PMSM FOC with single shunt current reconstruction

This application implements the sensorless field oriented control (FOC) of a permanent magnet synchronous motor (PMSM). It estimates rotor position from measured phase currents and a reduced order luenberger observer (ROLO). The feedback motor phase currents are calculated from single shunt current reconstruction methodology. This algorithm is implemented on a SAMC21J18A MCU.

Description

Permanent Magnet Synchronous Motor (PMSM) is controlled using Field Oriented Control (FOC). Rotor position and speed is determined using Reduced Order Luenberger Observer. The PMSM motor current current is determined by single shunt current reconstruction method. Motor start/stop operation is controlled by the switch and motor speed can be changed by the on-board potentiometer. Waveforms and variables can be monitored runtime using X2CScope.

Key features enabled in this project are:

• Single shunt current measurement
• Speed control loop

MCC Project Configurations

• ADC0-ADC1 Peripheral:

  • ADC0 and ADC1 are setup to operate independently
  • Both ADCs convert single ended inputs. DC bus current corresponds to time during T1 is sampled and converted by ADC0 and DC bus current corresponds to time during T2 is sampled and converted by ADC1.
  • Both ADCs are hardware triggered by events generated from TCC1 during its CC0 and CC1 compare match
  • Conversion Ready interrupt is generated by ADC1. At the time of this ISR, both DC bus corresponds to time slots T1 and T2 is ready.
• TCC0 Peripheral:
  • Configured to generate three pairs of complimentary PWM signals at a frequency of 8 kHz in “Dual Slope PWM with interrupt/event when counter = ZERO or counter = TOP”.
  • Period match interrupt is enabled for both ZERO and TOP counter match
    • Dead-time is enabled and set to
    • mclv2_sam_c21_pim.X. - 1uS
  • Non-recoverable Fault is enabled on EV1. When an event is detected on EV1, all PWM channels are held low.
• TCC1 Peripheral:
  • Configure to operate at a frequency of 5 kHz in “Dual Slope PWM with interrupt/event at ZERO match”.
  • CC0 and CC1 Compare match events are used to trigger ADC0 and ADC1 respectively.
• TC0:
  • Configured in One shot mode to trigger TCC0 and TCC1 simultaneously.
• EIC:
  • External Interrupt Controller detects a hardware over-current fault input and generates a non-recoverable fault event for TCC0, thereby shutting down the PWM in the event of an over-current fault.
• EVSYS:
  • Event System acts as an intermediary between event generator and event users.
  • Event generated by thr TC0 when the counter overflows, is used by hardware trigger to start TCC0 and TCC1
  • Event generated by the TCC1 when the counter reaches CC0 compare value, is used by the ADC0 as a hardware trigger source via the Event System.
  • Event generated by the TCC1 when the counter reaches CC1 compare value, is used by the ADC1 as a hardware trigger source via the Event System.
  • Event generated by the EIC upon over-current fault, is used by the TCC0 as a non-recoverable fault event via the Event System.
• DIVAS:
  • This demo uses “Divide and Square Root Accelerator” to perform 32-bit signed and unsigned division and 32-bit unsigned square root operations.
  • DIVAS is set to overload “Divide - / “ operator i.e. c = a / b; would use DIVAS accelerator for division without using a special API. However, square root operation would still require a special API. Refer to the DIVAS help for more details.
• SERCOM3 Peripheral:
  • SERCOM3 is configured in USART mode and is set to operate at 115200 bps.
  • This USART channel is used by the X2CScope plugin to plot or watch global variables in run-time. Refer to X2C Scope Plugin section for more details on how to install and use the X2CScope.

Control Algorithm

This application implements sensorless field oriented control (FOC) of permanent magnet synchronous motor (PMSM). It estimates rotor position from measured phase currents and a reduced order luenberger observer (ROLO). The algorithm is implemented on the SAMC21J18A MCU. The following section briefly describes the ROLO based sensorless FOC algorithm, software design and implementation.

