Method for adaptively optimizing the activation times and deactivation times in actuators of power electronics
The adaptive control method optimizes switch-on and switch-off times in power electronics actuators, addressing inefficiencies and compatibility issues by directly setting optimal times without additional components, resulting in reduced losses and improved performance.
Patent Information
- Application Number
- PCT/EP2024/000030
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-19
- Filing Date
- 2024-05-14
- Publication Date
- 2024-11-28
AI Technical Summary
Existing power electronics actuators face inefficiencies in switch-on and switch-off times due to unavoidable short circuits and voltage interruptions, leading to increased power losses and electromagnetic compatibility issues, with current methods either requiring additional components or causing unnecessary delays.
An adaptive control method using an extremal system to optimize switch-on and switch-off times directly, eliminating the need for additional components and minimizing signal delays, allowing for zero-voltage or zero-current switching with reduced losses and improved electromagnetic compatibility.
This approach reduces switching losses and electromagnetic interference by setting optimal switching times based on real-time conditions, ensuring efficient operation and robustness across varying operating points without additional hardware, thus enhancing the performance of power electronic systems.
Smart Images

Figure EP2024000030_28112024_PF_FP_ABST
Abstract
Description
[0001] Method for adaptive optimization of switch-on and switch-off times in power electronics actuators
[0002] Tobias Fink Rotkäppchenweg 31 70567 Stuttgart Germany tobifink@mail.de
[0003] May 11, 2024
[0004] Description
[0005] The invention relates to methods and devices for the adaptive optimization of the switch-on and switch-off times in actuators of power electronics.
[0006] State of the art
[0007] In most applications, power electronics actuators are constructed from one or more cells designed in the so-called half-bridge topology. Such a half-bridge consists of two series-connected switching elements with a center tap. To reduce power loss in the half-bridges, both switching elements are typically designed as controllable power semiconductors, even if the power electronics topology would in principle allow the use of non-controlled switching elements, such as diodes.
[0008] When controlling the controllable switching elements in a voltage-controlled half-bridge, the two critical cases in practice are those in which one of the switches is on and the other is off, and both switches are now required to change their previous switching state as directly as possible. Synchronous switching of both control signals of the power switches, i.e. synchronous hard switching, is not normally permitted, as this would cause a short circuit in the half-bridge during the switching state change and thus lead to unacceptable or at least undesirable short-circuit currents. Therefore, in practice the switching state change occurs in such a way that the switch that is on is first switched to the off state, and only then is the other, previously off switch switched to the on state.This is achieved according to the state of the art by not directly converting the switching state request into a switching state command for the circuit breaker, but rather by delaying the switching state command during a switch-on operation by a switch-on dead time. This switch-on dead time can be parameterized in common control circuits, but is not changed during operation or is only changed in a controlled manner. Therefore, the switch-on dead time must be selected to be at least long enough to ensure that no short circuit occurs in the half-bridge in all operating states, taking into account the signal propagation times and component tolerances, over the entire operating period. This approach ensures that the switching elements can always be operated within their specified range.However, this approach does not achieve the lowest possible losses by selecting the dead time based on the operating point, nor does it optimize for the best possible electromagnetic compatibility. This approach is used, among other things, in inverters for controlling electric motors.
[0009] When controlling the controllable switching elements in a current-controlled half-bridge, two cases are also critical in practice: one in which one of the switches is on and the other is off, and both switches are now required to change their previous switching state as directly as possible. Synchronous switching of both control signals of the power switches, i.e. synchronous hard switching, is not normally permitted, as this can briefly interrupt the current flow in the half-bridge and, as a result, lead to unacceptable or at least undesirable voltage surges at the switching elements. Therefore, in practice the switching state change occurs in such a way that the switch that is off is first switched to the on state and only then is the other, previously switched switch switched to the off state.This is achieved according to the state of the art by not directly converting the switching state request into a switching state command for the circuit breaker, but rather by delaying the switching state command during a switch-off operation by a switch-off dead time. This switch-off dead time can be parameterized in common control circuits, but is not changed during operation or is only changed in a controlled manner. Therefore, the switch-off dead time must be selected at least long enough to ensure that no interruption in the current flow in the half-bridge occurs in all operating states, taking into account the signal propagation times and component tolerances, over the entire operating period.
[0010] If the additional losses inherent in the described approach or the poor electromagnetic compatibility are intolerable, known methods for reducing switching losses are used. A fundamental distinction must be made between different approaches.
[0011] In one of these approaches, the switching elements are relieved of load during the switch-on and / or switch-off processes by so-called load-relief circuits, thereby reducing switching losses in the respective switching phase. For this purpose, additional elements are connected in series and / or parallel to the power electronic switching element, which appropriately influence the current and voltage curves during the switch-on and / or switch-off processes. Examples of such load-relief methods are described in DE 26 39 589 C2, DE 2641 183 C2, and DE 2644715 C2. Disadvantages of this approach are the considerable additional costs, as well as the required installation space and the weight of the load-relief circuit.
