Method and apparatus for detecting a single defective semiconductor switch based on the excitation current

DE102023205557B4Active Publication Date: 2025-05-08AUDI AG +1
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
DE102023205557
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Filing Date
2023-06-14
Publication Date
2025-05-08
Estimated Expiration
2043-06-14

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

Method for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches (210, 340) of a semi-controlled H-bridge circuit (100) for setting an excitation current of a separately excited synchronous machine, wherein the semi-controlled H-bridge circuit (100) a bridge branch (200) with one of the semiconductor switches T1 (210) and a diode D3 (230) connected in series in reverse bias and a further bridge branch (300) comprising a further diode D2 (320) in reverse bias in series with another of the semiconductor switches T4 (340) as well as a load branch (400) comprising an excitation winding (460) of the separately excited synchronous machine, wherein the load branch (400) connects a connection point 250) from one of the semiconductor switches T1 (210) and the diode D3 (230) in the one bridge branch (200) and with a further connection point (350) between the further diode D2 (320) and the further semiconductor switch T4 (340) in the further bridge branch (300), and wherein a current measuring device (480) for measuring the current excitation current is arranged in the load branch (400); wherein the semi-controlled H-bridge circuit (100) is coupled to a control device (1200) comprising an excitation current control (1220) which generates two pulse-width modulated control signals (611, 621) of the same frequency, phase-shifted by half a period, for controlling one of the semiconductor switches (210) and the other of the semiconductor switches (340), such that the excitation current flowing through the load branch (400) can be controlled via the pulse-width modulation of the control signals (611, 621) according to a requirement, wherein a duty cycle in normal operation is limited to a range of 50% to 100%, wherein the duty cycle specifies a ratio of the temporal pulse width of the signal level that puts the controlled semiconductor switch into a conducting state to the period of the control signal, and wherein the load branch (400) comprises a current measuring device (480) to measure the to measure excitation current, which is used as a feedback signal by the excitation current control (1220),, characterized by the fact that it is evaluated whether a change in the excitation current correlates its sign temporally with each of the nominal switching times (T1). ein , T4 aus , T4 ein , T1 aus ) one of the semiconductor switches (210, 340) changes, and if this is not the case, an error signal will be output indicating a defect in one of the semiconductor switches (210, 340), with a nominal switching time (T1) ein , T4 aus , T4 ein , T1 aus ) is a switching point at which one of the two semiconductor switches (210, 340) changes its switching state in an intact semi-controlled H-bridge circuit (100).
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The invention relates to a method for detecting a single permanently conductive, defective semiconductor switch of a semi-controlled H-bridge for setting an excitation current of a separately excited synchronous machine. Furthermore, the invention relates to a control device with a detection device for detecting a single permanently conductive, defective semiconductor switch of such a semi-controlled H-bridge for setting an excitation current.

[0002] A semi-controlled H-bridge circuit, also called a semi-controlled H-bridge, is often used as a control circuit for generating an excitation current of a separately excited synchronous machine. Fig. 1 schematically illustrates such a circuit. The circuit 100 comprises a bridge branch 200 and another bridge branch 300. One bridge branch 200 comprises a semiconductor switch 210, also referred to by the abbreviation T1. Connected in series with the semiconductor switch 210 is a reverse-biased diode 230, also referred to as D3. The further bridge branch 300 comprises a further diode 320 arranged in the reverse direction, also referred to as D2. Arranged in series with the further diode in the further bridge branch 300 is another semiconductor switch 340, also referred to as T4.

[0003] A load branch 400 is formed between a connection point 250 of one bridge branch 200 between one semiconductor switch 210 and one diode 230 and a further connection point 350 of the further bridge branch 300 between the further diode 320 and the further semiconductor switch 340. The load branch 400 comprises an excitation winding 460, for which an ohmic winding resistance 465 and a winding inductance 467 are shown as an equivalent circuit. Furthermore, a current measuring device 480 is formed in series with the excitation winding 460 in the load branch 400. The current measuring device 480 is designed to measure the excitation current I exc to measure. Across the excitation winding 460, an excitation voltage U exc The bridge branch 200 and the further bridge branch 300 are connected to one end 201, 301 with a positive pole 51 of a supply voltage U DCand connected to the opposite end 202, 302 with a negative terminal 52 of the supply voltage. A DC link capacitor 500 is arranged in parallel with the bridge arm 200. The supply voltage U is applied to this capacitor. DC In addition, a further current measuring device 180 is formed, which measures the total current I flowing through the half-controlled bridge circuit 100 DC A voltage measuring device 190 is arranged in parallel with the intermediate circuit capacitor 500 and measures the supply voltage U DC measures.

[0004] To obtain an excitation current I excIn accordance with a request, for example, from a driver in a vehicle powered by a separately excited synchronous machine, the semiconductor switches T1, T4; 210, 340 are each controlled with a pulse-width modulated control signal. When reference is made below to the semiconductor switches, this refers to the one semiconductor switch 210, T1 and the further semiconductor switch 340, T4. When reference is made to one of the semiconductor switches, this can be either the one semiconductor switch 210, T1 or the further semiconductor switch 340, T4.

[0005] The control signals 611, 621 are square waves with the same frequency, but shifted by half a period T PWM are offset from each other (see Fig. 2). A current flows through the semi-controlled semiconductor bridge when both one of the semiconductor switches 210, T1 and, at the same time, the other of the semiconductor switches 340, T4, are conductive.

[0006] The control signals 611, 621 each have the same duty cycle, which is 50% or greater. The duty cycle is the time ratio of the control signal at a level 612, 622, which switches the corresponding semiconductor switch 210, 340 controlled thereby into a conducting state, to the total duration of the period T PWM of the signal. The larger the pulse width 615, 625 of the control signals, the longer the periods during which current can be fed into the excitation winding 460.

[0007] In Fig. 2, five graphs 610-650 are shown schematically one below the other, each plotted against time 601 in units of period T PWMof the pulse-width-modulated signal. The upper two graphs 610, 620 indicate the pulse-width-modulated control signal 611 and 621 for controlling one of the semiconductors 210, T1 and another of the semiconductors 340, T4, respectively. A level 612, 622 is plotted against time 601. Since the semiconductor switch 210, T1 is closed and conductive when the control signal 611 has level 1, and closed and blocking when the level is 0, the graph 610 and, analogously, the graph 620 can also be understood as state graphs, and the control signals 611, 621 as state curves of the corresponding semiconductor switches T1, 210 and T4, 340.

[0008] The excitation voltage 631 is plotted in graph 630. In time ranges 635 in which the two control signals have level 1, a positive voltage, approximately U HVThe corresponding voltage is applied. Graph 640 shows the changing excitation current 641. This rises steeply in regions 646, 648 when both semiconductor switches 210, 340, T1, T4 are conductive, and falls somewhat more slowly in regions 647, 649, where only one of the semiconductor switches 210, 340 is closed. In an intact half-controlled H-bridge circuit, there are alternating current changes in the excitation current 641.

