nRF52840 custom board won't boot without debugger reset

[vincent.golle]

My nRF52840 custom boards seem to be unable to boot without RTT viewer started of JLink commander connected.

The hardware configuration is taken from the nRF52840 PS 1.0 and is circuit #3:

The problem occurs with and without softdevice, with my code but also the ble_peripheral/hrs example.

I have already checked that:

- Semihosting is disabled

- It occurs when logs are enabled & disabled

Strangely this seems to be connected to clocks, but happens even if the RC internal osc is selected as the source for LF clock.

Please help ?

[Håkon Alseth]

Hi,

 

Does this occur even if you use one of the precompiled .hex files in the SDK?

What SDK version and compiler/IDE are you using?

How are you loading the binary to the chip (nrfjprog, using an IDE, J-Link Commander)? If using nrfjprog, you should add a "nrfjprog --run" or "nrfjprog -r" afterwards as nrfjprog leaves the chip in a halted state.

 

Best regards,

Håkon

[vincent.golle]

I havn't tried with thr precompiled hex from the examples.

I am using GCC 6.3 q2 update as recommended.

The issue occurs with the custom board but not with my pca10056 V0.9.0

After some debugging I found the "stuck in boot" board came from an erroneous nrf_sdh_enable_request().

So this seems connected to Low frequency clocks not being able to start as expected if no debugger is connected

at boot.

In this configuration if I get past softdevice enabling, peripherals such as UART seem not to be working properly and

can't decode bytes...

[Stian Røed Hafskjold]

Hi, can you provide us with a better resolution picture of the schematics, and also the PCB layout? Thanks.

Does the application start if you use the internal LFRC clock, not the crystal?

What frequency are you using on the UART?

Are you using DCDC? Can you try to run the code without DCDC enabled?

[vincent.golle]

Hi Stian,

I will try to clarify as much as possible:

The board not booting issue was corrected using fix for errata 187 for USBD.

The remaining issue can be called: UART not working without debugger reset.

It happens using both the RC and the XTAL.

Symptoms are: UART only works after a debugger reset boot type.

Stange thing is that I no longer get any error..

The baud rate on the UART is 9600, and DCDC is currently disabled (NRFX_POWER_CONFIG_DEFAULT_DCDCEN and NRFX_POWER_CONFIG_DEFAULT_DCDCENHV are set to 0).

myStravaB_V2.pdf

[vincent.golle]

Attached is the bottom copper layer of the board.

The workaround for this issue is for now to trigger a watchdog reset if RESETREAS is 0.

myStravaB_V2_copper.pdf

[Stian Røed Hafskjold]

I'm not really sure what is going on here. So the UART is only working after a soft reset, and you can only see this symptom on your custom PCB?

When you enable the UART the first time, before the reset, do you get any error messages, and what happens when you try to transmit data? Did you check with a logic analyzer? Are the lines initialized at the correct levels, including flow control lines?