DAN DUGAN:
> >>As I recall they're good to at least 50K.
>>found it:
>>http://www.sounddevices.com/tech/7series-bandwidth.htm
>>Looks like 3 dB down is around 100K.
RAIMUND SPECHT:
>Dan, I think that the above frequency response plot applies to the analog
>preamplifier circuit only. The effective frequency response of a recorder =
is
>determined by the combination of the frequency responses of both the
>preamplifier and the A/D converter at a certain sample rate.
>I measured the approximate frequency response of the 722 at the 192kHz
>sample rate by using a sine signal sweeping from 6 to 128 kHz (fed into th=
e
>LINE input):
>
>http://www.avisoft.com/scratch/alias_722.gif
>
>The response I got is very similar to the published response of the Cirrus
>Logic A/D converter product family CS53xx at quad speed mode:
>
>http://www.avisoft.com/scratch/alias_CS53xx.jpg
>(taken from http://www.cirrus.com/en/pubs/proDatasheet/CS5381_F2.pdf , pag=
e
>9)
>
>So, I would agree with your initial suggestion that the SD722 is good to a=
t
>least 50kHz (at -10dB). However, aliasing can occur above about 60kHz.
Yes. Thanks for that study. I haven't paid much attention to high
sampling rate stuff, but it looks like Sound Devices chose a
1st-order gentle rolloff instead of the usual brick-wall sharp cutoff
filter. The rationale for that is usually given that such a filter
makes less phase distortion on the frequencies below it. But so far
as I know from perceptual studies, that doesn't make a damn bit of
difference up beyond audibility, anyway. I'd rather see just a higher
brick-wall filter that didn't alias.
-Dan Dugan
________________________________________________________________________
________________________________________________________________________
|