[0009] Finally, in the last graph 650, the total current I DC 651 flowing through the half-controlled half-bridge is plotted against time 601.

[0010] In the variant shown according to the state of the art, the two control signals 611, 621 are offset by half the period T PWM / 2. This offers the advantage of keeping the current fed into and out of the intermediate circuit capacitor 500 low. Furthermore, the frequency of the current ripple of the excitation current 641 is twice the switching frequency of the individual semiconductor switches when the semiconductor switches are functioning, which is advantageous for the ripple of the excitation current. Furthermore, the power loss is distributed evenly between the semiconductor switches and diodes. In addition to the center-centered pulse width modulation shown here, edge-related pulse width modulation can also be used, in which the control signals 611, 621 are also offset by half a pulse width modulation period.

[0011] When controlling the semiconductor switches 210, T1; 340, T4, it is essential that the voltage at the excitation winding normally changes each time one of the semiconductor switches 210, T1 or 340, T4 is switched on or off and is equal to the positive supply voltage, for example a traction network voltage, or equal to the negative sum of the forward voltages of the diode and semiconductor (approximately -2 V to -3 V). The current then flows through a series of exponential functions and, with a positive voltage at the excitation winding, rises relatively steeply, limited by the inductance 467, and falls relatively slowly in the short-circuit phase, i.e., when the semiconductor switch is closed and the semiconductor switch is open.

[0012] The previous description assumes that both semiconductor switches 210, T1; 340, T4 switch correctly according to the control signals 611, 621. Since the excitation winding 460 is usually designed as the rotor of a synchronous machine and often has a large inductance 467, it is necessary, in the event of a fault, to short-circuit the excitation winding 460 via one of the two semiconductor switches, either one of the semiconductor switches 210, T1 or the other of the semiconductor switches 340, T4. Due to the current direction, the respective diode 320 or 330 of the other bridge branch 200, 300 conducts. If the short circuit does not occur, the energy from the large inductance of the excitation winding of the rotor would flow into the intermediate circuit of the inverter. If, in this case, feedback occurs relatively simultaneously, including shortly before or.Shortly after the short circuit, the battery contactors are opened and the inverter is no longer connected to the traction network battery. The capacity of conventional DC link capacitors 500 is generally insufficient to prevent an overvoltage that could destroy a main inverter and / or the H-bridge circuit described here, as well as other HV components. Therefore, it is advisable to determine in advance if one of the semiconductor switches is defective.

[0013] In the event of a fault that one of the two semiconductor switches remains permanently open despite being triggered, the excitation current I exc within a short period of time, usually a few 100 ms, to 0. If both switches are permanently switched on, the supply voltage U DC as voltage of the excitation winding U excpermanently on, causing the current through the excitation winding to rise sharply. Overcurrent detection can reliably detect this error and shut off the supply voltage. A more difficult and critical situation is when one of the two semiconductor switches 210, 340 no longer switches and remains in a permanently conductive state, thus being defective, while the other semiconductor switch is still switching normally. The cause of such a defect can lie in the control circuit or even be a defect in one of the semiconductor switches 210, 340 itself.

[0014] DE 10 2014 102 869 A1 describes that, in a rotating electrical machine, a first switching element is provided for stopping the supply of an exciting current that excites a field winding. A detector detects, based on a voltage at an exciting current control terminal, how the field winding is supplied with the exciting current via the exciting current control terminal. An abnormality determiner determines, based on a detected result of the detector and a voltage induced across at least one single-phase stator winding, whether there is an abnormality in at least one second switching element connected to an output terminal of the at least one single-phase stator winding. A second drive controller controls the driving of the first switching element based on a result of the determination of the abnormality determiner.

[0015] EP 3 208 922 A1 relates to the operation of modular power converters, which is intended to be made safer. For this purpose, a method is proposed for testing a modular power converter comprising several modules connected in series, each of which has a half-bridge with two switches connected in series and a buffer capacitor parallel to the half-bridge. A voltmeter is arranged in the respective module directly parallel to the buffer capacitor. When one of the two switches is closed, a voltage waveform is recorded by the voltmeter. This voltage waveform is differentiated over time, and a test signal is generated from it. This allows testing to be performed at the module level.

[0016] DE 10 2016 203 355 A1 describes an electrical device, in particular an electrical machine such as a robot. The electrical device comprises an electrical load, an electronic switch which comprises an electronic switching element and a driver device that controls the electronic switching element, and a clocked power supply which has a power section with the electronic switching element and which is configured to generate an electrical supply voltage or an electrical supply current for the electrical load from an electrical voltage due to an alternate switching on and off of the one electronic switching element. The power section has a current path through which an electrical current associated with the electronic switching element of the electronic switch flows during operation of the power supply. The electrical device further comprises a pulse transformer.

[0017] DE 10 2020 112 704 A1 describes an electrical circuit arrangement for supplying current to a rotor of an electrical machine having a rotor winding, comprising a control unit, an input with a first potential connection and a second potential connection, and an output with a first connection and a second connection for connecting the electrical circuit arrangement to a start and an end of the rotor winding, wherein the first connection is connected to the first potential connection via a first switching element and to the second potential connection via a first diode, and the second connection is connected to the first potential connection via a second diode and to the second potential connection via a second switching element, wherein the first potential connection is connectable to a center tap of the rotor winding via a third switching element and the second potential connection is connectable to a center tap of the rotor winding via a fourth switching element,wherein the control unit is connected to the switching elements and the control unit is designed to switch the third and fourth switching elements permanently open during normal operation and, upon the occurrence of a defect in the first switching element or the second switching element, to switch the third switching element and / or the fourth switching element at least temporarily closed depending on status information describing an operating state of the electrical machine to be set via the electrical circuit arrangement.

[0018] DE 10 2014 018 665 A1 describes a converter with a control section featuring motion control and a power section including a power semiconductor switch. The power section has a safety monitoring system that provides a safety function.

[0019] The invention is based on the object of specifying a method and a device with which a detection of an individual defective semiconductor switch in a semi-controlled H-bridge for a separately excited synchronous machine in which one of the semiconductor switches is permanently conductively defective is detected in a simpler manner.

[0020] The object is achieved according to the invention by a method having the features of patent claim 1 and by a control device having the features of patent claim 9, as well as by a computer program product according to claim 11. Advantageous embodiments emerge from the subclaims.

