RE: What makes a good Analog/Digital Converter
2007/11/18 13:56:45
(permalink)
Ohhey pretty much laid it out. I don't know that there's any real "spec" or meaningful way for a consumer to distinguish good from bad based on data sheets or what have you, because it is in large part dependent on the particulars of the design and construction and how everything fits together.
Obviously signal integrity of the analog front-end is critical, and how you use the converters matters here too, since so many people with all-digital setups are recording at signal levels that would be considered extremely high for typical studio work. The 0dB "target" on a typical analog VU meter is often 18 or more dB lower than say the -6dBFS "target" level that many digital studios use as their record level. Clearly a circuit optimized to handle steady-state signals of 1 volt with headroom up to 16V would be somewhat different from a circuit optimized to handle steady-state 16V with similar logarithmic headroom (that's actually 256V peak!). I suppose for typical super-hot, home studio line level that is pushed close to 0dBFS, you would actually want a circuit optimized to handle something like 12V steady-state with only a little headroom. But again, we come back to this idea of design philosophy for the application. Any circuit designed to handle anything from 1 to 256 volts with equal aplomb is obviously going to be extremely expensive and somewhat comprimised throughout.
Another critical point is the cutoff filters, and again, how these are made is a matter of particulars. There is no such thing as a perfect filter, and it's up to the designer to decide what kinds of signal and application to design for, and up to the construction to determine how well the actual filter achieves the design goals.
Obviously clock stability is of paramount importance, and while a stable clock source is not very expensive or difficult to achieve in and of itself, the circuit that delivers the clock to the AD is as susceptible to noise and interference as any, so again you come back to the particular design and construction of the overall device. Moreover the PLL circuit to synch to external clock is typically somewhat independent of the internal clock circuit, so again, the "ideal" may be somewhat application-dependent.
And then you have all the stuff like error correction or oversampling or whatever that may or may not be present, and where it is, it is usually a matter of tolerating small comprimises in signal purity to avoid larger potential problems. Whether that is a good or bad thing and to what degree is a matter of real-world practice and implementatiton.
The good news is that even fairly inexpensive converters tend to perform extremely well for everyday sound quality in sensible real-world use. So if you are buying super-premium converters you are typically looking at stuff well above the price-performance knee and we're talking about fairly fine and subtle distinctions that may not even be audible "differences" to most listeners.
Something that is often easy for hobbyists to lose sight of is that mission-critical professional tools including studio gear are often made to be "better than perfect," because professionals are willing to pay a premium for certainty that may not necessarily translate into a meaningful real-world improvement in the end result.
Cheers.