Internal 8 MHz factory-trimmed oscillator STM32F10x

[yah996]

Posted on March 10, 2009 at 10:52

Internal 8 MHz factory-trimmed oscillator STM32F10x

[yah996]

Posted on May 17, 2011 at 13:05

Hello to everyone,

I have not seen any board design without an external 8MHz crystal yet. Even all sample boards that I know, are using an external oscillator. Is there any reason not use only the internal oscillator? Thanks for your advice, K.

[lil-vince]

Posted on May 17, 2011 at 13:05

Hi,

I think it´s for a better precision and with the internal clock you can´t have a SYSCLK higher than 64MHz ( (8/2MHz)*16 )

[jj]

Posted on May 17, 2011 at 13:05

The internal osc. is spec'ed at +/- 3% over temperature. This may have serious impact on ALL Async (i.e. Uart) communication, Timers etc.

Only real justification - imho - is when you ''really'' need another GPIO. Suggest that you build/test your Application first with crystal - only then would I re-test with internal osc. Should this ''saving'' cause reliability issues you will pay a terrible price - far in excess of that ''saved!''

[sofiene]

Posted on May 17, 2011 at 13:05

Hi yah996,

You can refer to the AN2868 ''STM32F10xxx internal RC oscillator (HSI) calibration '' that describes a method to minimize the HSI frequency errors. On the other hand, nothing prevents the boards that you mentionned can use the internal oscillator since the user has the choice between using the internal and the external oscillator. B.R. M3allem

[yah996]

Posted on May 17, 2011 at 13:05

Thanks for all the answers.

It's now clear to me, i did not realized the spec of, (+/- 3%) K.

[jj]

Posted on May 17, 2011 at 13:05

Final point - ''just in case:''

Like M3allem's calibration post. However - even after that time/effort - you are ''still'' susceptible to the relatively large ''drift w/temperature.'' There's a reason ''every'' Eval/Dev. board has a crystal...