[0021] The invention is based on the idea that the current flowing in the excitation winding of the separately excited synchronous machine is measured several times during a control cycle, the duration of which corresponds to the period of the control signals for the semiconductor switches of the semi-controlled H-bridge circuit. Due to the use of phase-shifted control signals, there are four switching operations in one control cycle: a switching-on of the first of the semiconductor switches T1 ein , switching off the first of the semiconductor switches T1 aus , switching on the semiconductor switch T4 ein , switching off the semiconductor switch T4 ausSince the semiconductor switches of the half-controlled H-bridge are equivalent and no hierarchical order exists, they will be referred to below as one of the semiconductor switches and the other of the semiconductor switches. Switching on is considered to be the induction of a conductive state. Switching off is considered to be the induction of an interrupted or open and thus non-conductive state. The current in the excitation winding increases when both semiconductor switches T1, T4 are in the conductive state. If one of the two semiconductor switches T1, T4 is open and the other of the two semiconductor switches T1, T4 is closed, the current in the excitation coil decreases. A control cycle thus has four switching times and correlated switching processes that occur in chronological sequence, for example, in the following order and with the following events: Switching on of one semiconductor switch T1 ein, switching off the semiconductor switch T4 aus , switching on the semiconductor switch T4 ein , switching off one of the semiconductor switches T1 aus The H-bridge circuit states of the intermediate time periods can be associated with the switching states of the semiconductor switches or the switching processes that bring about the switching states and result in: T1 aus up to T1 ein ; T1 ein up to T4 aus ; T4 aus up to T4 ein ; and T4 ein up to T1 aus For the sake of completeness, it should be noted that in a center-centered pulse width modulation (PWM), the first (last) time period T4 ein up to T1 auscontinues beyond the control cycle boundaries from the previous / into the next control cycle. Regardless of this fact, with edge-related or center-centered pulse width modulation of the control signals, there are always four consecutive time periods with alternating current changes in the excitation current. Current increase and current decrease alternate in chronological order if both semiconductor switches T1 and T4 switch correctly. The existence of this fact is checked by measuring the excitation current several times and monitoring the changes in sign of the current changes. Since switching operations do not take place if one of the semiconductor switches T1, T4 or a control circuit is defective, the terms nominal switching times and nominal switching operations are sometimes used to indicate that the information refers to the intended switching time or intended switching operation.In case of doubt, the nominal switching time and the nominal switching process are always meant, unless the context indicates otherwise. If the control and the H-bridge circuit are intact, every nominal switching time correlates with a real switching time, and every nominal switching process also correlates with a real switching process.

[0022] In particular, a method is provided for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches of a semi-controlled H-bridge for setting an excitation current of a separately excited synchronous machine, wherein the semi-controlled H-bridge circuit comprises a bridge branch with one of the semiconductor switches T1 and a diode D3 connected in series in the reverse direction and a further bridge branch comprising a further diode D2 in the reverse direction in series with another of the semiconductor switches T4 and a load branch comprising an excitation winding of the separately excited synchronous machine, wherein the load branch connects a connection point of one of the semiconductor switches T1 and the diode D3 in one bridge branch and with a further connection point between the further diode D2 and the further semiconductor switch T4 in the further bridge branch,and wherein a current measuring device for measuring the current excitation current is arranged in the load branch; wherein the semi-controlled H-bridge circuit is coupled to a control device comprising an excitation current control, which generates two equal-frequency, pulse-width-modulated control signals, phase-shifted by half a period, for controlling one of the semiconductor switches and the other of the semiconductor switches, so that the excitation current flowing through the load branch can be controlled via the pulse-width modulation of the control signals according to a request or a request signal (which alternatively serve as a reference variable), wherein a duty cycle during normal operation is limited to a range of 50% to 100%, wherein the duty cycle indicates a ratio of the temporal pulse width of the signal level that places the controlled semiconductor switch into a conducting state to the period of the control signal, and wherein the load branch comprises a current measuring device,by the excitation current I, exc to measure, which is used by the excitation current control as a feedback signal (controlled variable), wherein it is evaluated whether a current change in the excitation current changes its sign in a time-correlated manner with each of the nominal switching times of one of the semiconductor switches, and if this is not the case, an error signal is output which indicates a defect in one of the semiconductor switches, wherein a nominal switching time is a switching time at which one of the two semiconductor switches changes its switching state in an intact half-controlled H-bridge circuit.

[0023] In a preferred embodiment, an excitation current measurement value is recorded in time correlated with the, preferably at the, nominal switching times of each control cycle, and the current change of the excitation current is determined as a difference between the excitation current measurement values ​​recorded in the control cycles immediately following one another, correlated with the nominal control cycle switching times of one of the semiconductor switches, and evaluated to determine whether the resulting differences in the excitation current measurement values ​​are alternately positive and negative, and if this is not the case, the error signal is output.

[0024] In one embodiment, it is thus evaluated whether the differences thus determined alternately change their sign and, if this is not the case, an error signal is output.

[0025] For example, a difference is determined between the excitation current measured at the immediately preceding switching instant of one of the semiconductor switches and the excitation current measured at the current nominal switching instant of one of the semiconductor switches. The current changes determined in this chronological sequence continuously change their sign in an intact circuit.

[0026] In practice, however, it is difficult to always determine the current change precisely. Depending in particular on the traction network voltage, which depends on the battery charge level, the differential inductance of the excitation winding, the frequency of the control signals for the semiconductor switches, the set duty cycle of the control signals, and the signal quality in a measuring chain consisting of a measuring device for current detection and an analog-to-digital converter (ADC), the current change can lie within the resolution of the analog-to-digital converter. Since the switching times, for example, are sometimes close together in time, the current changes are sometimes only small in magnitude, i.e. the excitation current measured values ​​differ only slightly at successive points in time.If the difference is within the resolution range of an analog-to-digital converter (ADC), the excitation current differences determined in chronological order may not always change sign, even though the half-controlled H-bridge circuit is functioning correctly. The evaluation can then be improved by averaging. One embodiment therefore provides for averaging. For example, averaging over 1024 measured values ​​improves the resolution of the analog-to-digital converter by five bits.

[0027] In one variant, it is possible for the differences determined between consecutive nominal switching times, which correspond to each other in the control cycles executed one after the other, to be averaged over several control cycles. This means that a current change occurring between two identical nominal switching operations in successive switching cycles is derived by averaging the determined differences determined from the current measured values ​​correlated with the switching operations that limit the time range of the corresponding excitation current change in the switching cycles. Thus, in one variant, the excitation current measured value differences belonging to the same nominal switching operations are averaged.

[0028] Another variant provides that the excitation current measured values, which are recorded correlated with, in particular with, nominal switching times corresponding to each other in different control cycles, are averaged and the differences are determined on the basis of these averaged excitation current measured values.

[0029] Preferably, in both variants, a moving average is carried out over a predetermined number of control cycles.

[0030] Alternatively to one of the averaging methods, or in addition, an error signal can only be generated when a minimum number of sign changes between consecutive current changes have been missed in a number of consecutive control cycles, i.e., fewer sign changes have occurred than expected. This is equivalent to the number of sign changes between detected current changes at consecutive time intervals within a time period or a specified number of control cycles falling below a threshold value.

[0031] In a further development, the semiconductor switch is identified as defective if, after at least one nominal switching time in the control cycle, no change in the sign of the current change occurs. This means that, after a nominal switching operation of one of the semiconductor switches, no change in the sign of the current change occurs.

[0032] Preferably, an excitation current measurement is recorded in correlation with each of the nominal switching operations, preferably simultaneously or alternatively shortly after or alternatively shortly before. Differences in the excitation current are determined between the recorded excitation current measurement values ​​of switching operations that occur consecutively in the switching cycles, and it is evaluated whether the successive differences alternate in sign. The differences are assigned or can be assigned to the switching operations via the excitation current measurement values. Without loss of generality, for example, a difference is always assigned to the switching operation whose correlated excitation current measurement value (or its mean value) represents the minuend of the difference calculation. The difference results from the minuend minus the subtrahend.Without loss of generality, the difference is always calculated in such a way that the most recently measured excitation current value forms the minuend and the previously measured excitation current value forms the subtrahend. With this convention, an excitation current difference is positive if the excitation current increases between the two measurement times and switching operations.

[0033] To improve the resolution of the analog-to-digital converter that converts the excitation current measurements into digital signals, the excitation current measurements associated with the same nominal switching operations are averaged, or alternatively, the excitation current measurement differences associated with the same nominal switching operations are averaged. This averaging is preferably performed on a rolling basis. If the sign change does not occur or no longer occurs, an error signal is output.

[0034] If the excitation current measured value differences are each assigned to the switching operation whose excitation current measured value forms the minuend, and if there is no change in sign between two consecutively determined excitation current measured value differences, then the semiconductor switch to which the current change occurring earlier or the excitation current measured value difference indicating this current change is assigned is defective. If, for example, one of the semiconductor switches T1 is permanently conductively defective and the other semiconductor switch T4 is intact, then between the current change that occurred between the nominal switching times or nominal switching operations of switching on the other semiconductor switch T4 ein and switching off one of the semiconductor switches T1 aus and the one with ΔI exc_T1_aus and is given by ΔI exc_T1_aus = I exc_T1_aus - I exc_T4_ein , and the subsequent current change, which is expressed as ΔI exc_T1_ein = I exc_T1_ein- I exc_T1_aus , no change of sign takes place. The subsequent current change ΔI exc_T4_aus = I exc_T4_aus - I exc_T1_ein no sign change occurs. Thus, the defect can be assigned to the semiconductor switch to which the excitation current difference is assigned, for which no sign change subsequently occurs. According to this consideration, the excitation current difference is assigned to the semiconductor switch whose switching operation occurs simultaneously or closely correlated in time with the acquisition of the excitation current measured value, which forms the minuend of the excitation current difference.

[0035] If the excitation current changes or the excitation current measured value differences approximating them are assigned to the semiconductor switches whose switching process is assigned to the excitation current measured value that represents the subtrahend, then the semiconductor switch is defective to which the excitation current measured value difference is assigned that does not have a change of sign compared to the temporally preceding excitation current measured value difference.

[0036] An alternative way to detect the absence of sign changes is to determine the frequency or change frequency of the excitation current signal. If the semiconductor switches are intact, the change frequency is twice the switching frequency of the control signals. Except in the case of a symmetrical triangular current waveform, the higher harmonics of the fundamental frequency at which the excitation current signal changes also occur. In an intact state, the fundamental frequency is twice the switching frequency of the control signals. If the switching frequency abruptly appears as a change frequency in the spectrum, this indicates a defect in one of the semiconductor switches, in which the defective semiconductor switch is permanently conductive. It should be noted that, even in the case of a defect, components of the excitation current signal usually have twice the switching frequency.One embodiment therefore provides that, in order to determine whether the current change of the excitation current changes its sign in a time-correlated manner with each of the nominal switching times of one of the semiconductor switches (210, 340), a frequency of the excitation current change is monitored and checked, and the error signal is output when the excitation current change frequency corresponds to the frequency of the control signals or this occurs in the frequency spectrum of the excitation current.

[0037] Here too, the averaged excitation current measured values, preferably the moving average excitation current measured values, can be used to determine the excitation current change frequency.

[0038] Alternatively, however, the excitation current can also be continuously recorded in simultaneous steps with a sampling frequency that is at least four times, preferably at least 10 times, more preferably at least 50 times the switching frequency of the control signals.

[0039] To determine the change frequency of the excitation current signal, a transformation analysis, for example, in the form of a fast Fourier transformation, can be performed. From the result, the corresponding frequency components can be easily determined and monitored. This analysis is very robust. High-frequency and low-frequency interference in the excitation current signal hardly affects the evaluation result.

[0040] The frequency of change can also be monitored in other ways, for example with bandpass filters used together with threshold detectors for the filtered signal.

[0041] The excitation current measured values ​​recorded in chronological order or the averaged excitation current measured values ​​determined in chronological sequence result in an excitation current signal.

[0042] The defective semiconductor switch can be identified by evaluating the phase position of the excitation current signal relative to the control signals of the semiconductor switches. In the event of a fault, in which the frequency of the excitation current change equals the frequency of the control signals, the excitation current signal is in phase with the control signal of the non-defective semiconductor switch. There is a phase offset relative to the control signal of the defective semiconductor switch. One embodiment therefore compares the phase position of the excitation current signal relative to the phase position of the control signals of the semiconductor switches, whereby the semiconductor switch whose control signal exhibits a phase offset of the excitation current signal is identified as defective.

[0043] The invention can be implemented in program code, which is executed on a processor and, in conjunction with the semi-controlled H-bridge circuit, its excitation current measuring device and the control device for driving it, and the semi-controlled H-bridge circuit, carries out defect detection according to the method described above. In this case, the nominal switching times can be calculated separately from an excitation current control, which generates the control signals for the semiconductor switches. Alternatively or additionally, signals from the excitation current control, for example, the pulse-width modulated control signals, can be used to synchronize the acquisition of the excitation current measured values. If the evaluation is performed based on the frequencies of the excitation current signal, the excitation current measured values ​​are continuously acquired at short intervals as an excitation current signal.

[0044] The error signal can be output via an output unit, for example a control lamp, or as an electrical or electronic signal that is processed, for example, by other control units of the motor vehicle in which the separately excited synchronous machine is used.

