DE102023205557A1 - Method and device for detecting a single defective semiconductor switch based on the excitation current - Google Patents

Abstract

The invention relates to a method and a device for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches (210, 340) of a half-controlled H-bridge circuit (100) for setting an excitation current of an externally excited synchronous machine, wherein the half-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) and a current measuring device (480) for measuring the current excitation current;
wherein a control device (1200) generates two pulse-width modulated control signals (611, 621) of equal frequency and phase-shifted by half a period for controlling the semiconductor switches (210, 340), wherein it is evaluated whether a current change in the excitation current changes its sign in a temporally correlated manner with each of the nominal switching times (T1 _on , T4 _off , T4 _on , T1 _off ) of one of the semiconductor switches (210, 340), and if this is not the case, an error signal is output.

Description

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 an externally excited synchronous machine. In addition, 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.

A half-controlled H-bridge circuit, also called a half-controlled H-bridge, is often used as a control circuit for generating an excitation current of a separately excited synchronous machine. In 1 Such a circuit is shown schematically. The circuit 100 comprises a bridge branch 200 and a further bridge branch 300. One bridge branch 200 comprises a semiconductor switch 210, which is also referred to by the abbreviation T1. A reverse-biased diode 230, which is also referred to as D3, is connected in series with the semiconductor switch 210. The further bridge branch 300 comprises a further diode 320 arranged in the reverse direction, which is also referred to as D2. In series with the further diode, a further semiconductor switch 340, which is also referred to as T4, is arranged in the further bridge branch 300.

A load branch 400 is formed between a connection point 250 of the one bridge branch 200 between the one semiconductor switch 210 and the 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 diagram. A current measuring device 480 is also 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 . An excitation voltage U _exc is dropped across the excitation winding 460. The bridge branch 200 and the further bridge branch 300 are connected at one end 201, 301 to a positive pole 51 of a supply voltage U _DC and at the opposite end 202, 302 to a negative pole 52 of the supply voltage. An intermediate circuit capacitor 500 is arranged parallel to the bridge branch 200. The supply voltage U _DC is applied to this. In addition, a further current measuring device 180 is formed, which can measure the total current I _DC flowing through the half-controlled bridge circuit 100. A voltage measuring device 190, which measures the supply voltage U _DC, is formed parallel to the intermediate circuit capacitor 500.

In order to generate an excitation current I _exc in accordance with a request, for example from a driver in a vehicle that is driven by an externally excited synchronous machine, the semiconductor switches T1, T4; 210, 340 are each controlled with a pulse-width modulated control signal. When semiconductor switches are mentioned below, this means the one semiconductor switch 210, T1 and the other semiconductor switch 340, T4. When one of the semiconductor switches is mentioned, this can be either the one semiconductor switch 210, T1 or the other semiconductor switch 340, T4.

The control signals 611, 621 are square-wave signals with the same frequency, but offset from each other by half a period T _PWM (see 2 ). A current flows through the half-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.

The control signals 611, 621 each have the same duty cycle, which is 50% or greater. The duty cycle is understood to be the time ratio of the control signal at a level 612, 622, which switches the corresponding semiconductor switch 210, 340 controlled thereby into a conductive 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 time periods in which current can be fed into the excitation winding 460.

In 2 Five graphs 610-650 are shown schematically one below the other, each plotted against time 601 in units of the period T _PWM of the pulse width modulated signal. The upper two graphs 610, 620 indicate the pulse width modulated control signal 611 or 621 for controlling one of the semiconductors 210, T1 or another of the semiconductors 340, T4. 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 the level 1, and is 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.

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 _HV corresponding to The changing excitation current 641 is shown in graph 640. This rises steeply in areas 646, 648 when both semiconductor switches 210, 340, T1, T4 are conductive, and falls again somewhat more slowly in areas 647, 649 in which only one of the semiconductor switches 210, 340 is in a closed state. In an intact half-controlled H-bridge circuit, there are alternating current changes in the excitation current 641.

