Method for synchronizing control electronics to a rotating motor with sinusoidal control

Abstract

Method for synchronization in the sinusoidal control of already moving permanent magnet synchronous machines (PMSM), wherein the PMSM motors are controlled with at least one half-bridge via at least one controlled connecting line each, and wherein a dead time (6) is provided in the associated control signal when controlling at least one half-bridge, and a. where at the beginning of the process the machine connections are high-resistance and the machine is already rotating either by external drive or by its mass inertia and b. wherein the relative proportion of the dead time (6) in a control period (T) at the beginning of the synchronization process is different from the relative proportion of the dead time (6) in a control period (T) at the end of the synchronization process or c. wherein the relative proportion of the dead time (6) in a control period (T) at an earlier time of the synchronization process than the end of the synchronization process is different from the relative proportion of the dead time (6) in a control period (T) at the end of the synchronization process d. wherein, in the course of the synchronization, the relative proportion of the dead time (6) in the respective control periods (T) of at least one output signal of at least one half-bridge is gradually reduced until it has reached or fallen below a predetermined target value and e. wherein the relative proportion of the dead time (6) in a control period (T) of at least one output signal of at least one half-bridge is greater at the beginning of the synchronization process than at the end of the synchronization process.

Description

Introduction

[0001] The control of permanent magnet synchronous machines (PMSMs) is a task that always requires the generation of a rotating field that has a fixed relationship to the rotor field in terms of its phase position, amplitude, and field frequency. A number of methods for generating a corresponding rotating field exist in the state of the art. As arbitrary examples from the state of the art (SdT), DE 10 393 516 T5 and DE 10 2005 035 073 A1 can be cited as relevant documents regarding sinusoidal control. These solve the control task more or less effectively during steady-state operation, during load and speed changes, and during starting from standstill.

[0002] However, a major problem in the current state of the art is still synchronizing to a machine that is already rotating. In this operating case, no control takes place. The machine's connections have high resistance and the machine is already rotating, either due to an external drive or due to its inertia. Such operating cases occur, for example, in wind turbines, cooling fans or motors that drive vehicles. In the former case, the wind turbine is already set in motion, for example, by the wind. With cooling fans, this can happen, for example, through the airflow from the vehicle. In electric vehicles, this can be the operating case when the electric drive is switched on while the vehicle is already moving.

[0003] The problems described below typically occur when synchronizing to a rotating machine with sinusoidal control. This can occur, for example, with cooling fans, air conditioning fans, and various pumps, but also in other applications such as air conditioning damper controls in residential areas.

[0004] Generating a phase and frequency correct control is not a problem in itself, since both can be determined with sufficient accuracy either by evaluating sensor signals or, for example, measured voltages.

[0005] However, generating the correct amplitude for a smooth transfer of motor control by the control circuit remains a problem. Especially in high-power machines with low internal resistance, even very small deviations in the control amplitude compared to the amplitude specified by the motor lead to strong current increases during the transition from the high-impedance state of the motor to the normal controlled state. The steep current increase, in turn, leads to strong voltage changes in the supply voltage as well as to strong torque changes and possibly to noise, which is disruptive, for example, in automotive applications. The resulting parasitic strong torque changes can, in turn, lead to damage to the mechanical system connected to the motor. In existing control systems, for example,By measuring the amplitude or calculating it from the field speed using constants stored in the system's memory, an attempt is made to generate the most accurate amplitude possible at the time of takeover. However, this is not always successful, which leads to the aforementioned, sometimes unacceptable effects.

[0006] From DE 10 2008 059 052 A1 a method is known with which the rotor position of a synchronous machine with a permanently excited rotor is determined by measuring the stator currents and voltages by measuring the stator voltages during a measuring window (reference symbol T M DE 10 2008 059 052 A1). During this measurement window, the stator current is suspended. A person skilled in the art will preferably determine the position of this measurement window so that the rotor position is measured near the zero crossing of the stator current, i.e., during the dead time that is required anyway.

[0007] From DE 10 2010 030 239 A1, a method for starting a synchronous machine with a permanently excited rotor is known, in which the absolute dead time, e.g. measured in ms, changes, but not the relative dead time related to a control period. Due to the circuit diagram used for the non-sinusoidal control ( Fig. 2 of DE 10 2010 030 239 A1), the dead time used there is always one third of the control period, so that only two winding phases are always live (page 5, lines 1 and 2, left column of DE 10 2010 030 239 A1).

