On Sun, 17 Jul 2016, Robert Edge wrote:

If you don't believe me, do this simple test:

feed a 1kHz tone in to the input

set 'analog volume' control in software to 0dB

connect to ardour or your choice of recorder with decent metering.  set meter
point to 'in'

increase volume of test signal until you are getting 0dBFS.  no matter how much
you increase the level of the test signal from here it will continue to read
0dbFS.  turn the signal generator up a good few dB so you are well in to
clipping.

record a bit of this.  you will see a squared off wave.

now turn down the 'analog volume' control until you are seeing a nice comfortable
level.  do not change the level of the test signal generator.

record again.

the result will still be clipped.

in fact it will be identical to the first signal you recorded, just scaled down.

I mean, you guys do whatever you want, but that is what is actually happening in
your signal chain.

How do you know this is digital and not analog clipping? The result would look the same.

signal path:

AC -> resitor -> buffer -> digitally -> ADC
in    pad on     amp       controlled
      card                 analog gain

if the input signal after the input pad has p-p higher than amp's power rails allow, the signal will be squared off, just as you are seeing. The gain stage would be a part of the ADC chip BTW. The buffer amp is probably not, but rather a gain stage added to allow the use of mic level inputs. (M-Audio advertise it as a Mic Pre)

--
Len Ovens
www.ovenwerks.net