System for sensorless angle estimation for trapezoidal control and method thereof

Abstract

A system (800) and method for sensorless trapezoidal control of a brushless DC motor (802) provides efficient high-torque starting and low-speed operation during motor operation without the use of Hall effect sensors or encoders. The system and method also provide the ability to enhance the signal-to-noise ratio of motor starting and low-speed operation via increasing the supply voltage. A sampling architecture and a current-dependent inductance modeling architecture for the control system are also described.

Description

[0001] This typically involves systems and methods for sensorless angle estimation for controlling brushless DC motors (e.g., three-phase brushless DC motors), and more specifically for trapezoidal control. Background Technology

[0002] Trapezoidal control with Hall effect sensors is a widely adopted commutation technique for three-phase brushless direct current (BLDC) motors because these motors exhibit a trapezoidal back electromotive force (back EMF) waveform. This control method is used in several applications, such as fans, power tools, and compressors.

[0003] A BLDC motor comprises two components that can rotate relative to each other: a rotor and a stator. BLDC motor control may involve determining or estimating the rotational position of the rotor relative to the stator. The stator generates a magnetic field to rotate the rotor, and this magnetic field needs to be 90 degrees to the rotor on average. Various methods for BLDC motor control have mechanisms for determining this relative position. Summary of the Invention

[0004] In one example, a method is useful for sensorless trapezoidal control of a brushless direct current (BLDC) motor. Voltage and total current signals are measured from the BLDC motor, which has a rotor. The voltage and total current signals are processed using a processing circuitry system to estimate the angular position of the rotor or the commutation state of the BLDC motor. This processing uses a nonlinear motor model that provides at least one of voltage offset and voltage thresholds varying according to current level and current slope, or zero-crossing time intervals varying according to current level and current slope. Based on the estimated angular position or commutation state, the motor is controlled during startup.

[0005] Another example provides a system that can be used, for example, for sensorless trapezoidal control. The system includes a measurement circuitry configured to measure a voltage signal and a total current signal from a brushless direct current (BLDC) motor having a rotor. The system also includes a processing circuitry coupled to the measurement circuitry, configured to process the voltage and total current signals to estimate the angular position of the rotor or the commutation state of the BLDC motor using a nonlinear motor model. The nonlinear motor model provides at least one of a voltage offset and a voltage threshold varying according to current level and current slope, or a zero-crossing time interval varying according to current level and current slope. The system further includes a control circuitry coupled to the processing circuitry and configured to control the motor during startup based on the estimated angular position or commutation state. Attached Figure Description

[0006] Figure 1 This is a block diagram illustrating an example system for trapezoidal control using a Hall effect sensor.

[0007] Figure 2 This is a block diagram illustrating an example sensorless system used for trapezoidal control.

[0008] Figure 3 This is a cross-section of an example three-phase brushless DC motor.

[0009] Figure 4 This is a circuit diagram of an example three-phase inverter-motor interconnection.

[0010] Figure 5 It is a graph showing the commutation state of each phase and the corresponding inverse EMF signal.

[0011] Figure 6 This is a block diagram illustrating an example sensorless system used for trapezoidal control.

[0012] Figure 7 This is a block diagram illustrating an example sensorless system used for trapezoidal control.

[0013] Figure 8 This is a block diagram illustrating an example sensorless system used for trapezoidal control.

[0014] Figure 9 This is a graph illustrating the spectral separation of the anti-EMF and inductance-related voltage terms.

[0015] Figure 10 This is a flowchart of an example algorithm used for motor control.

[0016] Figure 11 This is a flowchart of another example algorithm used for motor control.

[0017] Figure 12 This is a timing diagram showing the sequential voltage sampling of an example.

[0018] Figure 13 This is a timing diagram demonstrating parallel voltage sampling in an example.

[0019] Figure 14 It is a graph showing the performance of an example of a method for sensorless trapezoidal control when a load is applied at full speed.

[0020] Figure 15 This demonstrates the effect of applying a load when the motor starts, compared to... Figure 14 A graph showing the performance of an example of the same sensorless trapezoidal control method.

[0021] Figure 16This is a graph showing the performance of an example method for sensorless trapezoidal control when a load is applied at full speed.

[0022] Figure 17 This demonstrates what happens when a load is applied during motor startup, such as... Figure 16 A graph showing the performance of an example method for sensorless trapezoidal control.

[0023] Figure 18 This is a flowchart of an example method for sensorless trapezoidal control. Detailed Implementation

[0024] Figure 1 An example system 100 is depicted for trapezoidal control using Hall effect sensors (not shown) in a motor 102. Three Hall effect sensors output H... a H b and H c Position information is provided for commutation timing. System 100 uses an internal commutation loop cascaded with one or more external loops for current, speed, or phase control. In system 100, the internal loop includes a motor 102 (and specifically its Hall effect sensor), a commutation controller 104, and an inverter 106, while the external loop includes a motor 102, a speed estimator 108, a speed controller 110, and an inverter 106. The speed controller 110 is based on an estimated ω generated from the rotor angle from the speed estimator 108. est And input reference rotor angle ω ref Provides a duty cycle signal V indicating the desired speed of motor 102. pu In applications such as power tools and electric bicycles, an external circuit may not exist. In these applications, the human operator directly controls the duty cycle signal V via throttle input. pu The commutation circuit creates a rotating stator magnetic field vector that guides the rotor, thereby creating positive torque to counteract the resistive load torque. The commutation controller 104 receives its input from Hall effect sensors in the motor 102 to provide gate drive input to the inverter 106.

