> A digital limiter would be post conversion, therefore missing the
> point of a limiter entirely, which would be to avoid crashing the
> ADC. It would need to be analog to be of any use.
<<Scott,
Not if you over-sample. Two bits extra give you 12DB more headroom,
for instance.>>
I don't follow this. Oversampling doesn't buy you extra bits of dynamic ran=
ge in the conversion, it gets you higher frequency response, although in ge=
neral practice oversampling was implemented in order to be able to apply a =
gentler rolloff slope near the Nyquist frequency in place of the brickwall =
filters previously employed, rather than for extending high end.
<< There is a 24 bit ADC onboard, so use it and select the
most appropriate 16 bits to record. That's still 78dB S/N ratio after
dithering. That would suggest the input sensitivity settings could be
in 6 dB stages. As I said, if recording at 24 bits, there is no excuse
for overshooting the ADC. Theoretically 24 bits (138dB range) actually
means you don't need a record level control, but no ADC can be that
good.>>
It's not the ADC that limits us to less than 24 bits of dynamic range. It's=
the molecular noise of conductors carrying analog into the converter. That=
gives us a theoretical maximum dynamic range equivalent to about 21 bits d=
igital. Nevertheless, a digital limiter working on the data stream post con=
version, while occasionally applicable in post-production, would be a bit o=
f a 5th wheel for a field recorder.
Scott Fraser
|