nRF9160 PWM

RE: nRF9160 PWM

paul_tanner — Tue, 26 Jan 2021 10:02:58 GMT

I did not find a pwm sample in nrfconnect sdk 1.4 so I thought I should try this. I see a reference to pwm.h and wondered where this should be?

On a related note there is no .overlay so how could it know which pins to use?

Regards.

RE: nRF9160 PWM

Håkon Alseth — Thu, 23 May 2019 07:03:50 GMT

Great to hear!

Cheers,

Håkon

RE: nRF9160 PWM

Gurkan — Wed, 15 May 2019 14:26:39 GMT

Thanks Hakon,

I've done the changes on my end,it's working all good now.

RE: nRF9160 PWM

Håkon Alseth — Wed, 15 May 2019 08:08:41 GMT

The fix for this is upstream, and will be merged in ncs soon:

https://github.com/zephyrproject-rtos/zephyr/commit/c0a15c8bd6764ce11a61e2cc3d9a0deedc2ec0b6

Kind regards,

Håkon

RE: nRF9160 PWM

Håkon Alseth — Mon, 13 May 2019 13:26:30 GMT

Sorry, I do not have an update on this. You can go directly on the nrfx_pwm driver (bypassing the zephyr shim layer), or see if the other pwm functions work better (pwm_pin_set_nsec)

Kind regards,

Håkon

RE: nRF9160 PWM

Gurkan — Fri, 10 May 2019 17:53:39 GMT

Hello,

Is there any timeline about fixing this bug?

Or any workaround for high frequency signals? I need to run my PWM signal over 10kHz

Thanks.

Gurkan

RE: nRF9160 PWM

Håkon Alseth — Mon, 29 Apr 2019 07:05:35 GMT

Thank you, I wish you the same.

The bug is reported, so the developers are now aware of the issue.

Kind regards,

Håkon

RE: nRF9160 PWM

Gurkan — Fri, 26 Apr 2019 14:30:20 GMT

Thank you Håkon,

Have a nice day.

Regards,

Gurkan

RE: nRF9160 PWM

Håkon Alseth — Fri, 26 Apr 2019 14:15:02 GMT

This is weird. If I input > 1000 on both period and pulse, it behaves "as expected", however; if I provide pulse=250 and period=500, it goes haywire.

Pulse is correct-ish, 252 us, while the period is way off at 753 us. I would expect DC of 50%, but get approx. 25%.

I'll report it internally.

Best regards,

Håkon

RE: nRF9160 PWM

Gurkan — Fri, 26 Apr 2019 13:06:49 GMT

The result was same with your hex too. You are right that was the cross talk. I haven't checked the schematics if the P0.2 at P17 is directly connected to the microcontroller. Apperantly P0.2 is controlling a digital switch which is connected to the P17 and led1.

I've switched the output of PWM_0 to another pin and its working well now.

On my previous message I said I cannot get PWM signal below %50 but it's reverse actually. I cannot get PWM signal above %50 duty cycle.

/**
 * @brief Set the period and pulse width for a single PWM output.
 *
 * @param dev Pointer to the device structure for the driver instance.
 * @param pwm PWM pin.
 * @param period Period (in microseconds) set to the PWM.
 * @param pulse Pulse width (in microseconds) set to the PWM.
 *
 * @retval 0 If successful.
 * @retval Negative errno code if failure.
 */
static inline int pwm_pin_set_usec(struct device *dev, u32_t pwm,
				   u32_t period, u32_t pulse)

With this function if I put period 250, the period of signal is 500us.

when the pulse 1, the duty cycle is 1%;

When the pulse 250, the duty cycle is 50%.

If I try to put something bigger that 250, I'm getting invalid pulse_cycle error.

Regards,

Gurkan

RE: nRF9160 PWM

Håkon Alseth — Fri, 26 Apr 2019 09:09:14 GMT

That looks like crosstalk. Are you sure you are scoping the correct pin? If you are using the nRF9160-DK, the LEDs are not routed to the pin lists, so you have to scope from the left side of the LED1.

Did you try the .hex in my .zip to see if that one worked?

Kind regards,

Håkon

RE: nRF9160 PWM

Gurkan — Thu, 25 Apr 2019 15:45:14 GMT

Hello Håkon,

I've tried your sample and I've realized that I actually have pwm signal on P0.2 but with a super small peak-peak voltage level. I'm attaching my scope output. After realizing that I've tried my first sample and the result is same, pwm signal exist but with a low voltage output.

Do you think it's a hardware issue?

My another question is about timings of signal?

In your example the duty cycle keep going up and between 0-255us. And the period is set to 255us.

/**
 * @brief Set the period and pulse width for a single PWM output.
 *
 * @param dev Pointer to the device structure for the driver instance.
 * @param pwm PWM pin.
 * @param period Period (in microseconds) set to the PWM.
 * @param pulse Pulse width (in microseconds) set to the PWM.
 *
 * @retval 0 If successful.
 * @retval Negative errno code if failure.
 */
static inline int pwm_pin_set_usec(struct device *dev, u32_t pwm,
				   u32_t period, u32_t pulse)

My scope shows the period of signal 500us and duty of signal keep changing between 50% to 100%. How can I get a pwm signal below 50%? If i try to give pulse time, bigger than period, I'm getting an error as "<err> pwm_nrfx: Invalid pulse_cycles 4096, > period_cycles 4080." as expected.

RE: nRF9160 PWM

Håkon Alseth — Thu, 25 Apr 2019 08:08:00 GMT

I played a bit around with the PWM, and here's my example:

devzone.nordicsemi.com/.../pwm.zip

Could you try this one and see if it works?

The build/zephyr/merged.hex includes spm (secure_boot). Try this .hex first to see if it works at your end as well, prior to building.

Kind regards,

Håkon

RE: nRF9160 PWM

Gurkan — Wed, 24 Apr 2019 14:50:23 GMT

CONFIG_PWM=y
CONFIG_LOG=y
CONFIG_PWM_LOG_LEVEL_DBG=y
# Stacks and heaps
CONFIG_MAIN_STACK_SIZE=4096
CONFIG_HEAP_MEM_POOL_SIZE=1024

This is prj.conf file

RE: nRF9160 PWM

Håkon Alseth — Wed, 24 Apr 2019 14:41:40 GMT

Hi,

How does your prj.conf look like?

Do you have "CONFIG_PWM=y" set?

Kind regards,

Håkon