[0045] Furthermore, a control device is proposed with a detection device for detecting a single permanently conductive, defective semiconductor switch of the semiconductor switches of a half-controlled H-bridge for setting an excitation current of a separately excited synchronous machine, wherein the half-controlled H-bridge a bridge arm with one of the semiconductor switches T1 and a diode D3 connected in series in reverse direction and a further bridge branch comprising a further diode D2 in reverse direction in series with another of the semiconductor switches T4 and a load branch comprising an excitation winding of the separately excited synchronous machine, wherein the load branch connects a connection point of one of the semiconductor switches T1 and the diode D3 in the one bridge branch and with a further connection point between the further diode D2 and the further of the semiconductor switches T4 in the further bridge branch, and wherein a current measuring device for measuring the current excitation current is arranged in the load branch; wherein the control device comprises an excitation current controller configured to generate two equal-frequency, pulse-width-modulated control signals, phase-shifted by half a period, for controlling one of the semiconductor switches T1 and the other of the semiconductor switches T4, such that the excitation current flowing through the load branch can be controlled via the pulse-width modulation of the control signals according to a requirement, wherein a duty cycle in normal operation is limited to a range of 50% to 100%, wherein the duty cycle indicates a ratio of a temporal pulse width of the signal level that places the correspondingly controlled semiconductor switch into a conductive state to the period of the control signal, and wherein the load branch comprises a current measuring device for measuring the excitation current, which is used by the excitation current controller as a feedback signal, characterized in that the detection device is designed to detect excitation current measured values ​​of the current measuring device and to evaluate the detected excitation current measured values ​​in the manner specified above in order to detect a lack of a change of sign between successive current changes between the various switching operations and to output an error signal if such a lack of expected change of sign is detected.

[0046] For this purpose, the detection device preferably comprises a processor on which program code can be executed, which executes the detection method in cooperation with the device described above.

[0047] The invention is explained in more detail below with reference to a drawing: Fig. 1 is a schematic diagram of a semi-controlled H-bridge for generating an excitation current; Fig. 2 different graphs, plotted against time in units of time per pulse width period, with the control signals for the semiconductor switches, a voltage drop across the excitation coil, a variation of the excitation current and the total current flowing through the entire circuit each graphically represented; Fig. 3 comparable graphs to those of the Fig. 2 in the event that one of the semiconductor switches is permanently conductively defective; Fig. 4 different graphs schematically showing the control signals and the excitation current signals for an intact H-bridge circuit state and a defective H-bridge circuit as well as transformed frequency spectra of the excitation current signals; Fig. 5 is a schematic view of a motor vehicle with a separately excited synchronous machine, which is controlled by means of a semi-controlled H-bridge via a control unit, which has a detection device for detecting a permanently conductive defective semiconductor switch; and Fig. 6 a schematic representation of a flow diagram of a method for detecting a permanently conductive defective semiconductor switch.

[0048] In Fig. 3 are analogous to the one described above Fig. 2 shows the signal curves and the measured values ​​determined or resulting for the operating case of a half-controlled H-bridge, in which one of the semiconductor switches, without restriction of generality, the one of the semiconductor switches 210, T1 according to Fig. 1, is permanently conductive and defective. In graph 820, the level 822 of the control signal 821 for the other semiconductor switch 340, T4 is plotted against time 801 in period lengths of the pulse-width modulated control signal. In the case of an intact semiconductor switch, the graph can also be understood as a state graph for the other semiconductor switch 340, T4. At level 0, the semiconductor switch 340, T4 is open, i.e., not conductive, and correspondingly at level 1, conductive and closed. The graph 810 accordingly shows the control signal 811 for one of the semiconductor switches 210, T1, which is shown in dashed lines. The level 812 is plotted against time 801. The switching state 813 of one of the semiconductor switches 210, T1 is shown as a solid line. This is permanently conductive, which corresponds to level 1. The third graph 830 shows the voltage U dropping across the excitation coil 460 exc831. Since one of the semiconductor switches 210, T1 is permanently conductive and a minimum duty cycle of 50% is specified for the control signal 811 of the semiconductor 340, T4, the supply voltage U DC for at least half the time on the excitation winding 460.

[0049] This causes a minimum excitation current I exc_min The fourth graph 840 shows the amplitude fluctuation of the excitation current I exc 841. The frequency of this current change is halved compared to the intact operating state. The fifth graph 850 represents the total current flow I DC 851.

[0050] In Fig. 2 and Fig. 3, the nominal switching times and nominal switching operations are shown in the entries “T1 ein ”, “T4 aus ”, “T4 ein ”, “T1 aus The excitation current changes ΔI occurring between these times exc_Tx_yyy(with x = 1 or x = 4 and yyy = on or yyy = off) are assigned here, without loss of generality, to the switching process with which the acquisition of the excitation current measured value is temporally correlated, at the same time as the excitation current measured value is acquired, which is the minuend in the calculation of the excitation current measured value difference. The current change Δ I exc_T4_aus in the time interval between the nominal switching time T1 ein of the nominal switching operation T1 ein Switching on one of the semiconductor switches 210 T1 and the nominal switching time T4 aus of the nominal switching operation T4 aus Switching off the other of the semiconductor switches 340 T4 is here, without loss of generality, the nominal switching process T4 aus assigned to the measurement and recording of the excitation current measured value I exc_T4_aus is correlated in time, which is the minuend of the excitation current measurement difference calculation: ΔI exc_T4_aus = Iexc_T4_aus - I exc_T1_ein . Switching times and the temporally correlated switching processes are each designated the same, with the designation indicating the semiconductor switch and its nominal state after the switching process, i.e., the target state. T1 ein means that one semiconductor switch T1 is switched into the on (conductive) state.

[0051] In chronological order, the excitation current measured value difference ΔI exc_T4_aus ,ΔI exc_T4_ein , ΔI exc_T1_aus , ΔI exc_T1_ein iteratively. Using a comma-separated index " ,n ”, where n represents a counting index for the switching cycles, the switching times can be assigned to switching cycles that correlate with the periods of the control signals.

[0052] In order to make a more reliable and robust statement, for example, an average of the corresponding excitation current measured value differences ΔI exc_T4_aus,n, ΔI exc_T4_ein,n ,ΔI exc_T1_aus,n , ΔI exc_T1_ein,n over a number N of control cycles, which are numbered with a natural number n = 1, 2, ... , N. The averaged excitation current measured value difference, for example ΔIexc¯_T4_ein, is obtained by summing the excitation current measured value differences that correspond to the switching process T4 ein are assigned, and division by the number N of summands, for example, for the switching time T4 ein assigned averaged excitation current measured value difference: ΔIexc¯_T4_in=1N∑n=1NΔIexc_T4_in,n.

[0053] Equivalently, the corresponding excitation current measured values ​​I exc Tx_yyy,n be averaged, ΔIexc¯_Tx_yyy=1N∑n=1NIexc_Tx_yyy,n. and the corresponding excitation current measured value differences ΔIexc¯_Tx_yyy be determined from the averaged excitation current measurements ΔIexc¯_Tx_yyy=ΔIexc¯_Tx_yyy−ΔIexc¯_Tz_www, e.g. ΔIexc¯_T4_on=Iexc¯_T4_on−Iexc¯_T4_off.