[0025] Figure 2An example system 200 for sensorless trapezoidal control is depicted. "Sensorless" means the system operates without using Hall effect sensors or encoders to determine the rotor-stator position relationship in a BLDC motor. However, even in a "sensorless" implementation, the voltage or current supplied by the BLDC can be sensed by a circuit system capable of making such measurements. Compared to sensored systems, sensorless implementations reduce system cost and wiring, and provide improved reliability. Similar to system 100, system 200 may also include two loops: one containing motor 202, speed estimator 208, commutation controller 204, and inverter 106, and the other containing motor 202, speed estimator 208, speed controller 210, and inverter 206. Figure 2 In the control architecture described herein that does not have Hall effect sensors, three-phase voltage measurement (in Figure 2 The middle label is V a V b V c The total current I supplied to the motor 202 can be measured from different nodes in the inverter 206. tot This can be provided to estimator 208. Estimator 208 uses these measurements to calculate the commutation timing. In the sensorless drive system 200, the measured electrical signal V can be used to calculate the commutation timing. a V b V c I tot The obtained rotor angle estimate ω est A commutation algorithm is provided for implementation in the commutation controller 204.

[0026] Figure 3 This illustrates an example cross-section of the phase winding shaft of a BLDC motor. The stator field vector moves (electrically) in 60° discrete steps. Figure 4 Explanation of three-phase inverter 402 (which can correspond to) Figure 1 Or 2 inverters 106 or 206) and BLDC motor 404 (which may correspond to Figure 1 Interconnection between the three terminals of motor 102 or 202 (or motor 2). Figure 4 In the process, the voltages at inverter nodes a, b, and c can be sensed and measured relative to ground to provide a voltage signal V. a V b V c A sensing resistor R can also be used. s An operational amplifier (not shown) measures the current on the bottom side of the inverter to provide the total current I. tot ,like Figure 4 As shown in the image.

[0027] Motor operation can be divided into six states, S0 to S5, with switching time intervals specified in Table 1. There are six reverse EMF zero crossings per electric rotation, one for each state, as shown below. Figure 5 As shown in the diagram. In each state, two half-bridges of the three-phase inverter 402 are active, that is, in each bridge, the gate input of one transistor remains high and the gate input of one transistor remains low, and the third half-bridge is floating, that is, the gate inputs of these transistors remain low.

[0028] For example, refer to Figure 4 In state S0, the gate input of transistor T6 remains high, the gate input of transistor T3 remains low, and the gate inputs of transistors T1 and T4 are driven by complementary pulse width modulation (PWM) logic. This PWM signal transmission (which can correspond to...) Figure 1 or 2 of V pu It can be directly controlled by the operator or by a speed / phase loop (i.e., Figure 1 Calculate (or the external loop of 2). Figure 4 The gate inputs of transistors T2 and T5 are kept low to drive the current in that phase to zero. This gate configuration corresponds to a specific magnetic field direction. When the rotor has rotated sufficiently, the motor needs to "commutate," that is, switch phases to generate a new magnetic field vector to drive the rotor to continue rotating. Without this commutation, i.e., without continuous reconfiguration of the magnetic field, the rotor will align with the static magnetic field and stop, which is not the desired behavior of a motor commanded to provide rotational torque to the load. The commutation states, which vary depending on the rotor angle, are summarized in Table 1. In systems with Hall effect sensors, similar to... Figure 1 The system shown in the figure has a Hall effect sensor output containing rotor position information used to determine the instant of transition from one state to another.

[0029] Table 1

[0030] A table illustrating the commutation status within the electrical angle range.

[0031] For each state and active transistor

[0032]

[0033] In sensorless systems used for trapezoidal control (similar to...) Figure 2 In a sensorless system, the measured phase voltage is used to estimate the angular position, which is used to determine the commutation state. Figure 5 This illustrates the ideal inverse EMF signal E corresponding to the six states of the electrical cycle (as shown by the dashed lines in the table). a E b E cThis highlights the anti-EMF zero-crossing during state S1. In the illustrated example, the commutation from S1 to S2 occurs at a commutation angle φ = π / 6 radians above the zero-crossing of phase B. Similarly, the commutation from S2 to S3 occurs at a commutation angle φ = π / 6 radians above the zero-crossing of phase A.

[0034] Rotor electrical angle θ r Controlling the moment of commutation, and can be achieved by, for example, calculating the corresponding reverse EMF signal E. a E b and E c And from the sensed phase voltage signal V a V b and V c Calculation or estimation. The mathematical equations for calculating anti-EMF are shown in Equations 1 through 3 below. Methods for sensorless trapezoidal control (similar to those by...) Figure 2 The method used in system 200 is assumed to be a non-salient pole and sinusoidal motor.

[0035] (1)

[0036] (2)

[0037] (3)

[0038] Phase voltage V a V b and V c Current I a I b and I c and the reverse EMF signal E a E b and E c Based on equations 1, 2, and 3, during state S0, the gates of T2 and T5 remain low to allow current I to pass through. c Drive to zero. Substitute I into Equation 3. c =0, and assume I c The derivative is zero. According to Equation 4, the measured voltage V... c Can be compared with the desired reverse EMF voltage E c and neutral point voltage V n Related.

[0039] (4)

[0040] Therefore, it can be based on V c and V n Calculate the inverse EMF signal E cThe neutral point voltage can be obtained by making the neutral point tap available and sensing the voltage. Alternatively, the neutral point voltage can be estimated as follows, thus eliminating the need for neutral point connection. Adding equations 1, 2, and 3, and assuming a sinusoidal motor (E... a +E b +E c =0),

[0041] (5)

[0042] Equations 4 and 5 can be used to estimate the neutral point voltage V during state S0 based on Equation 6. n And therefore the reverse EMF signal E c .

[0043] (6)

[0044] First, the zero-crossing of the inverse EMF signal can be estimated. Continuous zero-crossings can be used to estimate velocity. Furthermore, the commutation instant can be estimated based on the offset relative to the zero-crossing corresponding to the desired commutation angle φ.

