MOTOR-16 — Clarke and Park Transforms

FACT — AC traction motors operate on three-phase sinusoidal currents that continuously oscillate at high frequencies. However, standard linear Proportional-Integral (PI) controllers intrinsically struggle to track time-varying sinusoidal references, inevitably introducing steady-state amplitude error and phase lag. Clarke and Park transformations solve this dilemma by mapping three-phase AC quantities into an orthogonal rotating coordinate frame where sinusoidal waves collapse into pure time-invariant DC quantities.

In this lesson, we break down the coordinate transformations that form the mathematical engine of modern electric vehicle Field-Oriented Control (FOC):

1. The Control Dilemma: Tracking AC Quantities with PI Regulators
2. The Clarke Transform: Three-Phase to Stationary Orthogonal ($a-b-c \to \alpha-\beta$)
3. The Park Transform: Transitioning to the Synchronous Frame ($\alpha-\beta \to d-q$)
4. Physical Interpretation: Direct ($d$) vs. Quadrature ($q$) Axes
5. Inverse Transformations: Returning to the Physical Phase Domain
6. Real-Time Execution Pipeline in Inverter Microcontrollers

---

1. The AC Tracking Challenge

Consider an electric vehicle cruising at $150\text{ km/h}$ with an 8-pole ($p=4$) traction motor rotating at $10,000\text{ rpm}$. The electrical fundamental frequency of the stator currents is:

$f_e = \frac{p \cdot n}{60} = \frac{4 \times 10,000}{60} \approx 667\text{ Hz}$

The phase currents $i_a(t), i_b(t), i_c(t)$ reverse polarity 667 times per second.

A standard PI controller transfer function:

$C(s) = K_p + \frac{K_i}{s}$

provides infinite gain only at DC ($s = 0, \omega = 0$), guaranteeing zero steady-state error for constant setpoints. However, when presented with a sinusoidal signal at $667\text{ Hz}$, the integrator gain drops inversely with frequency ($|K_i / j\omega| \to 0$). The controller lags behind the actual physical waveform, causing phase distortion and degraded torque production.

The Solution: Attach the observer coordinate frame directly to the rotating rotor flux vector! Just as two passengers sitting inside a high-speed train appear stationary relative to each other, three-phase AC currents appear as completely static DC currents when observed from a coordinate frame rotating synchronously with the rotor field.

---

2. Clarke Transformation ($a-b-c \to \alpha-\beta$)

Formulated by Edith Clarke in the 1930s, this transform maps three spatially distributed phase axes separated by $120^\circ$ ($a, b, c$) onto a two-axis stationary orthogonal Cartesian frame ($\alpha, \beta$).

In a balanced three-phase system with an isolated neutral point, Kirchhoff’s Current Law dictates:

$i_a + i_b + i_c = 0$

Because the three phase variables are constrained by this relationship, the system possesses only two degrees of freedom. The third phase variable contains no independent information.

Clarke Reference Frame:
          beta
           ^
           |       Phase b
           |       /
           |      /
           |     /
           |    / 120°
           +---+----------------> alpha (Aligned with Phase a)
              / \
             /   \ 120°
            /     \
           /       \
      Phase c

Amplitude-Invariant Clarke Matrix:

To ensure that peak sinusoidal phase current equals the resultant vector magnitude in the $\alpha-\beta$ plane, the amplitude-invariant Clarke transform is formulated as:

$\begin{bmatrix} i_\alpha \\ i_\beta \end{bmatrix} = \frac{2}{3} \begin{bmatrix} 1 & -\frac{1}{2} & -\frac{1}{2} \\ 0 & \frac{\sqrt{3}}{2} & -\frac{\sqrt{3}}{2} \end{bmatrix} \begin{bmatrix} i_a \\ i_b \\ i_c \end{bmatrix}$

Substituting $i_a + i_b + i_c = 0$ simplifies real-time microcontroller calculation, requiring only two current sensors:

$i_\alpha = i_a$

$i_\beta = \frac{1}{\sqrt{3}} (i_a + 2 i_b) = \frac{i_a + 2 i_b}{\sqrt{3}}$

The $\alpha$ axis is fixed to the magnetic axis of Phase A, while the $\beta$ axis leads $\alpha$ by $90^\circ$ electrical. Both axes remain stationary relative to the stator.

---

3. Park Transformation ($\alpha-\beta \to d-q$)

Published by Robert H. Park in 1929, this transformation projects the stationary $\alpha-\beta$ orthogonal variables into a synchronous rotating reference frame ($d-q$) locked to the rotor’s permanent magnet flux vector.

