Fwiw, the Lavry Engineering ADC (AD122-MKIII) has, according to the
manufacturer, 127 dB unweighted dynamic range, equivalent to about 21
real bits and about 3 bits worth of digital padding, since
everything's in byte bunches. Both the gold-colored A/D and D/A use
an oven for stability, as well as on-board compensation for value
drift with age. These are "discrete" topologies, so the resistors
are cherry picked.
Andrew
On Apr 27, 2011, at 7:17 PM, George Brock-Nannestad wrote:
> From: Patent Tactics, George Brock-Nannestad
>
>
> Bob, I think you will agree that we are past the obituary stage.
>
> Basically, 20 bit resolution with linearity is an impossibility, as
> is 24
> bit, without a temperature controlled oven for all the resistors in
> the
> networks used in AD or DA conversion. All the resistors have to be
> burned-in
> for stability beforehand (easy when it is integrated), all
> capacitors have to
> be high-stability and low loss. When you have done that you can
> make 8½ or 9½
> digits and you are called Lucent. Alas, for audio purposes this
> precise
> voltmeter is only useful at fractional speed, because it takes
> about 1/10
> second to perform a measurement. It is clear that 20 and 24 bits
> can be
> realised and when in IC form to keep transmission distances short,
> quite
> fast. But it will not have the precision that would permit analog
> amplification of the least significant 8 bits to give them a value
> of the 8
> most significant and to retrieve the waveform. 24 bit may be
> practical for
> very many reasons, but it is what is called empty magnification in
> optics.
> Sample rate is much more important, but I would say that jitter is
> probably
> even more so.
>
> I have unfortunately not been able to locate the AES paper that is
> most
> revealing of the wheelings- and dealings that went on to make an
> audio sample
> rate that would work with video frame rates, but it is out there!
>
> Kind regards,
>
>
> George
>
>
>
>> -----Original Message-----
>> From Andrew Hamilton: "... although conversion, and jitter
>> attenuation,
>> has
>> come a long way, the format that CD offers falls short of the
>> relatively
>> (as
>> if) effortless and open sound that 48 k and higher sampling
>> offer. .."
>>
>> The Society of Motion Picture and Television Engineers along with the
>> Audio
>> Engineering Society did an extensive test of sample rate and bit
>> depth in
>> order to determine the minimum values capable of audible
>> transparency for
>> the purpose of establishing a standard production format. This
>> included
>> meticulously crafted double blind listening tests.
>>
>> What they came back with was 48 kHz. at 20 bits. Many designers
>> believe
>> audibly better filters remain theoretically possible up to around
>> 60kHz.
>> and
>> experiments have shown 21 bits to be necessary in order to
>> reproduce the
>> dynamic range of an orchestra as heard from a conductor's
>> podium. This
>> is
>> real science and not audiophile or skeptics' fantasy. Progress
>> doesn't
>> change the laws of physics and the limitations of anti-aliasing
>> filters
>> within the audible range below 20kHz.
>>
>> Bob Olhsson Audio Mastery, Nashville TN
>> Mastering, Audio for Picture, Mix Evaluation and Quality Control
>> Over 40 years making people sound better than they ever imagined!
>> 615.562.4346 http://www.bobolhsson.com http://audiomastery.com
|