STM32H563 Peripheral Saturation – Moving to Multi-MCU (Stepper + DC Load Slaves) – Architecture Review

[Prathapraj]

Hello ST Community,

I am working on a commercial vending machine controller and would like to get feedback from ST engineers and experienced STM32 designers regarding a system architecture change.

---

Problem Statement

Our current system uses a single MCU, but we are now hitting hard peripheral limits, mainly:

• Timer availability (stepper + DC motor control)

• GPIO interrupt count

• GPIO pin availability

• Increasing real-time jitter when multiple loads are active

This is not a software optimization issue anymore; it is a resource saturation issue.

Because of this, we are planning to move from a single-MCU architecture to a multi-MCU system.

---

Proposed Architecture

• Master MCU: STM32H563ZIT6

  • Handles:

    • All sensors (ADC / digital)

    • Control logic & state machine

    • Recipe execution

    • Safety checks

    • UI communication

  • No direct motor timing generation

• Slave MCUs (2x):

  • Dedicated load-handling controllers

  • One for stepper motors

  • One for DC motors / PWM-based loads

  • Slaves perform only deterministic actuation

  • No sensors, no business logic

• Inter-MCU Communication:

  • Considering CAN (FDCAN) as primary bus

  • RS485 as an alternative / secondary option

I have attached a block-level architecture diagram for reference.

Current Peripheral Usage on STM32H563 (New Approach)

Peripheral Instances Used Usage

ADC	1	12 channels
I2C	I2C1, I2C2, I2C3	3
TIM	TIM1, TIM2, TIM3, TIM4, TIM5, TIM8	6
UART / USART	USART1, USART2, USART3, UART5, UART7, UART8, UART9, UART12	8
DAC	DAC1	1
GPIO Interrupts	–	14
GPIO Inputs	–	20
GPIO Outputs	–	5
CAN	FDCAN1	1

Even with this split, the master MCU is still heavily utilized, which is why motor timing is fully offloaded to slave MCUs.

Questions for Discussion

1. Is this master + load-slave MCU split the correct long-term approach for vending machine / appliance-class products?

2. From an STM32 architectural perspective, are there recommended design patterns for timer- and interrupt-heavy motor systems?

3. CAN vs RS485 for multi-MCU control:

   • CAN seems preferable for arbitration and fault handling

   • Any STM32H5-specific considerations?

4. Are there common pitfalls (boot sequence, synchronization, fault recovery, debugging) that should be addressed early?

5. From a product lifetime standpoint (8–10 years), does this architecture look robust and maintainable?

---

I would appreciate feedback from:

• ST application engineers

• Developers with multi-MCU STM32 designs

• Anyone with experience in CAN-based industrial / appliance systems

Thank you in advance for your insights.

Best regards,
Prathapraj S,

Boatload Minds Pvt Ltd.

@stm32h5,can,motor-control,system-architecture,multi-mcu

[mƎALLEm]

Hello and sorry for the late reply,

Not possible to answer all of your questions and propose an exhaustive answer. But my proposal why don't you scale up the STM32 by using a more performant MCU such as STM32H7. STM32H7 dual core (Ex: STM32H745) can be a good candidate for your application. It can reach up to 480MHz of system clock in LDO mode at VOS0.

And the partitioning of your application will be as the following:

CM7 is controlling the motors / Graphics if available

CM4 is managing the sensors and external communication: FDCAN, USARTs, RS485 etc ..

Please refer to the application note AN5557 "STM32H745/755 and STM32H747/757 lines dual-core architecture" / especially the section  4. Application partitioning examples.

The figure from that application note:

Hope that I've provided an idea on how to proceed.

Good luck.