[0045] However, the method described above for sensorless trapezoidal control of BLDC motors has several limitations. Due to the inverse EMF signal E... a E b or E c The magnitude of the inverse EMF signal is proportional to the rotor speed, therefore at sufficiently high rotor speeds, the inverse EMF signal is a feasible signal for rotor position estimation. However, at low speeds, especially during motor startup and transition to sufficient speed, the inverse EMF signal is small or zero, and does not provide rotor position information for reliable commutation. Therefore, for example, the method described above (i.e., Figure 2 The method is only effective when the anti-EMF term dominates at higher motor speeds and can fail at low speeds (e.g., during motor startup). Furthermore, for trapezoidal wound and salient-pole motors, the voltage equation includes additional terms.

[0046] Compared to the methods described above for sensorless control, the system and method for sensorless trapezoidal control of BLDC motors described herein can provide low-speed estimation by employing the positional dependence of the motor inductance. This behavior is due to the fact that even at zero speed, PWM current transients result in a mutual inductance component of the total voltage. Equations 7 and 8 represent the voltage equations for the motor's floating (phase C) and drive terminals (phases A and B) during state S0, and are even valid for trapezoidal anti-EMF salient-pole motors, unlike equations 1 to 3.

[0047] (7)

[0048] (8)

[0049] Table 2 summarizes the symbols used in the above equations.

[0050] Table 2

[0051] Symbols and annotations

[0052]

[0053]

[0054] The floating terminal voltage contains rotor position information as the sum of the following two key terms: (a) mutual inductance (the third term in Equation 7, i.e., the term starting with √3) and (b) anti-EMF (the fourth term in Equation 7, i.e., the term starting with ω). r (Items at the beginning). When the half-bridge drive phase C is in a high-impedance state, I c ≈0, and the first two terms of equation 7 can be ignored. Permanent magnet motors exhibit salient pole effect due to the difference between the d-axis and q-axis inductances (L2≠0). When the inverse EMF term ω... r *λ m F(θ r +2π / 3) can be ignored, and the positionally dependent mutual inductance term √3L2sin(2θ) is used. r +π / 3)*(dI a The mutual inductance term exists even at zero and low speeds. It is a high-frequency term due to the high-frequency PWM switching current, with harmonic content close to the PWM carrier frequency. The anti-EMF term is a low-frequency term with harmonic content close to the motor's electrical frequency. The anti-EMF and mutual inductance voltage terms can be separated from the frequency spectrum by having both high-frequency and low-frequency signal paths. Extraction of low-frequency and high-frequency components can be achieved via analog or digital filtering, i.e., before or after the three-phase voltage signal is sampled by the ADC.

[0055] In Equation 7, the phase current dI in the mutual inductance term is A. a The rate of change of / dt over time (which can also be called the "current slope" of the phase current A) can be implicitly determined from Equation 8, because V ab Equal to DC bus voltage ( Figure 4 V in dc Therefore, the left side of equation 8, V ab This is known to the estimator because it is what is being driven. For example, the total current I can be estimated within the PWM cycle. tot Multiple samples are sampled to help estimate the current slope.

[0056] By increasing dI a / dt can increase the level of the sensed signal, and thus improve the SNR. If the motor makes L2 in Equation 7 very small, then V varies according to the inductance L2. cn This will also be very small, meaning the SNR depends on the motor and could be very small. Therefore, when the motor starts up or drops below a threshold motor speed (and subsequently remains at this low speed for a duration thereof), the DC bus voltage fed to the PCM input can be dynamically increased to increase the SNR of the measured and estimated voltage signal. For example, this can be based on the measured signal V... a V b V c and I tot To estimate V cn Angle θ r This is the main unknown to be determined, as it is essential for the rotating motor. The inductance L2 varies with the current. Therefore, the effect of inductance variation on the current level can be captured. For example, a lookup table can store different commutation voltages that vary with the current. For instance, to obtain this lookup table, the motor can rotate with the sensor, recording all voltages and currents under different commutation conditions, and storing these values ​​in the lookup table. Then, the current I at different levels... tot The inductance L2 is known, and I a and dI a Given that / dt is already determined, sin(2θ) r +π / 3) can be used at low speeds (e.g., by...). Figure 6 The estimator 608 in the digital logic or other digital logic estimates V. cn / (√3*L2dI a / dt), thus obtaining the rotor electrical angle θ r The estimation. Such a lookup table can be programmed by the user within the estimator 608 (i.e., in...). Figure 7 The digital logic (either within a 708 or 808) or the estimator or digital logic can be programmed (e.g., pre-programmed) using lookup tables for various BLDC motors, and the appropriate lookup table can be selected by the user during motor-controller integration.

[0057] Instead, instead of estimating the rotor electrical angle θ r The digital logic in the estimator can be configured to estimate the voltage at which a commutation event occurs, which is a transition between states S0, S1, etc., such as... Figure 5 and Figures 9 to 12 The upper chart shows (voltage and θ) r (Related). In such alternatives, the motor can be driven to a known rotor electrical angle θ. rFurthermore, all commutation voltages can be recorded and saved to a lookup table, which is then used by the estimator. Similarly, such lookup tables can be programmed by the end user (i.e., the motor-controller integrator) or selected by the end user from various available lookup tables during motor-controller integration.

