strange signal / noise on PWM. Noise diaspears when connecting to JLINK! WHY?

[MaxRPEng]

Hi,

i am using a PWM signal to control the brightness of a LED. When the brightness is at zero, i have a random flickering on the LED.

This signal is also present at the pin itself. The strange thing, as soon i connect a JLINK probe the signal is perfect zero. So no noise anymore.

As soon i remove the JLINK the flickering is back.

Annoing magic.

Any idea why the JLINK influnces the behaviour of the nRF52?

thank you

Max

[haakonsh]

One source could be the DCDC regulator or the clocks. The debugger will force the nRF52's regulator into LDO mode and keep the HFCLK running. 

Try disabling the DCDC regulator. If that does not work try to enable Constant Latency (force HFCLK on, see Sub power modes). 

See in the POWER Registers, see DCDCEN register, and the TASKS_CONSTLAT. 

[MaxRPEng]

Hi, thank you for your fast answer: DCDC is disabled.

nrfjprog  --memrd 40000578
0x02625C42: 0000                                      |..|

I turned the CONSTLAT on: NRF_POWER->TASKS_CONSTLAT = 1;

But i still have that strange random signal on my PWM output.

[haakonsh]

Hmm, that's strange. 

Can you share the PWM code? 

-edit:
It would also be useful to scope the PWM with the debugger enabled and not.

[MaxRPEng]

PWM init:

void pwm_init(void)
{
    /* Config PWM 0*/
    nrf_drv_pwm_config_t const config0 =
    {
        .output_pins =
        {
            LED_SPOT , // channel 0
            LED_WIDE ,  // channel 1
 
        },
        .irq_priority = APP_IRQ_PRIORITY_LOWEST,
        .base_clock   = NRF_PWM_CLK_8MHz,
        .count_mode   = NRF_PWM_MODE_UP,
        .top_value    = 100,
        .load_mode    = NRF_PWM_LOAD_INDIVIDUAL,
        .step_mode    = NRF_PWM_STEP_AUTO
    };
    APP_ERROR_CHECK(nrf_drv_pwm_init(&m_pwm0, &config0, NULL));

}

Update duty cycle:

// Set duty cycle between 0 and 100%
void pwm_spot_update_duty_cycle(uint8_t duty_cycle)
{
    uint8_t invert = 100 - duty_cycle;
    static uint8_t pwm_test_spot; 
    // Check if value is outside of range. If so, set to 100%
    if(invert >= 100)
    {
        seq_values.channel_0 = 100;
    }
    else
    {
        seq_values.channel_0 = invert;
    }

    pwm_test_spot = 
    
    nrf_drv_pwm_simple_playback(&m_pwm0, &seq, 1, NRF_DRV_PWM_FLAG_LOOP);
}

// Set duty cycle between 0 and 100%
void pwm_wide_update_duty_cycle(uint8_t duty_cycle)
{
    uint8_t invert = 100 - duty_cycle;
  
    // Check if value is outside of range. If so, set to 100%
    if(invert >= 100)
    {
        seq_values.channel_1 = 100;
    }
    else
    {
        seq_values.channel_1 = invert;
    }
    
    nrf_drv_pwm_simple_playback(&m_pwm0, &seq, 1, NRF_DRV_PWM_FLAG_LOOP);
}

[MaxRPEng]

unwanted signal: always the same width, but random frequency

This channel should be always zero!

this is the signal of the second channel. works correct.

[haakonsh]

Try using nrf_drv_pwm_sequence_values_update() instead of nrf_drv_pwm_simple_playback() to change the duty cycle.

[MaxRPEng]

according some examples i implemented the following:

static nrf_pwm_values_individual_t seq_values[];

 seq_values[0].channel_0 = 50;
  seq_values[0].channel_1 = 50;

  nrf_pwm_values_t new_pwm_values;

  new_pwm_values.p_individual = seq_values;

nrf_drv_pwm_sequence_values_update(&m_pwm0, 0, new_pwm_values);

But now i dont have any PWM output !