[0054] This is preferably done on a sliding scale, so that the most recently recorded and considered value from the averaging process is replaced by the most recently measured corresponding value. For averaging, for example, the values ​​of the last k control cycles or, alternatively, the excitation current measured values ​​recorded within a predefined past time period are always evaluated. If the excitation current measured value differences are averaged, the excitation current measured value differences determined for the previous k control cycles are averaged, or the excitation current measured value differences determined within a predefined past time period are averaged.

[0055] In Fig. 4 shows the graph 640 of the detected excitation current signal 641 for an intact half-controlled H-bridge circuit 100, the graph 920 of its Fourier-transformed spectrum 921, the graph of the detected excitation current signal 841 for a defective half-controlled H-bridge circuit 100, and the graph 940 of its Fourier-transformed spectrum 941, as well as the two graphs 610, 620 of the (nominal) control signals 611 and 621, one below the other. The first graph 640 corresponds to the one with the same name. Fig. 2 the third graph 840 is equal to the one with the same name Fig. 3. Based on the Fourier-transformed spectra 921 and 941, it is clearly visible that the frequency of the main components 923, 943, with which the excitation current changes, is halved when one of the semiconductor switches enters the permanently conductive defective state. In contrast to the Fourier-transformed spectrum 921 of the intact semiconductor circuit, in which a fundamental frequency 922 of the main component 923 is recognizable, which is twice the switching frequency 2f PWM of the control signals 611, 612, the switching frequency f appears as the fundamental frequency 942 in the Fourier-transformed spectrum 941 of the defective semiconductor circuit. PWM of the control signals 611, 612 of the semiconductor switches 210, 340, where the main component 943 can be seen. The frequency component at twice the switching frequency 2f PWMThe control signals 611, 612 are significantly reduced, for example, their intensity is halved. It should be noted that the Fourier-transformed spectra 921, 941 are greatly simplified. The frequency components at higher harmonics of the switching frequency f PWM the control signals 611, 612 are not shown in the graphs 920 and 940, for example, for reasons of simplification.

[0056] Other transformations, in particular discrete transformations, can also be used, which utilize functions better adapted to the approximately triangular shape of the excitation current signal, in order to perform an optimal frequency analysis that reliably provides the main frequency of the excitation current signal.

[0057] By comparing the phase position of the excitation current signal 841 for the defective state of the half-controlled H-bridge circuit 100 with the phase positions of the control signals 811 and 821, it can be seen that the phase position 845 of the excitation current signal 841 corresponds to the phase position 625 of the control signal 621 of another of the semiconductor switches T4 340. There is an offset in the phase position 615 relative to the control signal 611 of one of the semiconductor switches T1 210. Thus, the other of the semiconductor switches T1 210 is permanently conductive and defective. The phase positions 845, 615, 625 are each given, for example, by the maximum / minima of the signals of the corresponding periodic signal waveforms.

[0058] In Fig. 5 schematically shows a motor vehicle 1. This has a separately excited synchronous machine 10 as its drive machine. An excitation current for the excitation winding, which is generally arranged in the rotor, is provided via a semi-controlled H-bridge circuit 100, which is controlled via a control device 1200. The electrical energy originates from a traction battery 50, which is connected, among other things, to the H-bridge circuit 100. The traction battery 50 has a positive pole 51 and a negative pole 52. The control device 1200 detects a request signal 21 from a driver, for example via an accelerator pedal 20. The control device 1200 comprises a detection device 1240, which is designed to detect an individual defective semiconductor switch in the semi-controlled H-bridge circuit 100 and to output an error signal 31.This can be output, for example, via a control lamp in an instrument cluster or any other display device 30 in the vehicle. The error signal 31 can also be output as an electrical signal or radio signal for further processing in the control unit 1200 or other control units (not shown) of the motor vehicle 1.

[0059] Identical technical features are provided with the same reference numerals as in the other figures and are not explained in detail here again. In particular, the H-bridge circuit 100 is similar to the one shown in Fig. H-bridge circuit 100 shown in Figure 1.

[0060] The control device 1200 comprises an excitation current control 1220. This has a control unit 1221, which, based on the detected request signal 21, generates a target current signal I exc, sollA signal generating device 1222 of the excitation current control 1220 generates the control signals StS T1 and StS T4 These are the pulse-width-modulated control signals 611, 621, which are phase-shifted by half a pulse-width period. These control the semiconductor switches T1, 210 and T4, 340. The two semiconductor switches 210, 340 are designed, for example, as IGBTs. Starting from the supply voltage U DC a medium voltage falls on the excitation winding Uexc¯=UDC(2DC−1)ab. DC is the duty cycle. The resulting average excitation current Iexc¯=Uexc¯Rexc=UDC(2DC−1)Rexc. Here, R exc the ohmic resistance of the load branch or the excitation winding 460. The excitation current I measured by the current measuring device 480 exc, which is detected as a controlled variable by the control device 1200 and its excitation current control 1220, the excitation current I exc be regulated according to the request signal 21 or the request accordingly.

[0061] The detection unit 1240 is described below together with the Fig. 6, a method for detecting a permanently conductive defective semiconductor switch in the semi-controlled H-bridge 100 is described in more detail. The detection device records measured actual values ​​I in chronological order. exc of the current measuring device 480 2020, which are referred to as excitation current measurements. As can be seen from the graphs 640, 840 of the Fig. 2 and Fig. 3, the excitation current I exc continually.

[0062] The detection device is designed to use the recorded excitation current measured values ​​I excTo determine the change in the excitation current sign. In one embodiment, the sign of the excitation current change is determined iteratively based on the recorded excitation current measured values.

[0063] For this purpose, for example, the excitation current measurements recorded in chronological order are compared. If the subsequently measured excitation current measurement is smaller than the previously measured excitation current measurement, the sign is negative; otherwise, it is positive.

[0064] Alternatively or additionally, the sign can be derived from calculating the difference between the consecutively acquired excitation current values. The sign of the excitation current change corresponds to the sign of the determined difference if the previously acquired excitation current value is used as the subtrahend and the subsequently acquired excitation current value is used as the minuend.

[0065] Since in a switching cycle the excitation current change changes its sign with each switching operation in an intact H-bridge circuit 100, it is advantageous that the excitation current measured values ​​are recorded in a time-correlated manner with the nominal switching operations, ie in a time-correlated manner with the nominal switching times of the control cycles.

[0066] For this purpose, the detection device can receive synchronization information or generate it itself 2010. As synchronization information, the detection device can receive, for example, the control signals for the two semiconductor switches. Alternatively, the detection unit 1240 can detect the request signal 21 in a similar way to the excitation current control 1220 and derive the nominal switching times from this, taking into account the detected excitation current measured values.