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

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 that the current that is fed into and out of the intermediate circuit capacitor 500 is kept low. Furthermore, the frequency of the current ripple of the excitation current 641 is twice as high as the switching frequency of the individual semiconductor switches when the semiconductor switches are functioning, which is advantageous for the ripple of the excitation current. In addition, the power loss is distributed evenly between the semiconductor switches and diodes. In addition to the center-centered pulse width modulation shown here, an 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.

When controlling the semiconductor switches 210, T1; 340, T4, it is important 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 (approx. -2 V to -3 V). The current then flows through a series of exponential functions and increases relatively steeply when the voltage at the excitation winding is positive, 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.

The previous description assumes that both semiconductor switches 210, T1; 340, T4 switch correctly in accordance with the control signals 611, 621. Since the excitation winding 460 is usually designed as the rotor of a synchronous machine and often has a high inductance 467, in the event of a fault it is necessary 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. Depending on the direction of the current, 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 high 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 converter is no longer connected to the traction network battery, the capacity of standard intermediate circuit capacitors 500 is generally not sufficient to prevent an overvoltage with the destruction of a main inverter and/or the H-bridge circuit described here and possibly other HV components. It is therefore sensible to determine in advance if one of the semiconductor switches is defective.

In the event of a fault that one of the two semiconductor switches remains permanently open despite being controlled, the excitation current I _exc will drop to 0 within a short period of time, usually a few 100 ms. If both switches are permanently switched on, the supply voltage U _DC is permanently present as the voltage of the excitation winding U _exc , so that the current through the excitation winding increases sharply. An overcurrent detection can reliably detect this fault and switch off the supply voltage. A more difficult and critical case is when one of the two semiconductor switches 210, 340 no longer switches and remains in a permanently conductive state and is defective, while the other semiconductor switch still switches normally. A cause for such a defect can lie in the control circuit or it can be a defect in one of the semiconductor switches 210, 340 itself.

The DE 10 2014 102 869 A1 describes that in a rotating electric machine, a first switching element is provided for stopping a 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 a drive of the first switching element based on a result of the determination of the abnormality determiner.

The EP 3 208 922 A1 concerns the operation of modular power converters, which is to be made safer. For this purpose, a method is proposed for testing a modular power converter that has 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 curve is recorded by the voltmeter. This voltage curve is differentiated according to time, and a test signal is generated from it. This allows the test to be carried out at module level.

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 consumer, an electronic switch which comprises an electronic switching element and a driver device which controls the electronic switching element, and a clocked power supply which has a power section with the electronic switching element and which is designed to generate an electrical supply voltage or an electrical supply current for the electrical consumer from an electrical voltage due to the one electronic switching element being alternately switched on and off. The power section has a current path through which an electrical current associated with the electronic switching element of the electronic switch flows when the power supply is in operation. The electrical device further comprises a pulse transformer.

The DE 10 2014 018 665 A1 describes a converter with a control section having a motion control and a power section comprising a power semiconductor switch. The power section has a safety monitoring system that provides a safety function.

The invention is based on the object of specifying a method and a device with which a detection of a single defective semiconductor switch in a half-controlled H-bridge for a separately excited synchronous machine in which one of the semiconductor switches is permanently conductively defective can be detected in a simpler manner.

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 and by a computer program product according to claim 11. Advantageous embodiments emerge from the subclaims.

