STM32WBA54 LSE not stable

[gkkr]

Hello, 
There’s probably a solution, but I haven’t been able to fix it. Could you help me?

In the circuits I’ve designed recently, the LSE crystal either doesn’t work at all or its operation is unstable. Generally, when debugging, it gets stuck at the SystemClock_Config(); section and won’t proceed.

I’ve designed circuits using different MCUs and have encountered this problem across the board.  In my latest circuit design (4-layer FR-4), I used an STM32WBA54CGU6 MCU and, for the crystal, the ABS07-LR-32.768KHZ-6-1-T recommended in the AN2867 Application Note. I’m using an 8pF C0G 0402 package capacitor as the load capacitor. I also tried a 6.8pF capacitor, but it didn’t work at all. The 8pF capacitor is also unstable. I tried driving it in different modes (LOW, MEDIUM, MEDIUMHIGH, etc.), but it remains unstable. I didn’t adhere perfectly to the layout rules, but I haven’t experienced problems in previous circuits where I followed them even less strictly. I’m trying to keep as few signals as possible passing beneath the crystal, leaving it as a single ground plane wherever possible, and using vias for ground connections. I’ve attached the layout images.

I’ve also used other MCUs such as the STM32WL55CCU6 (2-layer FR-4), STM32WB55CGU6 (2-layer FR-4) and STM32L053R8T6 (2-layer FR-4). I encountered similar issues with those as well. 
Sometimes, when the circuit isn’t working, the problem goes away if I heat the crystal with hot air. then when gets cold again it doesnt work. I suspect it might be an impedance issue, but at 50k ohms, I think the crystals are actually quite low compared to the standard.
Has anyone else encountered a similar problem and found a solution.
 

stm32wba top layer

stm32wba 2nd layer gnd poligon

 

stm32wba 3rd layer (normally 3V3 power layer. Perhaps it would have been better if it hadn’t been underneath the crystal, but as there was a GND layer in between, I didn’t make an issue of it)

stm32wba bottom layer

 

2 layer STM32L0 top

2 layer stm32L0 bottom

 

[mƎALLEm]

Hello ​@gkkr and welcome to the ST community,

Your question is related to STM32 wireless product (STM32WBA54). So, it should be asked in STM32 MCUs Wireless forum board.

PS: I’ve already moved the post.

[Imen.D]

Hello ​@gkkr ,

Please check the STM32WBA5xxx errata-sheet to confirm whether your issue is described as a limitation, such as:

• LSE crystal oscillator may be disturbed by transitions on PC13
• LSE in bypass mode may not be functional.

Kindly let us know if the description in the errata reflects the problem you encounter.

[gkkr]

PC13 is not in use.

The LSE is not in bypass mode; it is operating in normal mode with a direct crystal connection.

 

While reviewing previous threads, I saw some people mention that there were errors like this in the code generated by Cubemx. Is that true? Is there such a bug? We generate the code using Cubemx.

[AScha.3]

Set LSE drive High.

Try with 2x 10pF caps. , COG or NPO .

And clean any remains/flux from soldering.

 

[Laurent LOUAZON]

Hello,

See also below wiki page.
HSE and LSE: How to select crystals and center oscillators with STM32WBAx - stm32mcu

I see no concern using ABS07-LR-32.768KHZ-6-1-T.
As it is a 6pF Xtal, expected load cap values should probably be around 6pF.
 

I also do not see big concern for layout (even if we do not recommend Vdd plane on layer 3 - better to route Vdd with lines in star connection).

 

Note you may also decide not to used LSE at all (you save the cost of Xtal and you only have a very small penalty of current consumption)
How to run an RF application on the STM32WBA without LSE crystal - stm32mcu

 

Regards

Laurent