Transport vs DAC - Page 2

Hmm...I wonder whether or not a belt-driven hard drive would sound better!

Seriously...I've bought into the analog voodoo of cables and amps and so forth, and I realize that those who've lived a lifetime dealing with the many nuances of analog equipment have a hard time accepting some of these ideas, but in the digital domain this mindset that every piece of the system contributes to the qualty of the sound is flawed. I've already explained that incredibly cheap CD players have the ability to perfectly read CDs at rates far beyond those required for music playback. Computers routinely move data losslessly at thousands of times the rate (e.g. off hard drives) that is required for music playback...this idea that somehow this data will sound better if the device reading it is more expensive or better engineered for data that happens to be musical in nature makes no sense to me at all.

By the way, a DVD Player is another example of a device that needs to read a great deal more data than a CD Player losslessly (read errors will result in digital artifacts all over the screen, or no picture at all...please don't confuse these with the artifacts you see when uncompressing lossy video data, by the way).

And then there is the ridiculous notion that coating a CD with some magic liquid or putting a sticker on it in order to improve the data in such a way that it will enable the CD Player to make this data sound better. Although I guess it's possible they damage the disc in such a way that the improperly read data sounds better (but this seems extremely unlikely).

Now onto expensive digital cables: Consider the monitor I am looking at right now as I write this message. It is an LCD monitor running in 32-bit mode at a resolution of 1600x1200 with a refresh rate of 60hz connected via a digital interface (DVI). This $10 (if that) cable is moving 1600x1200x4x60 bytes (440MB, or about 2/3 of a full compact disc) of data per second. Losslessly (i.e. if a single bit is messed up, I'd see flickering pixels on the screen). This makes a cable that needs to transmit less than 200K (1/2000 that required for video on my monitor) of data per second seem rather simple.

In short...in my opinion, the DAC is much more important for the sound and the transport is effectively meaningless (assuming it functions as well as your typical $30 computer CD drive). I'd rank the DAC second to the headphones (or speakers) in order of importance for sound, followed by the amp, power source, any pre-amps in the loop, cables, and, lastly, the transport. Compressed audio files (e.g. MP3s) or a damaged CD (e.g. with lots of scratches) will also detract from the sound, probably with a greater effect than all the components except the headphones/speakers.

Of course, I am not addressing the psychological effect of listening to a CD transport you just spent $5000 on. I am sure that influences the sound a great deal. The same goes for the delicate task of applying a magic liquid to a CD and immediately listening to it (with the smell of the liquid still lingering in the air) afterwards. I am sure I experience this effect with my analog components too (I remember unpacking the PS Audio Statement power cord and being shocked by the size of the thing and being fairly certain that it simply must sound better than my old generic power cable). And, to shed more light on my hypocrisy, I just bought a $50 toslink cable (well...my old one was all twisted and mangled looking...).

Also, just to clarify, most CD Players contain a DAC...so of course they are being used for much more than just reading the data off the disc (when using their analog outputs). These players may well justify their cost, based on the quality of the DAC inside them. I know I wouldn't turn down an Accuphase DP-85 if one was offered to me.

Well Scrith,

My limited experience in audio but much longer in IT tells me you are right. Also I havn't read any technically credible description of what could go wrong in a normally working transport, except jitter. And jitter is taken care of by the modern dacs. So what else could corrupt the bit stream?

Scrith is right. No difference between 30$ AV-710 and kilobucks transports if the latter don't do something to the signal. Transport shouldn't do anything to the signal. If it does, it becomes a part of the DAC.

You know the same logic of a transports not mattering because it is bit perfect, can also be applied to dacs. Dacs are all going to come to the same end result, if they were somehow changing something then it would result in a noticable flaw in the music. It comes down to signal quality and signal strength. I like how you all make bold statements like there is no other way to look at it "transports do not matter," maybe somebody should tell that to the reviewers at stereophile, hifichoice, absolute sound, sound-stage, tnt-audio, etc., etc., Im sure none of your conclusions are based on solid testing with equipment in your home.

Exactamundo! Logic tells us all sorts of things about how cables can't possibly make a difference, and transports can't possibly make a difference, etc., but people do report that they hear differences, and to say it is all placebo is silly, IMO. And I agree that it is farfetched to assume that all the reviewers at stereophile, hifichoice, absolute sound, sound-stage, tnt-audio, etc., etc., who have listened to various transports and say: "Don't let anybody tell you it's all "1's" and "0's" because a good transport does make a difference" are just hearing things. BTW, as far as my own biases are concerned, I'm not convinced myself yet that transports ever make a difference, or sometimes make a difference, or make a difference depending on the DAC (though I hope to reach some partial conclusions for my own purposes in the next few days based on actual listening experience). But I am convinced that many who say that transports (or cables) can't possibly make a difference haven't done the listening themselves. They are just basing it on logic or what they believe the science compels. Logic and science are certainly relevant and important considerations, but so is what people hear, and the suggestion that every single person who claims to have heard a difference between transports (including all the professional reviewers and thousands of audiophiles on audio asylum and elsewhere) is really just hearing things that aren't really there is too facile an explanation for what is being reported.

