nrf52 require reset form debugger to start

[patlas]

Hi, I am developing my own program based on serial example. Project is based on gcc compiler under eclipse. I have preapared my own board with rigado module (nrf52832). HW works correctly. But my program works only after programming device. After unplug / plug power it seems to stay in halt mode. But only after connecting JLink (from devboard) and click RESET / BOOT button on devkit board my device starts working properly. After reset the SWD line may be disconnected and everything works fine but after power restart the same approach with reset is required.

Instead pressing RESET/BOOT button I can send command reset_no_halt (rnh) over JLink.exe with the same result.

Is there any define statement or something similar that I should have set?

P.S. Such a problem does not occure with ones of example provided that why I am sure that HW is correct.

[Nguyen Hoan Hoang]

Look for rdimon in the linker settings.  Remove all of them.  rdimon requires the jtag to be connected to start. 

[Allen Foster]

Hi Nguyen,

There is no reference to rdimon or any type of semi-hosting debug. We are building using gcc.

This code works fine on the nRF52-DK but on our chip down solution it has problems. Once programmed, we follow the usual Nordic makefile convention of doing "nrfjprog -f nrf52 --reset" to restart the SoC with the updated application. Oddly this does not cause a restart and to begin execution I need to use JLinkExe to connect and the use 'r' at which point the SoC restarts and execution of the application begins. I cannot understand this issue as I thought nrfjprog was essentially a wrapper around JLinkExe?

[Nguyen Hoan Hoang]

Probably the reset command sequence of the nrfjprog that is different.  I guess you don't have the reset pin (p0.21) connected.  It would help to have a hardware reset pin.

[Allen Foster]

We currently have p21 pulled high with a 10k resistor,

[Nguyen Hoan Hoang]

Is that pin configured as reset pin ? If it does, try to connect that to the swd connector to see if that helps.

[Allen Foster]

It is just pulled high at the moment but yes, that is a good idea and I can break it out to the SWD connector. Thanks for responding!

[Allen Foster]

Okay, one interesting thing.

In my makefile at present, after programming I did nrfjprog -f nrf52 --reset and the device will not start.

However if I change this to nrfjprog -f nrf52 --pinreset then the device does start.

It is my understanding the --pinreset is a 'hard' reset whereas --reset is a 'soft' reset

I am beginning to think this might be something related to the RBP (Read Back Protection) on the nRF52. Although I cannot see where I explicitly enable RBP.

[Nguyen Hoan Hoang]

The RBP affect softreset only if it is enabled.  If not, it is not.  Something I noticed. When softdevice is present.  The softreset does not seem to work properly.  Probably nrfjprog has the same issue.  

[Allen Foster]

We eventually found the problem - which relates to the following.

https://devzone.nordicsemi.com/f/nordic-q-a/28472/automatic-control-of-mode-ldo-and-dcdc

We had a call to sd_power_dcdc_mode_set(NRF_POWER_DCDC_ENABLE); but without the necessary inductors required to make this work on our custom board.

Frustratingly, the custom board started up normally with SDK12 but not with SDK13 if sd_power_dcdc_mode_set(NRF_POWER_DCDC_ENABLE) was called.

Anyway, the boards are now working. Hopefully this is of help to someone else.

[Nguyen Hoan Hoang]

Odd, this doesn't explains why hard reset works in your previous observation.  If DC/DC was the cause, it should not work either ways.  Have you re-tried without pin reset connected ?