[0008] The state of the art involves measuring the properties of the motor and its current supply or its electromotive force, in which the rotor position can be measured by measuring the EMF by briefly not supplying current, i.e. a short-term, abrupt jump in the dead time to 100%.

[0009] Such a method is known from DE 10 2013 218 041 A1. However, the method of DE 10 2013 218 041 A1 uses block commutation (see Fig. 3 of DE 10 2013 218 041 A1), i.e., it does not provide sinusoidal control and is therefore not suitable for solving the task of synchronizing a sinusoidal control of PMSM machines during start-up. Due to the block commutation, DE 10 2013 218 041 A1 generates more harmonics than sinusoidal commutation and is therefore unsuitable for many applications. In the method of DE 10 2013 218 041 A1, a dead time T1 for block commutation is introduced in the region of the zero crossing of the sinusoidal EMF of the motor, during which the bridge is switched to high resistance for an extended period. The block commutation used in DE 10 2013 218 041 A1 differs from a sinusoidal control in the prior art in that the control period of the block commutation is equal to the rotation period of the EMF, which in turn is typically equal to the rotation period of the motor.After zero-crossing detection, according to the method of DE 10 2013 218 041 A1, a second dead time T2 is applied, which corresponds to the target dead time at zero crossing. In this way, the method of DE 10 2013 218 041 A1 ensures that the driving half-bridge remains at high resistance for a certain time, the current is zero, and the EMF voltage can be reliably observed during this period. The process is repeated periodically. This dead time enables optimal measurement of the EMF during block commutation. Using an iterative process, the dead time of block commutation is increased at zero crossing according to DE 10 2013 218 041 A1. In the method of DE 10 2013 218 041 A1, the motor current remains sufficiently limited if the motor voltage always remains within the range defined by the smaller dead time (reference symbol T2 of DE 10 2013 218 041 A1).The temporary increase in the zero crossing range during block commutation according to DE 10 2013 218 041 A1 does not support synchronization during startup, because in this range the deviation between the generated voltage and the actual voltage must remain relatively small due to the short dead time. However, the greatest deviation is to be expected precisely in the area of ​​the EMF maximums. This is precisely the case that should be avoided.

[0010] DE 19 846 831 A1 discloses a method for sensorless determination of the rotor position with sinusoidal control by generating artificial current gaps for EMF measurement. However, it does not solve the problem of how synchronization is achieved with an already moving PMSM when the rotor position is initially unknown. The claim set claimed here addresses precisely this problem by a targeted, step-by-step adjustment of the dead time for synchronization, which is not provided for in DE 19 846 831 A1.

[0011] With regard to the differences between the control of permanent magnet synchronous machines (PMSM) and permanent magnet brushless DC motors (BLDC), reference is made here to the dissertation "Electrical drives with permanent magnet excited machines in dynamic sensorless operation" by Bassel Sahhary, submitted to the Faculty of Electrical Engineering at the Helmut Schmidt University of the Federal Armed Forces Hamburg, Hamburg 2008, and specifically to its chapter 2.1.

[0012] The previously described problem of starting PMSM machines from standstill has not yet been solved for sinusoidal control. Furthermore, the known state of the art does not provide any indication of a solution to the described problem. Object of the invention

[0013] The object is to create a method for synchronizing an already running PMSM machine with sinusoidal control that does not exhibit any of the aforementioned problems and thus avoids them. This object is achieved by a method according to claim 1, a control system according to claim 13, and a machine according to claim 14. Description of the invention

[0014] In contrast to the prior art, the invention allows for trouble-free synchronization of a sinusoidal control to an already rotating PMSM machine. Furthermore, the otherwise necessary parameterization of the motor constant in the control system can be eliminated. The invention is explained below.

[0015] The essential inventive concept disclosed in this description is to achieve the above-described problem by using a special method that changes the dead time during synchronization by changing its relative proportion of the dead time in a control period. A control period has a temporal length T that corresponds to the inverse of the control frequency. The term control period is known in the prior art, for example, from DE 10 2010 038 557 A1. In the prior art, the term control period typically refers to the PWM period or the period of a space vector modulation control (SVM). In the case of sinusoidal space vector modulation, as known, for example, from DE 10 2013 017 654 B3, this is always shorter than the motor's rotation period according to the prior art, which must be distinguished from the latter.If the control period is equal to the rotation period, this control is referred to as block commutation according to the state of the art, which differs from the sinusoidal control according to the state of the art.

