6. Validate & Beyond - MPLAB Device Blocks for Simulink

Home > MPLAB Device Blocks for Simulink > Examples > Tutorial — Picooz Helicopter > 6. Validate & Beyond

Goal. Verify that the real hardware response matches the simulated prediction — and draw the conclusion that gives MBD its name.

The final check — does the real rotor behave like the model said it would?

With the controller running on the target, capture the closed-loop response the same way you captured the open-loop response on page 2: External Mode, Data Inspector, export to workspace. Overlay this hardware response with the simulation output from page 5.

Simulation vs. real hardware — large-step closed-loop response

The simulated and real curves lie on top of each other. No tuning iteration was performed on the hardware. This is the deliverable of Model-Based Design — a faithful simulation predicts reality, so you can explore the design space on your PC and commit the result to hardware with confidence.

The MBD promise, summarised

Simulate the model + controller faster than real time on the PC.
Generate the embedded C code automatically.
Validate on target — and the response matches the simulation.

Fidelity of the model is the enabler. A faithful model makes simulation a reliable prediction of reality.

Going further — where to explore next

This tutorial deliberately stopped at the basic closed-loop-validated controller. Here is where to go from here:

Tune the controller live with External Mode

Replace the fixed setpoint with a Slider in Simulink. Drag it at runtime — the motor follows. Add live read-out of the integral term and watch it compensate disturbances. Inject a step load (press the rotor briefly); watch the controller recover. All of this happens without recompiling.

See External Mode & PIL .

Multi-rate scheduling

The controller runs at 1 kHz; BEMF sampling runs at 20 kHz; UART logging runs at 100 Hz. Put each in its own triggered subsystem with distinct sample times — the toolbox’s Rate-Monotonic Scheduler handles real-time scheduling automatically.

See Scheduler & Multi-Tasking .

Port to another target

Change the Master block chip from dsPIC33AK512MPS506 to a dsPIC33CK, a PIC32MK or a SAM E70. The controller’s C code is unchanged. Peripheral pinning adapts.

Fixed-point vs. floating-point

The dsPIC33A has a hardware single-precision FPU — use floats guilt-free. On a dsPIC33C without FPU, the same controller works but at a higher CPU cost. The toolbox supports Fixed-Point Designer conversion for maximum efficiency.

The INSA Lyon 2022 course goes deep into this topic.

Apply to a different plant

The four-phase workflow (acquire → identify → control → validate) is identical for:

PMSM / BLDC motors — see Motor Control Examples and the ctrl-elec project .
Power converters — PFC on dsPIC33A is a future tutorial in this series.
Industrial axes, thermal systems, battery models — anywhere you have a measurable state and a controllable input.

Advanced controllers

Once Super-Twisting is familiar, explore:

Higher-order sliding mode — arbitrary convergence order for systems where chatter on \(\ddot{s}\) also matters.
Extended-state observers — estimate disturbances online and feed them forward.
Model Predictive Control (MPC) — solve an optimisation at each step; increasingly practical on modern dsPIC33A with FPU and 200 MIPS.

Source material and credits

Figures and walkthrough are derived from the 30-minute live demo delivered at École Centrale Nantes on 2025-12-04 (Model-Based Design: From Laboratory to Embedded Deployment).
INSA Lyon 2022 Rapid Control Prototyping course — 3-hour French reveal.js slide deck covering the Picooz + data types + multi-rate scheduling.
ctrl-elec project — advanced motor control research and teaching, INSA Lyon × Microchip.
ICEM 2024 Tutorial — academic reference (Delpoux, Kerhuel).

Original Picooz work: R. Delpoux (INSA Lyon, Laboratoire Ampère) and L. Kerhuel (Microchip Technology), with assistance from many students and interns over the 2012–2025 iterations.

You’ve finished the tutorial

You now know how to:

Acquire data on a running MCU using External Mode.
Fit a physics-based model to that data using base MATLAB or a paid toolbox — your choice.
Design a robust non-linear controller.
Simulate the closed loop before touching hardware.
Generate embedded C code and deploy it in one click.
Validate on hardware — and see that the simulation was right.

That is the entire MBD workflow. Every project you tackle next — PMSM, PFC, autopilot, whatever — is a specialisation of these six pages.

Back to the series index → Tutorial — Picooz Helicopter

Back → 5. Simulate & Deploy