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 half-controlled H-bridge circuit. Due to the use of phase-shifted control signals, there are four switching operations in a control cycle: switching the first of the semiconductor switches T1 _{on ,} switching the first of the semiconductor switches T1 off, switching the other semiconductor switch T4 _on , and switching the other semiconductor switch T4 _off . Since the semiconductor switches of the half-controlled H-bridge are equivalent and no hierarchical order exists, they are referred to below as one of the semiconductor switches and the other of the semiconductor switches. Switching on is regarded here as bringing about the conductive state. Switching off is regarded as bringing about 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 therefore has four switching points in time and correlated switching processes which occur in chronological sequence, for example in the following order and with the following events: switching on of one semiconductor switch T1 _on , switching off of another semiconductor switch T4 _off , switching on of another semiconductor switch T4 _on , switching off of one of the semiconductor switches T1 _off . The H-bridge circuit states of the time periods in between can be associated with the switching states of the semiconductor switches or the switching processes which bring about the switching states and result in: T1 _off to T1 _on ; T1 _on to T4 _off ; T4 _off to T4 _on ; and T4 _on to T1 _off . For the sake of completeness, it should be noted that with center-centered pulse width modulation (PWM), the first (last) time period T4 _on to T1 _off continues 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 time sequence when 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 sign changes 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, nominal switching times and nominal switching operations are sometimes referred to in order to to point out that the information refers to the intended switching time or intended switching process. 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, a real switching time correlates with every nominal switching time and a real switching process correlates with every nominal switching process.

In particular, a method is provided 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 an externally excited synchronous machine, wherein the half-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 externally excited synchronous machine, wherein the load branch connects a connection point of the 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 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 half-controlled H-bridge circuit is coupled to a control device which comprises an excitation current control which generates two pulse-width-modulated control signals of equal frequency and 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 in accordance with a request or a request signal (which alternatively serve as a reference variable), 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 which puts the controlled semiconductor switch into a conductive state to the period of the control signal, and wherein the load branch comprises a current measuring device to measure the excitation current I _exc , 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 has lost its sign temporally correlated 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.

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.

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

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

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

In one variant, it is possible for the differences determined between successive nominal switching times that correspond to one another in the control cycles carried out one after the other to be averaged over several control cycles. This means that a current change that occurs between two identical nominal switching operations in switching cycles carried out one after the other is derived by averaging the differences determined from the current measurements that are correlated with the switching operations that limit the time range of the corresponding excitation current change in the switching cycles. In one variant, the excitation current measurement value differences belonging to the same nominal switching operations are thus averaged.

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

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

As an alternative to one of the averaging methods or in addition, an error signal can only be generated if a minimum number of sign changes between successive current changes in a number of successive control cycles have not occurred, i.e. fewer sign changes have occurred than expected. This is equivalent to the number of sign changes between determined current changes at successive time periods within a time period or a specified number of control cycles falling below a threshold value.

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

Preferably, an excitation current measurement value is recorded in correlation with each of the nominal switching operations, preferably at the same time or alternatively shortly afterwards 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 formed in such a way that the excitation current measurement value measured last in time forms the minuend and the excitation current measurement value recorded previously forms the subtrahend. According to this convention, an excitation current difference is positive if the excitation current increases between the two measurement times and switching operations.

In order to improve the resolution of the analog-digital converter that converts the excitation current measurements into digital signals, the excitation current measurements assigned to the same nominal switching operations are averaged or, alternatively, the excitation current measurement differences assigned to the same nominal switching operations are averaged. This averaging is preferably carried out in a sliding manner. If the sign change does not occur or no longer occurs, an error signal is output.

If the excitation current measurement value differences are each assigned to the switching process whose excitation current measurement value forms the minuend, and if no sign change occurs between two consecutively determined excitation current measurement value differences, the semiconductor switch to which the current change occurring earlier in time or the excitation current measurement 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, there is no change in sign between the current change that lies between the nominal switching times or nominal switching operations of switching the other semiconductor switch T4 _on and switching the one of the semiconductor switches T1 _off , which is designated by ΔI _{exc_T1_off} and is given by ΔI _{exc_T1_off} = I _{exc_T1_off} - I _{exc_T4_on} , and the subsequent current change, which is designated by ΔI _{exc_T1_on} = I _{exc_T1_on} - I _{exc_T1_off} . There is also no change in sign for the subsequent current change ΔI _{exc_T4_off} = I _{exc_T4_off} - I _{exc_T1_on} The defect can therefore be assigned to the semiconductor switch to which the excitation current difference is assigned, for which no sign change subsequently occurs. According to this view, the excitation current difference is assigned to the semiconductor switch whose switching process takes place at the same time or closely correlated in time with the recording of the excitation current measurement value, which forms the minuend of the excitation current difference.