[0016] This is explained in more detail below: When controlling half-bridges, a dead time is used to prevent cross-current flow through the high-side driver transistor and the low-side driver transistor during the switching transition phases. Such cross-current would lead to unnecessary heating and / or even destruction of the half-bridge, which could only be avoided by a more resilient design and thus an increase in the cost of the half-bridge. This dead time, during which control is suspended to avoid cross-current flow, is minimized relative to the control period according to the state of the art, as it causes deviations between the desired and the actual duty cycle. The deviation of the actual duty cycle, in turn, depends on the load current direction.

[0017] When dead time is mentioned below, this always refers to the relative proportion of this dead time relative to the control period. It is therefore preferably expressed as a percentage of the control period.

[0018] The synchronization method according to the invention exploits this effect. At the beginning of the process for synchronizing the control to a machine that may already be rotating, it deliberately generates a significantly increased dead time, unlike prior art methods that typically minimize dead time. The "error" in the duty cycle generated as a result according to the prior art is used in the method as the "uncertainty" of the output voltage. This "uncertainty" is set in the method so that at the moment of synchronization it is at least as large as the uncertainty that the output voltage to be generated already has, e.g. due to measurement errors and parameterization errors. The aim is therefore to use an artificially greatly increased dead time to generate an output voltage with an "uncertainty range" such that the actual, exact output voltage to be generated lies within this "uncertainty range".The dead time is then slowly reduced to the minimum possible value.

[0019] This minimum possible value is detected when the value of a register containing the current dead time no longer exceeds a previously parameterized value stored elsewhere, e.g., in another register. This parameterized value depends only on the dimensioning of the half-bridges and not on the load or motor. Of course, it is also conceivable, for example, to set this target value externally or permanently using analogue technology and to compare this externally specified, previously parameterized value with the value of the register. Instead of a digital solution, a purely analog solution is also possible.

[0020] The slow reduction enables a continuous takeover with a smooth current and torque curve.

[0021] In addition, this method allows the elimination of the complex parameterization of the motor, so that, for example, without calibration processes in production or during commissioning, the approximate amplitude can be measured simply while the motor is rotating and the synchronization can be started with a correspondingly very long dead time.

[0022] The overall system thus shows a continuous torque and current curve when synchronizing the control to a rotating machine

[0023] In addition, some machine-specific parameterizations can be eliminated and replaced by a universal parameterization of the dead time generation.

[0024] In typical applications, the method according to the invention is preferably carried out in such a way that initially only the uncertainty of the amplitude is compensated. Typically, phase synchronization is a minor problem. This phase synchronization can be achieved, for example, by time measurements between the zero crossings of the measured voltages at the motor terminals. Nevertheless, an increased dead time is also very well suited to compensating for uncertainties in the phase. This is because it can be assumed that a motor that is already rotating will definitely change its speed. This can be caused, for example, by an external braking torque or even torque fluctuations. Typically, such an output signal can then only be generated after a corresponding time measurement. As a result, a small phase difference between the output voltage and the motor voltage can be expected.

[0025] It is obvious to the person skilled in the art that, based on the method, it is possible to operate a device that executes the method described above. In this case, the reduction of the dead time (6) of an output signal of a half-bridge and the correction of the phase / amplitude of this output signal can occur simultaneously.

[0026] In this way, a control system for a permanent magnet synchronous machine can be constructed in which the reduction of the dead time of an output signal of a half-bridge and the correction of phase / amplitude of this output signal occur simultaneously.

[0027] This control can be part of an electrical machine, which is typically a permanent magnet synchronous machine.

[0028] The invention is further explained with reference to the attached exemplary schematically simplified figures.

[0029] The figures show: Fig. 1 shows a comparison between EMF of the machine and generated voltages at the beginning of synchronization according to the state of the art Fig. Figure 2 shows the generated signals with dead time according to the state of the art (Note: For better visibility and better understanding by the reader, the dead time is greatly enlarged compared to reality and is therefore not shown to scale). Fig. Figure 3 shows the uncertainty range of a generated output voltage caused by the dead time according to the state-of-the-art technology. (Note: For better visibility and better understanding by the reader, the uncertainty range is greatly enlarged compared to the signal and is therefore not shown to scale.) Fig. 4 shows a signal generated according to the invention at the beginning of the synchronization Fig. 5 shows a signal generated according to the invention at the end of the synchronization Figure 1

[0030] Fig. Figure 1 shows the comparison between the first curve (1) of the electromotive force (EMF) of the machine and the voltages generated at the beginning of synchronization according to the state-of-the-art technology. The second curve (2) represents a voltage with a slightly too large amplitude. This voltage would lead to strong acceleration with a correspondingly high current and a strong torque change at the beginning of synchronization.

