softdevice_fault_handler error

[Gaétan]

Hi,

I have developed a application on nRF51822 that is detect eBeacon and read some information in Beacon and Scan reponse frames.

There is about 200 systems deployed.

I have 4 boards, that is reset periodicly after ~7 seconds ! Only 4 boards at this time, are this problem...

Whean I use debug: 

softdevice_fault_handler() is called ... and the error handler call NVIC_SystemReset();

The  softdevice_fault_handler() function is called with id=1, pc=0xf764, info=0...

I don't use timeslot.

What is the problem ? ... ?

Regards,

Gaétan

EDIT: In my projet I have configured scan_window = scan_interval = BLE_GAP_SCAN_WINDOW_MAX = 0x4000

Now I have changed this value to 2500 and now there no reset !

What is the max value for scan_interval and scan_windows ?

I have indentified that, if there are a lot of ebeacon or smartphone near  the nrf51822, the Reset appear more frequently !

There is a maximum of bluetooth detection in a scan_windows ?

Softdevice S130 V2.001

Regards,

[Kenneth]

Hi,

Is it possible that you are disabling (blocking) interrupts somewhere in your code and/or have any interrupts with priority 0 and/or use timeslot api?

It may also be related to accuracy of the LFCLK is out of specified tolerance.  It's important that the correct XTAL accuracy/ppm value for the crystal is set when enabling the SoftDevice.

Best regards,
Kenneth

[Gaétan]

No interrupt with priority = 0
I think they no blocking or disabling interrupt
No  timeslot used

The mounted XTAL is: https://www.digikey.fr/product-detail/fr/abracon-llc/ABS06-32.768KHZ-T/535-10104-1-ND/2089921 (20ppm)
I have tried to change clock_lf_cfg.xtal_accuracy to NRF_CLOCK_LF_XTAL_ACCURACY_250_PPM, and it's the same...

Regards,
Gaétan

[Kenneth]

This looks good. Have you tried scan window = 0x3FFF and scan interval = 0x4000?

I can check with the designers.

Best regards,
Kenneth

[Gaétan]

I have tried scan window = 0x3FFF and scan interval = 0x4000, 

No change, reset occur every ~10 seconds

[Kenneth]

Are you able to replicate this on a devkit and can send us a precompiled hex file we may try here?

I may check with the developers in any case.

Best regards,
Kenneth