[0058] Figure 6 An example of trapezoidal control for a BLDC motor is illustrated in system 600 based on an analog filter, which utilizes the frequency separation of the anti-EMF and mutual inductance voltage terms as described above. For example, it can use an anti-aliasing (AA) filter 612 to provide the low-frequency component. The sensed phase voltage is sampled in two separate paths (i.e., a low-frequency (LF) path and a high-frequency (HF) path) corresponding to the filtered and unfiltered paths, respectively. As used herein, the "high" and "low" frequency ranges vary depending on the motor's electrical time constant (L / R) and the PWM frequency. For example, the PWM frequency can be between eight and ten times the time constant. For example, the low-pass filter can be between approximately five and approximately eight times the time constant. These LF and HF data paths can be implemented in any of several physical forms. The LF path contains position information generated by the anti-EMF component, and the HF path contains position information generated by the sum of the anti-EMF and mutual inductance components. LF and HF data can also be obtained using a low-pass filter with a higher cutoff frequency, using a single data path (e.g., the LF path in the dual data path arrangement described above), such that the unfiltered path contains all the required LF and HF data. The LF data can then be obtained by digitally low-pass filtering the data from the unfiltered path. With both LF and HF paths, the LF data path contains only the inverse EMF term (the fourth term in Equation 7), while the HF data path contains only the mutual inductance term (i.e., the third term in Equation 7), allowing both signals to be used to determine the values ​​needed to estimate the position of motor 602.

[0059] Similar to systems 100 and 200, system 600 may also include two loops: one containing a motor 602, an anti-aliasing (AA) filter or other low-pass filter 612, an analog-to-digital converter (ADC) 614, a speed estimator 608, a commutation controller 604, and an inverter 606; and the other containing a motor 602, an AA filter or other low-pass filter 612, an ADC 614, a speed estimator 608, a speed controller 610, and an inverter 606. Based on the sensed electrical signal V... a V b V c I tot The obtained digital value estimate of the rotor angular velocity ω estThis can be provided to a digital circuit system that implements the commutation algorithm in a digital logic commutation controller 604. For example, the ADC 614 can be a single ADC with multiple channels sampling LF and HF data paths. The ADC can sample at speeds much higher than required by the motor, so one ADC can sample different paths sequentially. Alternatively, LF / HF filtering can be performed digitally by sampling the measured signal at a sufficiently high frequency to capture all desired high-frequency content (e.g., a sampling rate greater than kilosamples per second). Although in the illustrated example, only three voltage signals V are sensed. a V b V c This signal is then fed to an anti-aliasing filter 612 or an ADC 614, but in other instances, a larger number of voltage signals are sensed and provided.

[0060] As described in this paper, the benefits provided by the improved system / method for sensorless trapezoidal control can be understood in the context of drilling rigs, pumps, or any motor application where the motor experiences a heavy load at startup, when the speed is close to zero and the relative rotor-stator position is unknown. The fourth term of Equation 7 (i.e., with ω...) r The term at the beginning of Equation 7 will only exist after the motor starts rotating. Therefore, the system and method can use the third term of Equation 7 (i.e., the term starting with √3) to estimate the relative rotor-stator position during startup.

[0061] Figure 7 and 8 This section describes two examples of sensorless trapezoidal control architectures, in which... Figure 8 This describes the analog front-end implementation of the method described for sensorless trapezoidal control. Figure 7 In the sensorless architecture, the drive circuit system 704 includes a multiplexer switch 718 with a sampling capacitor, an ADC 714, digital logic 708, and a pre-driver circuit system 710. The latter prepares the digital output of the digital logic 708 for amplification by the power FET 706, serving as a pulse-width modulated drive signal for the motor 702. Therefore, the power FET 706 can correspond to... Figure 2 Inverter 206 or Figure 4 Inverter 402. For example, the pre-driver circuit system 710 may be implemented as a three-phase intelligent gate driver from Texas Instruments Incorporated, which has three current shunt amplifiers DRV8323. Figure 7 As seen on the right, the phase voltage V relative to ground a V b V cThe signal is filtered by low-pass filter 712 to remove high-frequency switching components, and the inverse EMF signal is reconstructed in an estimator contained in digital logic 708. Multiplexer switch 718 selects between a path for analog-to-digital conversion by ADC 714 and a path for subsequent estimation in digital logic 708. Although the illustrated example senses three voltage signals V... a V b V c However, other instances can sense and sample a larger amount of voltage signal. This is because the low-pass filter 712 filters out the mutual inductance component in the voltage signal. Figure 7 The current architecture is not particularly feasible for angle estimation during the start-up of the BLDC motor 702 under load. In an alternative instance not explicitly stated, the low-pass filter 712 can be excluded, allowing only the total spectrum of each phase voltage signal to be sampled. The total spectrum signal is then used to determine the commutation timing instant. A drawback of this approach is that the sum of the anti-EMF and mutual inductance components in the total signal makes determining the appropriate voltage threshold for the commutation timing during start-up challenging.

[0062] exist Figure 8 In the sensorless architecture, the drive circuit system 804 includes a multiplexer switch 818 with a sampling capacitor, an ADC 814, digital logic 808, and a pre-driver circuit system 810. The latter prepares the digital output of the digital logic 808 for amplification by a power FET 806, serving as a pulse-width modulated drive signal to the motor 802. For example, the pre-driver circuit system 810 can be implemented as a three-phase smart gate driver from Texas Instruments Incorporated, featuring three current shunt amplifiers, DRV8323. Figure 8 As seen on the right, the phase voltage V relative to ground a V b V c High-frequency switching components are removed by filtering with a low-pass filter 812. However, compared with... Figure 7 The architecture shown in the paper is the opposite. Figure 8 The analog front end provides both filtered and unfiltered versions of these three-phase voltage measurements to the multiplexer switch 818 and ADC 814 for selection and sampling. Therefore, the multiplexer switch 818 can select and sample all three phase voltages V. a V b and V c The selection is made between a low-pass filtered and an unfiltered version. The inverse EMF signal can then be reconstructed in an estimator contained in the digital logic 808. Although the illustrated example senses three voltage signals V a V b V cHowever, other instances can sense and sample a larger number of voltage signals.

