Method for controlling power semiconductors in an inverter

Abstract

Method for controlling power semiconductors in an inverter by means of a microprocessor controlling pulse width modulation (PWM), the method comprising the following steps: i Creating, calculating and / or selecting a numerical value as the first factor from a defined range of numbers; ii Calculating a clock frequency A of the pulse width modulation (PWM) by multiplying the first factor by the value of a basic switching frequency; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, calculation of a new clock frequency A by multiplying a newly created, calculated and / or selected numerical value from the defined numerical range as a new factor with the basic switching frequency; iv multiple repetition of step iii, in that after the expiry of a number X of clock cycles or after the expiry of a specific period T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the basic switching frequency by a newly created / calculated and / or selected numerical value; and v Selecting a probability density for the clock frequencies to be selected depending on the application mode or operating mode, including switching between the probability densities, omitting or immediately aborting certain clock frequencies known or identified as interfering frequencies and recalculating the clock frequency.

Description

[0001] The invention relates to a method for controlling power semiconductors in an inverter. The invention serves to improve electromagnetic compatibility (EMC) and is particularly applicable in electric refrigerant compressors of motor vehicles.

[0002] In the electric refrigerant compressors of a motor vehicle, a multi-phase, usually three-phase, rotating field is generated with the help of an inverter. This is achieved by applying a sine or sine-like current from the battery's direct current (DC) to the motor inductance via the periodic switching of power semiconductors in the inverter. State-of-the-art technology is to drive the power semiconductors with a constant clock frequency. Depending on the power class, a frequency between 5 and 30 kHz is selected for this purpose. The current and motor speed are regulated by controlling the switch-on times of the individual power semiconductors using pulse-width modulation.Pulse width modulation (PWM) generally refers to a type of modulation in which a technical quantity, for example an electrical voltage, changes between two values ​​and the duty cycle, i.e. the duration of a switch-on time in a time cycle (period), can change between these values.

[0003] The switching of power semiconductors generates interference at the inverse of the period. This frequency can be referred to as the fundamental frequency. Likewise, the switching of power semiconductors generates interference at the harmonics, multiples of the fundamental frequency, and thus has a significant impact on electromagnetic compatibility. Electromagnetic compatibility (EMC) refers to the ability of a technical device not to interfere with other devices through unwanted electrical or electromagnetic effects, or to be interfered with by other devices.

[0004] WO 2011 / 131 201 A1 describes a method for controlling a switching power supply, in which the switching power supply comprises a circuit for receiving a supply voltage and for outputting a switched voltage and a converter circuit which is coupled to the circuit and is controlled by the switched voltage to generate a required output voltage, the method comprising the following steps: • Providing a drive signal for the circuit to regulate the switching frequency of the circuit, • Applying spread spectrum modulation to the drive signal to reduce electromagnetic interference in the power supply, the method further comprising the following steps: • Monitoring the supply voltage for the circuit during operation of the circuit and • Stopping the operation of the circuit when the monitored supply voltage falls below an operating voltage threshold.

[0005] From US 2007 10 047 272 A1, a method based on pulse width modulation (PWM) is known in which clock frequencies are generated randomly, which are selected from frequency registers by means of a multiplexer and accordingly control an oscillator.

[0006] The object underlying the invention is to reduce EMC interference occurring when controlling power semiconductors in an inverter.

[0007] The object of the invention is achieved by a method having the features according to claim 1. Further developments are specified in the dependent claims.

[0008] The method according to the invention for controlling power semiconductors in an inverter is carried out by means of a microprocessor controlling pulse width modulation (PWM), the method comprising the following steps: i Creating, calculating and / or selecting a numerical value as the first factor from a defined numerical range, preferably from 0.1 to 1, particularly preferably between 0.3 and 1; ii Calculating a clock frequency A of the pulse width modulation by multiplying the first factor by the value of a basic switching frequency; iii after a number X of PWM clock cycles have elapsed or after a certain period of time T with the previously calculated clock frequency A, calculating a new clock frequency A by multiplying a newly created, calculated and / or selected numerical value from the defined numerical range as a new factor with the basic switching frequency; iv multiple repetition of step iii, wherein after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the basic switching frequency by a newly created / calculated and / or selected numerical value.

[0009] To implement random pulse width modulation (PWM), the clock frequency in the microprocessor, which controls the power semiconductors via driver circuits, is changed using the method described above.

[0010] According to a first embodiment of the invention, the method comprises the following steps i-iv: i Creating a lookup table with a number Y of random numbers from the defined number space, preferably by means of a random generator, whose probability density within the defined number space corresponds to a Gaussian distribution or a uniform distribution, and selecting the first random number of the table as the first factor; ii Calculating the clock frequency A of the pulse width modulation by multiplying the value of the fundamental switching frequency by the first factor; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, calculating a new clock frequency A by multiplying the next random number from the table by the value of the basic switching frequency; iv multiple repetition of step iii, in that after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency by the value of the random number following in the table, up to the end of the table.

