nRF5340 Internal RC Oscillator Question for DTM USB Configuration

[gfgf2400a]

Hi,

I used both SDK v2.3.0 and v2.4.0, I was able to get DTM example working over USB CDC ACM with nrf5340 DK PCA10095. I switched to my custom board which has no 32KHz XTAL between P0.00/XL1-P0.01/XL2. I changed both direct_test_mode\prj.conf and direct_test_mode\child_image\remote_shell\prj.conf to use internal RC oscillator to include:

CONFIG_CLOCK_CONTROL_NRF_K32SRC_RC=y
CONFIG_CLOCK_CONTROL_NRF_K32SRC_500PPM=y
CONFIG_CLOCK_CONTROL_NRF_K32SRC_SYNTH=y

My issue is right after I programmed the combined application and network image (merged_domains.hex) with nRF Connect Programmer, I was able to see the USB VCOM (when nRF Connect Programmer trace has "10:34:02.245 Getting serialport options from persistent store D0C417815ECA4DEC.pc-nrfconnect-programmer") for 7 seconds then disappeared.

I browsed on DevZone and saw a few questions about internal RC oscillator but I am not sure what is the root cause in my case. Please advise.

Regards

[Vidar Berg]

Hello,

For boards that don't include the 32K XTAL, we generally recommend selecting the LFRC source. It is not as accurate as LFSYNTH, but it gives you lower sleep currents since it does not rely on the HFCLK. You can use our online power profiler to compare the difference between LRFC and LFXO in terms of power consumption.

CONFIG_CLOCK_CONTROL_NRF_K32SRC_RC=y
CONFIG_CLOCK_CONTROL_NRF_K32SRC_500PPM=y
CONFIG_CLOCK_CONTROL_NRF_K32SRC_SYNTH=y

Here you are selecting both LFRC and LFXO, but it is only possible to select one. It is also important that you select the same clock source for all images in your project.

It should be sufficient to apply CONFIG_CLOCK_CONTROL_NRF_K32SRC_RC=y for each build

Best regards,

Vidar

[gfgf2400a]

Hi Vidar,

Thank you for the suggestion. With just CONFIG_CLOCK_CONTROL_NRF_K32SRC_RC=y alone, I am not able to see USB VCOM port from Windows device manager at all. With just CONFIG_CLOCK_CONTROL_NRF_K32SRC_SYNTH=y alone, Windows shows USB VCOM for about 7 seconds as before. I attach direct_test_mode/prj.conf here:

# Configure assertions
CONFIG_ASSERT=y
CONFIG_ASSERT_NO_COND_INFO=y
CONFIG_ASSERT_NO_MSG_INFO=y

CONFIG_HW_STACK_PROTECTION=y

CONFIG_CONSOLE=n
CONFIG_UART_CONSOLE=n

CONFIG_LOG=y
CONFIG_LOG_PRINTK=y
CONFIG_USE_SEGGER_RTT=y
CONFIG_LOG_BACKEND_RTT=y

# Use necessary peripherals
CONFIG_NRFX_TIMER0=y
CONFIG_NRFX_TIMER1=y
CONFIG_NRFX_TIMER2=y

CONFIG_FEM_AL_LIB=y

CONFIG_CLOCK_CONTROL=y
CONFIG_CLOCK_CONTROL_NRF=y
CONFIG_CLOCK_CONTROL_NRF_K32SRC_SYNTH=y

[gfgf2400a]

I attach \direct_test_mode\child_image\remote_shell\prj.conf here

CONFIG_STDOUT_CONSOLE=y

CONFIG_SERIAL=y
CONFIG_UART_INTERRUPT_DRIVEN=y
CONFIG_RING_BUFFER=y

CONFIG_MBOX=y
CONFIG_IPC_SERVICE=y
CONFIG_IPC_SERVICE_BACKEND_RPMSG=y

CONFIG_NRFX_UARTE0=y

CONFIG_HEAP_MEM_POOL_SIZE=4096

# Config logger
CONFIG_LOG=y
CONFIG_LOG_PRINTK=y
CONFIG_USE_SEGGER_RTT=y
CONFIG_LOG_BACKEND_RTT=y
CONFIG_LOG_BACKEND_UART=n

CONFIG_REMOTE_SHELL_TX_RING_BUFFER_SIZE=1024

CONFIG_BOARD_ENABLE_CPUNET=y

CONFIG_CLOCK_CONTROL=y
CONFIG_CLOCK_CONTROL_NRF=y
CONFIG_CLOCK_CONTROL_NRF_K32SRC_SYNTH=y

[Vidar Berg]

Hi,

To help narrow down the problem, could you please try to flash the hex files below to your app and net core to see if you get the same result? They were built from the original DTM sample in SDK v2.4.0 with the USB overlay. The only change I made was to disable the DCDC (just in case you don't have the inductors) and change the clock source to LFRC.

1778.GENERATED_CP_APPLICATION_merged_domains.hex

0383.GENERATED_CP_NETWORK_merged_domains.hex

Thanks,

Vidar

[gfgf2400a]

Hi Vidar, your loads work on my custom board. This is great. The EE on my custom board said we have the inductors. I will look into on how to disable DCDC. Please share the CONFIG to disable DCDC. Thank you so much for your helps. 

[Vidar Berg]

Hi,

That is great. I didn't really suspect the inductors to be the problem; I just disabled the DCDC regulators for good measure. It could be that the inductor(s) are mounted, but one is missing or has the wrong value. Typically, if you don't have any inductors mounted but enable the DCDC, the device will become "bricked" as the unstable supply will prevent you from re-programming it.

Relevant Kconfig symbols for the DCDC regulator are:

CONFIG_BOARD_ENABLE_DCDC_APP, BOARD_ENABLE_DCDC_HV, and CONFIG_BOARD_ENABLE_DCDC_NET. 

Ref. https://github.com/nrfconnect/sdk-zephyr/blob/main/boards/arm/nrf5340dk_nrf5340/Kconfig#L8 

[gfgf2400a]

Hi Vidar,

I removed the 32kHz on nRF5340 DK PCA10095. Without disabling DCDC regulators, I was able to see the same problem (USB VCOM disappeared after a few seconds) as on my custom PCBA.

I then added:

CONFIG_BOARD_ENABLE_DCDC_APP=n
CONFIG_BOARD_ENABLE_DCDC_HV=n
CONFIG_BOARD_ENABLE_DCDC_NET=n

to both direct_test_mode/prj.conf and \direct_test_mode\child_image\remote_shell\prj.conf. The problem still exist. Please let me know how you did disabling DCDC regulators exactly.

Thanks,

Dennis

[Vidar Berg]

Hi Dennis,

gfgf2400a said:

CONFIG_BOARD_ENABLE_DCDC_APP=n
CONFIG_BOARD_ENABLE_DCDC_HV=n
CONFIG_BOARD_ENABLE_DCDC_NET=n

It's sufficient to apply these configurations to the FW image(s) running on the app core since the regulators are controlled from the application domain. Applying the same settings for a netcore image won't have any effect.

I disabled the DCDC regulators by adding the configurations above to the \direct_test_mode\child_image\remote_shell\prj.conf file. 

 

Best regards,

Vidar