[0063] Because the system and method described herein separate the anti-EMF and mutual inductance components, the mutual inductance component (from the third term of Equation 7) can be used to determine the commutation timing during motor startup. After the motor reaches a specific desired speed, the anti-EMF component (from the fourth term of Equation 7) can be additionally used to determine the commutation timing. For example, the value of this desired speed can be set to a user-configurable threshold in digital logic 708 or 808.

[0064] The system and method described in this paper can use a nonlinear motor model for sensorless angle estimation for trapezoidal control. For example, such a nonlinear motor model can provide voltage thresholds, offsets, and zero-crossing time intervals that vary depending on the current level and current slope. The inductors L0 and L2 in Equations 7 and 8 vary depending on the current and the commutation advance angle φ. The salient-pole inductance L2 decreases with increasing current level. The drive voltage (e.g., V on the left side of Equation 8) is... ab Set to equal V DC Or 0 (corresponding to PWM on and off), the current slope in the drive phase (e.g., dI in Equation 8) a The inductance / dt) varies depending on the current level and the commutation advance angle φ. This dependence is explained in Equation 9, which simplifies to Equation 8 for low speeds and explicitly shows the dependence of the inductance on the current and commutation angle.

[0065] (9)

[0066] This slope affects the floating phase voltage relative to the neutral point, as given in Equation 7. The floating phase voltage relative to the neutral point can be estimated from the measured phase to ground voltage or its filtered derivative using Equation 6.

[0067] Substituting Equation 9 into Equation 7 and simplifying the equations for low speeds, the floating terminal voltage can be expressed according to Equation 10. In this equation, an additional phase offset term is introduced. (I a The effect of saturation on phase is described by φ.

[0068] (10)

[0069] In the nonlinear operating region, the phase-to-neutral point voltage varies not only with rotor position but also with current level and commutation angle. Therefore, one method for extracting rotor position information from phase measurements is to provide a nonlinear mapping of the position-dependent neutral point voltage variation for different current levels. This characterization can be achieved by driving the motor with sensors (e.g., encoders or Hall sensors) and establishing a model that correlates the floating neutral point voltage with current level and current slope.

[0070] Figure 9 Examples of this type of representation data are illustrated in the text. Figure 9 Plot the example phase-to-neutral point voltage waveforms for different speeds, varying according to the load (i.e., current) level. Specifically, Figure 9 The chart is plotted in degrees, showing the rotor electrical angle θ as it varies with voltage amplitude for different frequencies and loads. r Therefore, the inverse EMF voltage estimate varies with angle and load (i.e., current) level. By applying the Taylor series approximation, it can be shown that Equation 10 can be approximated as a first-order equation by a linear relationship between the slope and offset around the zero-crossing point for each commutation time interval. Therefore, from Figure 9 The graph allows for the creation of estimation equations for each state of the commutation cycle to predict the zero-crossing of the inverse EMF waveform. Furthermore, a set of estimation equations can be created at discrete current levels representing the motor's operating range. The six-level commutation states are also plotted on... Figure 9 middle.

[0071] Therefore, in some instances, the aforementioned nonlinear voltage contrast angle and current data (the basic model of which is given by Equation 10) can be modeled by establishing a table of one or both of the offset and zero-crossing duration and / or commutation voltage (i.e., voltage threshold) that vary according to rotor angle and current. In other instances, other known curve fitting techniques can be used for model fitting. The establishment of tables for offset, zero-crossing duration, and commutation voltage constitutes the training of the nonlinear motor model.

[0072] After building (i.e., training) the nonlinear motor model offline (i.e., before the normal operation of the motor to be controlled), during normal operation, the sensorless commutation algorithm can use voltage and current measurements as inputs and the stored nonlinear motor model to derive commutation decisions and thus control the motor.

[0073] Figure 10 and 11 The example commutation algorithms 1000 and 1100 are explained, which can calculate the commutation state based on a constructed nonlinear motor model. In each example algorithm, the voltage used to calculate the floating gate voltage to the neutral point (1004, 1104) can be a filtered or unfiltered phase voltage. In Equation 6, V... nThe calculations can be performed using analog or digital circuit systems.

[0074] exist Figure 10 In the commutation algorithm 1000, the voltage 1002 and total current from the brushless DC motor are measured, and an estimate of the floating gate voltage to the neutral point 1004 is calculated. For example, this calculation 1004 can be performed using Equation 6 above. The offset 1006 is determined based on the current measurement and a pre-calculated voltage offset table. For example, it can be determined by combining Equation 10 above and... Figure 9 The described method calculates this pre-calculated offset table. That is, the pre-calculated offset table can be a part of the aforementioned nonlinear motor model. The determined offset can then be added to the calculated estimate of the floating gate voltage to the neutral point (1008). In the operation of a brushless DC motor, the zero-crossing of this estimate can be determined for the current commutation state (1010). Based on the detected zero-crossing, the commutation voltage threshold (1012) can be calculated based on the current measurement and the pre-calculated voltage threshold table. This pre-calculated voltage threshold table can also be a part of the aforementioned nonlinear motor model. Based on reaching the voltage threshold, the commutation state can be switched (1014) to the next commutation state. This switching decision effectively determines the voltage threshold in the brushless DC motor (e.g., ...). Figure 4 (The diagram shows which of the six half-bridge drivers are turned on and which are turned off.) Then the current and voltage measurements are repeated 1002 times, and the entire process is repeated 1000 times, for example, at a rate generally exceeding any expected maximum physical speed of the motor, to provide the desired improved commutation control.