[0067] Preferably, the excitation current measurements are temporally correlated, preferably recorded immediately at the switching time or shortly before or after the nominal switching times, and used to derive the sign of the excitation current change. Shortly before or shortly after the nominal switching time means that the time interval between the measurement and the nominal switching time is less than one hundredth, preferably one thousandth, of the period length of the pulse-width-modulated control signals. Particularly preferably, measurements are taken simultaneously with the nominal switching operations.

[0068] The detection unit then checks in 2050 whether a change in the sign of the excitation current change is detected, correlated in time with the nominal switching operations, i.e., correlated in time with the nominal switching times. If this is the case, the process steps described so far are executed iteratively.

[0069] If the sign does not change, an error signal 2070 is output.

[0070] If the time intervals between switching operations are short, the changes in the excitation current are also small. Therefore, in a preferred embodiment, it is advantageous to perform an averaging 2030 when determining the sign of the excitation current change.

[0071] Here, the recorded excitation current measured values, which each correspond to the same switching operation, ie to one of the switching operations T1 ein , T4 aus , T4 ein and T1 aus, are averaged over a specified number of switching cycles, for example, 100 switching cycles, and the averaged excitation current measured values ​​are used to derive the sign of the excitation current change. Alternatively, the excitation current measured value differences that approximate the excitation current change can be averaged, again averaging the excitation current measured value differences that occur between the same nominal switching operations in the switching cycles.

[0072] Alternatively, it is also possible to average the signs determined from non-averaged excitation current measurements. In this case, the signs of the corresponding current changes in switching cycles are averaged, for example, by assigning the value "+1" to a positive sign and the value "-1" to a negative sign. These values ​​are summed. The sign of the sum is used as the averaged sign. To avoid an uncertain statement with the value "0," the averaging can be performed over an odd number of signs.

[0073] Preferably, a moving average is calculated over a previous number of control cycles. Alternatively, a time period can also be specified.

[0074] Alternatively or in addition to the averaging 2030, it can also be provided that the error signal is only generated if, within a time period or a further number of control cycles, a number of missing expected sign changes exceeds a threshold value 2060. Otherwise, the method is restarted iteratively.

[0075] When a fault signal is generated, one embodiment identifies the permanently conducting defective semiconductor switch 2080. This is the semiconductor switch for which no sign change is detected after its nominal switching operation.

[0076] The signal identifying the defective semiconductor switch is also output 2090. It can be used, for example, to diagnose the half-controlled H-bridge circuit.

[0077] In the example shown, it is determined that after the nominal switching operations T1 ein and T1 ausno change in the sign of the current change occurs. The signs of the excitation current changes or the excitation current differences ΔI exc_T1_aus = I exc_T1_aus - I exc_T4_ein , ΔI exc_T1_ein = I exc_T1_ein - I exc_T1_aus and ΔI exc_T4_aus = I exc_T4_aus - I exc_T1_ein are all positive. After the nominal switching operations of the semiconductor switch T4 340, a sign change occurs. This semiconductor switch T4 340 is therefore intact. Therefore, the other semiconductor switch T1 210 is defective.

[0078] As an alternative to the individual determination of the sign 2041 of the excitation current change for individual time periods, the sign changes can be determined by determining a change frequency or a main frequency of the change in the excitation current 2045. For this purpose, for example, a frequency analysis can be carried out using a transformation method, for example a fast Fourier transformation or a similar discrete transformation method 2046.

[0079] However, the frequency can also be determined in other ways, for example, through the use of filters. If a bandpass filter with a pass frequency in the range of twice the frequency of the control signals and another bandpass filter with a pass frequency in the range of the frequency of the control signals are used to filter the acquired excitation current measurement signal, the ratios of the signal strengths of the signals passed through the two filters can be used to determine whether the excitation current change occurs at twice the frequency of the control signals or at the frequency of the control signals.

[0080] Checking whether the sign changes correlate with the switching operations of the semiconductor switches of the semi-controlled H-bridge circuit 100 (2050) can be done by comparing the determined frequency with the frequency of the control signals (2055). If the main frequency component is approximately twice as large as the frequency of the control signals, the sign changes correlate with the switching operations. A sign change occurs for each switching operation. If the determined frequency or main frequency component is equal to the frequency of the control signals, one of the semiconductor switches is permanently conductive and defective.

[0081] Determining which semiconductor switch is defective can be done by determining the phase position of the excitation current signal relative to one of the control signals 2085. The defective semiconductor switch is the one whose control signal is out of phase. It will be understood by those skilled in the art that the nominal control signals are meant here. The detection unit can determine the failure of a control signal by evaluating the respective control signal. If one of the control signals remains permanently at level 1, this also leads to a permanently conductive semiconductor switch; however, the signal generating device 1222 is then defective.

[0082] Both the excitation current control 1220 and the detection device 1240 are preferably implemented by means of a program-controlled microprocessor 1260, which executes program code 1270 stored in the memory device 1250. It is understood that other embodiments may also include dedicated circuits that implement the excitation current control and / or the detection device. List of reference symbols 1 motor vehicle 10 separately excited synchronous machines 20 Accelerator pedal 21 Driving request signal 30 Display device 31 Error signal 50 traction battery 51 positive pole 52 Negative pole 100 H-bridge circuit 180 additional current measuring devices 190 Voltage measuring device 200 bridge branch 201 an end 202 opposite end 210, T1 semiconductor switch 230, D3 diode 250 connection point 300 additional bridge branches 301 an end 302 opposite end 320, D2 additional diode 340, T4 additional semiconductor switch 350 connection point 400 load branch 460 Excitation winding 465 ohmic winding resistance 467 Winding inductance 480 current measuring device 500 DC link capacitor 601 time 610 Graph 611 Control signal (T1) 612 levels 615 Phase position 620 Graph 621 Control signal (T4) 625 Phase position 622 levels 630 Graph 631 Excitation voltage 640 graphs 641 Excitation current 650 graphs 651 total current 801 time 810 Graph 811 Control signal (T1) 812 levels 813 Switching state 820 Graph 821 Control signal (T4) 822 levels 830 Graph 831 Excitation voltage 840 Graph 841 Excitation current 845 Phase position 846 offset 850 Graph 851 total current 920 Graph 921 Fourier-transformed spectrum of the excitation current signal (intact H-bridge circuit) 922 fundamental frequency 923 main component 940 Graph 941 Fourier-transformed spectrum of the excitation current signal (defective H-bridge circuit) 942 fundamental frequency 943 main component 1200 control unit 1220 Excitation current control 1221 Control device 1222 Signal generating device 1240 detection device 1250 memory 1260 microprocessor 1270 program code 2000 Flowchart of a method for detecting a defective semiconductor 2010 - 2090 procedural steps