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 to which the excitation current measured value difference is assigned that does not show a change in sign with respect to the excitation current measured value difference preceding it in time is defective.

An alternative way of detecting the lack 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 curve, 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 appears abruptly as a change frequency in the spectrum, there is a defect in one of the semiconductor switches, in which case this defective semiconductor switch is permanently conductive. It should be noted that, even in the case of a defect, parts 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 if the excitation current change frequency corresponds to the frequency of the control signals or if this occurs in the frequency spectrum of the excitation current.

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.

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.

In order to determine the change frequency of the excitation current signal, a transformation analysis can be carried out, for example in the form of a fast Fourier transformation. The corresponding frequency components can be easily determined and monitored from the result. This evaluation is very robust. Disturbing high-frequency and low-frequency interference in the excitation current signal hardly affects the evaluation result.

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

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.

The defective semiconductor switch can be identified by evaluating the phase position of the excitation current signal in relation to the control signals of the semiconductor switches. In the event of a fault in which the frequency of the excitation current change is equal to the frequency of the control signals, the excitation current signal is in phase with the control signal of the semiconductor switch that is not defective. There is a phase offset with 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, with the semiconductor switch being identified as defective if a phase offset of the excitation current signal is detected for its control signal.

The invention can be implemented in program code which is executed on a processor and, in conjunction with the half-controlled H-bridge circuit, its excitation current measuring device and the control device for controlling it and the half-controlled H-bridge circuit, carries out defect detection according to the method described above. 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 recording of the excitation current measured values. If the evaluation is carried out based on the frequencies of the excitation current signal, the excitation current measured values are continuously recorded at short time intervals as an excitation current signal.

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

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 an externally excited synchronous machine, wherein the half-controlled H-bridge
a bridge branch 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 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 control device comprises an excitation current control which is designed to generate two pulse-width-modulated control signals of equal frequency, phase-shifted by half a period, for controlling one of the semiconductor switches T1 and the other of the semiconductor switches T4, 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 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 which puts 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 to measure the excitation current which is used by the excitation current control 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 sign change between successive current changes between the various switching operations and to output an error signal when such a lack of expected sign change is detected.

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.

• 1 eine schematische Darstellung einer halbgesteuerten H-Brücke zum Erzeugen eines Erregerstroms;
• 2 unterschiedliche Graphen, aufgetragen gegen die Zeit in Einheiten der Zeit pro Pulsweitenperiode, wobei die Ansteuersignale für die Halbleiterschalter, eine über der Erregerspule abfallende Spannung, eine Variation des Erregerstroms sowie der durch die gesamte Schaltung fließende Gesamtstrom jeweils grafisch dargestellt sind;
• 3 vergleichbare Graphen zu denen der 2 für den Fall, dass einer der Halbleiterschalter dauerhaft leitend defekt ist;
• 4 unterschiedliche Graphen, die die Steuersignale und die Erregerstromsignale für einen intakten H-Brückenschaltungszustand und eine defekte H-Brückenschaltung sowie transformierte Frequenzspektren der Erregerstromsignale schematisch zeigen;
• 5 eine schematische Ansicht eines Kraftfahrzeugs mit einer fremderregten Synchronmaschine, die mittels einer halbgesteuerten H-Brücke über ein Steuergerät angesteuert wird, welches eine Detektionseinrichtung zum Erkennen eines dauerhaft leitend defekten Halbleiterschalters aufweist; und
• 6 eine schematische Darstellung eines Flussdiagramms eines Verfahrens zum Erkennen eines dauerhaft leitend defekten Halbleiterschalters.

The invention is explained in more detail below with reference to a drawing:

• 1 a schematic representation of a half-controlled H-bridge for generating an excitation current;
• 2 different graphs plotted against time in units of time per pulse width period, wherein 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 are each graphically shown;
• 3 comparable graphs to those of the 2 in the event that one of the semiconductor switches is permanently conductively defective;
• 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;
• 5 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
• 6 a schematic representation of a flow chart of a method for detecting a permanently conductive defective semiconductor switch.

