从电机速度控制到FOC：电流、转矩和嵌套回路

Change parameters and verify

Open the panel, then press Run to load Python. You can stop execution and reset parameters. Results are computed on this device. No Python installation is required.

Local execution steps below are optional for reproducing the source results; they are not required for the browser experiment.

The experiment controls are in English.

Open experiment panel in a new tab

Download reproduction source

Explore current coordinates and electrical angle

The panel above is a one-instant coordinate exercise. Supply actual current components in the true rotor frame, mechanical angle, pole-pair count, an electrical-angle error, and a hypothetical torque constant. It constructs balanced phase currents and transforms them back using the supplied estimated angle. These are imposed currents, not current references tracked by a controller. No motor, inverter or local Python is required.

The convention is explicit: α aligns with phase a; β is +90° counterclockwise from α. The true d axis follows rotor flux at θe=p θm, with fixed zero offset 0; q leads d by +90°. Positive angles are counterclockwise. p counts pole pairs, not poles. The display includes unwrapped electrical angle and its modulo-360° direction. Four pole pairs and 90° mechanical angle give 360° electrical angle, equivalent to 0° for the transforms. Changing p here changes the angle mapping only; the separately supplied torque constant stays fixed.

We use amplitude-invariant Clarke normalization with zero sequence fixed to zero. With R(θ)=[[cosθ,−sinθ],[sinθ,cosθ]], first form [iα,iβ]=R(θe)[id,iq]. Generate ia=iα, ib=−iα/2+√3 iβ/2, ic=−iα/2−√3 iβ/2. Thus ia+ib+ic=0. Clarke recovers iα=(2/3)(ia−ib/2−ic/2) and iβ=(ib−ic)/√3. These are instantaneous signed amperes, not phase RMS values. This amplitude normalization is not power invariant: ia²+ib²+ic²=(3/2)(id²+iq²).

The estimated angle is θ̂e=θe+Δθe. Park computes [id̂,iq̂]=R(θ̂e)ᵀ[iα,iβ]. Consequently id̂=id cosΔ+iq sinΔ and iq̂=−id sinΔ+iq cosΔ. At true id=0, iq=3 A and error +90°, the displayed estimated components become id̂=3 A, iq̂≈0. The original phase currents stay the same when only this angle error changes. The diagram shows the same stationary current vector with true and estimated axes; axis-arrow lengths are illustrative and do not encode current. All current scales are shared across A/B.

The inverse uses the same estimated angle: [iα′,iβ′]=R(θ̂e)[id̂,iq̂], then the same inverse Clarke equations. The maximum absolute reconstructed-phase difference should be small even at an incorrect angle. This residual measures algebraic consistency and floating-point roundoff; it cannot establish encoder alignment, sensor accuracy or a working FOC loop. Presets compare +/−90° errors, 180° error, mixed d/q current, zero current, negative q current, and electrical-angle wrapping. A zero vector hides every angle error.

A separate simplified nonsalient PMSM relation gives Te=Kt × true iq. Default Kt=0.10 N m/A and iq=3 A give 0.30 N m. Kt is defined for this amplitude-invariant peak-dq convention and includes the motor-specific conversion; it is not a phase-RMS constant to substitute blindly. The model omits reluctance torque and assumes constant flux. Angle error changes the displayed components but does not change imposed current or modeled torque. A real feedback loop using the wrong angle may change currents; that behavior is not simulated. Nonzero d current is a coordinate example, not a field-weakening or current-limit recommendation.

No PI control, PWM, DC-bus limits, voltage equations, speed dynamics, saliency, field weakening, sensor noise, heating, protection or hardware operation is modeled. Inputs are synthetic teaching assumptions. Use the separate PID Lab for controller behavior and the reducer Lab for mechanical transmission arithmetic.

Reference reviewed 2026-09-20: TI PMSM FOC report, sections 4–5. This exercise states its normalization and sign convention directly; it does not reproduce the report's motor-control implementation.