> >>> + avdd-supply: > >>> + description: > >>> + Analog power supply, voltage between AVDD and AVSS. When providing a > >>> + symmetric +/- 2.5V, the regulator should report 5V. > > > > Any precedence in tree for doing this? I thought we had bindings that required negative > > supplies to be specified separately if present - so this would need to be 2 > > supplies. e.g. > > https://elixir.bootlin.com/linux/v6.7-rc5/source/Documentation/devicetree/bindings/iio/adc/ti,adc12138.yaml#L37 > > > > Given that some serious thought... > > Splitting into positive and negative supplies would make sense if the chip > would have terminals for positive, negative and ground. Which it does not > have, there's only positive and negative (which the datasheet misleadingly > calls "analog ground"). > > The analog voltage supplied to the chip has no effect on its outputs, the > analog supply must be connected between the AVDD and AVSS pins. Its relation > to analog ground is not relevant, so whether the voltages are +5/0 or > +2.5/-2.5 or +4/-1 or whatever does not affect the output of the ADC, which > only reports the difference between its "p" and "n" input signals. It only > affects the range of the inputs, as it cannot measure (p or n) outside the > analog supply. > Ah. Not having seen any isolators, I assumed this device was running against a common ground with the digital side of things. AVSS to DGND is between -3 and 0.2 so I think there is a ground even if it's not directly connected. Given this only operates in differential mode the fact both signals are referenced to +-vref which is from VASS doesn't matter in the end and any offset goes away. So yes I think I agree. This should be a single supply. Jonathan