Monday, January 16, 2012
Delta Sigma DAC chips...
I wrote this blog post a while ago, and so I add this preface to say that over time and after gaining more listening experience with different chips, I don't believe that all delta sigma DAC chips sound that bad. I now firmly believe that some delta sigma chips sound very good. But, I am still yet to hear one that to my ears sounds as analog as earlier more costly breed of DAC chips.
Today in electronics everything is made small, and as a consequence, compromises must be made to make everything fit in a small package, and run off one power rail.
Delta Sigma DACs all have internal op-amps to provide a voltage output.
Op amps are not evil, however there are good sounding op-amps and average sounding ones. Good ones cost more than most delta sigma DAC chips themselves. One of the problems with most delta sigma designs, is that the on board op-amp is not of the best quality. There is no option to take the current output from the chip. Hence we are stuck with the voltage output and consequently, the "sound' of the op amp, that the manufacturer gives us.
The reason delta sigma was developed was to get the chip size smaller and reduce manufacturing costs. From what I can tell, it is not in any way shape or form to obtain better sound.
If the object was to obtain better sound then the manufacturers would improve laser trimming of the R2R ladder network, however this costs a lot of money, and the chips stay big! However the PCM1704 is an exception to this rule being a small surface mount 24bit 96khz R2R chip.
The Wikipedia entry on Delta Sigma makes numerous comments about the "low cost" of Delta Sigma DACs.
Most (but certainly not all) manufacturers are incredibly lazy these days, they just want to use plug-packs for power supplies, and put little thought into designing a good power supply. They also want to minimize part count and maximise profits. R2R chips are more complex, and require more attention to power supply quality, but in return provide sound which can't be matched by any delta sigma design. What I mean by this is that low level details can be more easily heard which results in greater tonal accuracy, and gives a more natural sound only comes from an R2R DAC.
R2R DAC chips have a resistor for each bit. CD is 16 Bit, so hence an R2R DAC has 16 (actually 32 resistors) per channel. Check Wikipedia for a full explanation of R2R conversion.
What this means is that rather than having feedback loops and high frequency oversampling clocks to produce an approximation of the digital data, the R2R DAC will turn on each required bit in the digital "word", and the exact amount of current will flow from the chip, to reconstruct the analog waveform. The method is perfect in its simplicity, and this is reflected in the sound quality.
It might seem that R2R DACs are old technology, but actually they were the "flagship" of audio DAC design. Intelligent people recognise that when electronics companies present new technology such as CD players, Plasma TVs, and bluray players, the first generation equipment is expensive, over engineered, and a lot of effort is put into it to make it the best that they can. It needs to look and sound excellent in order to convince the consumer to buy it. Once the "flagship" product has been purchased by a few people and recieved some very good reviews, then the manufacturers start to make the technology cheaper and more disposable, relying on the reputation of the "flagship" product. If you need any proof of this then look at the first generation plasma TVs which in many cases are still running to this day. Most new plasma TVs will fail in 1-3 years. Many vintage CD players had at least a 20 + year life span.
At this point you may be reading this and thinking that early CD players received mixed reviews, and you are correct.
But what seems to have happened is that somewhere the consumer was lead to believe that the DAC chip itself was the reason behind these mixed reviews. In doing so, the baby was thrown out with the bath water.
I believe the sonic failings of some early CD players to be in the design of the analog output filter, and simply that not enough effort was put into listening to the player and making the "tweaks" needed to make them sound really good. Having said all of this, there were early CD players that did sound great out of the box. Early reviewers were used to analog sound, which has certain characteristics. When presented with digital audio they must have been in absolute shock, and not surprisingly were quick to point out the differences, negatives, and highlight the "dryness" of the CD sound in comparison to vinyl. For anyone to expect any two different mediums to sound identical is foolhardy, just as expecting an electrostatic speaker to sound like a conventional cone speaker, or expecting a horn tweeter to sound like a dome tweeter.