facing system config crystal and RTC Crystal problem in STM32G0B1CET6

[ashutamar]

Hi Team,
I am currently working with the STM32G0B1CET6 microcontroller and using two Abracon crystal parts, ABL-8.000MHZ-B2 and AB38T-32.768KHZ-B. However, I am encountering issues with the crystal when I use heavy ( like code more than 50% ram & flash of this STM-g0) percent of my code in STM. Attached is the schematic for your reference., 
 

Thank you,
Ashu Tamar

[STOne-32]

Dear @ashutamar ,

Can you please details what kind of issues , waveforms of signals. Try to output value of the LSE or HSE on MCO pin to understand the case . Also the PCB routing . I would recommend to try to isolate where these issues are coming ?

R10 is not necessarily and may create issues - it is not recommended . For LSE 32,768Khz to know the selected drive level with LSE - I see such crystal is 12,5pF CL and is not compatible with lower drives .

 

this is our reference App Note : https://www.st.com/resource/en/application_note/an2867-oscillator-design-guide-for-stm8afals-stm32-mcus-and-mpus-stmicroelectronics.pdf

hope it helps you .

STOne-32 

[Tesla DeLorean]

I'm hard pressed to draw a relationship between code size and crystal function.

Focus on other potential issues as more direct causes.

Make sure you actively clear auto/local variables.

Make sure you use appropriate flash wait states for speed and voltage considerations.

 

[AScha.3]

>I'm hard pressed to draw a relationship between code size and crystal function.

:) :)

"crystal " has nothing to do with your  code size.

So you do something in your code, that is very wrong.

+

What are the "issues with the crystal  " ? which crystal (you show both) ?

How you know, its the crystal and not your code ??

[waclawek.jan]

 

Maybe consequence of this erratum?

What exactly are the symptoms?

JW

[Tesla DeLorean]

That sounds like linear code execution issue where the data feed to the prefetch wraps within the bank due to lack of broader context. ART related? Or Exception/IRQ has a start address close to the boundary

 

ie 0x08007FFE rolling into 0x08008000 if the banks were 32KB (adjust for architecture..), instead fetching 0x08000000