Slightly overclocking STM32G070 without external oscillator

[Slina]

It runs at 64MHz, and I need it running at 65 MHz. I've managed to create 65 MHz clock in PLL using multipliers/dividers in STM32 Cube IDE, but I'm not able to test it before I order the PCB-s, which I'm not intending to do before I confirm this would work. I managed to generate a code after frequency set, although I did get a warning that 64 is max frequency. I understand that there are MCU-s that would do the job, but I can get STM32G070 very cheaply and would like to try it if it's possible. I also understand the risks, but it's only 1.5% overclock.

So the question is: would the generated code work despite the STM32 Cube IDE warning?

[Ozone]

Not sure what the problem is.

Surely most microcontroller will work beyond their specified clock limit, much more than 1.5%.

The maximum value the vendor gives is valid even for the worst case (e.g. 85°C ... 105°C).

So, at room temperature, you might get away with 30% or more above, without problems.

But there is no guarantee, and no warranty.

[TDK]

> would the generated code work despite the STM32 Cube IDE warning?

It's going to depend on your code and your environment, specifically temperature, in which you're operating.

Based on my experience I would expect it to work fine with 1.5% overclocking if you're running at ambient. But you'll need to test for yourself.

[Tesla DeLorean]

I'd lean toward it being viable, the flash wait states provide some margin. would need to see if additional wait states could be set.

Generally one would try to find the ceiling to determine where the critical paths lay. Here it would likely be the prefetch/line caching mechanism.

ST provides themselves a pretty wide berth.

For the longest time ARM stopped people from quoting high numbers for CM0 as it would eat away at market share for CM3/CM4 designs.

I've got some ICs here with CM3 clocking at 384 MHz in current process geometries out of TSMC

And I've seen NXP claim 200 MHz for a CM0 with a RAM-ONLY memory space.

[Slina]

Thanks for all your answers, but I still didn't get the answer to the main question: will the clock be set accordingly despite the IDE warning that the clock settings are outside allowed boundaries? I can't check it yet since I only have STM32H7 nucleo board, and not yet STM32G070.

[Ozone]

> but I still didn't get the answer to the main question: will the clock be set accordingly despite the IDE warning that the clock settings are outside allowed boundaries?

Most probably.

The IDE just accesses a part database, and checks your settings against the datasheet value. It gives the same warning if you try 250MHz.

But best is to try it out.

[TDK]

> will the clock be set accordingly despite the IDE warning that the clock settings are outside allowed boundaries?

Yes it will. You'll have to acknowledge a warning. This is different from your original question, which was in fact answered:

> would the generated code work despite the STM32 Cube IDE warning?

[Slina]

I wasn't very clear. Sorry about that.