Re: Is 44.1 KHz sample-rate enough? Yes and No.

"Chris Hornbeck" <chrishornbeckremovethis@[EMAIL PROTECTED]
> wrote in message 
news:c7qf145eqg1akbh8kpri0u1j252hlmgg6h@[EMAIL PROTECTED]
> On Wed, 30 Apr 2008 03:30:18 GMT, "Chronic Philharmonic"
> <karl.uppiano@[EMAIL PROTECTED]
> wrote:
>
>>Fifteen years ago I attended a seminar put on by Crystal Semiconductor
to
>>talk about their A/D and D/A converters. I was primarily interested in 
>>audio
>>applications, but they also discussed instrumentation devices. These
were
>>sigma-delta noise-shaping devices capable of true 24-bit resolution (in
>>truth, their 24 bit audio devices were not, at least at that time). 
>>Anyway,
>>they still had a problem with the instrumentation devices at the LSB 
>>level.
>>It seems the thermocouple effects of simply soldering the parts together
>>caused a temperature dependent DC offset which could not be ignored,
even
>>though the converters were self-calibrating. From that, it became
obvious 
>>to
>>me that 24 bits (audio or not) was getting to the theoretical level of
>>dimini****ng returns, requiring near superhuman efforts to keep the
>>resolution stable. I doubt the theoreticals have changed much in fifteen
>>years.
>
> Instrumentation classically means a much lower bandwidth
> than audio, but without definitions, anything can mean
> anything.
>
> This whole thread is in response to a grad-student-level
> troll, so all discussion that truly informs, like yours,
> is good stuff. _Potential_, as opposed to "practically
> available/ real world" dynamic range is constrained by
> temperature and by the Carnot cycle.
>
> Dynamic range and information are related by their mulplicative
> terms: bandwidth and bit depth.
>
> And, as others have said better, humans are constrained
> by their own noise at one end and by hearing damage at the
> other. Numbers larger than 130dB (as defined as within a wide
> audio bandwidth, maybe 20-20K Hz just to be conventional) are
> purely fictional in this context.

Yes, the instrumentation devices had a significantly lower effective 
bandwidth than the audio devices. However, the main consideration in this 
case was that the instrumentation devices had to accurately encode DC 
voltage levels, whereas the audio devices contained a high pass filter
that 
removed DC, and with it, any stray DC offsets. What I took away from all 
that was, 24 bit encoding put us firmly at the lower limits of the 
fundamental ability of of analog technology (or even straight wire for
that 
matter) to represent the signal in the first place.