Re: [RFC 0/4] iio: light: add driver for Broadcom APDS9999

[Jonathan Cameron <jic23@kernel.org>]

On Mon, 11 May 2026 17:19:54 +0200
"Jose A. Perez de Azpillaga" <azpijr@gmail.com> wrote:

> On Mon, May 11, 2026 at 01:21:12PM +0100, Jonathan Cameron wrote:
> > On Mon, 11 May 2026 12:14:42 +0200
> > "Jose A. Perez de Azpillaga" <azpijr@gmail.com> wrote:
> >  
> > > This series adds support for the Broadcom APDS9999 ambient light
> > > sensor. The APDS9999 is a digital proximity and RGB sensor with
> > > ALS capability, featuring I2C interface, programmable gain and
> > > integration time.
> > >
> > > This chip is considered by Broadcom to be the 'successor' to the
> > > APDS-9960 which has technically been discontinued.
> > >
> > > The initial driver implements the ALS/Lux functionality using the
> > > green channel, which approximates the human eye spectral response.  
> >
> > That's not normally a good approximation unless it is also has
> > very bad colour curves for green!  Better than any other channel
> > on it's own but still not good. If we had calibrated sRGB it would be:
> >
> >  Something like: 0.2126R+0.7152G+0.0722B
> > for example.
> >
> > The lux vs white light and lux vs incandecent light are fairly linear
> > so 'maybe' it's good enough but it's a bit of a stretch to describe it
> > as an illuminance measure which is what we'd need for it to be measured
> > in lux.  
> 
> the datasheet states: "Approximates human eye response with green
> channel" and "Uses optical coating technology to emulate human eye
> spectral response."
Hehe. Marketing. They provide the light curves that show it approximates
very well the green part of human eye spectral response and very badly the
rest of it! :)

> 
> that's the reason why I chose to use green channel.
> 
> > So I'm on the fence at the moment.  We might be better off teaching userspace
> > code to do it's own approximation if it has an RGB sensor.  
> 
> I'll also add IIO_INTENSITY channels for R, G, B in v2 so userspace can
> comapute its own weighted lux if it prefers.
> 
> --
> jose a. p-a
>