In 3 are analogous to the one described above 2 the signal curves and determined or resulting measured values for the operating case of a half-controlled H-bridge are shown, in which one of the semiconductor switches, without restriction of generality the one of the semiconductor switches 210, T1 according to 1 , is permanently conductively defective. In the 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 used as a state graph for the other the semiconductor switch 340, T4. At level 0, the semiconductor switch 340, T4 is open, ie not conductive, and correspondingly conductive and closed at level 1. The graph 810 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 _exc 831 dropping across the excitation coil 460. 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 is applied to the excitation winding 460 for at least half the time.

This causes a minimum excitation current I _{exc_min} to flow in the excitation winding 460. The fourth graph 840 shows the amplitude fluctuation of the excitation current I _exc 841. A frequency of this current change is halved compared to the intact operating state. The fifth graph 850 shows the total current flow I _DC 851.

In 2 and 3 the nominal switching times and nominal switching processes are shown using the entries "T1 _on ", "T4 _off ", "T4 _on ", "T1 _off ". The excitation current changes ΔI _{exc_Tx_yyy} (with x = 1 or x = 4 and yyy = on or yyy = off) occurring between these times are assigned here, without loss of generality, to the switching process with which the recording of the excitation current measured value is temporally correlated, at the same time as the excitation current measured value is recorded, which is the minuend when calculating the excitation current measured value difference. The current change ΔI _{exc_T4_off} in the time interval between the nominal switching time point T1 _on of the nominal switching process T1 _on switching on one of the semiconductor switches 210 T1 and the nominal switching time point T4 _{off of} the nominal switching process T4 _off switching off the other of the semiconductor switches 340 T4 is assigned here, without loss of generality, to the nominal switching process T4 _off , with which the measurement and recording of the excitation current measured value I _{exc_T4_off} is temporally correlated, which is the minuend of the excitation current measured value difference calculation: ΔI _{exc_T4_off} = I _{exc_T4_off} - I _{exc_T1_on} . Switching times and the switching processes temporally correlated with them are each referred to in the same way, whereby the designation indicates the semiconductor switch and its nominal state after the switching process, i.e. the target state. T1 _on means that the semiconductor switch T1 is switched to the on (conductive) state.

The excitation current measurement value difference ΔI _{exc_T4_off} , ΔI _{exc_T4_on} , ΔI _{exc_T1_off} , ΔI _{exc_T1_on} thus occur iteratively in chronological order. Using an index " _n " separated by a comma, 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.

In order to be able to make a more reliable and robust statement, for example, an average of the corresponding excitation current measured value differences ΔI _{exc_T4_out,n} , ΔI _{exc_T4_in,n} , ΔI _{exc_T1_out,n} , ΔI _{exc_T1_in,n} is carried out 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 ΔI _{exc_T4_ein} , is obtained by summing the excitation current measured value differences _assigned to the switching process T4 on and dividing by the number N of summands, for example, the following applies to the _average excitation current measured value difference assigned to the switching time T4 on: $\Delta I_{\overline{exc}\_\mathrm{<figure-callout\; id="4"\; label="T"\; filenames="DE102023205557A1\_0004.png,DE102023205557A1\_0005.png"\; state="\{\{state\}\}">T</figure-callout>}4\_ein}=\frac{1}{N}{\displaystyle \sum {}_{n=1}^N\Delta I_{exc\_\mathrm{<figure-callout\; id="4"\; label="T"\; filenames="DE102023205557A1\_0004.png,DE102023205557A1\_0005.png"\; state="\{\{state\}\}">T</figure-callout>}4\_ein,n}}.$

