Re: [PATCH v1 1/2] dt-bindings: iio: adc: avia-hx711: add avia,hx710b compatible

[Jonathan Cameron <jic23@kernel.org>]

On Sun, 19 Apr 2026 07:17:50 +0200
Andreas Klinger <ak@it-klinger.de> wrote:

> Hi,
> 
> David Lechner <dlechner@baylibre.com> schrieb am Sa, 18. Apr 16:46:
> > On 4/18/26 12:05 PM, Piyush Patle wrote:  
> 
> [...]
> 
> > >  
> > >    Specifications about the driver can be found at:
> > >    http://www.aviaic.com/ENProducts.aspx
> > > @@ -23,11 +33,12 @@ properties:
> > >    compatible:
> > >      enum:
> > >        - avia,hx711
> > > +      - avia,hx710b
> > >  
> > >    sck-gpios:
> > >      description:
> > >        Definition of the GPIO for the clock (output). In the datasheet it is
> > > -      named PD_SCK
> > > +      named PD_SCK.  
> > 
> > Save the cleanups for a separate patch to keep the adding HX710B changes clear.
> > 
> > I'm guessing the existing binding for HX711 is quite old because it is quite
> > incomplete.
> > 
> > It has avdd-supply, but is missing vsup-supply and dvdd-supply.
> > 
> > It should probably also have a way to describe how the rate pin is wired.
> > 
> > And it should have a clocks property instead of clock-frequency.  
> 
> The real meaning is a wait time until the DOUT is stable. As a submitted the
> driver many years ago my suggestion of a wait time property was not accepted
> because it would have introduced a new property which didn't exit in those days.
> The suggestion was to name it clock-frequency because it already existed. This
> clock-frequency made the driver also a little bit more complicated because at
> the end we needed a waiting time and not a frequency.
> 
> Today i see there is as "wait-delay" property already introduced at other
> bindings. This would also simplify the driver a bit.
> 
> @robh@kernel.org, @krzk+dt@kernel.org, @conor+dt@kernel.org:
> Would this change in the binding be acceptable?
> 
> If yes, i could prepare a driver and binding patch separate of this patchset to
> clean it up.

You would need to maintain backwards compatiblity with the old binding.
Might be fine to deprecate it.

Jonathan

> 
> > It would make sense to have two clocks, on for XI/XO and one for PD_SCK.
> > The second one being optional because of sck-gpios.
> > 
> > HX710B has many fewer pins, so we will need an:
> > 
> > allOf:
> >   - if:
> >       properties:
> > 	compatible:
> > 	  const: avia,hx710b
> > 
> > section that sets anything for pins that chip doesn't have to false, like
> > vsup-supply.
> > 
> > HX710B also has a vref-supply that HX711 doesn't have. (Unless these are the
> > same thing by a different name?)
> > 
> >   
> > >      maxItems: 1
> > >  
> > >    dout-gpios:
> > > @@ -43,6 +54,9 @@ properties:
> > >        Definition of the regulator used as analog supply
> > >  
> > >    clock-frequency:
> > > +    description:
> > > +      Bit-bang clock frequency on PD_SCK. Keep the PD_SCK high time below
> > > +      the chip power-down threshold.  
> > 
> > I suspect that this was meant to be the crystal frequency (XI/XO), not PD_SCK
> > since sck-gpios already exists for PD_SCK  
> 
> see above
> 
> [...]
> 
> Best regards,
> 
> Andreas
>