Re: [ARSCLIST] Sampling Theory (was Fred Layn's post on the Studer list re: Quantegy)

[Jerome Hartke <jhartke@xxxxxxxxxxxx>]

> -----Original Message-----
> From: Association for Recorded Sound Discussion List
> [mailto:ARSCLIST@xxxxxxx] On Behalf Of Mike Richter
> Sent: Monday, January 17, 2005 6:16 PM
> To: ARSCLIST@xxxxxxxxxxxx
> Subject: Re: [ARSCLIST] Sampling Theory (was Fred Layn's post on the
> Studer list re: Quantegy)
>
> At 03:03 PM 1/17/2005 -0500, Dave Bradley wrote:
> >I find it interesting that people point to subjective studies about what
> >can be sensed above 20 KHz, or above 15 KHz for that matter, and say that
> >because digital doesn't go there it's inferior. Take note that the
> >cartridge on your turntable isn't going there either
>
> But it is. That is, the falloff from uniform reproduction may be steep
> when
> given sine waves, but waveshape is still approximated. The Nyquist limit
> is
> abrupt - above it are only artifacts - and it is made more so by preceding
> circuitry if the engineer is competent. Square waves at two-thirds the
> 'maximum' frequency are far from square, but they do have the right
> components, the lower harmonics are not terribly out of phase and overall
> they approximate the input waveform at least grossly where the sampled
> signal does not.
>
> I've had people make the same argument on tape until I showed them that I
> could read the bias frequency from the tape on a Tandberg - a signal an
> octave above the supposed limit of response. I'm sure the Studer will do
> the same. It was many db down, but still easily isolated from the noise
> and
> easily measured.
>
>
> Mike
> --
> mrichter@xxxxxxx
> http://www.mrichter.com/

It seems that discussions assume rather simple filtering, while linear
phase, constant time delay, filter designs have been available for decades.
Such computer-aided designs can have both sharp cutoffs and low phase
distortion. The techniques are there; it remains for the circuit designer to
take the trouble to do it right.

There are many other opportunities for poor implementation, including noisy
D/A converters that mix analog and digital grounds, use of a noisy
reference, or poor power supply decoupling that introduces intermod. Don't
blame the fundamental design for sloppy implementation.

Jerry
Media Sciences, Inc.