|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
65.19.76.104
In Reply to: RE: that doesn't prove anything about 16 bit vs 24 bit. posted by fmak on April 22, 2014 at 08:28:07
I guess I must conclude from your answer that you have not done the kinds of tests that involve different dithering schemes, otherwise you could easily have given specific answers to my questions.
Your point about the Rotel is, exactly what?
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Follow Ups:
I once spent a year playing with the Meridian 518 that leading recording guys like Green Room used and wrote about.
You clearly doesn't know what the Rotel 990 can do wit dither schemes. Go find out first before more questions as my time is valuable.
The Rotel 990 CD player uses the Pacific Microsonics PMD 100 HDCD decoder chip, which provides a number of dither modes. The user interface of the Rotol exposes these to the user. I am familiar with how this chip works, having investigated it several years ago while researching how HDCD works and as part of a general project to understand the theory and practice of dither.
The Rotel user manual is quite terse, but the PMD 100 chip specification provides considerable detail about the dither function. From reading the chip specification and being familiar with the HDCD patents I concluded that there are three reasons why dither might be useful in a product such as the Rotel 990 CD player where this chip is used: 1. The PMD 100 includes a digital volume control When used to attenuate a signal there will be additional bits generated by the multiplications. If dither is not used, this will result in distortion. 2. The PMD 100 includes an HDCD decoder which produces a 24 bit output. If the decoder is being used with a 20 bit DAC chip (as in the Rotel) then there will be distortion added unless dither is applied. 3. The use of dither can improve the sound of multi-bit DACs by causing their low-level errors to be randomized, but this comes at the expense of added noise. Because the last two reasons are specific to a particular product design employing the chip, there are configuration options for dither. As the chip manual describes, the best choice may depend on the converter chip and analog amplifiers further downstream:
Dither Modes
The PMD-100 provides 8 different output dithers in Program mode. All dither levels are available for output data
widths from 16 to 24 bits at 8 Fs and 4 Fs oversampling rates only. Dither must be disabled if the 2 Fs oversampling rate
is selected (as well as 24 bit mode must be selected when the 2 Fs oversampling rate is selected). There are seven levels
of high-frequency weighted dither (modes 0-6) plus minimum amplitude white triangular PDF dither (mode 7). Modes
0 and 7 are minimum amplitude dithers which correct quantizing errors only, whereas modes 1 through 6 are increasing
levels of high-frequency weighted dither designed to smooth out non-linearity errors in multi-bit DAC’s (Modes 0 or 7
are appropriate for single-bit DAC’s). The HF weighted dithers put the dither energy above the audio spectrum, where
most of it is later filtered out by the analog low-pass filter following the DAC. In general, multi-bit DAC’s behave better
with high levels of dither, but some analog circuits following the DAC may have problems with transient intermodulation
distortion (TIM) when confronted with high levels of high frequency energy. The best dither level for a particular circuit
implementation must be determined empirically.
Dither Notes:
1) Dither level 6 is the highest level of high-frequency weighted dither available with the PMD-100.
2) In Stand-Alone Mode, dither mode 0 (minimum high frequency weighted) is available. Setting pin 4 DITH low
will disable all dither including programmed dither and should be used only for test purposes.
The significance of this is that Keith Johnson was well aware that the characteristics of digital signal processing interact with all the downstream analog electronics, including amplifiers as well as converter chips. This means that it is not possible to "optimize" a dither function outside of the context of a particular playback chain. In particular, it is not possible for Archimago to "optimize" the use of dither (or non-use thereof) for a 24 bit to 16 bit conversion. Suggesting or implying otherwise is incorrect and inappropriate. Audiophiles should be aware that their "knowledge" of what is "optimum" is highly specific to the particular components used in their system. Great engineers, such as Keith Johnson, have always known this.
One can approximate all of the dither modes in the Rotel CD player by appropriate configuration of audio editors, so the customary audiophile dismissal is inapplicable: "If you haven't heard product X you are not qualified to comment on this subject."
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
what I said in one line. The Rotel enables dither to be changed according to user preference. It also changes SQ significantly depending on setting.
What was your wobble about?
Perhaps I was too subtle for your mind. I was explaining that I did not have to listen to the Rotel to understand the effects to be heard, as I had already explored these in great detail some time ago.
More important was the explanation provided by Keith Johnson (presumably as he was the brains behind PM) as to how dither choices interact with downstream electronics. Again, nothing new here, but apparently it's not a concept that everyone appreciates, especially those who complain that developers of source components are not properly "optimizing" their products, something that is impossible outside of the context of the complete playback chain. This issue of system interaction is the key to justifying Archimago's choice of simple dither for his experiment.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Quite the opposite and you admitted that the effects can be great.
Subtle in several hundred unrelated words? Ah,mental experiments are of no use in audio.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: