STM32L073z-eval 32.768 LSE issue

[AChas.1]

Hello, i have some trouble with my L073z-eval board. I try to use RTC with LSE and found it very inaccurate. I checked the frequency using TIM21 (capture-compare mode with remapped LSE as input) and see very instabbility value - some times value goes below 29.700 kHz. I check this algorithm with LSI and this work fine (stable result). In additional LSE very sensitive to touch by finger or any capacitive stuff. This immediately affected to LCD Glass and frequency. I checked gerber and dont found any ground connected to crystal's shield....

[waclawek.jan]

The EVAL boards sometimes have multiple connections to a pin, perhaps through jumpers, resistors or solder bridges - isn't something connected to the LSE pins here?

Also, try to change LSEDRV and see if that results in any change.

JW

[AChas.1]

Crystal directly connect to board through 0603 0 Ohm resistors, solder bridges (which redirect pins to external header) are open. I changed LSEDRV to 11b (default in CubeMX was be 00b). Now crystal less sensetive to touch by finger even possible to measure frequencies by osciloscope but just in 1 of two pin. If i connect osciloscope probe to another pin LSE is stop.

[AChas.1]

Some additional info:

1. Low drive - frequency instabbility and value dance between 508-510 Hz (at the RTC callibration output) measured by osciloscope
2. Medium low drive - best accuracy 512-512.8 Hz
3. Medium hight drive - good stability but frequency is 520-520.8 no bounce
4. Hight drive - worst case 520 - 529 Hz