Code generation for STM32 MCUs using MATLAB® and

simulink®

March 2020
 Systems check

• Mandatory Software
• From Mathworks
• MATLAB®, Simulink and Embedded Coder (2018B)
• From STMicroelectronics
• STM32CubeMX (5.6.0)
• STM32-MAT/TARGET toolkit to develop STM32 applications (STM32MatTarget_5.6.0_setup.zip)

• One of following Toolchain

• EWARM from IAR (8.32.3)
• MDK-ARM from Keil (5.29.0)
• STM32CubeIDE from STMicroelectronics (1.3.0)

• Hardware
• Any STM32 based electronic application board with SWD/JTAG connection.
• STLinkV2/V3 or 3rd party debug probe
• Virtual Com Port implemented between STM32 and Computer

2
 Key information

• Arm®/ST/MathWorks initiative
• Optimized Cortex-M code for Embedded Coder based on Arm’s CMSIS DSP SW library
• Simulink Peripheral blockset provided by ST based on STM32Cube abstraction layers supporting most of
STM32 portfolio
• Compiler / IDE integration for MATLAB, based on Arm Keil , IAR and STM32CubeIDE

• Enables Model-Based Design Workflow for STM32 MCUs

• Create executable models
• Generate code automatically
• Automate software integration

• Reduces cost and time, improves quality

• ~33% cost & time reduction in ECU development
• ~50% fewer specification errors

3
 Why matlab/simulink on STM32 ?

• Simulation maybe of interest when creating a mathematical algorithm

• Matlab/Simulink enable STM32 simulation and on chip algorithm execution
• The algorithm can use the actual STM32 peripherals, ending up with a complete
processing 100% done in the MCU, with no written lines of code and a complete
model-driven approach, less error prone

4
 Quick description of tools

Toolchain

High level language for complex

Matlab Collection of embedded A toolchain from one of our
calculations STM32Cube
software components, highly partners is required to compile
Embedded and link C code generated by
portable from one STM32 to
Software Embedded Coder,
another
STM32CubeMX and STM32Cube
Graphical development environment embedded software
Complete environment of simulation
Simulink
and implementation of embedded
systems. Configuration software tool
on the PC, able to generate
STM32CubeMX
initialization C code based
on user choices
C code generation for embedded
Embedded systems.
Coder Embedded system interface

5
 Using scenarios and tools
• Step 1: Pure simulation MATLAB/
• Everything done on the PC Simulink

• Step 2: Processor-in-the-loop (PIL) MATLAB/ STM32Cube

Embedded
• Algorithm fully executed on STM32 MCU Simulink Software
Core
Toolchain
• Data (Input or output) exchanged between Embedded
STM32CubeMX
MATLAB/Simulink and STM32 MCU via UART Coder
*: used only for UART

• Step 3: Everything on STM32 MATLAB/ STM32Cube

Embedded
Simulink
• Data (input or output) gathered from STM32 Software

peripherals (ADC, Timers, …) and algorithm fully Embedded Toolchain Core

executed on STM32 STM32CubeMX
Coder ADC

• Step 4: External mode TIM

• Same behavior than step 3. Data monitoring from I2C

Simulink via UART. Not possible to modify STM32

configuration during External Mode.

6
Step 1: pure simulation
 MATLAB® & simulink

• Not a subject for this presentation : You do your model ready, simulate and test it
on the PC.

• General purpose for MATLAB®/Simulink

• Doesn’t need STM32-MAT/TARGET toolkit.

• Contact MathWorks for MATLAB®, Simulink trainings

8
Step 2: processor-in-the-loop
 PIL overview
• PIL is used to run MATLAB® algorithm on STM32 device
• To see if STM32’s computational results are numerically
equivalent to PC-based simulation results.
• To measure directly the time to execute the Simulink
model steps (profiling)
• PIL simulation uses serial port to send data from Simulink to
STM32 and receive Simulink processed data back from
STM32
• PIL simulation doesn’t run in «real-time»
• Simulink is «master» sending asynchronously data to STM32 through
serial port.
• STM32 device waits for data from Simulink, process data (in real-time)
and sends result back to Simulink through serial-port.

• PIL simulation doesn’t process real data from STM32

peripherals.
• STM32 peripherals (ADC etc…) are not used.
• Only STM32 USART peripheral is used to communicate with Simulink.

10
 Simulink PIL start

• Set simulation duration time and click run simulation green button
• Example: Simulate 1 second (given 1000 values from testdata, one for each 1 ms)

11
 Code generation overview

• PIL referenced model runs into STM32 target while the simulation model runs on
PC
• Automatic conversion of PIL model to .c/.h files
• Automatic call to STM32CubeMX to create project
• Manually build and download project to STM32 target from selected IDE

12
Step 3: everything on STM32
 Handle data through real peripherals

• Configure your peripherals through a bridge with the STM32CubeMX tool

• Flash your device and run !
• More info at www.st.com/stm32-mat-target

14
 Build application

• Generate code for the Simulink application

• Press «Build Model» to automatically generate C code and IDE project.

15
 MATLAB® code generation

• STM32 peripherals driver code is generated in .c/.h

files which name is created using name of Simulink
model and peripheral name.

16
 STM32CubeMX code generation

• STM32CubeMX process
• STM32CubeMX is automatically called from MATLAB®
when «Download Application» has been selected
from the Model Configuration window.

• STM32CubeMX generates configuration code.

• STM32CubeMX adds necessary library files. .c/.h library files from HAL STM32 libraries.

• STM32CubeMX generates toolchain project including files generated from MATLAB® Click « Open Project » to
automatically open project
using selected toolchain.

• The project generated by STM32CubeMX can be opened

17
Step 4: external mode
 Simulink external mode

• Monitoring
• As long as STM32 application is running, Simulink model parameters can be edited

• Monitoring means that only parameters can be modified. C generated code can't be modified. For
example, it is not possible to replace x with + as x C code function has been generated.

19
 Summary

• New Support Package for MATLAB, Simulink, and Embedded Coder

• Generate optimized code
• Automate build, integration and verification steps

• Now with ST and MathWorks you can:

• Accelerate time to market
• Reduce development costs
• Improve product quality

• …Join the Design Community

20
 Summary

Thank you

© STMicroelectronics - All rights reserved.

The STMicroelectronics corporate logo is a registered trademark of the STMicroelectronics
group of companies. All other names are the property of their respective owners. 21