STM32 F4 crystal calculation and discrepancy with NUCLEO boards

[ELowe]

Hi!

I am building PCB around STM32F469ZET6.

Now LSE and HSE crystals and accompanying capacitors need to be chosen.

AN2867 says capacitors values should be CL1 = CL2 = 2 x (Cl - Cs)

Cl – crystal’s load capacitance

Cs – stray capacitance of oscillator circuit (mystical value equal to 5 pF)

….so if I choose crystal with Cl = 12,5 pF, I get:

CL1 = CL2 = 2 x (12,5 pF – 5 pF) = 15 pF.

Nice………but before going further with Gm’s, I chose to check my theory on freely available NUCLEO Board, just to check my understanding is correct.

If I look at NUCLEO-F439ZI’s schematics (this is F4 , but not mine):

https://www.st.com/resource/en/schematic_pack/nucleo_144pins_sch.zip

….I see that ST chose for this board to use LSE crystal NX3215SA-32.768KHZ-EXS00A-MU00525, which has Cl = 12,5 pF, so following AN2867, capacitors C36 and C35 should have been:

C36 = C35 = 2 x (12,5 pF – 5 pF) = 15 pF

….but ST chose to use 2 pF ??!?!?!?!?!?

Why?

[STOne-32 · Accepted solution]

Dear @ELowe ,

Thank you for the question to all community members . You need to stick strictly to AN2867 , please note that STM32F469 is among our F4 series generation 2 for LSE , where we introduced low and high drive modes to support Crystal up to 9pF CL , you can refer to table 7 and see compatible crystal not exceeding 1.5uA/V in gm crit . 12,5 pF CL Crystals are not compatible with this MCU . 
our Nucleo boards are generic for many MCUs  built with a custom crystal.

So if we summarize , you can assume Cs between 3pF to 5pF maximum , using a crystal  of CL=9pF in high drive mode ( not by default ) you need a CL1=CL2 between 12pF to 8pF - the exact value will depend on your PCB and is the one that output :32,768KHz with the best accuracy and best PPMs .

hope it helps you .

STOne-32