Re: [PATCH 1/4] dt-bindings: iio: adc: add bindings for AD4691 family

[David Lechner]

On 3/9/26 3:57 AM, Sabau, Radu bogdan wrote:
> 

...

>>>> Regarding 'mode', I agree that it should be something that could be modified
>>>> at run-time, especially since all register modes (CNV_CLOCK, CNV_BURST,
>>>> AUTONOMOUS and SPI_BURST) rely on the same principles of reading the
>>>> ADC result from the registers, the main difference being that PWM on the
>>>> CNV pin is required for CNV_CLOCK and CNV_BURST, but the board design
>>>> stays the same. Perhaps this PWM can be initialized at start-time and only
>>>> be used when CNV modes are being used. This would mean mode can
>>>> become an IIO attribute that could be set by the user at run-time.
>>>
>>> More likely, it would be two different ways of doing a buffered read,
>>> so maybe two different buffers? Or just pick the "best" one and only
>>> implement that mode.
>>
>> I 'think' burst mode is really an oversampling thing as you read back from
>> either the output of an averaging filter or an accumulator.
>> I doubt there is reason to support both the oversampled and raw
>> readings at the same time.
>>
> 
> I agree with that, perhaps burst modes (CNV_BURST and SPI_BURST) could
> be dropped. Also, AUTONOMOUS mode could be used for read_raw since
> its compatible with both CNV_CLOCK and MANUAL hardware setups.
> So, then if the 'pwm' is present in the device tree, CNV_CLOCK
> will be used for buffered readings and if not, MANUAL will.
> This will indeed not only simplify the driver, but also keep two important
> modes up for choice for the user.
> 
> Please let me know your thoughts on this approach.
> 
Sounds sensible to me.