2. Acquisition with External Mode - MPLAB Device Blocks for Simulink

Home > MPLAB Device Blocks for Simulink > Examples > Tutorial — Picooz Helicopter > 2. Acquisition with External Mode

Goal. Record how the motor reacts to a known PWM input signal, so we have the data needed to fit a mathematical model in the next step.

💻 Try the interactive version. This chapter also exists as an auto-graded hands-on lesson built with MATLAB’s R2026a Course Builder. You start from a minimal Simulink model containing just the Master and UART blocks; three auto-graded tasks guide you through adding the Chirp, the PWM HS FEP and the ADC HS 12b with correct wiring and parameters. Each task shows green/red feedback immediately.

In MATLAB R2026a or newer, from the Command Window:

>> picCourse

This opens the lesson in the Course Builder runtime. On R2025b or older releases, picCourse prints a pointer back to this page — the Course Builder framework is not available there.

External Mode — the diagnostic tool you cannot live without

External Mode is the Simulink feature that turns MBD from “compile-flash-print-debug” into a real-time instrument. Once enabled:

Your Simulink model runs on the MCU.
Simulink on the PC is connected live over UART (using the industry-standard XCP-on-SxI protocol).
Scopes, Data Inspector and parameter tuners inside Simulink update in real time — while the MCU is running, without recompile.

Drag a gain slider to change a PI gain → watch the motor response on the scope instantly. Start recording → export the captured timeseries straight to the MATLAB workspace for analysis. That is what we will do for the Picooz.

See External Mode & PIL for the deep-dive.

The acquisition Simulink model

⬇ tutorial_picooz_acquire.slx ⬇ tutorial_picooz_acquire_extmode.slx

The model is intentionally minimal:

[ Chirp Generator ] ──► [ PWM HS FEP block ] ──► (motor on hardware)
                                                    │
                                                    ▼
                                             [ ADC HS 12b block ] ──► [ External Mode scope ]
                                                                         ↑
                                                                  (live-streamed to PC)

Everything runs on the MCU; the scope on your PC is just a viewer for the data the MCU is sending over UART.

The main blocks

Block	Role	Reference
Master	Target chip + clock + fuses	—
PWM HS FEP	20 kHz PWM carrier, duty driven by the chirp signal	—
ADC HS 12b	Sample BEMF during the PWM-OFF window	—
UART Configuration	External Mode transport	—
Chirp	Simulink Sources library — swept sine from 0.1 Hz to 10 Hz over 30 s	—

Why a chirp, not a step?

A chirp (swept-sine) excites every frequency in the 0.1 Hz–10 Hz range in one 30-second experiment. Its autocorrelation is low and it visits every operating point — exactly the property we need for grey-box identification.

Compare:

Input signal	Frequency content	Suitable for ID?
Step	Only excites the dominant time constant	✗ Poor — misses fast dynamics
Multi-step sequence	Better, but discrete	~ Ok, many experiments
White noise / PRBS	Excites everything	✓ Good — industry standard
Chirp	Controlled sweep, smooth, bounded	✓ Excellent — and visually interpretable

For educational clarity chirps win: you can see on the scope which frequencies are still able to drive the rotor and which ones the mechanical inertia filters out.

Step-by-step — run the acquisition

Open tutorial_picooz_acquire_extmode.slx.
In the Master block, confirm the target chip matches your board (default: dsPIC33AK128MC106).
Check PWM settings: 20 kHz carrier, single channel, duty = block input.
Check ADC settings: triggered on PWM-OFF window, 12-bit resolution.
Select Simulation → Mode → External.
Press Monitor & Tune (recent release) or Run on hardware (older releases).

Within ~10 seconds the toolbox:

Generates C code from your Simulink model.
Compiles with XC-DSC.
Flashes the MCU.
Starts the model running on the target.
Opens a live scope view of the streamed signal.

Capturing data to MATLAB

Open the Data Inspector, select the BEMF signal, and Export → To workspace. You now have timeseries u(t) (the chirp) and omega_meas(t) (the measured BEMF) in your MATLAB workspace — ready for identification.

Reference dataset shipped with the tutorial

If you don’t have hardware handy (or just want to reproduce the tutorial exactly), the bundled dataset below was captured on a real Picooz + Curiosity dsPIC33AK512MPS512 with this exact acquisition model. Use it unchanged through the rest of the series.

⬇ tutorial_picooz_chirp.mat (clean (t, u, omega_meas) for identification) ⬇ tutorial_picooz_chirp_log.mat (raw Simulink Dataset log) ⬇ tutorial_picooz_process_log.m (raw-to-clean conversion script)

What’s in tutorial_picooz_chirp.mat:

Variable	Size	Description
`t`	6001 × 1	Time vector, 5 ms period, 30 s total
`u`	6001 × 1	PWM duty cycle, chirp in [−0.3, +0.3]
`u_abs`	6001 × 1	`abs(u)` — the rotor only responds to magnitude, single-quadrant drive
`omega_meas`	6001 × 1	Rotor speed proxy (BEMF-derived voltage) in [0, 3.6] V

The waveform:

Observation: the rotor accelerates quickly at low chirp frequency (left side), then has visibly reduced amplitude at high frequency (right side) — the mechanical low-pass due to rotor inertia + aerodynamic drag. This is exactly what identification will quantify.

External Mode — advanced features to explore

These go far beyond the basic acquisition above and are worth knowing even after you finish the tutorial:

Feature	When it pays off
Live parameter tuning	Place a Slider Gain in your controller and drag it at runtime. Explore a PI gain sweep in seconds rather than minutes.
Log-to-disk with picgui	Simulink Data Inspector buffers in RAM. For multi-hour experiments, the blockset’s `picgui` tool streams the signals straight to disk. See Data Visualization .
Triggered capture	Start recording on an event (rising edge of a digital signal). Useful for transient capture of one-shot startup sequences.
Per-signal log rate	Current at 20 kHz, setpoint at 10 Hz — each signal captured at its own rate, keeping the UART bandwidth in check.
XCP-on-SxI	The underlying transport is a standard automotive XCP protocol. Any CAN-bus calibration tool (CANape, INCA, …) can talk to your firmware.

What’s next

With t, u and omega_meas in hand, we can fit a mathematical model of the motor. The next page shows five different ways to do it — most of them needing no MATLAB add-on toolbox.

Next → 3. Identification — Five Methods

Back → 1. Hardware