tests: drivers: dac channel structure has a buffered parameter boolean

[FRASTM]

Add the '.buffered' field to the dac_channel_cfg structure when testing the DAC.
This boolean parameter is initialised to 'true' to PASS the test.
It follows the #57730

Fixes #58931

Signed-off-by: Francois Ramu [email protected]

[FRASTM]

$ west build -p auto -b nucleo_wl55jc tests/drivers/dac/dac_loopback/
when pin A2 is connected to A4 on the arduino connector CN8 of the nucleo_wl55jc board

*** Booting Zephyr OS build v3.4.0-rc2-101-g7b2034aaecc4 ***                    
Running TESTSUITE dac_loopback                                                  
===================================================================             
START - test_dac_loopback                                                       
 PASS - test_dac_loopback in 0.011 seconds                                      
===================================================================             
TESTSUITE dac_loopback succeeded                                                
                                                                                
------ TESTSUITE SUMMARY START ------                                           
                                                                                
SUITE PASS - 100.00% [dac_loopback]: pass = 1, fail = 0, skip = 0, total = 1 dus
 - PASS - [dac_loopback.test_dac_loopback] duration = 0.011 seconds             
                                                                                
------ TESTSUITE SUMMARY END ------                                             
                                                                                
================================
```===================================             
PROJECT EXECUTION SUCCESSFUL                                                    

[martinjaeger]

The issue does not seem to be the not enabled buffer, but incorrect output configuration. We just didn't realize this before, as an enabled buffer automatically connects the DAC to the external pin.

I think we should use LL_DAC_SetOutputConnection to fix the driver instead.

[FRASTM]

The issue does not seem to be the not enabled buffer, but incorrect output configuration. We just didn't realize this before, as an enabled buffer automatically connects the DAC to the external pin.

I think we should use LL_DAC_SetOutputConnection to fix the driver instead.

Do you mean LL_DAC_SetOutputBuffer( ... OutputBuffer ... )
plus LL_DAC_SetOutputConnection( LL_DAC_OUTPUT_CONNECT_GPIO)

[martinjaeger]

Do you mean LL_DAC_SetOutputBuffer( ... OutputBuffer ... ) plus LL_DAC_SetOutputConnection( LL_DAC_OUTPUT_CONNECT_GPIO)

Yes, I think that would be the correct way. What do you think?

[FRASTM]

Do you mean LL_DAC_SetOutputBuffer( ... OutputBuffer ... ) plus LL_DAC_SetOutputConnection( LL_DAC_OUTPUT_CONNECT_GPIO)

Yes, I think that would be the correct way. What do you think?

That's better, but initializing the .buffered = true in the tests/drivers/dac/dac_loopback/src/test_dac.c is mandatory.

diff --git a/drivers/dac/dac_stm32.c b/drivers/dac/dac_stm32.c
index bfbdc5003f..c49b6ef20e 100644
--- a/drivers/dac/dac_stm32.c
+++ b/drivers/dac/dac_stm32.c
@@ -111,6 +111,11 @@ static int dac_stm32_channel_setup(const struct device *dev,
 		table_channels[channel_cfg->channel_id - STM32_FIRST_CHANNEL],
 		output_buffer);
 
+	/* By default, the output pin is considered as GPIO output pin */
+	LL_DAC_SetOutputConnection(cfg->base,
+		table_channels[channel_cfg->channel_id - STM32_FIRST_CHANNEL],
+		LL_DAC_OUTPUT_CONNECT_GPIO);
+
 	LL_DAC_Enable(cfg->base,
 		table_channels[channel_cfg->channel_id - STM32_FIRST_CHANNEL]);
 

[martinjaeger]

I have tested with nucleo_l073rz and unfortunately, it does not have the LL_DAC_SetOutputConnection and compilation fails. Looks like there are some differences between the MCU series.

Regarding the original discussion whether the output buffer is needed or not: I have checked the impedance of the DAC and the ADC in the STM32L072 datasheet.

DAC

ADC

They are quite close and without a capacitor the DAC impedance does indeed seem to be too high to drive the ADC. I have added a 100 nF capacitor and it works even without the buffer.

As we don't have the cap in standard Nucleo boards I agree to enable the buffer, now that I understand why it doesn't work without the buffer.