STM32CubeMX allows max CPU frequency even with VOS low; intentional or not?

[PhucXDoan]

I'm using the NUCLEO-H7S3L8 and the max CPU frequency is 600MHz.

By default, CubeMX sets VOS is set to 0 (low).

But if this were the case, wouldn't it be violating the datasheet where it states 400MHz is the max CPU clock?

I've been using 600MHz clock without ever setting VOS high and I haven't gotten any issues, so I'm not sure whether or not it's fine to leave VOS alone.

[Sarra.S]

Hello @PhucXDoan, 

Thank you for reporting this behavior, I have submitted a ticket for further investigation by STM32CubeMX team (Internal ticket number: 200435) 

[filipxsikora]

Hello,

This is actually "correct". At least from the standard H7 line PoV, the 'Power Regulator Voltage Scale 0' was always the highest VOS, achieving the highest CPU Freq. The scale is inverse, the lower the number, the higher is the regulator output voltage and therefore higher achievable CPU Freq.. For some reason.

And sice the H7RS line is just a derivate of the original H7, same core, etc. I'm guessing they just went with it as it was.

If you check the source code:

https://github.com/STMicroelectronics/stm32h7rsxx-hal-driver/blob/7ca2e07ca21bc66b53654e845b4c85c884343b60/Src/stm32h7rsxx_hal_pwr_ex.c#L350 

It is described in the comment of the function which sets the VOS.

[PhucXDoan]

I think it is indeed backwards, but the documentation is also a bit inconsistent about this.

If the RM says VOS=0 is "low level", this would naturally map to mean "VOS low", and as the DS page I shown above, "VOS low" implies FCPU MAX of 400MHz. Funny stuff.

[Sarra.S · Accepted solution]

Hello again @PhucXDoan and @filipxsikora, 

This is indeed a naming issue, the RM will be corrected with: 

VOS0 is high level

VOS1 is low level 

So, I confirm that there is no issue in CubeMX.