[0075] exist Figure 11 In the commutation algorithm 1100, the voltage and total current from the brushless DC motor are measured at 1102, and an estimate of the floating gate voltage to the neutral point is calculated at 1104. For example, this calculation at 1104 can be performed using Equation 6 above. The offset at 1106 is determined based on the current measurement and a pre-calculated offset table. For example, it can be determined by combining Equation 10 above and... Figure 9 The described method calculates this pre-calculated offset table. That is, the pre-calculated offset table can be a portion of the aforementioned nonlinear motor model. The determined offset can then be added to the calculated estimate of the floating gate voltage to the neutral point 1108. In the operation of the brushless DC motor, this estimated zero-crossing can be determined for the current commutation state 1110. Based on the detected zero-crossing, the time interval for commutation can be calculated 1112 based on the current measurement and the pre-calculated duration table (i.e., the zero-crossing time interval), and a counter is started. This pre-calculated duration table can also be a portion of the aforementioned nonlinear motor model. Based on the counter reaching the calculated time interval value (i.e., as calculated in 1112), the commutation state can be switched 1114 to the next commutation state. This decision determines the brushless DC motor (e.g., Figure 4Which of the six half-bridge drivers shown in the diagram are turned on and which are turned off. The current and voltage measurements 1102 are then repeated, and the entire process 1100 is cycled, for example, at a rate generally exceeding any expected maximum physical speed of the motor, to provide the desired improved commutation control.

[0076] Figure 12 and 13 The timing diagram illustrates the operation of two different instance architectures used for sampling voltage and current. Figure 12 This represents a structure that sequentially obtains voltage and current samples within time intervals when the current slope is positive. Figure 13 This indicates an architecture where voltage samples are acquired simultaneously, followed by current samples acquired sequentially. Three-phase voltage and two current samples are... Figure 12 and 13 The examples illustrate why Figure 12 The diagram shows five arrows per cycle (one arrow represents three sequential voltage measurements, and the other two arrows represent two sequential current measurements). Figure 13 The diagram shows three arrows per cycle (one arrow represents three parallel voltage measurements, and the other two arrows represent two sequential current measurements). In each instance, two current samples can be used to determine the current slope. These sampling instances are specific to three-phase motors but are applicable to motors with other numbers of phases.

[0077] Figure 14 and 15 Explanation at full speed ( Figure 14 ) at startup ( Figure 15 A sensorless trapezoidal control system (similar to) under 1.4 times rated load. Figure 2 The performance of a sensorless trapezoidal control system. In the described method, in Figure 14 and 15 The reversal moments (i.e., S0, S1, S2, S3, S4, or S5) shown in the corresponding top chart are determined based on a fixed electrical angle offset (e.g., 30°) relative to the anti-EMF zero crossing. This offset is calculated based on velocity estimation. Figure 14 As shown, when the motor is already at full speed before load is applied (starting from 0.2 seconds), the actual commutation closely matches the ideal commutation. However, this method fails during motor startup when reliable speed estimation is unavailable, such as... Figure 15 As shown in the figure, a rated load is applied during startup, and the actual commutation state does not converge with the ideal commutation state (e.g., Figure 15 (As shown in the top chart).

[0078] Figure 16 and 17 Explain when the full speed ( Figure 16 ) at startup ( Figure 17 A sensorless trapezoidal control system (similar to) under 1.4 times rated load. Figures 6 to 8 The performance of a sensorless trapezoidal control system. Figure 16 and 17 In the described method, the commutation instant can be determined based on applying a threshold voltage to the reconstructed floating phase to neutral point voltage. For example, during motor startup, a threshold can be applied to the mutual inductance voltage alone, thereby eliminating the dependence of commutation on the unknown motor speed. After the motor reaches the desired speed, a threshold can be applied to the inverse EMF component, as this component has sufficient signal-to-noise ratio (SNR) at medium to high speeds. The threshold can be generated from a lookup table that maps changes in the threshold voltage to current-varying inductance in the saturation region. Therefore, as... Figure 16 and 17 As shown in the corresponding top chart, this method exhibits excellent convergence between actual and ideal commutation at both low and high speeds, even without Hall effect sensors or encoders.

[0079] Figure 18 A method 1800 for sensorless trapezoidal control is described. Voltage signals (e.g., three such phase voltage signals) and at least one total current signal are measured from a BLDC motor and can be provided 1802, for example, to an analog-to-digital converter for digital sampling. For example, the voltage signals and total current signals can be measured from or provided to an inverter circuit system coupled to the BLDC motor (e.g., a three-phase BLDC motor). These measured signals can be processed, or signals derived from these signals can also be processed. As an example of signals derived from these measured signals, before or after digital sampling, the voltage signal can be separated 1804 into a first spectral component containing high-frequency signal content and a second spectral component excluding high-frequency signal content. For example, the voltage signal can be separated into a first component containing high-frequency signal content and a second component excluding high-frequency signal content by filtering the three-phase voltage signal using an anti-aliasing filter.

[0080] Based on the measured signals and / or signals derived therefrom (e.g., frequency-component separated voltage signals and total current signals), an estimate of the rotor angular position or commutation state of the 1806 motor is calculated during the startup of the BLDC motor. For example, a digital logic circuit system (e.g., Figure 6 The estimator 608 or Figure 7 This estimate can be calculated using a digital circuit system (either 708 or 808 in the digital logic circuit system). The estimation calculation may involve determining the commutation voltage value or the current-dependent inductance value from a lookup table stored in the digital logic circuit system.

[0081] In this description, the term "start-up" is defined as any operating region in which the angular velocity of the motor is below a speed threshold, thereby giving the mutual inductance term in the motor model (i.e., the third term in Equation 7 above) a majority influence (i.e., dominance) on the estimated position or commutation state. Conversely, if the angular velocity of the motor is above the speed threshold, the influence of the mutual inductance term decreases, and the anti-EMF term in the motor model (i.e., the fourth term in Equation 7 above) has a major influence (i.e., dominance) on the estimated position or commutation state. Therefore, in this description, the term "start-up" includes any period of time when the angular velocity of the motor (e.g., rotor speed) is below the speed threshold (even beyond initial operation).