[0012] A fundamentally different known approach involves inserting a resonant circuit into the main current path of the switching elements using a resonant capacitance and / or a resonant inductance. The resonant circuit modifies the current and voltage curves at the switching element in such a way that so-called zero-voltage switches (ZVS) or zero-current switches (ZCS) can be realized, in which switching losses are significantly reduced. However, the resonant circuits cause increased peak loads while the switching element is switched on and thus inevitably lead to oversizing of the switching element. Examples of such methods are described in DE 26 50673 A1 and DE 26 514 92 A1.
[0013] Another known method for generating zero-voltage switching elements in voltage-controlled half-bridges or full-bridge circuits is to logically combine the switching state command vector from the switching state request vector using a minimum switch-on dead time required for safety reasons with an additional enable signal that takes the voltage change at the switching elements into account. In this way, the switching elements are only switched on when the zero-voltage conditions have been reached. However, this does not bring about the optimal switching time, since the signal propagation times and the switching times of the switching elements inherently result in delayed switch-on. An example of such a method can be found in the publication by L. Cong and H. Lee, “A 2MHz 150-to-400V input isolated DC-DC bus converter with monolithic slope-sensing ZVS detection achieving 13ns turn-on delay and 1.6W power saving”, 2018 IEEE International Solid - State Circuits Conference - (ISSCC), San Francisco, CA, USA, 2018, pp. 382-384.
[0014] Disclosure of the invention
[0015] According to the invention, a method for the optimized control of actuators of power electronics is provided, in which the switch-on times and the switch-off times are adaptively adjusted by a control or an extremal system so that the power loss in the switching elements is as low as possible.
[0016] The advantage of the method according to the invention compared to the relief circuits and resonant circuits mentioned in the prior art is that no additional switching elements have to be inserted into the main current path.
[0017] An advantage of the method according to the invention compared to the controlled methods mentioned in the prior art for realizing a zero-voltage switching element is that, due to the principle, there is no unnecessary time delay of the switching-on and switching-off signals, which leads to unnecessary additional losses in the method according to the prior art.
[0018] These two main advantages of the method according to the invention can be achieved by setting the optimal switching times of the actuator via a dedicated control system, which is used specifically for this task and which is used in parallel to the actual control system for the desired output variable or, if necessary, several output variables of the overall system. The control system for setting the optimal switching times regulates the switching times of the switching elements of the actuator, for example, in such a way that the lowest switching power loss occurs during a switching process in the actuator and that the switching times are reset to the new optimal value in the event of a deviation from this state, as typically occurs, among other things, when the operating point of the control system changes or when a parameter of the controlled system changes, i.e.to the optimized switching time, which must then be set again and at which the lowest switching losses occur under the changed conditions.
[0019] A further aim for the optimal selection of the switching times of the switching elements of the actuator can, for example, be to switch at a time when conditions with regard to the electromagnetic compatibility of the actuator occur that are as easy to control as possible, ie, for example, that conducted interference emissions that are as easy to control as possible occur during the switching process. To this end, the switching time can be selected such that the absolute voltage change that is as small as possible and, in addition, the edge steepness for this voltage change at the switching element of the actuator is also as small as possible. This is achieved in particular when the switching element is switched at a time when no voltage drop across the switching element is achieved. The method according to the invention is suitable for setting the switching times such that a zero-voltage switching element is realized.The method according to the invention is also capable of setting the earliest possible time for the realization of the zero-voltage switching process.
[0020] The method according to the invention differs from a first category of prior art methods, in which so-called relief circuits and / or resonant circuits are used, in that power electronic switching elements can be used directly without additional effort, and this applies to both unipolar and bipolar switching elements. This means that, in particular, no additional components such as capacitors, chokes, or transformers are required.
[0021] The method according to the invention differs from a second category of prior art methods, in which the switching of the switching element of the power actuator is triggered causally and thus sequentially based on the detection of the presence or attainment of a zero-voltage situation, in that it is a closed-loop control rather than a closed-loop control, and in that, unlike the prior art, the method according to the invention is fundamentally capable of setting the optimal switching time. With conventional methods, this is inherently impossible due to signal propagation times that occur both when a zero-voltage situation is detected and when the switching command for the switching element of the actuator is generated, as well as due to dead times that occur in the switching element from the application of the switching signal until the actual implementation of the switching process.
[0022] The method according to the invention makes it possible for the first time to implement a control system for setting the optimal switching times for use in actuators with power electronic switching elements, without requiring additional load-relieving circuits and / or resonant circuits for the actuator. Different optimization criteria can be used to select the switching time.
