current hardly increases with increasing voltage V DS The onset of saturation depends on the gate voltage V G s • Curves for different gate voltages V are shown Gs = 5V, 6V, 7V, 8V and 10V. Also shown is the boundary between the ohmic region and the pinch-off region, which increases with increasing gate voltage V GS to higher drain-source voltages V DS shifts . If a load current of e.g. approx. 250A flows, then for the MOSFETs that are operated at their nominal gate voltage of 15V, this corresponds to the drain-source voltages V shown in Fig. 4a in area 1 DS of approximately 0.7V. In contrast, the gate voltages reduced by the Zener diodes are V GS = 7V operated MOSFETs M2 and M5 near the pinch-off region at a drain-source voltage V DS of approximately 1.2V. The corresponding on-state resistance is shown in Fig. 4b (areas 1 and 2, respectively). For the gate voltage V GS = 15V operated MOSFETs it is approx . at approx . 2 . 8 mQ, for those with V GS = 7V operated MOSFETs M2 and M5 at approx . 4 . 8 mQ .

[0038] MOSFETs M2 and M5 operate at a gate voltage of only 7V and thus generally have higher losses, resulting in higher operating temperatures and longer temperature cycles, which reduces both their service life and their MTBF (mean time between failures). These MOSFETs will therefore fail sooner than those operated at higher gate voltages. By reducing the gate voltage using a Zener diode, it is determined which MOSFET will fail due to wear. Such a MOSFET is also referred to below as a sacrificial MOSFET.

[0039] This measure increases the total on-resistance of the entire switch (in Fig. 4b by approximately 2 mQ per sacrificial MOSFET). However, this increase in the on-resistance RDSon is slight and has only an insignificant effect on the performance characteristics of the switch. On the one hand, only two switch modules (with MOSFETs M2 and M5) in the series connection are operated with a slightly increased RDSon, and on the other hand, the other MOSFETs, now operating ideally, compensate for this increase. For example, with six MOSFETs connected in series, the RDSon is increased in an example case from 6 x 3.6 mQ = 21.6 mQ to 5 x 3.6 mQ + 4.0 mQ = 22.0 mQ. By operating the five other MOSFETs at 15 V, their RDSon is also reduced somewhat, even if this is not clearly evident from Fig. 4a and 4b. This could result in the same, if not a slightly reduced, total on-state resistance or total RDSon.In any case, the change is smaller than the scatter of this parameter stated in the data sheet and is therefore rather negligible.

[0040] A sacrificial MOSFET always operates with a significantly lower gate voltage than the rest of the series circuit. This has the following consequences: a. This MOSFET now switches to the active region at a specified current, which can be read off the datasheet, and builds up a counter voltage. b. The total counter voltage can be specifically adjusted, since all MOSFETs in this series circuit that operate above this MOSFET (between the MOSFET with Zener diode and the load connection, or further away from the driver than the MOSFET or the circuit module with Zener diode) are forced to switch to the active region as soon as the MOSFET builds up voltage, which is determined by the installation of the Zener diode.

[0041] A type of "sacrificial MOSFET" is therefore created which has higher losses than the rest during normal operation, but ensures that all the others can operate with optimized losses at this operating point. In contrast to short-circuit current limitation without a Zener diode, the individual MOSFETs no longer run into the pinch-off region one after the other and gradually increase the counter voltage, but the sacrificial MOSFET determines the drain current solely through its output characteristic, which leads to saturation. The short-circuit current limitation has therefore become voltage-independent. The 10% higher losses of the sacrificial MOSFET then also lead to a 10% higher chip temperature, which is reflected in the load change through higher temperature swings. As a result, this MOSFET ages faster than the other MOSFETs installed in the series circuit (quantifiable by the so-calledRaindrop model according to which large temperature changes lead to faster aging, but many small temperature changes also cause the component to age, see "Li fetime calculation for power modules, application and theory of models and counting methods", K. Mainka, M. Thoben, O. Schilling, Engineering Proceedings of the 2011 14th European Conference on Power Electronics and Applications Li fetime calculation for power modules, application and theory of models and counting methods). The higher temperature also increases the probability of failure of the sacrificial MOSFET. This one MOSFET is also subject to particular stress in the event of a short circuit or overload. All lower MOSFETs (i.e. MOSFETs between the reference potential of the driver and the sacrificial MOSFET) are always operated in the ohmic range and are therefore in an operating state with optimal power loss.All MOSFETs between the sacrificial MOSFET and the load connection, both the circuit part for blocking the positive and the negative voltage direction (i.e. MOSFETs further away from the driver than the sacrificial MOSFET) are switched off when the sacrificial MOSFET is saturated, because the transition to the pinch-off region creates a voltage on the drain-source path of the sacrificial MOSFET, causing the gate voltage of these MOSFETs to fall to zero. This means that the gate voltages of these MOSFETs become smaller than the threshold voltage, which corresponds to the "off" state. Thus, only the sacrificial MOSFET sees full load current at full drain-source voltage when it switches to short-circuit current limitation, which means additional thermal stress for this MOSFET alone and also leads to faster aging of this MOSFET.

[0042] Alternatively, it is possible to deliberately over-design the sacrificial MOSFET (e.g., lower on-resistance RDSon) and thus increase it to the same MTBF as the rest of the circuit. As a result, the circuit would no longer consist of one MOSFET type, but of two.

