HSE clock hardware design clarification

[CProt]

Hello everyone,

I am designing a board using STM32L496RE and I'm currently in the HSE clock design phase.

I have chosen a 16MHz crystal with CL 18pF. I have been following the AN2867 (https://www.st.com/content/ccc/resource/technical/document/application_note/c6/eb/5e/11/e3/69/43/eb/CD00221665.pdf/files/CD00221665.pdf/jcr:content/translations/en.CD00221665.pdf) and the formula at the end of page 12, setting CL=18pF, CL1=CL2. According to the datasheet of STM32L4x (https://www.st.com/resource/en/datasheet/stm32l496re.pdf), bottom of page 163, I can assume that CS=10pF.

Having done the calculations, I end up with CL1=CL2=16pF. However, I have looked at two other designs (not published by ST) where they have used the load capacitance in both branches of the oscillator, which makes me wonder if I should just set CL1=CL2=18pF.

Thank you, any clarification is appreciated,

Chris

[Tesla DeLorean]

I suspect the L4 is more forgiving than earlier STM32 families as the design/process technology has evolved.

I would expect CL1=CL2=8-9pF

Discuss your design specifics with the FAE supporting your account (see local sales office to pin one down)

Be prepared to test as built performance and change the BOM as necessary. Consider testing expectations on a NUCLEO board by removing MCO clock and adding components.

[MikeDB]

All capacitors that small valued are the same physical size so get a selection pack and find the value that sets the frequency the most accurately for your particular PCB layout. You'll find the stray capacitance of the board often swamps other capacitances anyway.

[Tesla DeLorean]

Indeed

Often a poorly characterized design will still start/oscillate but can take longer than necessary. One should benchmark the start time of a new implementation, and also when getting PCB from new supplier.

[CProt]

Hello,

Thank you for your replies. I am assuming most of this will be trial and error then. I have noticed that many higher frequency crystals require load capacitance of <10pF which make that formula unusable anyway.