[0011] According to an advantageous development of this embodiment, steps ii to iv are repeated after reaching the end of the table. This means that after the table has been traversed from beginning to end, the system jumps back to the beginning of the table after using the last random number.

[0012] According to a second embodiment of the invention, the method comprises the following steps i-iv: i Calculating a number sequence R limited within the defined number space using a mathematical formula and calculating a first element of the number sequence R or generating a random number using a random generator implemented in the microprocessor, in each case during the runtime of the microprocessor; ii Calculating the clock frequency A of the pulse width modulation by multiplying the value of the basic switching frequency by the first calculated element of the number sequence R or the first generated random number; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, calculating a new clock frequency A by multiplying the value of the basic switching frequency by the value of a next calculated element from the number sequence R or a next generated random number; iv multiple repetition of step iii, wherein after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency by the value of a new element of the number sequence R or a new random number.

[0013] Preferably, the mathematical formula contains one or more non-linear functions, such as SIN, COS and / or TAN and / or their inverse functions ARCSIN, ARCCOS, and / or ARCTAN and / or log and / or e.

[0014] The calculation of the numerical sequence R and / or its elements or the random number is preferably performed either in a separate unit or in the microprocessor controlling the pulse-width modulation. Particularly preferred is the calculation of the numerical sequence and / or its elements or the random numbers using an external input variable. A random process, preferably noise, such as resistance noise or shot noise, or a temperature profile can be used as the input variable for calculating the numerical sequence and / or its numerical values ​​or the random numbers. However, the calculation can also be performed without an external input variable.

[0015] A random number generator is generally understood to be a process that generates a sequence of random numbers. A fundamental distinction is made between non-deterministic and deterministic random number generators. Non-deterministic random number generators, i.e., random number generators that produce different values ​​under identical initial conditions, include, in particular, physical random number generators, which use physical processes, such as thermal resistance noise, to generate random numbers. Deterministic random number generators generate pseudorandom numbers and are therefore usually called pseudorandom number generators. The number sequences consisting of so-called pseudorandom numbers are much easier for computers to generate and are available in virtually all higher-level programming languages. Suitable sources include, for example, MS Excel™, MATLAB™, and the Linux™ kernel.

[0016] According to a third embodiment of the invention, the method proceeds in the following steps i-iv: i Generation of a random input variable for calculating a clock frequency in an electronic circuit with the aid of a random oscillator, so that a random number in the defined number range is available at any time; ii calculating a clock frequency A by multiplying the value of the basic switching frequency by the value of the random number at a first time; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, calculation of a new clock frequency A by multiplying the value of the basic switching frequency by the value of the random number available at that time; iv multiple repetition of step iii, wherein after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency by the value of the new random number present at that time.

[0017] The random input variable can in turn be generated by processing a random physical process, for example resistance noise or shot noise.

[0018] According to the method according to the invention, certain clock frequencies that are known or identified as interference frequencies are omitted or immediately aborted, with the clock frequency being recalculated. Omitting in this context means that individual, specific clock frequencies cannot be calculated / called up / generated. This is advantageous, for example, if the frequency to be omitted or a harmonic thereof would lead to severe interference in the system, for example, in a vehicle.

[0019] As an example, consider a component in a vehicle that malfunctions when exposed to disturbances at a frequency of 50 kHz. The third harmonic of a PWM clock frequency of 16.6 kHz would correspond to this frequency of 50 kHz. To prevent the malfunction, the value that would lead to a clock frequency of 16.6 kHz can be removed from the loopup table.

[0020] The advantage of the invention is that the random or quasi-random selection of the inverter's clock frequency, as described above, smooths out the interference spectrum. While the power density of the interference spectrum remains the same, the number and severity of the interference harmonics decreases significantly. This measure can reduce the requirements of the physical EMC filter of the inverter. This can reduce manufacturing effort, costs, and space requirements.

[0021] According to the invention, the method further comprises an additional step v, in which, depending on the application mode or operating mode, a probability density is selected for the clock frequencies to be selected, wherein switching between the probability densities can also occur. In the first embodiment mentioned, for example, two lookup tables with different probability densities can be stored for this purpose. In the second embodiment mentioned, the formula for calculating the number sequence R and its elements, i.e., quasi-random numbers, changes. In the third embodiment mentioned above, switching in the circuit of the random oscillator is conceivable. The probability density is selected between a uniform distribution and a Gaussian distribution of the clock frequencies, wherein switching between uniform and Gaussian distribution of the clock frequencies is possible.