Claims

[1] Method for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches (210, 340) of a semi-controlled H-bridge circuit (100) for setting an excitation current of a separately excited synchronous machine, wherein the semi-controlled H-bridge circuit (100) a bridge branch (200) with one of the semiconductor switches T1 (210) and a diode D3 (230) connected in series in the reverse direction and a further bridge branch (300) comprising a further diode D2 (320) in reverse direction in series with another of the semiconductor switches T4 (340) and a load branch (400) comprising an excitation winding (460) of the separately excited synchronous machine, wherein the load branch (400) connects a connection point (250) of one of the semiconductor switches T1 (210) and the diode D3 (230) in the one bridge branch (200) and to a further connection point (350) between the further diode D2 (320) and the further of the semiconductor switches T4 (340) in the further bridge branch (300), and wherein a current measuring device (480) for measuring the current excitation current is arranged in the load branch (400); wherein the half-controlled H-bridge circuit (100) is coupled to a control device (1200) comprising an excitation current control (1220) which generates two pulse-width-modulated control signals (611, 621) of equal frequency and phase-shifted by half a period for controlling one of the semiconductor switches (210) and the other of the semiconductor switches (340), so that the excitation current flowing through the load branch (400) can be controlled via the pulse-width modulation of the control signals (611, 621) according to a requirement, wherein a duty cycle in normal operation is limited to a range of 50% to 100%, wherein the duty cycle indicates a ratio of the temporal pulse width of the signal level that puts the controlled semiconductor switch into a conductive state to the period of the control signal, and wherein the load branch (400) comprises a current measuring device (480) in order to to measure the excitation current used by the excitation current control (1220) as a feedback signal, characterized by that it is evaluated whether a change in the excitation current correlates its sign with each of the nominal switching times (T1 ein , T4 aus , T4 ein , T1 aus ) of one of the semiconductor switches (210, 340) changes, and if this is not the case, an error signal is output which indicates a defect in one of the semiconductor switches (210, 340), wherein a nominal switching time (T1 ein , T4 aus , T4 ein , T1 aus ) is a switching time at which one of the two semiconductor switches (210, 340) changes its switching state in an intact half-controlled H-bridge circuit (100). [2] Method according to claim 1, characterized bythat an excitation current measurement value is recorded in time correlated with the nominal switching times of each control cycle, and the current change of the excitation current is determined and evaluated as a difference between the excitation current measurement values ​​recorded in the control cycles in immediate succession correlated with the nominal switching times of the semiconductor switches (210, 340). [3] Method according to one of the preceding claims, characterized by that an averaging is carried out when determining the sign of the current change of the excitation current. [4] Method according to claim 3, characterized by that the excitation current measured values, which are recorded in correlation with nominal switching times corresponding to one another in different control cycles, are averaged and the excitation current changes are determined on the basis of these averaged excitation current measured values. [5] Method according to claim 3, characterized bythat the differences determined on the basis of excitation current measured values ​​which are recorded in correlation with each other in different switching cycles corresponding to nominal switching times are averaged. [6] Method according to one of the preceding claims, characterized by that the semiconductor switch is determined to be defective, correlates with whose nominal switching operations no change in the sign of the excitation current change occurs [7] Method according to claim 1, characterized bythat the change of sign is carried out based on a determination of a change frequency or main frequency component of an excitation current signal formed from the excitation current measured values, wherein the excitation current signal represents the current changes of the excitation current, wherein a temporal correlation of the change of sign with the switching operations exists when a determined frequency or main frequency component has at least approximately twice the value of a frequency of the pulse-width modulated control signals, and the error signal is output when the determined frequency or main frequency component has a value which corresponds at least approximately to the frequency of the pulse-width modulated control signals. [8] Method according to claim 7, characterized bythat the semiconductor switch is determined to be defective, correlates with whose nominal switching operations no change in the sign of the excitation current change occurs or with whose control signal the excitation current signal has a phase shift. [9] Control device (1200) with detection device (1040) for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches (210, 340) of a semi-controlled H-bridge (100) for setting an excitation current of a separately excited synchronous machine, wherein the semi-controlled H-bridge (100) a bridge branch (200) with one of the semiconductor switches T1 (210) and a diode D3 (230) connected in series in the reverse direction and a further bridge branch (300) comprising a further diode D2 (320) in the reverse direction in series with a further one of the semiconductor switches T4 (340) and a load branch (400) comprising an excitation winding (460) of the separately excited synchronous machine, wherein the load branch connects a connection point (250) of the one of the semiconductor switches T1 (210) and the diode D3 (230) in the one bridge branch (200) and with a further connection point (350) between the further diode D2 (320) and the further one of the semiconductor switches T4 (340) in the further bridge branch (300), and wherein a current measuring device (480) for measuring the current excitation current is arranged in the load branch (400); wherein the control device (1200) comprises an excitation current control (1220) which is designed to generate two pulse-width modulated control signals (611, 621) of equal frequency and phase-shifted by half a period for controlling one of the semiconductor switches T1 (210) and the other of the semiconductor switches T4 (340), so that the excitation current flowing through the load branch (400) can be controlled via the pulse-width modulation of the control signals (611, 621) according to a requirement, wherein a duty cycle in normal operation is limited to a range of 50% to 100%, wherein the duty cycle indicates a ratio of a temporal pulse width of the signal level (612, 622), which puts the correspondingly controlled semiconductor switch (210, 340) into a conductive state, to the period of the control signal (611, 621), and wherein the load branch (400) comprises a current measuring device (480) to measure the excitation current, which is used by the excitation current control (1220) as a feedback signal, characterized by , that the detection device (1040) is designed to detect excitation current measured values ​​of the current measuring device (480) and to evaluate whether a current change in the excitation current correlates its sign with each of the nominal switching times (T1 ein , T4 aus , T4 ein , T1 aus) of one of the semiconductor switches (210, 340) changes, and if this is not the case, to output an error signal indicating a defect in one of the semiconductor switches (210, 340), wherein a nominal switching time (T1 ein , T4 aus , T4 ein , T1 aus ) is a switching time at which one of the two semiconductor switches (210, 340) changes its switching state in an intact half-controlled H-bridge circuit (100). [10] Control device (1200) according to claim 9, characterized by that the detection device (1240) comprises a program-controlled microprocessor and a program memory with program code, wherein when the program code is executed on the processor, the steps are carried out according to a method according to one of claims 1 to 8. [11] Computer program product comprising program code which, when executed on a microprocessor in cooperation with the control device (1200) according to one of claims 9 or 10 and the semi-controlled H-bridge circuit (100) described therein, carries out the method according to one of claims 1 to 8.

Citation Information

Patent Citations

  • converter

    DE102014018665A1

  • Rotating electric machine with a function for detecting anomalies in a switching element

    DE102014102869A1

  • Electrical device with a switched mode power supply and method for checking the power supply of the electrical device

    DE102016203355A1

  • Method for operating an electrical circuit arrangement, electrical circuit arrangement and motor vehicle

    DE102020112704A1

  • Module self-check in a modular multicell converter

    EP3208922A1