Executing this rotation requires precise, instantaneous knowledge of the rotor electrical angle ($\theta_e$), supplied by an automotive resolver or absolute encoder:

Park Transformation Frame Alignment:
              q-axis (Quadrature / Torque)
               ^
               |       beta
               |      /
               |     /
               |    /   d-axis (Direct / Flux)
               |   /   ^
               |  /   /
               | /   / theta_e (Rotor Electrical Angle)
               +----+-------------> alpha

Park Transformation Matrix:

Applying a planar rotation matrix by angle $\theta_e$ yields:

$\begin{bmatrix} i_d \\ i_q \end{bmatrix} = \begin{bmatrix} \cos\theta_e & \sin\theta_e \\ -\sin\theta_e & \cos\theta_e \end{bmatrix} \begin{bmatrix} i_\alpha \\ i_\beta \end{bmatrix}$

In algebraic scalar form:

$i_d = i_\alpha \cos\theta_e + i_\beta \sin\theta_e$

$i_q = -i_\alpha \sin\theta_e + i_\beta \cos\theta_e$

---

4. Physical Meaning of $d-q$ Variables

In steady-state operation at constant speed and torque, $i_d$ and $i_q$ are completely time-invariant DC quantities!

1. Direct Axis Current ($I_d$): Aligned parallel to the rotor magnetic flux axis. Controls magnetization of the machine:
   • $I_d = 0$: Natural permanent magnet flux is unperturbed (standard SPMSM base speed mode).
   • $I_d < 0$: Injected counter-flux weakens net air-gap flux (Field Weakening), enabling high-speed operation above base speed.
2. Quadrature Axis Current ($I_q$): Positioned $90^\circ$ electrical ahead of the direct axis. Directly controls the electromagnetic torque output.

ANALOGY — The Park transform decouples the complex, coupled AC machine into two independent SISO (Single-Input Single-Output) channels, behaving exactly like a separately excited DC machine with independent field ($I_d$) and armature ($I_q$) control.

---

5. Inverse Transformations (Inverse Park & Inverse Clarke)

After the PI controllers process errors in the DC domain to generate required voltage demands ($V_d^*, V_q^*$), these commands must be mapped back to the physical three-phase stator windings.

Inverse Park Transform ($d-q \to \alpha-\beta$):

$\begin{bmatrix} V_\alpha^* \\ V_\beta^* \end{bmatrix} = \begin{bmatrix} \cos\theta_e & -\sin\theta_e \\ \sin\theta_e & \cos\theta_e \end{bmatrix} \begin{bmatrix} V_d^* \\ V_q^* \end{bmatrix}$

Inverse Clarke Transform ($\alpha-\beta \to a-b-c$):

$V_a^* = V_\alpha^*$

$V_b^* = -\frac{1}{2} V_\alpha^* + \frac{\sqrt{3}}{2} V_\beta^*$

$V_c^* = -\frac{1}{2} V_\alpha^* - \frac{\sqrt{3}}{2} V_\beta^*$

---

6. Real-Time Execution Pipeline in Inverter Microcontrollers

In automotive traction inverters (e.g. Infineon AURIX TC3xx or TI C2000), this mathematical loop runs continuously at the PWM carrier rate ($10\text{ kHz} - 20\text{ kHz}$, loop time $50 - 100\ \mu s$):

       [ Phase Currents: ia, ib ] ----> (Clarke) ----> [ i_alpha, i_beta ]
                                                              |
       [ Resolver: theta_e ] ------------------------> (Park)
                                                              |
                                                              v
   I_d_ref ----(+)---> [ PI Controller ] ---> V_d*          [ i_d, i_q ]
                ^                                             |     |
                |---------------------------------------------+     |
                                                                    |
   I_q_ref ----(+)---> [ PI Controller ] ---> V_q*                   |
                ^                                                   |
                |---------------------------------------------------+
                                |
                                v
               [ V_d*, V_q*, theta_e ]
                                |
                                v
                          (Inverse Park)
                                |
                                v
                     [ V_alpha*, V_beta* ]
                                |
                                v
                      (SVPWM Modulator)
                                |
                                v
                     [ 6x Gate Drive PWMs ]

---

7. Summary

• Tracking high-frequency three-phase AC currents directly with linear PI controllers is impossible due to inevitable phase lag.
• The Clarke transform projects three-phase $120^\circ$ currents into a two-axis stationary orthogonal frame ($\alpha-\beta$).
• The Park transform uses instantaneous rotor electrical angle ($\theta_e$) to rotate the frame into the synchronous $d-q$ coordinates.
• In the $d-q$ frame, AC waveforms collapse into constant DC values, allowing decoupled control of magnetic flux ($I_d$) and torque ($I_q$).
• Inverse Park and Inverse Clarke transforms convert controller outputs back into physical stator voltages for Space Vector PWM modulation.

Next Lesson

• MOTOR-17 — Field-Oriented Control (FOC): Closed-loop current regulation, cross-coupling decoupling networks, and tuning high-bandwidth 10 kHz traction current loops.