During mono audio capture over PDM, we observed a current draw of about 1.5 mA.

[Adityanath Pandey]

Our setup:

• Board: nRF5340 DK
  
• SDK: NCS v2.6.2
  
• Microphone: MMICT5837-00-012
  
• Interface: PDM
  

During mono audio capture over PDM, we observed a current draw of about 1.5 mA.
My questions are:

• Is there any way to reduce this current consumption further?
  
• How much current does the PDM peripheral itself typically draw in mono mode on the nRF5340 (excluding microphone current)?

[Turbo J]

Expected behaviour.

You need to keep the HF clocks running in order to be able to capture the audio data. That keeps a fairly sizeable portion of the digital chip area powered on and not being able to sleep at all.

[Adityanath Pandey]

We read microphone data by calling dmic_trigger(dmic, DMIC_TRIGGER_START) before starting,

and dmic_trigger(dmic, DMIC_TRIGGER_STOP) afterward. This process is repeated every 100 ms.
However, when we parse the recorded file, we hear a “beep” sound.

My question: How long does dmic_trigger() take to execute, and could this delay be the cause of the beep sound?

[Edvin]

From where do you call dmic_trigger()? In an interrupt? And what sort of buffer are you using? Single? Double?

Best regards,

Edvin

[Adityanath Pandey]

We are not calling dmic_trigger() from an interrupt.

We use it in our application code, before and after reading data. The actual microphone data is obtained through the dmic_read() API.
Our read function looks like this:
int mic::readdata(void *pcm_data, uint16_t *len) {
int ret;
void *buffer;
uint32_t size = 0;

ret = dmic_read(dmic, 0, &buffer, &size, READ_TIMEOUT);
if (ret < 0) {
LOG_ERR("read failed: %d", ret);
return ret;
}

memcpy(pcm_data, buffer, size);
*len = size;

k_mem_slab_free(&mem_slab, buffer);

return ret;
}

[Edvin]

And from where are you calling mic::readdata()?

Best regards,

Edvin