aactuator
ONE JOINT / ONE CLOSED LOOP

From command to motion.

One live joint. See the currents, follow the switching, and account for the energy.

Live joint C++ / WASM

Loading C++ engine…
Same running joint · select a stage to inspect itWaiting for the engine

01 / Set the command

Select a stage. Its explanation and measured values stay on this page.

Waiting for the first snapshot.
Clarkeabc → αβParkαβ → dqId*, Iq* − Id, IqCurrent errorPI / predictiveVd, Vq · limitInverse transforms→ SVPWM
CHANGE THE SCENARIO
● Commands applied

Live signals

Phase currents ia / ib / ic · A
iaibic
d–q currents Measured dq · A
IdIqIq*
SVPWM duties A / B / C · 0–1
ABC

Same run · 1 kHz sampled histories. Averaged plant: not switching-current ripple.

QDD / 6:1 compliant drive0.000 s
Drag to orbit · scroll to zoom · click a partIllustrative geometry · not vendor CAD
Ready when the engine loads

This run: averaged inverter · one output joint · compliant two-sided stop. No resolved switching edges or general 3D collisions.

Plant stateReferencerad · latest 4 simulated seconds

Plant state versus reference · 1 kHz telemetry. The feedback stage above shows the held ADC observation; this chart does not resolve PWM edges.

SVPWM → inverter

— V— AInverter DC terminals
DC+DC−

6PWM: timer supplies complementary signals and dead time.

Frozen-duty reconstruction from this joint; not a second integration or actual Vgs. OFF can still conduct through a diode. Resolved switching experiment ↗

Current sensor → ADC → feedback

Shunt / gainia [A]Analog filterbandwidthADC + delay→ FOC
True iaAnalog iaHeld ADC ia

Noise is added at capture, after the analog filter. ADC samples current; a DAC is not a motor-current smoothing stage.

Motor torque–speed

LIVE PATH

Live: gearbox-input torque vs motor RPM, not a rated envelope. Calculate replaces this path with a separate shaft dyno.

Independent default dyno · 28 points

Efficiency comparison

Live η: —

Live η: 200 ms DC→external-load means; gaps at stall, regeneration or storage transients. Reference curves: independent DC→motor-shaft cold dyno, not this live load.

Temperature · this run

WindingCaseFET

Physical simulation seconds. No accelerated heat or invented PCB temperature.

Losses · 200 ms mean

Await energy samples

Copper, inverter conduction/dead-time proxy, friction and gear dissipation. Iron and switching-energy losses: not modeled.

Run interpretation & diagnostics

What is happening in this run?

Loading the joint

Live interpretation will appear here.

External load / stop reaction—
Winding / FET temperature—
Voltage saturation—
Temperature belongs to this run. Switching scopes, 600 s heating and efficiency sweeps below are separate calculations, not extra views of this joint.

Go deeper, without mixing experiments

Independent calculation

Switching, heat & motor maps

Run a fresh C++ experiment with its own settings. It does not inherit or control the live joint above.

Open the calculation bench →
Read / inspect saved results

Understand the signals

Learn the equations or inspect reproducible native traces. These reference plots do not update with this joint.

Control theory → · Signals & losses →
Recorded MuJoCo examples 5 event loops · not the live joint

Previously recorded C++–MuJoCo experiments, including 3D impacts. Play, pause or open a full recording. Each loop restarts the event; it is not continued live motion. Recording provenance ↗