[0031] The third curve 3 represents a voltage with a slightly too low amplitude. This would lead to a large, undesirable braking torque at the beginning of synchronization. As a result, a high, undesirable current also arises through the driving half-bridge (not shown). The two operating cases, represented by the second curve (2) and third curve (3), thus lead to thermal stress on the driving half-bridges in the state of the art, which is why they must be designed larger to avoid damage. In addition, increased mechanical stress can arise from the resulting torque changes, which must also be absorbed by increased and therefore more expensive dimensioning to avoid damage. Figure 2

[0032] Fig. Figure 2 illustrates the generation of exemplary control signals for an exemplary half-bridge of the power circuit according to the state of the art for controlling the machine. The half-bridge should consist of a power switch for the positive supply voltage, the high-side driver, and a power switch for the negative supply voltage, the low-side driver. In this example, the power switch for the positive supply voltage switches the positive supply voltage to the output of the half-bridge when its control signal (4) is positive. Fig. 2, the fourth curve (4) represents the control signal for the power switch to provide a positive supply voltage at the output of the half-bridge. In this example, the power switch for the negative supply voltage switches the negative supply voltage to the output of the half-bridge when its control signal (5) is positive. Fig. 2, the fifth curve (5) thus represents the control signal for the power switch to provide a negative supply voltage at the output of the half-bridge.

[0033] The seventh curve (7) is an example of the resulting voltage curve of the output voltage of the half-bridge with an output current flowing out of the half-bridge. The eighth curve (8) is an example of the resulting voltage curve of the output voltage with an output current flowing into the half-bridge.

[0034] Between the control pulses of the half-bridge, there is a period of time in which neither the control signal of the power switch for the negative supply voltage (low-side driver) nor the control signal of the power switch for the positive supply voltage (high-side driver) are active. This period is the dead time (6). It is of course conceivable, but typically not sensible, for a residual cross-current other than zero to flow through the bridge during this dead time (6), or for one of the two switches to not be completely turned off. The dead time (6), as an undesirable side effect according to the state of the art, ensures that there are time intervals (9) in which the instantaneous output voltage corresponds to either the positive or the negative supply voltage, depending on the sign of the output current. If the output current is zero, the output voltage can assume any value during these time intervals. Figure 3

[0035] Fig. Figure 3 shows the curve of a generated output voltage in the presence of a dead time according to the state of the art. It contains an "uncertainty region" (10), which is undesirable according to the state of the art and surrounds the actual output voltage, which, since it is unknown, is not shown. For better representation, the uncertainty region is greatly enlarged compared to the generated signal and is not shown to scale. Within this uncertainty region, the output voltage can assume any value depending solely on the output current at the respective time. If the current direction out of the half-bridge were always positive, the actual output voltage would typically run along the upper limit of the hatched region (10). If the current direction were always negative, the actual output voltage would typically run along the lower limit of the hatched region (10).

[0036] A change in current direction leads to a change in the actual curve between the lower and upper limits. Values ​​within the hatched area are assumed when the output current is zero or during a current direction transition. Figure 4

[0037] The method according to the invention exploits this previously undesirable behavior to generate a gently increasing motor current and thus a gently increasing torque during synchronization. For this purpose, a very long dead time is selected at the beginning of synchronization. This ensures a correspondingly large uncertainty range (11). This uncertainty range (11) is generated according to the invention such that the actual EMF of the motor (1) always lies within the uncertainty range (11). If the amplitude of the EMF is completely unknown, it is even possible to start with a dead time of 100%. The synchronization process can therefore be started for the affected output signal of the respective half-bridge with a dead time of 100% or more than 99% and / or 98% and / or 96% and / or 90% and / or 80% and / or 60% and / or 40% and / or 10% and / or 5%.

[0038] According to the invention, the uncertainty range can also be used to include remaining uncertainties regarding the phase of the output voltage to be generated.

[0039] In the course of synchronization, the dead time is reduced slowly and gradually according to the invention, in contrast to the prior art, until it reaches the target value.

[0040] According to the invention, the amplitude and / or phase of the output voltage can also be continuously corrected until a certain load current flows or a certain load torque is generated before the dead time is reduced. This can be done in such a way that no sign changes are generated in the load torque during the dead time reduction.

[0041] In addition, the dead time reduction can also be interrupted at any time to adjust amplitude / phase and then further reduce the dead time.