[0023] In another, particularly advantageous embodiment of the method according to the invention, the optimal switching times of the actuator are set using an extremal system instead of a control system. The extremal system is again used specifically for this task and is used in parallel with the control system for the desired output variable or, if necessary, several output variables of the overall system. Such an extremal system is easier to implement in practice than a control system and is characterized by a very high level of robustness.
[0024] The method according to the invention can be used in power electronic actuators to implement zero-voltage switches or zero-current switches. The same advantages described for zero-voltage switches apply to the latter.
[0025] In a control system, it can certainly make sense to design some of the switching elements as zero-voltage switches and others as zero-current switches. In some control systems, it can also be sensible to implement the most favorable design depending on the operating point or the type of control. An example of this is a boost converter in its canonical design. Such a boost converter can be operated or controlled in various ways. Typically, the so-called "continuous conduction mode" is used, in which the current in the boost converter choke fluctuates around its mean value with a small alternating component. In another operating mode, the so-called "boundary conduction mode", the current through the boost converter choke is adjusted with very high alternating components so that the current reaches zero in each period.In such an operating mode, the switch connected to the positive pole of the output voltage can be designed as a zero-voltage switch and the switch connected to the negative pole of the output voltage can be designed as a zero-current switch.
[0026] In modern electrical energy conversion systems, the power electronics are often implemented as so-called multiphase converters. This has the advantage that the individual converter cells can be very small and that, with suitable, synchronized control of the individual converter cells, either better dynamics (faster response to setpoint changes) and / or lower AC components in the output voltages or output currents of the overall arrangement can be achieved. The method according to the invention can be applied without restrictions to arrangements in which power electronic actuators are implemented in parallel. This applies both to synchronized operation of these actuators and to non-synchronized, independent operation.In modern systems for electrical energy conversion, the switching elements in the actuators are not used as individual active switching elements in conjunction with another passive switching element, but rather at least in an active half-bridge topology or even in an active multilevel topology. The reason for this is that active switching elements have significantly lower forward power dissipation than passive switching elements such as diodes. As a result, actuators with significantly lower power dissipation can be realized with topologies that use active semiconductor switches. The method according to the invention is capable of optimizing the switching times in both half-bridge topologies and multilevel topologies.
[0027] They show:
[0028] Fig. 1 is a schematic block diagram of a power electronic system with a controlled system, a controller and the integration of the optimization method into the control system.
[0029] Fig. 2 shows the logical signal flow in power electronic actuators, with which the control signals for the power electronic switches are generated from the switching state command vector.
[0030] Fig. 3 Equivalent circuit diagrams of various power electronic switches with which zero-voltage or zero-current switches can be described.
[0031] Fig. 4 shows the schematic structure of an actuator in a half-bridge arrangement.
[0032] Fig. 5 shows the schematic structure of an actuator in a multilevel converter arrangement.
[0033] Fig. 6 shows the basic structure of an exemplary control unit in which the switch-on dead time and the switch-off dead time can be set independently of each other.
[0034] Fig. 7 shows the basic temporal course of the signals of the exemplary control unit.
[0035] Fig. 8 shows the temporal signal curve of a special design of the detection circuit for half-oscillation-relieved zero-voltage switching elements in a half-bridge arrangement. Fig. 9 shows three fundamentally different temporal voltage curves at a zero-voltage-relieved power electronic switching element for a relievable operating point with different selected switch-on dead times.
[0036] Fig. 10 shows a power electronic actuator consisting of half-oscillation-relieved zero-voltage switches in a half-bridge arrangement and a simplified equivalent circuit diagram for a large load inductance.
[0037] Fig. 11 shows an exemplary structure of a dead time setting device.
[0038] Fig. 12 shows an exemplary structure of a dead time adjustment system.
