1 ACCEPTED SOLUTION
Accepted Solutions
Ok, so it sounds like you want multiple samples of each channel?
You'll need a queue to have the DMA copy to each queue. On each callback, you increment the queue pointer and call HAL_ADC_Start_DMA which will now point to the next queue index. Below is an example code that i just tested.
// main.h
typedef struct
{
uint16_t data[2]; // the 2 channels
}ADC_Data_t;
typedef struct
{
ADC_HandleTypeDef *hadc;
ADC_Data_t queue[3];
RING_BUFF_STRUCT ptr;
bool rdy;
}ADC_t;
// main.c
ADC_t adc_buf =
{
.hadc = &hadc1,
.ptr.SkipOverFlow = true
};
int main(void)
{
HAL_ADCEx_Calibration_Start(&hadc1);
HAL_ADC_Start_DMA(&hadc1, (uint32_t*)adc_buf.queue[adc_buf.ptr.index_IN].data, 2);
while(1)
{
if(adc_buf.rdy)
{
adc_buf.rdy = false;
// do something with data.
}
}
}
void HAL_ADC_ConvCpltCallback(ADC_HandleTypeDef *hadc)
{
int status = 0;
if(hadc == adc_buf.hadc)
{
status = RingBuff_Ptr_Input(&adc_buf.ptr, 3); // sample 3 times
if(status)
{
adc_buf.rdy = true;
RingBuff_Ptr_Reset(&adc_buf.ptr);
}
else
{
HAL_ADC_Start_DMA(&hadc1, (uint32_t*)adc_buf.queue[adc_buf.ptr.index_IN].data, 2);
}
}
}
this is what the debug window will look like. The HAL_ADC_ConvCpltCallback will increment a pointer. If it doesn't rollover (status =0) then i call HAL_ADC_Start_DMA again which is pointing to the next queue index. If it does roll over, it finished it's 3 samples and so I set a flag.
DMA is in Normal mode. Continuous Conversion mode is disabled.
I attached the RingBuffer driver.
@Karl Yamashita wrote:
@ricky-fleet wrote:Hi Karl, thanks for the reply! Perhaps something I did not clarify well was that I'm trying to do more than one ADC conversion at a time using DMA peripheral to memory. When I set it up this way, it appears to only perform one conversion then stop. If I disable the linked list configuration but keep continuous conversion enabled, I do have the option to use HAL_ADC_ErrorCallback() to halt the ADC, but that doesn't seem like an ideal solution.
So this reply seems opposite of your original post.
Maybe a flowchart would clear up what you really want to happen?
Hi @Karl Yamashita ! I think I have a solution but to make sure I'm understanding it correctly and not creating issues later, here is a quick flow chart to describe what I was trying to do (I know it is a little simplified with the actual ADC and DMA interactions). In my case, the DMA buffer can handle many conversions and I only want it to stop once the buffer is full (as described by the size in the HAL_ADC_Start_DMA() call).
Ok, so it sounds like you want multiple samples of each channel?
You'll need a queue to have the DMA copy to each queue. On each callback, you increment the queue pointer and call HAL_ADC_Start_DMA which will now point to the next queue index. Below is an example code that i just tested.
// main.h
typedef struct
{
uint16_t data[2]; // the 2 channels
}ADC_Data_t;
typedef struct
{
ADC_HandleTypeDef *hadc;
ADC_Data_t queue[3];
RING_BUFF_STRUCT ptr;
bool rdy;
}ADC_t;
// main.c
ADC_t adc_buf =
{
.hadc = &hadc1,
.ptr.SkipOverFlow = true
};
int main(void)
{
HAL_ADCEx_Calibration_Start(&hadc1);
HAL_ADC_Start_DMA(&hadc1, (uint32_t*)adc_buf.queue[adc_buf.ptr.index_IN].data, 2);
while(1)
{
if(adc_buf.rdy)
{
adc_buf.rdy = false;
// do something with data.
}
}
}
void HAL_ADC_ConvCpltCallback(ADC_HandleTypeDef *hadc)
{
int status = 0;
if(hadc == adc_buf.hadc)
{
status = RingBuff_Ptr_Input(&adc_buf.ptr, 3); // sample 3 times
if(status)
{
adc_buf.rdy = true;
RingBuff_Ptr_Reset(&adc_buf.ptr);
}
else
{
HAL_ADC_Start_DMA(&hadc1, (uint32_t*)adc_buf.queue[adc_buf.ptr.index_IN].data, 2);
}
}
}
this is what the debug window will look like. The HAL_ADC_ConvCpltCallback will increment a pointer. If it doesn't rollover (status =0) then i call HAL_ADC_Start_DMA again which is pointing to the next queue index. If it does roll over, it finished it's 3 samples and so I set a flag.
DMA is in Normal mode. Continuous Conversion mode is disabled.
I attached the RingBuffer driver.