[0082] Then, based on the estimated angular position or commutation state, the 1808 motor is controlled during startup. Such control can be provided by generating a drive signal (e.g., a PWM signal) based on the estimated angular position or commutation state.

[0083] Method 1800 may further include providing a pulse width modulation (PWM) drive current signal to the motor, the drive current signal comprising a PWM cycle, and then sampling a voltage signal or a signal derived therefrom (e.g., a voltage signal with frequency components separated) at the corresponding rising edge of the PWM cycle of the drive current signal (e.g., as shown in the image). Figure 12 and 13 (As shown in the image). For example, such as... Figure 13 As shown, the phase voltage signal, or the signal derived from it, can be sampled simultaneously for 1802 seconds. Furthermore, as... Figure 12 and 13 As demonstrated in both examples, multiple samples of the total current signal can be sampled within a PWM loop. The current slope can then be estimated by a digital logic circuit system based on these multiple samples from the PWM loop.

[0084] The angular position or commutation state of the 1806 motor can be estimated in part by modeling the variations in voltage and total current signals that depend on the motor's inductance to provide a characterizing inductance model. Method 1800 can then further include determining the commutation instant from the voltage signal (or a signal derived therefrom), the total current signal, and the characterizing inductance model. As described in more detail above, the extracted inverse EMF component can be used to train the inductance model.

[0085] During motor startup and low-speed operation in method 1800, a first commutation instant can be estimated (e.g., via a digital logic circuit system) based on determining that the calculated mutual inductance voltage has exceeded a mutual inductance voltage threshold. A second commutation instant can be estimated (e.g., via a digital logic circuit system) based on determining that the rotor angular velocity has exceeded a rotor speed threshold, using a digital logic circuit system to estimate the second commutation instant based on determining that the calculated anti-EMF voltage has exceeded a voltage threshold. In this way, method 1800 transitions from inductance-based commutation instant determination to anti-EMF-based commutation instant estimation for commutation timing. The voltage threshold can be adjusted based on current measurements and a lookup table that takes into account the nonlinear variation of the motor inductance with current saturation.

[0086] Furthermore, the DC bus voltage can be increased during motor startup or in response to determining that the rotor angular velocity has dropped below a rotor speed threshold, thereby increasing the signal-to-noise ratio of the calculated mutual inductance voltage at low motor speeds.

[0087] The systems and methods for ladder logic control described herein can be manufactured in various ways for application in motor systems. For example, the system and methods can be implemented by a driver chip containing voltage and current sensing circuitry coupled to a separate chip containing digital read-only memory (ROM) with instructions for determining precise rotor-stator relative positions. In another example, analog driver circuitry and digital ROM circuitry can be combined into a single chip. In yet another example, the system can be implemented using a general-purpose microprocessor coupled to a suitable analog front-end circuitry.

[0088] The system and method for trapezoidal control described herein achieve sensorless start-up under load by employing rotor position information not only in the inverse EMF component but also in the inductance within the appropriate speed range. By performing position sensing using only motor voltage and current, the system and method eliminate the need for a board containing, for example, three Hall effect sensors or one or more encoders, thus freeing up board space and wiring. The disclosed and claimed system and method can utilize parallel three-phase voltage sampling to obtain the signal required to estimate the rotor electrical angle or equivalent commutation moment for enhanced trapezoidal control without sensors, and has improved low-speed performance compared to other sensorless methods. The disclosed and claimed system and method can also provide a DC voltage boost to the PWM drive signal during startup to maximize SNR, and thus further improve low-speed control performance. The disclosed and claimed system and method can further utilize a lookup table representation of the control input and commutation relationship to determine commutation criteria that vary according to the provided current level and motor speed. Therefore, the disclosed and claimed system and method provide a low-cost solution for applications such as power tools and electric bicycles by eliminating sensor and wiring costs. Compared to another sensorless approach for trapezoidal control of BLDC motors, the system and method offer improved angle tracking, reliability, reduced return and repair costs, and repeatability of stall torque.

[0089] In this description, the term "based on" means at least partially based on. In this description, the term "coupled" means an indirect or direct wired or wireless connection. Therefore, if a first device, element, or component is coupled to a second device, element, or component, the coupling can be direct coupling or indirect coupling via other devices, elements, or components and connections. Similarly, devices, elements, or components coupled between a first component or location and a second component or location can be directly connected or indirectly connected via other devices, elements, or components and / or couplings. Devices "configured to" perform tasks or functions can be configured by the manufacturer at manufacturing time (e.g., programmed and / or hardwired) to perform functions and / or can be configured (or reconfigured) by the user after manufacturing to perform functions and / or other additional or alternative functions. Configuration can be achieved through firmware and / or software programming of the device, through the construction and / or layout of the device's hardware components and interconnections, or a combination thereof. Furthermore, a circuit or device referred to as containing certain components can instead be configured to couple to those components to form the described circuit system or device. For example, a structure described as comprising one or more semiconductor elements (e.g., transistors), one or more passive elements (e.g., resistors, capacitors, and / or inductors), and / or one or more sources (e.g., voltage and / or current sources) may instead comprise only semiconductor elements within a single physical device (e.g., a semiconductor die and / or integrated circuit (IC) package), and may be configured to couple at least some of the passive elements and / or sources during manufacturing or, for example, after manufacturing by an end user and / or a third party, to form the described structure.

[0090] Modifications are possible in the described embodiments, and other embodiments are possible within the scope of the claims.

