Re: [PATCH v7 5/6] dt-bindings: iio: adc: ad7192: Add AD7194 support

[Conor Dooley]

On Fri, May 10, 2024 at 09:21:37AM -0500, David Lechner wrote:
> On 5/10/24 5:05 AM, Alisa-Dariana Roman wrote:
> > On 30.04.2024 20:21, Conor Dooley wrote:
> >> On Tue, Apr 30, 2024 at 07:29:45PM +0300, Alisa-Dariana Roman wrote:
> >>> +      diff-channels:
> >>> +        description:
> >>> +          Both inputs can be connected to pins AIN1 to AIN16 by choosing the
> >>> +          appropriate value from 1 to 16.
> >>> +        items:
> >>> +          minimum: 1
> >>> +          maximum: 16
> >>> +
> >>> +      single-channel:
> >>> +        description:
> >>> +          Positive input can be connected to pins AIN1 to AIN16 by choosing the
> >>> +          appropriate value from 1 to 16. Negative input is connected to AINCOM.
> >>> +        items:
> >>> +          minimum: 1
> >>> +          maximum: 16
> >>
> >> Up to 16 differential channels and 16 single-ended channels, but only 16
> >> pins? Would the number of differential channels not max out at 8?
> > 
> > Hello, Conor! I really appreciate the feedback!
> > 
> > The way I thought about it, the only thing constraining the number of channels is the reg number (minimum: 0, maximum: 271). 272 channels cover all possible combinations (16*16 differential and 16 single ended) and I thought there is no need for anything stricter. I added items: minimum:1 maximum:16 to make sure the numbers are from 1 to 16, corresponding to AIN1-AIN16.
> > 
> > Please let me know what should be improved!
> > 
> > Kind regards,
> > Alisa-Dariana Roman.
> > 
> 
> Having looked at the datasheet for this and other similar chips, I agree
> that this reasoning makes sense. Some of the similar chips that have fixed
> channel assignments still have, e.g. a channel where + and - are both
> AIN2 (I assume for diagnostics). So I think it makes sense to allow for
> doing something similar here even if the most common use cases will
> probably have at most 16 channels defined in the .dts.

Actually, I think there were a bunch of whiffs on this one by either
misreading the property in question (me) or not realising that I had done
that and trying to explain what the possible combinations are.
Looking at it now, I dunno wtf I was smoking because there's no way that
this would be a functional binding if the min/max in the quote above
constraining the number of channels. I can hardly blame y'all for that
though, I am supposed to know how bindings work after all...

Attachment: signature.asc
Description: PGP signature