und die entsprechenden Erregerstrommesswertedifferenzen ΔI _{exc_Tx_yyy} aus den gemittelten Erregerstrommesswerten bestimmt werden ΔI _{exc_Tx_yyy} = I _{exc_Tx_yyy}- I _{exc_Tx_www}, z. B. ΔI _{exc_T4_ein} =I _{exc_T4_ein} -I _{exc_T4_aus}.Equivalently, the corresponding excitation current measurements I _{exc_Tx_yyy,n} can be averaged, $\Delta I_{\overline{exc}\_Tx\_yyy}=\frac{1}{N}{\displaystyle \sum {}_{n=1}^N\Delta I_{exc\_Tx\_yyy,n}}$

and the corresponding excitation current measurement differences ΔI _{exc _Tx_yyy} can be determined from the averaged excitation current measurements ΔI _{exc _Tx_yyy} = I _{exc _Tx_yyy} - I _{exc _Tx_www} , e.g. ΔI _{exc _T4_ein} =I _{exc _T4_ein} -I _{exc _T4_out} .

This is preferably done on a sliding basis, so that the value recorded and taken into account furthest back in time is replaced by the corresponding value measured most recently. For averaging, the values of the last k control cycles or, alternatively, the excitation current measured values recorded in a given previous period of time 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 accordingly, or the excitation current measured value differences determined in a given previous period of time.

In 4 are 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 are shown below each other. The first graph 640 is equal to the one with the same name 2 the third graph 840 is equal to the one with the same name 3 . Based on the Fourier-transformed spectra 921 and 941, it is clearly visible that the frequency of main components 923, 943, with which the excitation current changes, is halved when one semiconductor switch goes into 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 can be identified, which corresponds to twice the switching frequency 2f _PWM of the control signals 611, 612, in the Fourier-transformed spectrum 941 of the defective semiconductor circuit, the switching frequency f _PWM of the control signals 611, 612 of the semiconductor switches 210, 340 is added as the fundamental frequency 942, at which the main component 943 can be identified. The frequency component at double the switching frequency 2f _PWM of the control signals 611, 612 is significantly reduced, for example, its 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 of the control signals 611, 612 are not shown in the graphs 920 and 940, for example for reasons of simplification.

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

From the comparison of 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 matches 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 compared to the control signal 611 of one of the semiconductor switches T1 210. The other of the semiconductor switches T1 210 is therefore permanently conductively defective. The phase positions 845, 615, 625 are given, for example, by the maxima/minima of the signals of the corresponding periodic signal curves.

In 5 a motor vehicle 1 is shown schematically. This has an externally excited synchronous machine 10 as the drive machine. An excitation current for the excitation winding, which is usually 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 comes from a traction battery 50, which is connected to the H-bridge circuit 100, among other things. 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 in order to be further processed in the control unit 1200 or other control units (not shown) of the motor vehicle 1.

The same 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 in 1 shown H-bridge circuit 100.

Hierbei gibt R_exc den ohmschen Widerstand des Lastzweigs bzw. der Erregerwicklung 460 an. Über den von der Strommesseinrichtung 480 gemessenen Erregerstrom I_exc, der als Regelgröße von der Steuereinrichtung 1200 und deren Erregerstromregelung 1220 erfasst wird, kann der Erregerstrom I_exc gemäß dem Anforderungssignal 21 bzw. der Anforderung entsprechend geregelt werden.The control device 1200 comprises an excitation current control 1220. This has a controller unit 1221, which outputs a target current signal I _exc, target based on the detected request signal 21. A signal generation device 1222 of the excitation current control 1220 generates the control signals StS _T1 and StS _T4 to match this. 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 as IGBTs, for example. Starting from the supply voltage U _DC, an average voltage U _exc = U _DC (2DC-1). DC is the duty cycle. The resulting average excitation current $I_{\overline{exc}}=\frac{U_{\overline{exc}}}{R_{exc}}=\frac{U_{DC}\left(2DC-1\right)}{R_{exc}}.$

Here, R _exc indicates the ohmic resistance of the load branch or the excitation winding 460. The excitation current I _exc can be regulated according to the request signal 21 or the request via the excitation current I _exc measured by the current measuring device 480, which is recorded as a controlled variable by the control device 1200 and its excitation current control 1220.