[0022] Further details, features, and advantages of embodiments of the invention will become apparent from the following description of exemplary embodiments with reference to the accompanying drawings. They show: Fig. 1: EMC measurement report of an inverter, state of the art; Fig. 2: Distribution of the clock frequency over time (uniform distribution); Fig. 3: a schematic representation of a first embodiment of the method for controlling power semiconductors; Fig. 4: a schematic representation of a second embodiment of the method for controlling power semiconductors and Fig. 5: a schematic representation of a third embodiment of the method for controlling power semiconductors and Fig. 6: an EMC interference spectrum with random pulse width modulation (PWM).

[0023] The Fig. Figure 1 shows an EMC measurement report of an inverter switching at a 20 kHz clock frequency, or more precisely, a typical recording of the EMC measurements of an electric air conditioning compressor with a fixed clock frequency. The disturbing harmonics of the clock frequency in the range of 100 kHz to 1 MHz are clearly visible. All harmonics are multiples of the clock frequency. The EMC interference is expressed in decibel microvolts (dBµV).

[0024] According to the state of the art, the clock frequency is constant, for example, with a clock frequency value of 20 kHz. In random mode, i.e., when using random pulse width modulation (PWM), which is used according to the invention, the clock frequency changes within a certain variation range of the percentage deviation of the fundamental frequency at short time intervals, for example, every 50 µs. Fig. In the embodiment shown in Figure 2, the clock frequency varies within a range from a lower clock frequency limit of 16 kHz to an upper clock frequency limit of 20 kHz, with the average clock frequency then being 18 kHz and the maximum deviation being 20%. This results in a clock frequency distributed over time, as in Fig. 2, where a uniform distribution is indicated.

[0025] To implement random PWM, the clock frequency of the PWM is changed randomly or quasi-randomly in the microprocessor that controls the power semiconductors. There are various implementations for this.

[0026] After a first version, which was Fig. 3 is shown schematically, in a first step i, a lookup table with a number Y of random numbers (random values ​​1 to 6) from the defined number range is created, preferably by means of a random generator and particularly preferably in a number range from 0.5 to 1. The probability density of the random numbers corresponds to a Gaussian distribution or a uniform distribution within the defined number range. The first random number from the table is selected as the first factor, and in a second step ii, the clock frequency A of the pulse width modulation PWM is calculated by multiplying the value of the basic switching frequency, for example 20 kHz, by the first factor.After a number X of clock cycles have elapsed or after a specific period of time T with the previously calculated clock frequency A, a third step iii involves calculating a new clock frequency A by multiplying the next random number from the table by the value of the basic switching frequency. A further step involves repeating the third step several times until the end of the table is reached. This means that after a number X of clock cycles have elapsed or after a specific period of time T with a previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency by the value of the next random number in the table (random number N+1). After the end of the table is reached, the above-mentioned steps ii to iv are repeated.This means that after the table has been traversed from beginning to end, after using the last random number in the table, the process returns to the beginning of the table and the subsequent clock frequency is again calculated using the first random number.

[0027] Furthermore, it is possible to provide for a selection of the probability density of the frequencies depending on the application and operating mode. For example, switching between uniform and Gaussian distribution of the factors for calculating the clock frequencies can be performed.

[0028] According to a second variant, which is Fig. As shown schematically in Figure 4, a random number is calculated in a first step during the microprocessor's runtime. A sequence of numbers R limited within the defined number range can be calculated using a mathematical formula, with the random number being determined as an element. The random number can also be calculated using a random generator implemented in the microprocessor, also during the microprocessor's runtime. According to the illustrated embodiment, the sequence of numbers and / or its elements or the random numbers are calculated using an external input variable.

[0029] In the second step, the clock frequency A is then calculated by multiplying the value of the basic switching frequency with the first calculated element from the number sequence R or the first random number generated by the random generator, and the pulse width modulation is output with this clock frequency.

[0030] After a number X of clock cycles have elapsed or after a specific period of time T has elapsed at the previously calculated clock frequency A, a new clock frequency A is calculated by multiplying the value of the basic switching frequency by the value of a next calculated element from the number sequence R or a next random number generated by a random generator, and the pulse width modulation PWM is output at this clock frequency. This process is then repeated as often as desired. This means that after a number X of clock cycles have elapsed or after a specific period of time T has elapsed at the previously calculated clock frequency, a new clock frequency at which the pulse width modulation PWM is output is calculated by multiplying the value of the basic switching frequency by the value of a new element of the number sequence R or a new random number.