[0043] A calculation example for the expected gate voltages of the individual MOSFETs can be found in Fig. 5a and 5b. The tables shown in these figures are again based on the use of a MOSFET with a forward resistance RDSon of 3.9 mΩ (in the linear range), such as the Infineon MOSFET IPT039N15N5. In the table in Fig. 5a, the driver voltage is 11 V, and in the table in Fig. 5b, it is 15 V. The individual MOSFETs or switch modules are numbered in the first column. The list corresponds to an increasing distance from the driver, ie the data in the first row corresponds to the MOSFET closest to the driver (e.g., MOSFET M1 in Fig. 3), the data in the second row to the next MOSFET (e.g., MOSFET M2 in Fig. 3), etc. Columns 2-5 list the gate voltages U Gs for different load currents (50A, 100A, 200A and 400A). In the last row, the total nominal on-resistance RDSon of the MOSFETs is listed according to number (i.e. number in the first column multiplied by the on-resistance RDSon of 3.9mQ, e.g. 2*3.9mQ = 7.8mQ, 3*3.9mQ = 11.7mQ etc.). For the values ​​shown for the gate voltages U GValues ​​shown in larger font correspond to MOSFETs in the linear range, and all values ​​shown in smaller font correspond to MOSFETs in the pinch-off range, for which the resistance is correspondingly increased (i.e., greater than 3.9 mΩ), as illustrated in Fig. 4b. All MOSFETs with a gate voltage shown in smaller font will, in reality, build up a counter voltage. If the sum of the counter voltages equals the driving voltage, equilibrium will be established, and the resulting current rise in the short circuit will be zero. The circuit thus actively and intrinsically limits the short-circuit current as a function of the applied driving voltage.

[0044] Fig. 5a shows values ​​without a Zener diode (or without a sacrificial MOSFET); in Fig. 5b, MOSFET 6 is designed as a sacrificial MOSFET. This is based on an RDSon of 3.9 mΩ and a gate signal of 15 V. It can be seen that at a current of 50 A, MOSFET 6 has a significantly lower gate voltage than MOSFET 7, but still operates in the linear region. At 200 A, it switches to the pinch-off region. According to Fig. 4a, the MOSFET used with a 6.1 V gate voltage cannot carry a 200 A drain current. It therefore builds up voltage, which reduces the gate voltage of all subsequent MOSFETs in the series circuit, and they also build up a counter voltage. MOSFETs 7 to MOSFET 11 therefore also switch to the active region. The achievable counter voltage at 200A is therefore the sum of all individual counter voltages of the MOSFETs, usually the limiting voltage of the overvoltage limiter installed in parallel.If the Zener diode is placed in the gate path of MOSFET 7, the countervoltage is reduced by the countervoltage of a single MOSFET. If it is placed in the gate path of MOSFET 5, the countervoltage is increased by the countervoltage of a single MOSFET. This opens up a multitude of possibilities for adjusting the required countervoltage and the value of the maximum short-circuit current.

Claims

Patent claims 1. Semiconductor switch with - a plurality of circuit modules connected in series, each formed with - a first transistor (M3) having a source terminal (source!), a control terminal (Gate3) and a drain terminal, wherein - the source terminal (source!) is a source terminal or an emitter terminal, the control terminal (gate!) is a gate terminal or a base terminal and the sink terminal is a drain terminal or a collector terminal, and - a second transistor (Q2) having a source terminal, a control terminal and a drain terminal, wherein - the source terminal is connected to the control terminal (gate!) of the first transistor (M3), - the drain terminal is connected to the source terminal of the first transistor (M3), and - with a driver (V2) which is connected via connections to the control terminals of the second transistors (Q1, Q2, Q3, Q4, Q5, Q6) of the individual circuit modules, wherein for a subset of the plurality of circuit modules a device for reducing the control terminal voltage of the first transistor is incorporated in the respective connection to the driver.

2. Semiconductor switch according to claim 1, characterized in that the device for reducing the control terminal voltage is a Zener diode.

3. Semiconductor switch according to one of claims 1 or 2, characterized in that the subset consists of one or two circuit modules.

4. Semiconductor switch according to one of the preceding claims, characterized in that the first transistors of the circuit modules are identical.

5. Semiconductor switch according to one of the preceding claims, characterized in that the circuit modules of the subset of the plurality of circuit modules are designed with a device for reducing the control terminal voltage of the first transistor for a simplified replacement of the module or the first transistor of the module compared to the other circuit modules.

6. Semiconductor switch according to one of the preceding claims 1 to 3, characterized in that the first transistors of the subset of circuit modules differ from the other first transistors with regard to their design. 7 . Semiconductor switch according to claim 6 , characterized in that - operating conditions for the semiconductor switch are specified, and - the first transistors of the subset are designed for a lifetime comparable to that of the other first transistors under the specified operating conditions. 8 . Semiconductor switch according to one of the preceding claims, characterized in that - it has an even number of circuit modules, - the first half of the circuit modules are connected in series one after the other with the same switching direction, and - the second half of the circuit modules are connected in series one after the other with the opposite switching direction, - two of the circuit modules are connected in series in opposite switching directions, and - for a circuit module of the first half of circuit modules and for a circuit module of the second half of circuit modules, a device for reducing the control terminal voltage of the first transistor is introduced in the respective connection to the driver, wherein the position of the two circuit modules with respect to the number of circuit modules between them and the circuit module with a corresponding switching direction, which connects to a circuit module with an opposite switching direction, corresponds to one another.

Images