The detection unit 1240 is described below together with the 6 A method for detecting a permanently conductive defective semiconductor switch in the half-controlled H-bridge 100 is described in more detail in the schematic flow diagram shown in FIG. records actual values I _exc measured in chronological order by the current measuring device 480 2020, which are referred to as excitation current measured values. As can be seen from the graphs 640, 840 of the 2 and 3 The excitation current I _exc changes continuously.

The detection device is designed to determine the change in sign of the excitation current change based on the recorded excitation current measured values I _exc . In one embodiment, the sign of the excitation current change is thus determined iteratively based on the recorded excitation current measured values.

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

Alternatively or additionally, the sign can be derived from a difference between the excitation current measured values recorded one after the other. The sign of the excitation current change corresponds to the sign of the difference determined if the excitation current measured value recorded earlier in time is used as the subtrahend and the excitation current measured value recorded subsequently is used as the minuend.

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 measurement values are recorded in a time-correlated manner with the nominal switching operations, i.e. in a time-correlated manner with the nominal switching times of the control cycles.

For this purpose, it can be provided that the detection device receives synchronization information or generates it itself 2010. The detection device can receive the control signals for the two semiconductor switches as synchronization information, for example. 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.

Preferably, the excitation current measurement values are correlated in time, preferably recorded immediately at the switching time or shortly before or shortly 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 at the same time as the nominal switching processes.

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

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

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

In this case, the recorded excitation current measured values, each of which belongs to the same switching operation, i.e. to one of the switching operations T1 _on , T4 _off , T4 _on and T1 _off , can be averaged over a predetermined number of switching cycles, for example 100 switching cycles, and the averaged excitation current measured values can be 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, whereby the excitation current measured value differences that occur between the same nominal switching operations in the switching cycles are again averaged.

Alternatively, it is also possible to average the signs that are determined from non-averaged excitation current measurements. In this case, the signs of the current changes that correspond to one another 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. In order to avoid an uncertain statement with the value "0", the averaging can be carried out over an odd number of signs.

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

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

When a fault signal is generated, an embodiment identifies the continuously conducting defective semiconductor switch 2080. This is the semiconductor switch after whose nominal switching operation no sign change is detected.

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.

In the example shown, it is determined that after the nominal switching operations T1 _on and T1 _off, no change in the sign of the current change occurs. The signs of the excitation current changes or the excitation current differences ΔI _{exc_T1_off} = I _{exc_T1_off} - I _{exc_T4_on} , ΔI _{exc_T1_on} = I _{exc_T1_on} - I _{exc_T1_off} and ΔI _{exc_T4_off} = I _{exc_T4_off} - I _{exc_T1_on} are all positive. After the nominal switching operations of the other semiconductor switch T4 340, a change in sign occurs in each case. This other semiconductor switch T4 340 is therefore intact. The other one of the semiconductor switches T1 210 is therefore defective.

As an alternative to the individual determination of the sign 2041 of the excitation current change for individual time periods, the sign change 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.

However, the frequency can also be determined in other ways, for example by using 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 recorded excitation current measurement signal, it can be determined from the ratios of the signal strengths of the signals passed through the two filters whether the excitation current change occurs at twice the frequency of the control signals or the frequency of the control signals.

Checking whether the sign changes correlate with the switching operations of the semiconductor switches of the half-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 are correlated with the switching operations. A sign change takes place 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 conductively defective.

A determination of which of the semiconductor switches is defective can be made by determining a phase position of the excitation current signal relative to one of the control signals 2085. The semiconductor switch that is defective 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, but the signal generating device 1222 is then defective.

Both the excitation current control 1220 and the detection device 1240 are preferably implemented by means of a program-controlled microprocessor 1260 on which program code 1270 is executed, which is stored in the memory device 1250. It is understood that other embodiments may also have dedicated circuits that implement the excitation current control and/or the detection device.