[0031] According to a third variant, which is Fig. As shown schematically in Figure 5, a random input variable for calculating a clock frequency in an electronic circuit is generated using a random oscillator, so that a random number within the defined number range is available at any time. Thus, at a first time, a clock frequency A of the pulse width modulation is calculated by multiplying the value of the basic switching frequency by the value of the random number. The pulse width modulation is output at this clock frequency.

[0032] After a number X of clock cycles have elapsed or after a certain period of time T has elapsed with the previously calculated clock frequency A, a new clock frequency A is calculated by multiplying the value of the basic switching frequency by the value of the random number available at that time, whereby the pulse width modulation is output at this clock frequency.

[0033] This process is then repeated any number of times. That is, after a number X of clock cycles has elapsed, or after a specific period of time T at the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the base switching frequency by the value of the new random number available at that time. This new clock frequency is used to output the pulse width modulation.

[0034] The advantage of the invention is that the random selection of the clock frequency smooths out the interference spectrum. The power density of the interference spectrum, determined by the area under the curve, remains the same, but the number and severity of the interference harmonics decreases significantly, as the Fig. 6 compared to Fig. 1. This is the result of the randomization of the PWM, which leads to the distribution of the harmonics.

Claims

[1] Method for controlling power semiconductors in an inverter by means of a microprocessor controlling pulse width modulation (PWM), the method comprising the following steps: i Creating, calculating and / or selecting a numerical value as the first factor from a defined range of numbers; ii Calculation of a clock frequency A of pulse width modulation (PWM) by multiplying the first factor by the value of a basic switching frequency; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, a new clock frequency A is calculated by multiplying a newly created, calculated and / or selected numerical value from the defined number space as a new factor with the basic switching frequency; iv. repeated multiple times of step iii, whereby, after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the basic switching frequency with a newly created / calculated and / or selected numerical value; and v Making a selection of a probability density for the clock frequencies to be selected depending on the application mode or operating mode, whereby switching between the probability densities can also take place, whereby certain clock frequencies that are known or identified as interference frequencies are omitted or immediately aborted, whereby a recalculation of the clock frequency takes place. [2] Method according to claim 1, characterized by , that steps i- iv proceed as follows: i Creating a table with a number Y of random numbers from the defined number space, whose probability density within the defined number space corresponds to a Gaussian distribution or a uniform distribution, and selecting the first random number of the table as the first factor; ii Calculation of the clock frequency A of the pulse width modulation (PWM) by multiplying the value of the basic switching frequency by the first factor; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, a new clock frequency A is calculated by multiplying the next random number from the table by the value of the basic switching frequency; iv. Repeat step iii multiple times, wherein after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency with the value of the random number following in the table, until the end of the table. [3] Method according to claim 2, characterized by , that after reaching the end of the table, steps ii to iv are repeated. [4] Method according to claim 1, characterized by , that steps i-iv proceed as follows: i. Calculation of a number sequence R limited in the defined number space using a mathematical formula and calculation of a first element of the number sequence R or generation of a random number using a random number generator implemented in the microprocessor, each during the runtime of the microprocessor: ii Calculation of the clock frequency A of the pulse width modulation (PWM) by multiplying the value of the basic switching frequency with the first calculated element from the number sequence R or the first generated random number; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, a new clock frequency A is calculated by multiplying the value of the basic switching frequency with the value of the next calculated element from the sequence of numbers R or the next generated random number; iv. repeated step iii multiple times, wherein after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency with the value of a new element of the number sequence R or a new random number. [5] Method according to claim 4, characterized by that the mathematical formula contains one or more nonlinear functions. [6] Method according to claim 4 or 5, characterized by , that the calculation of the number sequence R and / or its elements or the random number takes place either in a separate unit or in the microprocessor controlling the pulse width modulation (PWM). [7] Method according to any one of claims 4 to 6, characterized by that the calculation of the number sequence and / or its elements or random numbers is carried out using an external input variable. [8] Method according to claim 1, characterized by , that steps i-iv proceed as follows: i Generation of a random input variable for calculating a clock frequency in an electronic circuit using a random oscillator, so that a random number in the defined range of numbers is available at any given time, ii Calculation of a clock frequency A by multiplying the value of the basic switching frequency with the value of the random number at a first time point; iii after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency A, a new clock frequency A is calculated by multiplying the value of the basic switching frequency with the value of the random number available at that time; iv. repeated step iii multiple times, wherein after the expiry of a number X of clock cycles or after the expiry of a certain period of time T with the previously calculated clock frequency, a new clock frequency is calculated by multiplying the value of the basic switching frequency with the value of the new random number available at that time. [9] Method according to any one of claims 1 to 8, characterized by, that the probability density is selected between a uniform distribution and a Gaussian distribution of the clock frequencies.

Images