[0042] According to the invention, the reduction of dead time and the correction of phase / amplitude can also be carried out simultaneously. Figure 5

[0043] Fig. Figure 5 shows the EMF curve (1) and the generated signal at the end of the synchronization process (12). The dead time is reduced to a typically specified minimum possible value (target value). This target value is typically specified by the controller design and / or by an external control signal and / or by programming.

[0044] The remaining uncertainty range at the end of the synchronization process (12) is correspondingly small. The control generates a signal that is appropriate for the speed and load without causing significant torque or current fluctuations.

Claims

[1] Method for synchronization in the sinusoidal control of already moving permanent magnet synchronous machines (PMSM), wherein the PMSM motors are controlled with at least one half-bridge via at least one controlled connecting line each, and wherein a dead time (6) is provided in the associated control signal when controlling at least one half-bridge, and a. where at the beginning of the process the machine connections are high-resistance and the machine is already rotating either by external drive or by its mass inertia and b. wherein the relative proportion of the dead time (6) in a control period (T) at the beginning of the synchronization process is different from the relative proportion of the dead time (6) in a control period (T) at the end of the synchronization process or c. wherein the relative proportion of the dead time (6) in a control period (T) at an earlier time of the synchronization process than the end of the synchronization process is different from the relative proportion of the dead time (6) in a control period (T) at the end of the synchronization process d. wherein, in the course of the synchronization, the relative proportion of the dead time (6) in the respective control periods (T) of at least one output signal of at least one half-bridge is gradually reduced until it has reached or fallen below a predetermined target value and e. wherein the relative proportion of the dead time (6) in a control period (T) of at least one output signal of at least one half-bridge is greater at the beginning of the synchronization process than at the end of the synchronization process. [2] Method according to one of the preceding claims, wherein the relative proportion of the dead time (6) in a control period (T) for at least one control signal of at least one half-bridge is set such that the resulting uncertainty range (11, 12) of the output signal of this half-bridge is generated such that the actual EMF of the motor (1) always lies within this uncertainty range (11, 12). [3] Method according to claim 3, wherein a relative proportion of the dead time (6) in a drive period (T) is started for at least one output signal of at least one half-bridge with a relative proportion of more than 99% and / or 98% and / or 96% and / or 90% and / or 80% and / or 60% and / or 40% and / or 10% and / or 5%. [4] Method according to one of the preceding claims, wherein remaining uncertainties with regard to the phase of an output voltage to be generated of at least one output signal of at least one half-bridge are enclosed by an uncertainty range (11, 12) by a correspondingly large selected relative proportion of the dead time (6) in a drive period (T) of this output signal of this half-bridge. [5] Method according to one of the preceding claims, wherein the dead time (6) is reduced slowly and gradually during the method to the minimum possible value. [6] Method according to claim 1, wherein the target value is specified by the design of the controller and / or by an external control signal and / or by programming. [7] Method according to one of the preceding claims, wherein first the amplitude and / or the phase of the output voltage of at least one output signal of at least one half-bridge is corrected until a load current whose magnitude or amplitude is greater than a predetermined load current flows or a torque occurs which has a magnitude greater than a predetermined torque, before the relative proportion of the dead time (6) in a drive period (T) of this output signal of this half-bridge is reduced. [8] Method according to claim 7, wherein the relative proportion of the dead time (6) in a control period (T) of at least one output signal of at least one half-bridge is reduced such that during the reduction of the relative proportion of the dead time (6) in a control period (T) of this output signal of this half-bridge no sign changes in the torque are generated. [9] Method according to one or more of the preceding claims, wherein the reduction of the relative proportion of the dead time (6) in a drive period (T) of at least one output signal of at least one half-bridge is interrupted at at least one point in time and the amplitude and / or phase of this output signal of this half-bridge is adjusted after this interruption and the relative proportion of the dead time (6) in a drive period (T) of this output signal of this half-bridge is further reduced after this adjustment. [10] Method according to one or more of the preceding claims, wherein the reduction of the relative proportion of the dead time (6) in a drive period (T) of at least one output signal of a half-bridge and the correction of phase / amplitude of this output signal take place simultaneously. [11] Control for a permanent magnet synchronous machine which carries out a method according to one or more of the preceding claims. [12] Control for at least one half-bridge, which is suitable and intended to carry out a method according to one or more of claims 1 to 10 as part of a device according to claim 11. [13] An electrical machine which is a permanent magnet synchronous machine, comprising a control according to claim 11 or at least a control according to claim 12.

Images