[0039] Fig. 13 shows the exemplary structure of an extremal system for dead time optimization. In a controlled power electronic system (15) shown in Fig. 1, the vector of controlled variables (13) is adjusted by a controller (8) to the setpoints predetermined by the setpoint vector (12). For this purpose, the controller receives the information (10) about the state variables from the controlled system (9) and, if necessary, additional information about the manipulated variables and other variables relevant to the description of the controlled system. The desired switching state vector (5) determined by the controller is not passed on directly to the actuator (1), but is converted by a control unit (3) into an optimized switching state command vector (6) such that the switching elements (2a), (2b) of the actuator (1) are operated in an optimized manner.For this purpose, information (11) is supplied to the control unit (3), with which the optimal switching on and off times of the switching elements (2a), (2b) of the actuator (1) are determined. The information (11) for the control unit (3) is determined in the acquisition unit (4) from the information (10) of the controlled system (9) and the control signals (19a), (19b) of the switching elements (2a), (2b). The actuator (1) provides the manipulated variable vector (7) for the controlled system (9). The optimization method (14) is implemented by the acquisition unit (4), the control unit (3) in conjunction with the actuator (1). To implement the method (14), in principle, no changes need to be made to the controller (8) or to the topology of the main current path of the power electronic system (15) compared to a conventional system. This is a significant advantage over the prior art.Compared to a conventional system, the power electronic system (15) can be operated with reduced switching losses in the actuator (1) and / or with reduced electromagnetic interference emissions when using the method (14). The method (14) is capable of compensating for the overall effect of both parameter fluctuations in the controlled system (9), such as those caused by temperature-dependent resistors or non-linear, current-dependent inductances, and parameter fluctuations in the manipulated variable vector (7), such as those caused by battery voltage fluctuations due to the battery's state of charge, as well as parameter fluctuations in the actuator (1) and in the control unit (3), which arise, for example, due to operating point-dependent switching times and aging.This ensures a high level of robustness in the optimization of the control system against changes due to the operating state as well as against non-ideal properties and tolerances of the components used over the entire service life of the power electronic system.
[0040] The power electronic switches (18a), (18b) shown in Fig. 2 are generally not directly controlled by the logic signals supplied by the switching state command vector (6) or its elements (6a), (6b). Instead, they are usually amplified and, if necessary or desired, galvanically isolated. Both amplification and galvanic isolation are represented here by the control elements (17a), (17b). The control signals (19a), (19b) amplified by the control elements (17a), (17b) can switch the power electronic switches (18a), (18b) on and off more quickly and thus with lower losses. This is no different from the prior art.
[0041] Due to their parasitic properties, the real behavior of the power electronic switches (18a), (18b) is always described by one of the equivalent circuits shown in Fig. 3 in (23) to (27). Therefore, the method for optimizing the switching on and off times can, in principle, be applied to all power electronic switches. The equivalent circuits each contain a capacitance and an inductance. In addition, all equivalent circuits—with the exception of the equivalent circuit of a diode (25)—contain a switching element that can be switched on and off. A distinction is made between variants with and without an antiparallel diode for the switching elements that can be switched on and off.
[0042] The equivalent circuit diagrams are divided into so-called zero-voltage switches (23), (24), (25) and so-called zero-current switches (26), (27).
[0043] In general, for the zero voltage switches (23), (24), (25) the voltage U SEacross the capacitance has a limited rise time during turn-off. This means that the turn-off processes of such switches are always passively relieved. When turning on, however, it must be ensured that the voltage U SE ≃ 0 V in order to benefit from the switching relief. The special feature of the half-oscillation-relieved zero-voltage switch (23) is that the voltage U SE can never become significantly less than zero, since the diode then conducts. This is not the case with the fully oscillation-loaded zero-voltage switch (24). If a switch is not actively operated or only a diode is used, it is a passive zero-voltage switch (25).
[0044] Analogously, for the zero current switches (26), (27) it is generally true that the rate of rise of the current I SE is limited by the series inductance during switching on. During the switching off process, however, this current I SE≃ 0 A, otherwise the one-way valve must absorb the current chopping voltage time area when switching off, which is necessary to reduce the current in the inductance. If a diode is connected in antiparallel to the one-way valve, then this is a fully oscillation-balanced zero-current switch (26). In this case, negative currents I SE permissible at the time of switch-off, as these can be carried by the diode. This is not the case with a half-oscillation-balanced zero-current switch (27). Here, the current must be zero at the time of switch-off to avoid a current chopping voltage across the one-way valve.
[0045] Fig. 4 shows a half-bridge arrangement consisting of two zero-amplitude-relieved switches of any type. Zero-voltage switches (23) to (25) and zero-current switches (26), (27) are available in half-wave (23), (27) and full-wave (24), (26) versions, both active (23), (24), (26), (27) and passive (25), and can be combined as desired. It is significant that there is an output for connecting a load. However, the possibility of feeding back power from the load connection to the voltage or current source remains. An example of a power electronic system that utilizes this feature is a boost converter.
[0046] Fig. 5 shows a so-called multilevel converter, which is constructed similarly to a half-bridge and, by design, can provide more than two different manipulated variable levels. Various zero-level switch types can also be combined as desired.
[0047] Fig. 6 shows the basic logical principle of the control unit (3) as well as an exemplary design of the control unit. In this exemplary design, the control unit (3) consists of two delay elements with variable delay and an AND connection of the delayed logic signals (20) and (21) for each element of the switching state command vector (6). Due to the symmetrical structure, the principle behind it is that the longer set dead time is always the switch-on dead time and the shorter the switch-off dead time. The dead time specification device (22) generates the switch-on and switch-off dead times from the information (11) provided by the detection unit (4). This can be done, for example, cyclically for each switching operation or averaged over several switching periods.
