FOC (PMSM/BLDC) – Start-up and low-speed tuning

Motor state diagram for sensorless FOC start-up and low-speed operation

What we do

We make sensorless field-oriented control (FOC) behave as if you had a position sensor – especially at start-up and low speed, where estimators usually struggle. We design robust observers, clean start-up sequences and handover logic so your PMSM / BLDC drive delivers smooth torque, low acoustic noise and reliable restarts.

Outcomes we aim for

  • Safe start-up under load without stall, kickback or harsh transients.
  • Stable low-speed torque with minimal ripple and cogging.
  • Seamless transition from open-loop to closed-loop in sensorless mode.
  • Extended operating window across temperature, DC-bus sag and saliency.
  • Clear error recovery paths: detect loss-of-sync and restart gracefully.

Services

  • Observer design and tuning – SMO, MRAS / Luenberger, PLL / back-EMF, EKF where budgets allow; bandwidth versus noise trade-offs and delay compensation.
  • Start-up sequences – initial alignment, I–F ramp or open-loop V/f, current-limited ramps and seamless handover criteria into closed-loop operation.
  • Low-speed enhancements – high-frequency injection (HFI) for saliency tracking, demodulation filters and anti-cogging feed-forward.
  • Parameter adaptation – Rs(T) and flux linkage correction, inertia / friction identification, DC-bus feed-forward and field-weakening integration.
  • Ripple and acoustics – SVM timing, dead-time compensation, MTPA biasing, dq cross-coupling compensation and notch / averaging filters for speed estimation.
  • Protection and recovery – loss-of-lock detection, stall handling, re-sync strategies plus brown-out and low-temperature behaviour.
  • Data and test harness – logging of estimators, currents, voltages and dq variables; script-driven sweeps to validate the operating window.
  • Integration – MCSDK / HAL / LL modules, ISR-safe estimator updates and LUTs for temperature / speed regions.

Your deliverables

  • Sensorless control pack: tuned observer parameters, start-up / handover logic and safety thresholds (header / CSV plus notes).
  • Code patches: estimator modules, HFI demodulation (if used) and handover state machine for CubeIDE / MCSDK (HAL / LL).
  • Validation report: start-up success matrix, low-speed torque plots, acoustic / ripple metrics and corner-case results.
  • Test scripts and logs: reproducible sweeps (Python / MATLAB) and an acceptance checklist.
  • Handover session (60–90 minutes) – walkthrough and Q&A with your team.

Technology stack

  • Motors: IPMSM / SPMSM with surface or interior saliency; BLDC in FOC or 6-step to FOC migration.
  • Observers: SMO, MRAS / Luenberger, back-EMF PLL, EKF; optional HFI (sine or square) for demanding low-speed specifications.
  • Peripherals: TIM1 / TIM8 with centre-aligned PWM, injected ADC sampling synchronised to PWM, DMA pipelines, OPAMP / COMP, DFSDM.
  • MCUs: STM32F3 / F4 / F7, G4, H7, U5.
  • Tooling: X-CUBE-MCSDK, STM32CubeIDE / CubeMX, HAL / LL and FreeRTOS-compatible projects.

Engagement flow

  • Discovery (30 minutes) – mechanics, load profile and observed symptoms.
  • Baseline and data – capture start-ups, low-speed runs and estimator traces.
  • Design and integration – select observers, implement start-up / handover logic and add low-speed aids (HFI if needed).
  • Validation – window sweeps, temperature and DC-bus corners, plus abuse tests.
  • Handover – parameters, code, reports and team training.

What we need from you

  • Motor datasheet (pole pairs, Rs, Ld / Lq if known, Ke / Kt), inverter schematic and current-measurement method (shunt / Hall / LEM).
  • DC-bus range, PWM frequency, ADC sampling windows and estimated inertia / friction if available.
  • Operating window: min / max speed and torque plus start-up load conditions.
  • KPIs such as ≥ 99 % start-up success, low-speed ripple ≤ X %, restart time ≤ Y s.

Example packages

  • Sensorless audit – root-cause analysis, observer / start-up plan and acceptance tests.
  • Tuning sprint – observer tuning plus start-up / handover integration and initial validation.
  • Production hardening – HFI or advanced low-speed aids where needed, corner robustness, documentation and training.

Example use cases

  • E-bikes and scooters – hill starts without kickback and quiet crawl speeds.
  • Robotics / AGV / AMR – precise low-speed positioning and smooth micro-movements.
  • Pumps and compressors – start-up under load, sag-tolerant restarts and reduced thermal stress.
  • Fans / HVAC – whisper-quiet operation at low RPM with stable torque.

FAQ

Can you run sensorless at zero speed?
We use HFI to track saliency at standstill and near-zero speed, then blend to back-EMF / SMO as speed rises.

Is HFI always required?
No. Many SPMSM / BLDC systems meet their requirements with tuned SMO / MRAS and robust start-up / handover logic; HFI is reserved for tough low-speed specs or strong saliency.

Will this fit my MCU budget?
Yes – estimators are ISR-tight. HFI adds moderate CPU load; we size it to your loop time.

How do you avoid loss-of-sync?
With plausibility checks on back-EMF / observer residuals, adaptive bandwidth and torque-limited ramps; on errors we fall back to a safe state and restart under control.

Related services: STM32 real-time optimization , Motor drive protection , Electric drive systems .

Contact us