Different capture time using PING-PONG mode with ADC+DMA

Actually the data I am getting from the scope and processed in MATLAB:

Time_between_blocks (toggle) =

               0.000511515
                0.00051236
               0.000511515
               0.000512365
                0.00051152
                0.00051236
                0.00051152
                0.00051236

diff(Time_between_blocks)

ans =

     8.44999999999861e-007
    -8.44999999999752e-007
     8.49999999999723e-007
    -8.44999999999752e-007
     8.39999999999782e-007
    -8.39999999999782e-007
     8.39999999999782e-007

The grpah was the difference with the central (expected signal), so there is a PHASE adding in it.

Thank you Amit.

PAk said:

O Red: Time to capture the block (measured)

i.e. is when the ADC sample is taken, or at the end of a DMA block?

Chester Gillon said:

How is the measured "capture" time taken?

i.e. is when the ADC sample is taken, or at the end of a DMA block?

End of DMA block at the beggining of the interrupt.
I am attacking an external DDS which provides an EoC signal, which is pretty stable, that is one reference I use (blue x on the plot).

In case you owuld like to test it, every block is 128 samples at 250KHz.....using:

ulPeriod_timer = 480;
dutyCycle_timer = 240;

	TimerLoadSet(TIMER1_BASE, TIMER_A, ulPeriod_timer - 1);
	TimerMatchSet(TIMER1_BASE, TIMER_A, dutyCycle_timer); 
	
	

Amit Ashara said:

Hello PAk

Note that when processing a Ping Pong buffer, the CPU is processing an if-if condition in the interrupt handler. Is the GPIO toggle always in the unconditional part of the interrupt handler or in the conditional part of the interrupt handler?

Regards
Amit

First thing after the block check, otherwise I would get a toggle every ADC conversion:

ulMode = uDMAChannelModeGet(UDMA_CHANNEL_ADC2 | UDMA_ALT_SELECT);

		if (ulMode == UDMA_MODE_STOP)
		{

			toggle();


...
}

Amit Ashara said:

Hello PAk

Toggle the GPIO as the first thing to be done in the interrupt handler, outside of the if statement(s)

Regards
Amit

If I do that I get a toggle for every ADC sample!!

Amit Ashara said:

Hello PAk,

That is not correct unless the transfer size is 1. If the transfer size is N (where N is more than 1), then the interrupt must be fired at N samples.

Regards
Amit

Transfer Size is (2 * NUM_SAMPLES) since I am reading two channels. Then, why am I getting an interrupt every pair of ADC sample?

Amit Ashara said:

Hello PAk

And what is the NUM_SAMPLES you are reading from the ADC controller and at what interval is ADC is sampling?

Regards
Amit

Hello Amit, I corrected the issue with the interrupt. I was using ADCIntClear and ADCIntEnable, instead of ADCIntClearEx and ADCIntEnableEx.

Now the time jitter has improved decreased, however, the time is not constant!!!(The toggle is at the beggining of the interrupt - reed mark below):