Claims

1. A method for sensorless control, the method comprising: Measurement of voltage and total current signals obtained from a brushless DC (BLDC) motor with a rotor; Using a processing circuit system, in response to the measurement, the angular position of the rotor or the commutation state of the motor is determined according to a nonlinear motor model, wherein the nonlinear motor model provides a voltage offset in a lookup table (LUT) and at least one of the following, varying according to current level and current slope: a voltage threshold or a zero-crossing time interval, the determination further comprising: Determine the voltage at the floating terminal relative to the neutral point voltage. The offset is determined based on the total current signal and the amount of voltage offset in the LUT; The determined offset is added to the determined voltage of the floating terminal; and The motor is controlled during startup in response to a determined angular position or a determined commutation state.

2. The method according to claim 1, wherein the motor is a three-phase motor, and the voltage signal includes a three-phase voltage signal.

3. The method of claim 1, wherein the LUT includes a voltage threshold that varies based on the current level and the current slope, and the determination further includes: Determine the zero crossing of the commutation state; and The commutation voltage threshold is calculated based on the total current signal and the voltage threshold in the LUT.

4. The method of claim 1, wherein the LUT includes a zero-crossing time interval that varies according to the current level and the current slope, and the determination further includes: Determine the zero crossing of the commutation state; and The commutation time interval is calculated based on the total current signal and the zero-crossing time interval in the LUT.

5. The method of claim 1, further comprising sampling the voltage signal and the total current signal using an analog-to-digital converter, and filtering the voltage signal using an anti-aliasing filter to separate the voltage signal into a first component containing high-frequency signal content and a second component excluding the high-frequency signal content.

6. The method of claim 1, further comprising: A pulse width modulation (PWM) drive current signal is provided to the motor, the drive current signal having a PWM cycle; and The voltage signal is sampled at the rising edge of the PWM cycle.

7. The method according to claim 6, wherein: Simultaneously, the voltage signal or the signal derived therefrom is sampled; Multiple samples of the total current signal are obtained in one of the PWM cycles; and The current slope is determined by the processing circuit system based on the plurality of samples of the total current signal.

8. The method of claim 6, wherein the LUT is generated by driving the motor using a sensor so that the neutral point floating voltage is related to the current level and the current slope.

9. The method of claim 1, further comprising transitioning from an inductor-based mode to an anti-EMF-based mode for commutation timing via the following: During the startup period, the processing circuitry system determines the first commutation instant based on the determination that the calculated mutual inductance voltage has exceeded a voltage threshold; and Using the processing circuit system, the second commutation moment is determined based on the determination that the rotor's angular velocity has exceeded the rotor speed threshold and the calculated reverse EMF voltage has exceeded the voltage threshold.

10. The method of claim 9, wherein the voltage threshold is adjusted based on the measurement of the total current signal and the LUT that takes into account the nonlinear variation of the motor inductance with current saturation.

11. The method of claim 9, further comprising increasing the DC bus voltage during the startup period, or in response to determining that the angular velocity of the rotor has dropped below a rotor speed threshold, so as to increase the signal-to-noise ratio of the mutual inductance voltage when the angular velocity of the rotor drops below the rotor speed threshold.

12. A system for sensorless control, the system comprising: A measurement circuit system adapted to be coupled to a brushless direct current (BLDC) motor having a rotor, the measurement circuit system being configured to measure voltage signals and the total current signal of the BLDC motor; A processing circuit system coupled to the measurement circuit system is configured to determine, in response to the measurement, the angular position of the rotor or the commutation state of the motor based on a nonlinear motor model, wherein the nonlinear motor model provides a voltage offset based on a lookup table (LUT) and at least one of the following varying based on current level and current slope: a voltage threshold or a zero-crossing time interval; the processing circuit system is further configured to: Determine the voltage at the floating terminal relative to the neutral point voltage. Based on the total current signal and the amount of voltage offset in the LUT, the offset is determined, and The determined offset is added to the determined voltage of the floating terminal; and A control circuit system coupled to the processing circuit system and adapted to be coupled to the BLDC motor, the control circuit system being configured to provide motor control signals during motor startup in response to a determined angular position or a determined commutation state.

13. The system of claim 12, wherein the motor is a three-phase motor, and the voltage signal comprises a three-phase voltage signal.

14. The system of claim 12, wherein the processing circuitry is configured to determine the angular position or the commutation state by determining a commutation voltage value or a current-dependent inductance value from the LUT.

15. The system of claim 12, wherein the motor is configured to be driven using a pulse width modulation (PWM) drive current signal having a PWM cycle, and the system further includes an analog filter or digital frequency filter circuit system configured to sample the voltage signal or a signal derived therefrom at the rising edge of the PWM cycle.

16. The system according to claim 15, wherein: The analog filter or the digital frequency filter circuit system is configured to: Simultaneously, the voltage signal or the signal derived therefrom is sampled; and Multiple samples of the total current signal are obtained in one of the PWM cycles; and The processing circuitry is configured to determine the current slope based on the plurality of samples of the total current signal.

17. The system of claim 12, wherein the LUT includes a voltage threshold that varies according to the current level and the current slope, and wherein the processing circuitry is further configured to: Determine the zero crossing of the commutation state; and The commutation voltage threshold is calculated based on the total current signal and the voltage threshold of the LUT.

18. The system of claim 12, wherein the LUT includes a zero-crossing time interval that varies according to the current level and the current slope, and wherein the processing circuitry is further configured to: Determine the zero crossing of the commutation state; and The commutation time interval is calculated based on the total current signal and the zero-crossing time interval of the LUT.

19. The system of claim 17, wherein the LUT is generated by using a sensor to drive the motor so that the neutral point floating voltage is related to the current level and the current slope.

20. The system of claim 12, wherein the processing circuitry is configured to transition from an inductor-based mode to an anti-EMF-based mode for commutation timing via: During the startup period, the first commutation instant is determined based on the determination that the calculated mutual inductance voltage has exceeded a voltage threshold; and The second commutation moment is determined based on the fact that the rotor's angular velocity has exceeded the rotor speed threshold and the calculated reverse EMF voltage has exceeded the voltage threshold.

Images