[0048] Fig. 7 shows how, in the exemplary embodiment, the corresponding switching state command vector element (6) is generated by logical operation from the time profile of a switching state vector element (5) with the internal logic signals (20) and (21). Based on the example shown, it is clear that, due to the logic operation, the longer dead time is the switch-on dead time and the shorter dead time is the switch-off dead time.
[0049] Fig. 8 shows an example of the typical signal curve of the control signals (19a), (19b) of the switching elements (2a), (2b) as well as the voltage curves U SE, a and U SE , b the switching elements and their time derivatives and for a half-bridge arrangement of half-oscillation-relieved zero-voltage switches (23).
[0050] The first switching operation involves a switch-off operation of the switching element (2b) and a switch-on operation of the switching element (2a). The subsequent switching operation involves a reverse change of the switching states.
[0051] The first switching operation occurs too late compared to the optimal switching time, since the diode antiparallel to the switchable and switchable switching element has already begun to conduct at time ttot, opt, a. This can be seen both from the fact that the voltage U SE, a has already dropped below 0V, and that the voltage change at the switch is approximately
[0052] The second switching process occurs too early compared to the optimal switching process, because the voltage U SE , b at the switching element (2b) still has a significantly positive value and the voltage change is still significantly smaller. Premature switching on causes a partially hard switching on process.
[0053] For the exemplary time course of the load current I Last The switch-on times for both switching elements (2a), (2b) could be optimized. The acquisition unit (4) determines the switch-on and switch-off times of the switching elements (2a), (2b) from the control signals (19a), (19b). In addition, the acquisition unit determines the information (11) required for the optimization process from the information (10) of the controlled system and the manipulated variable.
[0054] These can be for zero-voltage switching elements depending on the optimization method used, e.g. the voltage curve U SE and / or the temporal voltage change at the switching elements and the load current I Last be.
[0055] For zero-current elements, this can be, for example, the current curve I SEand / or the temporal current change as well as the voltage at the load output.
[0056] Depending on the design of the optimization procedure, for example, the temporal progressions can be determined or, in a particularly suitable design, the values at the switching times can be determined.
[0057] All effects that can be influenced by the switching on and off times and that can be recorded can be used as optimization criteria.
[0058] In Fig. 9 there are three examples of the time course of the voltage U SE of a half-oscillation-relieved zero-voltage element around the switch-on time.
[0059] The solid line (10a) is the ideal, desired line, where exactly when U is reached SE = 0V the switching element is switched on.
[0060] In the dashed curve (10b), the switch is turned on too early. This results in an unnecessarily high voltage gradient due to partial hard switching.
[0061] In the dashed-dotted curve (10c), the switch is turned on too late. As a result, the load current has already commutated into the diode before the switching process.
[0062] The two cases (10b) and (10c) lead to unnecessary additional losses and increase the electromagnetic emissions.
[0063] In Fig. 10, the left-hand part shows a half-bridge arrangement of half-oscillation-loaded zero-voltage switches. In addition to the switching elements, a simple embodiment for detecting the voltage change during a switching operation of the switching element is arranged in parallel via an RC element. The right-hand part shows a simplified equivalent circuit diagram of this arrangement, which assumes that the load inductance is much larger than the two inductances of the zero-voltage switching elements. These can then be neglected. With this simplification, the voltage U can be applied to the RC elements instead of the voltage U ZVS the switching elements each have the voltage U SE or at the resistance of the RC element the voltage change be recorded.
[0064] Fig. 11 shows an exemplary embodiment of a dead-time specification device (22). In general, this device can vary and optimize both the on- and off-dead times separately. The optimization is performed for both dead times by a dead-time optimization device (28a), (28b).
[0065] Fig. 12 shows an exemplary dead-time optimization device (28a), (28b). Using the information (11) generated by the acquisition unit (4), the switch-on or switch-off dead time can be optimized using a dead-time controller or a dead-time extreme system (31). The switching load-relief possibility verifier (29) checks, depending on the existing topology of the power electronic system, whether load relief is possible for the upcoming switching operation. For a half-bridge arrangement of half-oscillation-relieved zero-voltage switches, the criterion for whether optimization of the switching times with regard to minimizing switching losses is possible is the sign of the load current I LastIf this is positive, optimization can be performed for switching element (2b), but not for switching element (2a). The reverse applies for negative signs. If optimization of the switching times of the switching element in the upcoming switching operation is possible, the switching relief possibility verifier (29) issues an enable signal to the dead-time optimizer (31) and transfers this, together with the information (11), as optimization information (30). If optimization of the switching times of the switching element in the upcoming switching operation is not possible, the switching relief possibility verifier (29) does not issue an enable signal to the dead-time optimizer (31). In addition, the switching relief possibility verifier (29) issues a substitution command (33) to the dead-time substituter.If the dead-time optimizer (31) receives an enable signal in the optimization information (30) for optimizing the switch-on or switch-off dead time for the upcoming switching operation, the dead-time optimizer (31) executes its optimization algorithm. This can generally be a control algorithm or the algorithm of an extremal system.
