stm32L4 : How to generate triangular wave using DAC DMA, uneven slopes

Question

I am trying to generate a triangular wave of frequency 8kHz using DAC and DMA. DAC is triggered using a timer so that the DAC speed will be 1 MSPS. I am working on stm32L476 discovery board. I am using stm32CUBEMX for code initialization. My configurations are as follows:

In stm32CUBEMX :

 Clock configuration : system clock = 80MHz 
                                     APB1 and APB2 peripheral clocks = 20 MHz
                                     APB1 and APB2 timer clocks = 40 MHz
                                     PLL source mux : HSI
                                     PLLM = /1
                                     *N = x10
                                      /R = /8
                                     PLLCLK selected
    Timer : Prescalar = 39
                UP counter
                Period = 1
                So that output frequency is 1 MHz to trigger the DAC.
    DAC :
              Output buffer : Enable
              Trigger : TIM7 Event out
              DMA(option inside DAC configuration in cubeMX): DAC channel 2(PA5), Half word, circular mode, priority = very             high, memory(ticked) in cubeMX

I am using ac6SW4stm32(system workbench) for writing code. I created an array like this for triangular wave.

const uint16_t val[]={130,260,390,520,650,780,910,1040,
1170,1300,1430,1560,1690,1820,1950,2080,
2210,2340,2470,2600,2730,2860,2990,3120,
3250,3380,3510,3640,3770,3900,4030,4095,
4030,3900,3770,3640,3510,3380,3250,3120,
2990,2860,2730,2600,2470,2340,2210,2080,
1950,1820,1690,1560,1430,1300,1170,1040,
910,780,650,520,390,260,130,0};

In main, I added the following three statements. These statements start timer, DAC and DMA respectively.

HAL_TIM_Base_Start(&htim7);
HAL_DAC_Start(&hdac1,DAC_CHANNEL_2);
HAL_DAC_Start_DMA(&hdac1, DAC_CHANNEL_2, (uint32_t*)val, 64, DAC_ALIGN_12B_R);

At the output, I am not getting proper triangular wave. Frequency of the wave is correct but the wave is not perfectly triangular. It has different rising and falling slopes(unequal edges).

Answer 1

You have not enough samples to have the triangular wave. 64 samples is far not enough. (single step is >100mV). If you want to keep 1MSPS the maximum is 125 samples which is not good enough in my opinion.

Your DAC can be driven with much more samples per second than 1MSPS. My record (using external fast opamp) is 5.5MSPS. With the internal buffer on you can try up to 2-3MSPS.

Waveform was generated using STM32F446 and the oscillogram was taken by my prototype STM32F303 10MSPS oscilloscope.

This is the oscillogram of the saw 8kHz with 32 samples:

This is the oscillogram of the saw 8kHz with 256 samples:

TIM6 -> DIER |= TIM_DIER_UDE;
TIM6 -> PSC = PSC_Value;
TIM6 -> ARR = ARR_Value;
TIM6 -> CR2 |= TIM_CR2_MMS_1;

DAC -> CR = DAC_CR_DMAEN1 | DAC_SR_DMAUDR1 | DAC_CR_TEN1 | DAC_CR_BOFF1;
DAC -> CR |= DAC_CR_EN1;

DMA1_Stream5 -> NDTR = Nsamples;
DMA1_Stream5 -> PAR = (uint32_t)&(DAC -> DHR12R1);
DMA1_Stream5 -> M0AR = (uint32_t)Buff;
DMA1_Stream5 -> CR = (DMA_SxCR_TEIE | DMA_SxCR_CHSEL | DMA_SxCR_CIRC | DMA_SxCR_DIR_0 | DMA_SxCR_EN | DMA_SxCR_PSIZE_0 | DMA_SxCR_MSIZE_0 | DMA_SxCR_MINC | DMA_SxCR_PL_0);

TIM6 -> CR1 |= TIM_CR1_CEN;

PS the code is for 446RE