How to choose between AIN0 and in_voltage0_raw when reading ADC?

Is there any reason for using the AINn files in /sys/devices/ocp.3/helper.15
rather than the in_voltageN_raw files in /sys/bus/iio/devices/iio:device0 ?

I.e. is there any (useful) processing done between the raw 12-bit
integer value in the raw file and the integer 'millivolts' value in
the AINn values?

As far as I can see one just loses a bit or two of resolution by using
the AINn values. The fact that it gives the input voltage in mV is
pretty useless as one is never actually measuring the voltage at the
input, it's almost bound to be scaled in some sort of fashion. So
taking the AINn value and mutiplying it by some factor to give the
required result is no different from multiplying the raw value by a
different factor to give the same result (except the loss of some
resolution).

That is, of couse, unless some sort of processing (smoothing, time
weighted average, etc.) is done between the raw value and the AINn
value.

cl@isbd.net wrote:

Is there any reason for using the AINn files in /sys/devices/ocp.3/helper.15
rather than the in_voltageN_raw files in /sys/bus/iio/devices/iio:device0 ?

I.e. is there any (useful) processing done between the raw 12-bit
integer value in the raw file and the integer 'millivolts' value in
the AINn values?

As far as I can see one just loses a bit or two of resolution by using
the AINn values. The fact that it gives the input voltage in mV is
pretty useless as one is never actually measuring the voltage at the
input, it's almost bound to be scaled in some sort of fashion. So
taking the AINn value and mutiplying it by some factor to give the
required result is no different from multiplying the raw value by a
different factor to give the same result (except the loss of some
resolution).

That is, of couse, unless some sort of processing (smoothing, time
weighted average, etc.) is done between the raw value and the AINn
value.

No answer anyone?

OK, can anyone tell me how/where the values in the AINn files at
/sys/devices/ocp.3/helper.15 are produced? There must presumably be a
little bit of code somewhere doing it.