[0066] At its output, the dead time optimizer (31) transfers the optimized dead time (32) for the upcoming switching operation to the dead time substituter (34). In this case, the dead time substituter outputs the optimized dead time (32) as an output signal at the output of the Dead time optimization device (28a), (28b).
[0067] Fig. 13 shows an example of a particularly suitable design of the dead-time optimizer with the implementation of an extremal system. A half-oscillation-relieved zero-voltage switch is used as the switching element. The algorithm of the dead-time optimizer (31) is intended to optimize the switch-on dead time of the switching element so that the switch-on losses are as low as possible. The solid curve (36) describes the switch-on losses of the switching element as a function of the switch-on dead time i. tot, ein - If the switch-on dead time is less than the optimal switch-on dead time t tot, opt , the switching element is partially switched on hard.
[0068] The resulting unnecessary switching losses are characterized by the energy still stored in the parallel capacitance of the switch at the time of switching. This results in the quadratic curve shown in the area t tot> ein < i tot, opt .
[0069] If the switch-on dead time is greater than the optimal switch-on dead time t tot, opt The load current has already completely commutated into the diode before the switching process. This causes forward losses in the diode. Assuming a constant load current, the power loss in the diode is constant. The unnecessary turn-on losses exhibit a linear progression over the dead time.
[0070] The dot-dash curve (35) shows the basic relationship between the switch-on dead time and the voltage U applied to the switching element at the time of switch-on. SE .
[0071] The dashed curve (37) shows the basic relationship between the switch-on dead time and the voltage change prevailing at the switching element at the switch-on time. The algorithm of the extremal system has the task of Adapt the switch-on dead time so that t tot , opt is set.
[0072] If the switching element is operated with a shorter dead time than t tot, opt operated, the voltage U SE > 0 and the time derivative is significantly negative.
[0073] If the extremal system detects that at a switch-on time with a set switch-on dead time, the voltage U SE > 0 and / or the time derivative is significantly negative, the algorithm increases the switch-on dead time for the next upcoming switch-on process.
[0074] If the extremal system detects that at a switch-on time the voltage U SE < OV and / or the time derivative, the algorithm increases the Switch-on dead time for the next upcoming switch-on process.
[0075] This procedure results in the extremal system moving directly toward the optimal operating point in the event of a suboptimally set switch-on dead time, i.e., during operation with unnecessary additional losses. After reaching the optimal switch-on dead time, it oscillates around it with a trial movement. The optimization system finds the optimal switch-on time for each operating point of the power electronic system (15). In the event of dynamic changes to the optimal switch-on dead time, for example, due to a change in the operating point, the extremal system automatically adjusts the switch-on dead time to the respective optimum. In this way, the extremal system robustly adapts the switch-on times.
[0076] In the dead time optimizer (31), control algorithms can also be used in a comparable manner to perform the dead time optimization. For this purpose, in the dead time optimizer (31), for example, a control method can be used to calculate the dead time from the difference between the actual switch-on dead time t tot, ein and the optimal switch-on dead time t tot , opt A control signal for adjusting the switch-on dead time can be generated. Generally known control algorithms can be used for this purpose.
[0077] The difference between the actual switch-on dead time t tot, ein and the optimal switch-on dead time t tot, opt can be determined as follows:
[0078] If the switch-on dead time t tot, ein is greater than the optimal switch-on dead time t tot, opt / tot, opt ~ / tot, on is determined as follows: The time period is recorded which starts from the moment at which the voltage U SE < 0V or the time derivative is reached, it takes time until the switch-on command (19a), (19b) for the switching element (2a), (2b) is given.
[0079] If the switch-on dead time is smaller than the optimal switch-on dead time t tot, opt , is determined as follows: At time t ein , in which the switch-on command (19a), (19b) for the switching element (2a), (2b), the voltage and its time derivative recorded. With good approximation, the following applies: This provides the control system with a signal for the control deviation for dead time optimization.
[0080] Examples of implementation
[0081] Specific embodiments of the present disclosure will be described below with reference to the accompanying drawings. In the embodiments, like parts among the embodiments, which are assigned like reference numerals, are omitted or simplified to avoid redundant description.
[0082] 1. Embodiment: Method for optimizing the control of an actuator in a half-bridge arrangement, in which the electronic power switches are designed as zero-voltage switches and the setting of the optimal switch-on time is carried out using an extremal system:
[0083] The method can optimize all switching instants of the switching elements at which the current would naturally commutate to this switching element without active switching. These switching instants are selected by the switching relief possibility verifier (29).
