Re: [PATCH 2/2] iio: temperature: ltc2983: Add support for ADT7604

[Jonathan Cameron <jic23@kernel.org>]

On Mon, 11 May 2026 07:52:17 +0000
"Stan, Liviu" <Liviu.Stan@analog.com> wrote:

> On Fri, 8 May 2026, Nuno Sá wrote:
> > > > > A leak detector is a sensor like the raindrop detection modules, whose
> > > > > resistance decreases as more of its surface is contacted by liquid. In this
> > > > > case, for example, if half of the sensor's surface is in contact 
> > > > > with a liquid, the ADT7604 should, with an appropiate user-provided 
> > > > > resistance-vs-coverage table, report 50% coverage in the IIO channel.  
> > > >
> > > > And, AFAIU, depending on the coverage the resistance changes and that's
> > > > how we get into the indirect temperature variations?  
> > >
> > > If I understand your comment correctly, you're implying that there are
> > > indirect temperature variations caused by resistance changes, and 
> > > that's what the chip is measuring? The chip does not measure any actual
> > > temperature for this sensor type, the value in the temperature result 
> > > register bank is solely a result of the chip's interpolation of the 
> > > user-defined custom table.  
> >
> > I meant that's what the table is trying to accomplish? That's why I kind
> > of put it as "indirect". My understanding was the chip does not measure
> > any temperature as well.  
> 
> Oh, okay, I understand now. Yes, this is what the table is trying to accomplish.
> 
> On Sat, 9 May 2026, Jonathan Cameron wrote:
> > > I meant that's what the table is trying to accomplish? That's why I kind
> > > of put it as "indirect". My understanding was the chip does not measure
> > > any temperature as well.
> > >  
> > Ok. So what are our options here?  Present it as simple resistance and leave
> > userspace to figure it out or add a new channel type? To me feels like new
> > channel type makes sense.  
> 
> The current approach presents it as IIO_TEMP since the chip outputs coverage
> (using the custom table interpolation) via the temperature result bank, not the
> resistance bank, but I agree a new channel type makes sense. Should I
> create a specific type like IIO_COVERAGE_PERCENT or would a general
> IIO_PERCENTAGE be better?

For ABI purposes we don't care where it comes from.

We already have some 'ratio' type measurements like concentration which are
percentages and similar to those I think we need some indication of 'what'
is being measured given it's unit free.  Hence IIO_COVERAGE_PERCENT seems
the better choice to me.

Jonathan


> 
> Thanks,
> Liviu