HSE "overclocking" ?

[LCE]

Heyho,

anybody ever put more than 50 MHz from an active oscillator to an F767 / H723 / H735 HSE input?

I need a VCO there, and I have a few with 51.2 MHz, so less than +5% over.

I'm pretty sure it won't damage anything, but before I start (de-) soldering these terrible pin-less packages, I better ask.

[Uwe Bonnes]

It will probably work in normal ambient. I would not use it for a mass production.

[AScha.3]

good point.

i used a 45,1584 M oscillator , no problem.

and look at pin speed: io port can drive 220MHz ...so input should do same. (at 11 core speed setting)

[Tesla DeLorean]

My expectation is that it will go higher, significantly. The reason it gets tested/characterized up to 50 MHz is for the Ethernet use case.

Perhaps when it comes to testing, build a couple of units with 66 MHz XCO's and run over temperature in a chamber. What you're not going to be able to do, that ST would, is cherry pick devices from the process window corners to use as test subjects.

Alternatively you might also see how much headroom the design has, by pushing in a clock via a frequency generator.

[FBL]

Hello @LCE 

The external clock signal has to respect the I/O static characteristics as referenced in datasheets. Also, we recommend consulting the application note: Oscillator design guide for STM8AF/AL/S, STM32 MCUs and MPUs - Application note

[Tesla DeLorean]

This is more of a fHSE max discussion about CMOS sources, say TCXO, VTCXO, OCXO, etc

[LCE]

Thank you all for your input!

I also think that if the input is specified for up to 50 MHz, its actual capability is probably much higher.
Even more so with an active oscillator, so that the STM's crystal circuit is out of the equation.

Will try...