[0084] The half-bridge arrangement shown in Fig. 4 is implemented in the present embodiment by electronic power semiconductors, e.g., with bipolar switching elements (IGBTs, bipolar transistors) or unipolar switching elements (MOSFETs, HEMTs). The equivalent circuit diagrams (23), (24), (25) shown in Fig. 3 are used to describe the properties of these switching elements. The inductances and capacitances shown there describe the parasitic properties of real components. If this is advantageous for the application, these properties can be enhanced by adding additional components by increasing the resulting capacitance and / or inductance values. The advantage of the invention disclosed here, however, is that the parasitic elements of the switching elements are already sufficient for the function of the optimization method. Fig. 3 shows the quantities I relevant to the method. ZVS(current through the zero-voltage switching element) and U SE (Voltage across the zero-voltage-relieved switch or its anti-parallel diode).
[0085] The complete switching cycle of the half-bridge arrangement used according to Fig. 4 comprises the following switching operations: The description begins with the switching element 2a switched on and the switching element 2b switched off. In this state, the complete load current I indicated in Fig. 4 flows. Last via switching element 2a. If switching element 2a is switched off and I Lastpositive, the voltage across the switching element 2b to be switched on begins to drop (see Fig. 9). Depending on the choice of switch-on time, one of the curves 10a, 10b, or 10c results. The corresponding switching power loss can be seen in curve 36 in Fig. 13. The task of the method is to choose the switch-on time such that a voltage curve according to curve 10a results. If switching element 2a is switched off and I Last If the result is negative, the switching relief possibility verifier (29) does not issue a release to the dead-time optimizer (31). Additionally, a substitution command (33) is issued to the dead-time substitute, and a minimum dead time for switching on switching element 2b is specified as an alternative.
[0086] When switched on, the entire load current flows through switching element 2b. When switching element 2b is switched on, the system behaves analogously to that described for switching element 2a, with the dependence on the sign of the load current being reversed.
[0087] If, during a relievable switch-on process, one of the switching elements is switched on via the voltage detection of U SE or the recording of the temporal change of the voltage found that the switch-on time was too early (Fig. 9, curve 10b), U SE at the switching time greater than 0, or was negative, the extremal system increases the switch-on dead time for the next switch-on process
[0088] If, however, it is determined that the switch-on process was too late (Fig. 9, curve 10c), U SE at the switching time less than 0, or was approximately 0, the extremal system reduces the switch-on dead time f tot for the next switch-on process. Due to this behavior, a small oscillation of the switch-on dead time around the optimal switch-on time occurs in principle.
[0089] 2. Example: Method for optimising the control of an actuator in a half-bridge arrangement, in which the electronic power switches are designed as zero-voltage switches and the setting of the optimum switch-on time is carried out using a control system:
[0090] The method can optimize all switching instants of the switching elements at which the current would naturally commutate to this switching element without active switching. These switching instants are selected by the switching relief possibility verifier (29).
[0091] The half-bridge arrangement shown in Fig. 4 is implemented in the present embodiment by electronic power semiconductors, e.g., with bipolar switching elements (IGBTs, bipolar transistors) or unipolar switching elements (MOSFETs, HEMTs). The equivalent circuit diagrams (23), (24), (25) shown in Fig. 3 are used to describe the properties of these switching elements. The inductances and capacitances shown there describe the parasitic properties of real components. If this is advantageous for the application, these properties can be enhanced by adding additional components by increasing the resulting capacitance and / or inductance values. The advantage of the invention disclosed here, however, is that the parasitic elements of the switching elements are already sufficient for the function of the optimization method. Fig. 3 shows the quantities I relevant to the method. ZVS(current through the zero-voltage switching element) and U SE (Voltage across the zero-voltage-relieved switch or its anti-parallel diode).
[0092] The complete switching cycle of the half-bridge arrangement used according to Fig. 4 comprises the following switching operations: The description begins with the switching element 2a switched on and the switching element 2b switched off. In this state, the complete load current indicated in Fig. 4 flows Last via switching element 2a. If switching element 2a is switched off and I Lastpositive, the voltage across the switching element 2b to be switched on begins to drop (see Fig. 9). Depending on the choice of switch-on time, one of the curves 10a, 10b, or 10c results. The corresponding switching power loss can be seen in curve 36 in Fig. 13. The task of the method is to choose the switch-on time such that a voltage curve according to curve 10a results. If switching element 2a is switched off and I L ast is negative, the switching relief possibility verifier (29) does not issue a release to the dead-time optimizer (31). In addition, a substitution command (33) is issued to the dead-time substitute, and a minimum dead time for switching on switching element 2b is specified as an alternative.
[0093] When switched on, the entire load current flows through switching element 2b. When switching element 2b is switched on, the system behaves analogously to that described for switching element 2a, with the dependence on the sign of the load current being reversed.
