STM32H755XIH6 Custom Board Lockup When Switching Clock Source — Recovery Needed via Boot Pin

[sarun]

Hello everyone,

We are using a custom STM32H755 controller board, and we are encountering an issue where the controller locks up whenever we switch the core clock source (for example, switching from HSI to HSE).

According to the evaluation board user manual, in case of an LDO configuration mismatch, the controller can enter a lockup state, and the suggested recovery method is to pull the Boot pin high and erase the flash. However, on the evaluation board (STM32H755 Nucleo/Eval), we have never faced this lockup while switching clock sources.

What could be potential reasons causing such behaviour on our board?

Any guidance or suggestions would be really helpful!

Thanks in advance!

 

[mƎALLEm]

Hello,

---

@sarun wrote:

According to the evaluation board user manual, in case of an LDO configuration mismatch, the controller can enter a lockup state, and the suggested recovery method is to pull the Boot pin high and erase the flash. However, on the evaluation board (STM32H755 Nucleo/Eval), we have never faced this lockup while switching clock sources.

---

That issue is related to a wrong power config selection: the H755 EVAL board is SMPS config by HW and if you set the power config to LDO by software you will face this lockup state. 

So the question is: what is the power config you have on your custom board?: LDO or SMPS?

What do you mean by : "switch the core clock source (for example, switching from HSI to HSE)." 

Is it on the fly? if yes how?

If no, you need to provide the board's schematic + the clock config (preferably in an ioc file).

 

 

[sarun]

Hi @mƎALLEm ,

We are using the SMPS configuration.

This issue occurs during reflashing, not during normal operation (on-the-fly).

Sorry, I can't share the schematic on a public platform.
Is it possible to raise a support ticket for this?

[mƎALLEm]

Please share at least the HSE part from the schematics + your ioc file.

Sorry it's not possible to raise an internal ticket for this kind of request as we can manage it in the community (at least at this stage).