Roam
Banned - aka aerius
- Joined
- Mar 1, 2007
- Posts
- 476
- Likes
- 21
Quote:
You know what's funny? After the single-bit delta-sigma chips were introduced, engineers came to realized that it was complete crap. So they began work on the multi-bit delta-sigma chips, last I heard they were up to around 8 bits. Not coincidentally, the more bits the delta-sigma chips had, the more accurate they became and the better they sounded. Funnily enough, the more bits there are, the more it resembles the old style R2R DAC chips.
Quote:
Raw output from a multi-bit delta-sigma DAC for a sine wave
Same for a 16 bit R2R DAC
No amount of filtering & signal processing magic is going to make the final output from the former as good as that of the latter. With an actual music signal, it gets even worse for the delta-sigma chip as the waveform no longer follows a predictable pattern. The interpolation & reconstruction algorithms can't deal with the unpredictable changes found in real music.
Quote:
It is indeed part of the problem. Digital conversion theory is poorly understood by most manufacturers; it's not surprising since the maths involved are quite advanced and pretty much impossible to understand for anyone without a PhD in the field. Even math PhDs will say it's "non-trivial", which is academic speak for "this problem is insolvable".
With regards to power supply and analog section design, this is another area where most people are frankly clueless. Most people think Blackgates, bigger capacitors, fancy regulators, and the latest "audiophile" op-amps are the solution. They aren't. But that's not something I want to get into right now.
Originally Posted by Ori /img/forum/go_quote.gif That's only partially true. Like all products fabricated on silicon, it takes time and expertise to optimize a design. You will find that the quality of the final product is directly proportional to the R&D invested in product engineering after the initial design was released. |
You know what's funny? After the single-bit delta-sigma chips were introduced, engineers came to realized that it was complete crap. So they began work on the multi-bit delta-sigma chips, last I heard they were up to around 8 bits. Not coincidentally, the more bits the delta-sigma chips had, the more accurate they became and the better they sounded. Funnily enough, the more bits there are, the more it resembles the old style R2R DAC chips.
Quote:
I have to say that only a few years ago I too was partial to parallel DACs. I feel that newer designs have reached a point where the converter chip itself is not anymore the weakest link in a DAC system. |
Raw output from a multi-bit delta-sigma DAC for a sine wave
Same for a 16 bit R2R DAC
No amount of filtering & signal processing magic is going to make the final output from the former as good as that of the latter. With an actual music signal, it gets even worse for the delta-sigma chip as the waveform no longer follows a predictable pattern. The interpolation & reconstruction algorithms can't deal with the unpredictable changes found in real music.
Quote:
We still have manufacturers who invest enormous amonuts of money in archaic transports and proprietary digital processing, while employing a cheaper approach to the analog section or the power supply. Maybe that's why digital is "only" where it is today... |
It is indeed part of the problem. Digital conversion theory is poorly understood by most manufacturers; it's not surprising since the maths involved are quite advanced and pretty much impossible to understand for anyone without a PhD in the field. Even math PhDs will say it's "non-trivial", which is academic speak for "this problem is insolvable".
With regards to power supply and analog section design, this is another area where most people are frankly clueless. Most people think Blackgates, bigger capacitors, fancy regulators, and the latest "audiophile" op-amps are the solution. They aren't. But that's not something I want to get into right now.