Using PDM for 2 Microphones in Stereo Mode

[bardia3000]

Hello, I'm trying to use two microphones in stereo mode of PDM (nrf52832, SDK17). Unfortunately, I have a major problem with my Right microphone. It seems that "nrfx_pdm_buffer_set" just put the Left one in the buffer.

This is my initialization for PDM in main function:

int main(void) {
    .
    .
    ...
    nrfx_pdm_config_t pdm_config = NRFX_PDM_DEFAULT_CONFIG(PDM_CLK_PIN, PDM_DATA_PIN);
    pdm_config.mode = PDM_MODE_OPERATION_Stereo;
    pdm_config.gain_l = PDM_GAIN;
    pdm_config.gain_r = PDM_GAIN;
    pdm_config.clock_freq = PDM_PDMCLKCTRL_FREQ_1000K; // 15625 Samples per Second
    pdm_config.edge = PDM_MODE_EDGE_LeftFalling;
    pdm_config.interrupt_priority = APP_IRQ_PRIORITY_HIGH;
    err_code = nrfx_pdm_init(&pdm_config, pdm_event_handler);
    APP_ERROR_CHECK(err_code);
    .
    .
    .
}

This is "pdm_event_handler" function and "pdm_buffer" array:

int16_t pdm_buffer[PDM_BUFFER_SIZE_SAMPLES];

static void pdm_event_handler(nrfx_pdm_evt_t const *const p_evt) {
    if (p_evt->buffer_requested) {
	    nrfx_pdm_buffer_set(pdm_buffer, PDM_BUFFER_SIZE_SAMPLES);
    } 
    if (p_evt->buffer_released) {
    	do_something(pdm_buffer);
    }
}

I tried many methods, including:

1. Make the type of "pdm_buffer" array uint32_t instead of int16_t.

2. Make "pdm_buffer" array 2 dimensional (int16_t pdm_buffer[2][PDM_BUFFER_SIZE_SAMPLES]).

3. Double the "PDM_BUFFER_SIZE_SAMPLES" constant in the buffer set ("nrfx_pdm_buffer_set(pdm_buffer, 2 * PDM_BUFFER_SIZE_SAMPLES)").

But non of them work.

[Edvin]

Hello,

Sorry for the late reply. 

What kind of microphones are you using? And I suspect that there is some sort of way you need to tell the microphone whether it is the left or the right device. Is that correct? Does both of the microphones work if you tell them that they are the left device (one at the time, of course). Does either of them work if you tell them that both are the right (not left) microphone?

Best regards,

Edvin

[bardia3000]

Both microphones are IM69D130 (digital MEMS microphone). Both of them are working properly and the connection between LR pin and GND or VCC defines which one is the right one and which one is the left one.

[bardia3000]

Yes, I set one to GND and one to VCC, Also I've checked it via multimeter. This must be something about my codes.

[Edvin]

So my impression is that you have wired them like this:

https://infocenter.nordicsemi.com/topic/com.nordic.infocenter.nrf52832.ps.v1.1/pdm.html?cp=5_2_0_42_4#hardware_example

Have you tried to use a logic analyser to analyze the signal pins?

On the microphone that is configured to have a Right channel output, does it output no data, or does it output left data even though it is configured to output right channel data?

BR,
Edvin

[bardia3000]

Yes I have wired them correctly, and here is my PDM Clock and Data captured from a logic analyzer. At both falling and rising edges I captured different values.

[Edvin]

I am not sure what I am looking at, or what it is supposed to look like.

You have set "pdm_config.edge = PDM_MODE_EDGE_LeftFalling;" So that means that the data we see here (all the changes) are from the left channel, since they are directly after the clock is falling from high to low. No change on low to high, do you agree?

And the microphone behaves like this both if you connect LR to GND or VDD? It always changes on HiToLo?

BR,

Edvin

[bardia3000]

I do not agree. As I can see PDM_DATA has both 0 and 1 in falling and rising of PDM_CLOCK. There must be a problem with my code or SDK as I mentioned before.

I don't understand what you are seeking for when you say: "And the microphone behaves like this both if you connect LR to GND or VDD? It always changes on HiToLo?"

I used 2 microphones on my own-designed PCB and I want to use them in stereo mode. My code doesn't detect anything from the right microphone (all data is captured from left one). Please take a look at my code or the problem may happened due to a problem in nrfx driver.

[Edvin]

Perhaps I misunderstand something, or don't know how PDM is supposed to work, but all of the changes on the data line are changing directly after the clock goes high to low:

Doesn't that suggest that it is only sending data on the falling clock?

BR,
Edvin