Re: [PATCH 2/2] iio: light: Add support for Capella cm36686 and cm36672p sensors

[David Lechner <dlechner@baylibre.com>]

On 2/2/26 6:04 AM, Erikas Bitovtas wrote:
> 
> 
> On 2/1/26 8:45 PM, David Lechner wrote:
>> On 2/1/26 11:03 AM, Erikas Bitovtas wrote:

Pro tip: when you reply, trim out the not relevant parts like this.
It was hard to find your actual reply.

...

>>> +static const struct iio_chan_spec cm36686_channels[] = {
>>> +	{
>>> +		.type = IIO_LIGHT,
>>> +		.info_mask_separate = BIT(IIO_CHAN_INFO_RAW) |
>>
>> IIO_LIGHT should have IIO_CHAN_INFO_SCALE to convert raw to convert the
>> raw value to lux. If we don't have that info due to lack of documentation,
>> then we should add a comment to explain that. Or maybe we can make an
>> educated guess?
> No documentation on this is available, but we know that ASUS in their driver for
> cm36686 use 160ms integration time for light and, after applying their custom
> calibration functions, divide received raw value by 25 if the device is cm36686,
> or by 20 if it's cm36283. Maybe from there we can extrapolate the lux value like
> this:
> 80ms -> raw / 12.5
> 160ms -> raw / 25
> 320ms -> raw / 50
> 640ms -> raw / 100
> Not sure how accurate that would be, however.
> Xiaomi gets IIO_CHAN_INFO_SCALE by multiplying a value retrieved from a device
> tree property called "als_trans_ratio", which is not documented anywhere.
The ASUS driver is an input driver, not IIO, so not sure that helps.

We could start with hard-coding the als_trans_ratio in this driver for the
scale (and write comments explaining where it came from and what we know
and don't know). Then at least we would have a scale. And if anyone needs
something more accurate and reverse-engineers it we could fix it up later
with a DT property or something like that.