PWM Tolerance

[WesC]

nRF52840

NCS V 2.2.0

Using the NRFX drivers natively trying to set the PWM to 333uS.

over 20 different units i am seeing this vary form 326 to 335

what is the expected accuracy (tolerance) of the PWM output.

Is there an example of the PWM in NCS that uses the NRFX drivers and not the Zephyr Drivers

[WesC]

I found this example, but haven't had a chance to test it.  This is very similar to my code.  I will run tests, but can you do similar to see how well it holds tolerance to 333us.ncs-nrfx-pwm-simple-example-master.zip

would this be the suggested way to use the NRFX drivers for the PWM.

If i use Zephyr and use something like PWM-LEDs for some other reason work better?

finally how do i determine if i am running from the HFCLK?

[Jared]

Hi Wes,

Can you check what HFCLK source the PWM is using? Is it HFINT or HFXO? You can check the HFSTAT register to what hf clock source that is used and if the HFCLK is running or not. At least in nRF5SDK, the PWM module would use the less accurate HFINT if the HFXO wasn't started explicitly beforehand. The tolerance of the PWM depends on the clock source that is used. 

Examples using the nrfx drivers directly can be found here.

regards

Jared 

[WesC]

Thanks.  If not set and using the HFINT.  How do i set this to use the HFXO in Zephyr?

[Jared]

You start the HFXO manually either by writing directly to the register:

// Start 64 MHz crystal oscillator.
NRF_CLOCK->EVENTS_HFCLKSTARTED = 0;
NRF_CLOCK->TASKS_HFCLKSTART = 1;

// Wait for the external oscillator to start up.
while (NRF_CLOCK->EVENTS_HFCLKSTARTED == 0)
{
// Do nothing.
}

Or by doing as clock_init() does here.

regards

Jared