[0094] If, during a relievable switch-on process, one of the switching elements is switched on via the voltage detection of U SE or the recording of the temporal change of the voltage, it was found that the switch-on time was too early (Fig. 9, curve 10b), U SE was greater than 0 or negative at the switching time, the control system increases the Switch-on dead time t tot,ein based on its control algorithm. This requires the deviation of the switch-on time t tot,ein from the ideal switch-on time i tot,opt , which results from the voltage across the switching element at the time of switching on and the temporal change of the voltage at the time of switch-on can be approximately calculated. If, however, it is determined that the switch-on process was too late (Fig. 9, curve 10c), U SE at the switching time less than 0, or was approximately 0, the control system reduces the switch-on dead time based on its control algorithm. This requires the deviation of the switch-on time from the ideal switching time which is preferably is to be determined by measuring the time period after reaching zero voltage until the switch-on time.
Claims
Patent claims 1. Method for reducing the switching energy and reducing the electromagnetic emissions of an actuator of a power electronics system, wherein the switch-on times and / or the switch-off times of the switching elements of this actuator are adjusted on the basis of an optimization criterion by a control system or an extremal system.
2. Method according to claim 1, wherein the optimization is carried out in principle independently of the control system for controlling the output variable and independently of the topology of the main current path or the main current paths of the power electronics.
3. Method according to claim 1 or claim 2, wherein the switching elements of the power electronic actuator are designed as zero-voltage switches and / or as zero-current switches.
4. Method according to one of claims 1 to 3, wherein a plurality of actuators, detection units and control units are designed in parallel and these actuators are controlled in a synchronized manner.
5. Method according to one of claims 1 to 3, wherein a plurality of actuators, detection units and control units are designed in parallel and these actuators are controlled individually, ie not synchronized.
6. Method according to one of claims 1 to 5, wherein the switching elements are arranged in a half-bridge topology.
7. Method according to one of claims 1 to 5, wherein the switching elements are arranged in a multilevel topology.
8. Method according to one of claims 1 to 7, wherein the detection unit for zero voltage switching elements - the current I ZVS and the voltage U SE or their temporal derivative or - the current I ZVS and the voltage ( U ZVSor their temporal derivative or - depending on the respective circuit topology, one or more sufficiently accurate, easier to record substitute variables are recorded or determined, sampled with the edge change of the control signal during the switching-on process of the switching elements and the control unit determines from this whether and / or how much the switching element was switched on too early or too late in accordance with an optimization criterion and the recording unit for zero-current switching elements - the current I ZCS or its time derivative and the voltage or - the current I SE or its time derivative and the voltage or - depending on the respective circuit topology, one or more sufficiently accurate, easier to record substitute variables are recorded or determined, sampled with the edge change of the control signal during the switch-off process of the switching elements and the control unit determines from this whether and / or how much the switching element was switched off too early or too late in accordance with an optimization criterion.
9. Method according to one of claims 1 to 8, wherein the control unit specifies the optimized switching on and off times of the switching elements individually with a control system or an extremal system by determining the switching dead time t tot, ein and the switch-off dead time i tot, aus determines by which the switching state commands for the switching elements are delayed compared to the switching state desired vector determined by the controller.
10. The method according to claim 9, wherein for the switching elements in each case either only the switch-on dead time t tot, einor only the switch-off dead time t tot, aus is optimized and the non-optimized dead time is specified in a controlled manner.
11. The method according to any one of claims 1 to 10, wherein at least one of the switching elements is operated with the optimization method and at least one of the switching elements is operated without the optimization method.
12. The method according to one of claims 1 to 11, wherein at least one of the switching elements is operated with a controlled, predetermined switch-on and switch-off dead time at operating points where optimization of the switching dead times is not possible due to the circuit topology.
13. The method according to one of claims 1 to 12, wherein the optimization criterion consists in operating the switching elements in such a way that the switching operation fulfills the zero-voltage and / or zero-current criterion at the earliest possible time.
Citation Information
Patent Citations
Arrangement without principle-related losses to relieve electrical or electronic one-way switches from their power loss stress when switching off
DE2639589C2
Device without inherent losses to relieve electrical or electronic one-way switches of their power loss stress when switching off
DE2641183C2
Device for relieving electrical or electronic one-way switches from high power loss stress during switch-on - in which a non-zero current previously flowing in another circuit path changes over to the relevant one-way switch - and from excessive blocking voltage stress at the end of switch-off
DE2644715C2
Low loss electric or electronic switch circuit - has electric load reduction system between four points of switch operating circuit, consisting of chokes, diodes and capacitors
DE2650673A1
One-way electronic switch - has inductance to limit rate of current rise at switch-on and reduce power loss
DE2651492A1
