Problem with timer interrupt

[RShre.2]

I am using HAL_TIM_PeriodElapsedCallback(TIM_HandleTypeDef *htim) to generate a sine wave based on DDS and I have also implemented ADC to get the value and perform some calculation. My sampling rate is 100KHz so my timer interrupt should trigger every 10us.

void HAL_TIM_PeriodElapsedCallback(TIM_HandleTypeDef *htim)

{

HAL_GPIO_WritePin(GPIOA, GPIO_PIN_6, GPIO_PIN_SET);

.................

few codes calling dac and adc

HAL_GPIO_WritePin(GPIOA, GPIO_PIN_6, GPIO_PIN_RESET);

}

so the execution time is around 4.8us which means the callback function should wait till it reaches 10us and new trigger start if I understand it correctly.

But is it possible that the cycle time for one session exceeds 10us?

The high pulse width is fine, increases with increase in lines of code, but the low pulse width seems to not decrease beyond. This delays my timer interrupt trigger. What is actually happening behind this callback function?

[Tesla DeLorean]

What's the TIM setting for PSC / ARR ? Remember these are N-1 values, ie last value before wrapping to zero.

Using ST's HAL with it's circuitous routing and call backs can add overhead of it's own.

Consider using DMA to feed pattern buffer values into ARR or CCRx registers at UPDATE

[waclawek.jan]

> What is actually happening behind this callback function?

Cube is open source, you can/should literally read it to find out what is happening, or also single-step if you want.

JW

[Tinnagit]

the timer period is not update, Is this your problem?

I think that you can use timer in PWM mode and update a Pulse value by your ADC value.

it should work.

[S.Ma]

While we don't know which STM32 it is, and how the sinewave is generated, I would use Timer to generate the sample frequency for the DAC output generation (timer output pwm trigger signal to wire to the DAC trigger input pin), the DAC is DMA sweeping a waveform cyclically (the sine wave)

Then, if you want to make an ADC measurement, to make it simple, use the same timer PWM and feed it to the ADC trigger pin, which will feed a DMA cyclic buffer for you to sweep.

[Tinnagit]

I think that the frequency synchronization between DAC generation and ADC Sampling is not necessary.

You have to sampling ADC at lease double frequency of media signal frequency and you 'll got the necessary data from signal already.

so you can make PWM from it at any frequency you want after it.

[RShre.2]

it's STM32-G071RB and I am using DDS technique to generate a sine wave. It's not the synchronization having the problem I assume. Just that the time taken for the code to be executed inside the timer interrupt is acting a bit weird. Even though the execution time is less than 10us which is the time period of my interrupt, it is delaying the interrupt. the question is why is my cycle time not fixed. This happened even when I was not using ADC.

[RShre.2]

i didn't understanding regarding using DMA to feed pattern but I just read a bit about overhead. Maybe I can try accessing the timer directly through registers first and see if that makes any difference. If not, i will try your method. I will get back with an update.