It will be understood by those skilled in the art that only an exemplary embodiment is described here.

list of reference symbols

1

motor vehicle

10

separately excited synchronous machine

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 device

190

voltage measuring device

200

bridge branch

201

an end

202

opposite end

210, T1

semiconductor switches

230, D3

diode

250

connection point

300

another bridge branch

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

intermediate circuit capacitor

601

Time

610

graph

611

control signal (T ₁ )

612

level

615

phase position

620

graph

621

control signal (T ₄ )

625

phase position

622

level

630

graph

631

excitation voltage

640

graph

641

excitation current

650

graph

651

total current

801

Time

810

graph

811

control signal (T ₁ )

812

level

813

switching state

820

graph

821

control signal (T ₄ )

822

level

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

QUOTES INCLUDED IN THE DESCRIPTION

This list of documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions.

Cited patent literature

• DE 10 2014 102 869 A1 [0014]
• EP 3 208 922 A1 [0015]
• DE 10 2016 203 355 A1 [0016]
• DE 10 2014 018 665 A1 [0017]

Claims (11)

Method for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches (210, 340) of a half-controlled H-bridge circuit (100) for setting an excitation current of an externally excited synchronous machine, wherein the half-controlled H-bridge circuit (100) comprises 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 another of the semiconductor switches T4 (340) and a load branch (400) comprising an excitation winding (460) of the externally excited synchronous machine, wherein the load branch (400) has 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 half-controlled H-bridge circuit (100) is coupled to a control device (1200) which comprises 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 which 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 which is used by the excitation current control (1220) as a feedback signal, characterized in that 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 (T1 _on , T4 _off , T4 _on , T1 _off ) of one of the semiconductor switches (210, 340), 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 _on , T4 _off , T4 _on , T1 _off ) 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). procedure according to claim 1 , characterized in that 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 and evaluated as a difference between the excitation current measurement values recorded in the control cycles immediately following one another correlated with the nominal switching times of the semiconductor switches (210, 340). Method according to one of the preceding claims, characterized in that an averaging is carried out when determining the sign of the current change of the excitation current. procedure according to claim 3 , characterized in that the excitation current measured values, which are recorded correlated with, in particular 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. procedure according to claim 3 , characterized in that the determined differences, which are determined on the basis of excitation current measured values which are each recorded in correlation with one another in different switching cycles corresponding to nominal switching times, are averaged. Method according to one of the preceding claims, characterized in that the semiconductor switch is determined to be defective, with whose nominal switching operations no change in the sign of the excitation current change occurs procedure according to claim 1 , characterized in that the sign change 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 sign change with the switching operations exists when a determined frequency or main frequency component has at least approximately twice a 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 at least approximately corresponds to the frequency of the pulse-width modulated control signals. procedure according to claim 7 , characterized in that the semiconductor switch which 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 offset. Control device (1200) with detection device (1040) for detecting a single permanently conductive defective semiconductor switch of the semiconductor switches (210, 340) of a half-controlled H-bridge (100) for setting an excitation current of an externally excited synchronous machine, wherein the half-controlled H-bridge (100) has 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 another of the semiconductor switches T4 (340) and a load branch (400) comprising an excitation winding (460) of the externally excited synchronous machine, wherein the load branch has 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 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 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 in 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 changes its sign in a temporally correlated manner with each of the nominal switching times (T1 _on , T4 _off , T4 _on , T1 _off ) of one of the semiconductor switches (210, 340), and if this is not the case, to output an error signal which indicates a defect in one of the semiconductor switches (210, 340), wherein a nominal switching time (T1 _on , T4 _off , T4 _on , T1 _off ) is a switching time at which one of the two semiconductor switches (210, 340) at a intact half-controlled H-bridge circuit (100) changes its switching state. Control device (1200) according to claim 9 , characterized in 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 the Claims 1 until 8 be executed. Computer program product comprising program code which, when executed on a microprocessor in cooperation with the control device (1200) according to one of the Claims 9 or 10 and the half-controlled H-bridge circuit (100) described therein, the method according to one of the Claims 1 until 8 executes.

Images