Field Oriented Control is the technique used to achieve the decoupled control of torque and flux. This is done by transforming the stator current quantities (phase currents) from stationary reference frame to torque and flux producing current components in rotating reference frame using mathematical transformations. The Field Oriented Control is implemented as follows:

1. Reconstruct the motor phase currents based on the DC bus current measurement.
2. Transform the three phase currents from three phase stationary reference frame (u,v,w) into two phase stationary reference frame (alpha, beta) using the Clarke transformation.
3. Estimate (sensorless) or measure (sensored) the rotor position angle.
4. Transform the stator currents from a two phase stationary reference frame (alpha, beta) into a two phase rotating reference frame (d,q) using the Park transformation. The rotating reference frame rotates synchronously with the rotor axis and therefore requires rotor angle information.
5. Clarke and Park transformation allows the transformation of three phase AC stator currents into an equivalent two phase DC currents which decouples the flux (id) and torque (iq) producing components of the stator currents.
6. The stator current torque (iq) and flux (id) producing components are regulated independently by two independent PI controllers. These PI controllers generate an output voltage reference in d,q reference frame which is transformed into three phase stationary frame AC voltages using the inverse park and inverse clarke transformations.
7. In order to increase the DC bus utilization, the three phase sinusoidal AC output voltages are translated into space vector modulated voltages and applied across the stator windings using a Three Phase Half-Bridge Inverter.

The following block diagram shows the software realization of the FOC algorithm.

Rotor Position and Speed Estimation:

The electrical rotor position is derived from back EMF voltage estimated by Reduced Order Luenberger Observer which is implemented in position_and_speed_estimation function. Back EMF voltage is a function of rotor speed. At lower rotor speeds, the estimated back EMF has poor signal to noise ratio and as a result does not provide accurate rotor angle estimation. Therefore, the FOC algorithm also integrates an open loop ramp-up profile for motor startup. The reference speed is incremented linearly until the required minimum reference speed is achieved such that the estimated back EMF has a high enough signal to noise ratio for accurate rotor angle estimation, at which point the algorithm uses estimated rotor angle for further control. For more details about ROLO, refer to Application note AN2590

Single Shunt Current Reconstruction:

The single shunt current reconstruction algorithm is used to estimate the three phase currents by using a single shunt connected across the DC link path.The DC link current input from the shunt resistor along with the knowledge of applied PWM at a particular instant of time, the three phase motor current is calculated as shown below:

The single shunt current reconstruction requires a minimum active vector time ( T1, T2 ) for shunt current to be stabilized before it can be sampled as shown above. To avoid shunt current ripples issue to creep into the ADC measurement, the calculated SVPWM is adjusted asymmetrically . Asymmetric adjustment of SVPWM for different scenarios is described below:

a. T1 is less than minimum time window for current measurement:

b. T2 is less than minimum time window for current measurement:

c. Both T1 and T2 are less than minimum time window for current measurement:

The implementation details of single shunt current reconstruction can be referred from the Application note AN1299

Software Design

The following figure shows the flow chart of the implemented algorithm:

Field Alignment - ALIGN:

In this mode, a pre-defined value of the current is asserted across the Q axis of the rotor for a pre-defined length of time, in order to align the rotor to a known angle of 90 electrical degrees. The magnitude of the current and the length of the time for which it is applied depends upon the electrical and mechanical time constant of the PMSM motor drive. Electrical time constant of the motor is a function of R and L values of the motor windings, whereas the mechanical time constant of the motor drive is primarily a function of the static load on the motor shaft.

Open Loop Control - STARTING:

In this mode, the speed of the PMSM motor is gradually ramped up using an open loop control. During this mode, the rotor angle is derived from the asserted open loop speed reference. This derived rotor angle would be lagging from the actual rotor angle. The speed is ramped up linearly to a minimum value required for the ROLO-based estimator to estimate the alpha and beta axis back EMF of the PMSM motor with sufficient accuracy. Rotor angle information is extracted from arctan(BEMFbeta/ BEMFalpha).

Close Loop Control - RUNNING:

In this mode, the estimated rotor angle obtained from ROLO-based estimator is used to perform field oriented control of the PMSM motor.

Timing Diagram:

Timing Diagram: The three critical blocks of the MCU which are involved in the motor control operations are:

• ADC - It measures motor phase currents at the start of the PWM cycle.
• CPU - Based on motor phase currents measurements, the CPU executes the control algorithm to calculate the PWM duty cycle needed to apply required voltage across motor windings.
• TCC0 - It generates the PWM signals needed to apply required voltage across motor windings.

The following timing diagram, shows the chronological role played by the ADC, CPU and TCC0 in the execution of FOC algorithm.

Software Configuration

The FOC algorithm is used across different application fields. In order to get the optimal control of the PMSM motor, the motor specific parameters, board specific parameters and application parameter macros have to be updated in the software. The following section describes how to update both motor and application specific parameters in the the project. These parameter macros can be found in userparams.h header file which can be found under logical path: Header Files > config > >userparams.h in MPLABX IDE.

Setting motor control PWM frequency and dead-time:

PWM frequency is set by configuring the timer period of the TCC0 in terms of TCC0 clock counts. This frequency needs to be communicated to other sections in the algorithm by defining the “Period Value” (set in MPLAB Harmony Configurator) as a macro in userparams.h. For example, in order to achieve a PWM frequency of 8KHz in center aligned mode with peripheral clock frequency of 48MHz, the period value defined in MHC is 3000. Similarly, dead-time is configured in terms of TCC0 clock counts in the MHC. The same value needs to be defined as macro in userparams.h.

Setting motor specific parameters:

Set the following motor specific parameters in the userparams.h file.

Setting board specific parameters:

Set the following board specific parameters in the userparams.h file.

Setting PI Controller parameters:

Depending on the type of motor used, and the corresponding application PI controller parameters should be updated in the userparams.h file.

Speed PI controller gains:

Current PI controller gains for D and Q axis currents:

Debug & Optimization Modes :

This example provides compile time re-configuration options using #define macro directives to enable or disable different debug and optimization modes. These macro directives can be accessed in the userparams.h header file.

• Current PI Tuning - CURPI_TUN
  • This mode allows tuning of the Current PI controllers. This mode generates a step current reference and depending upon the actual current response, user can accordingly adjust current PI gains.
  • This mode can be enabled by defining the macro CURPI_TUN in userparams.h header file
  • In this mode, the step current reference is asserted across the D axis
  • Macro CUR_STEP_AMP defines the step size of the asserted D axis current reference in amperes and macro CUR_STEP_TIM defines the length of the step in seconds
• Open Loop Mode - OPEN_LOOP_FUNCTIONING
  • This mode is used to debug the open loop startup of the motor. This mode could also be used in tuning of Current PI controllers
  • In this mode, the motor operates in open loop rotor angle mode. In other words, the rotor angle used by the algorithm to assert current will be different than actual rotor angle and therefore, the asserted current may not be orthogonal to the rotor magnetic axis, resulting in lower torque generation.
  • This mode can be enabled by defining the macro OPEN_LOOP_FUNCTIONING in userparams.h header file
  • The torque current reference in this mode is defined by the macro START_CUR_AMP
  • Depending upon the load inertia and startup acceleration time (STUP_ACCTIME_S), if the generated torque is not sufficient, then it may result in stalling of the motor
  • START_CUR_AMP and STUP_ACCTIME_S need to be modified to ensure that the motor starts up successfully in open loop mode without stalling. If the motor stalls and the PWM signals are shutdown, the algorithm enters into a critical fault mode. The MCU needs to be reset to recover from the critical fault.
• Torque Mode - TORQUE_MODE
  • In this mode, as the name suggests, the algorithm operates in torque mode. The torque reference is obtained from the potentiometer.
  • This mode can be enabled by defining the macro TORQUE_MODE in userparams.h header file
  • The potentiometer min - max input is scaled to generate torque current reference in the range of TORQUE_MODE_MIN_CUR_AMP - MAX_CUR_AMP
  • Depending upon the load, if the torque is too low then it could cause the motor to stall and algorithm to enter a critical fault mode. The MCU needs to be reset to recover from the critical fault.
• Control Frequency to PWM Frequency = 1:1 - CTRL_PWM_1_1
  • Defining this macro, executes current and speed controllers every other PWM cycle.
  • Note: Please account for CPU bandwidth availability while setting Control to PWM Frequency Ratio as 1:1. If the ISR execution is not completed within the same PWM cycle, it could lead to erratic motor behavior
• Executing from RAM - RAM_EXECUTE
  • In order to speed up execution, the frequently called motor control related functions can be placed and fetched from RAM during execution.
  • Fetching instructions from RAM is faster than fetching instructions from Flash, as fetches from RAM can occur at CPU clock speed, unlike the Flash, which requires wait states to be asserted.
  • This speed optimization feature can be enabled by defining the macro RAM_EXECUTE in userparams.h header file.
  • Enabling this mode would result in a larger data memory footprint.
  • Instruction breakpoint will not be asserted on an instruction which is being executed from RAM. It is recommended to disable this macro while debugging this algorithm using instruction breakpoints.
  • Exercise caution while using this feature as the functions in RAM are vulnerable to get overwritten due to stack overruns or unguarded buffer overruns.

Development Kits

MCLV2 with SAMC21 Family Motor Control PIM

Downloading and building the application

To clone or download this application from Github, go to the main page of this repository and then click Clone button to clone this repository or download as zip file. This content can also be downloaded using content manager by following these instructions.

Path of the application within the repository is apps/pmsm_foc_rolo_1shunt_sam_c21 .

To build the application, refer to the following table and open the project using its IDE.

---