I disagree. Since a DAC generates an analog signal, there are many, many variables that influence their output (an analog electrical signal representing audio) that are not a factor with a transport (the output of which is a digital signal):

• the algorithm used to create the analog signal from the digital data
• the circuitry that generates and then passes the analog signal to the output(s)
• the quality of the power being used to generate the analog signal
• the ability of the DAC to generate the analog signal at the proper frequency
• the buffering provided by the DAC to overcome any timing issues (jitter) in the input data stream
• the quality of the output connectors on the DAC
• the type of analog signal being output by the DAC (single-ended, balanced, etc.)
• any pre-amp capabilities the DAC supports (switching between multiple inputs and outputs, for example).
• many things that I'm not remembering off the top off my head, I'm sure.

There appears to be some fundamental confusion, I think, between the nature of digital and analog signals. An analog signal is used as is by an amplifier, errors and all. It is highly sensitive to interference and the electrical properties of the path it moves along. A digital signal has some room for errors (depending on the ability of the reader to work around imperfect input signals). I assume this was one of the reasons for creating digital signals in the first place...to overcome the limitations of analog signals.

I am not an expert on this stuff at all (I'm a software engineer), I'm just passing along information I've picked up along the way about electrical engineering. I am pretty sure we have some electrical engineers around here that could shed additional light on this subject (and probably make me look pretty foolish).

All you jitter skeptics need to read these:
aos's rant and Everything You Ever Wanted to Know About Jitter but were Afraid to Ask.

~m

Nice article. It seems to be mostly concerned with the problems with ADCs (and the CD and MD recorders that contain them) than with DACs, however. I don't see any issues related to the reliability of a CD transport being used to output a digital signal from a digital source.

Agreed. Nice article, but irrelevant in this conversation.

layman's (that would be me) explanation of jitter

Scrith, I know exactly how you feel!!!

Getting a bit-perfect output from a CD player is easy. So what's all this talk about bit-perfect digital sources having different "sounds"??

But... because so many people, from many different sources, have experienced a difference between transports, I had to look into it further. So, I've spent the better part of my free time for the last week researching this topic. Now call me a loony for doing that, but here's my conclusion of what it boils down to:

TIMING.
This calculation is probably inexact, but if you are transmitting 16 bits @ 44.1kHz, that is 2^16 * 44100 = 2890 Mbits/sec. Or, 346 picoseconds per bit. When transmitting data, you generally don't care if a bit arrives a few picoseconds earlier or later. (i.e. has "jitter") But when transmitting music digitally, you DO care if this happens, because that means the D/A conversion will happen a few picoseconds too early or too late. At the end of the day, that translates into a low-level distortion of the analog signal. And, given sensitive enough equipment and an attentive enough listener, it can be perceived.

EDIT: I now think the calculation I listed must be incorrect. I copied it off another web site, but - I believe the proper calculation is simply 16 bits * 44100 kHz = 705.6 kbits/sec. The other number just sounds way too high, and unless I misunderstand how CD data is stored, this calculation makes more sense anyways. 1 / (705.6 kbps) translates into 1417 nanoseconds/bit.

In a sense, there is an analog component to the digital signal -- and that is the TIMING of the bits. In other words, a bit could be mistimed by any number of picoseconds. As long as the bit is not grossly mistimed, it would still technically be "bit-perfect", but yet it would not result in a flawless D/A conversion.

A reasonable counterargument might be that this seems like it should be easy enough to fix... but apparantly, from what I've read, there's no one perfect solution. I have read about several different "workarounds", but they all add cost and complexity to the DAC circuitry, and they all have their own associated problems and imperfections. (upsampling, oversampling, slaving your CD clock, etc.) In any case, the concensus seems to be that, the less jittery your digital signal is to start with, the better your analog signal is going to sound in the end.

Here is some of the stuff I've read, which may be helpful. As a warning, some of it is pretty dense, and the explanations are not quite up to "textbook" clarity, but at least they do provide some theoretical framework to help explain this phenomenon.

The Extremist DAC - jitter explanation
LessLoss' explanation of their DAC solution
Bel Canto's explanation of why their DAC is the best (PDF)

Finally, for reference, I think something on the order of 0.1 nanoseconds of jitter can be perceived, however there is no definite number because -- this gets into more electrical engineering than I understand -- the amount of jitter in a system is actually signal dependant. Meaning, it is not a "white" noise - there are different patterns to it depending on what bitstream is currently transmitting. AND, to make things yet more confusing, our ear/brain interprets different patterns in different ways. (lack of soundstage, lack of detail, lack of "impact", etc.)

best-
Jeremy

It would be nice if we could packetize and buffer the whole process. Transport spinning 52x reads the 5 minute track bit-perfect in about ten seconds, and sends the data over a HDMI link to a DAC with say 64MB of high-speed low-latency buffer that the DAC then pulls out of.

With the relatively low bandwidth of audio and the kind of fast interconnects we have at our disposal, it should not be a problem to buffer a huge amount of audio data and not have to worry about buffer underflows or overflows. We have fast readers, fast bitperfect interconnects and big RAM pools. Why aren't we buffering and reclocking everything?

Dude, I don't know!

I have been trying to find out, but haven't found a straight answer yet.

Jeremy

Maybe transports do matter maybe they don't, maybe dacs matter maybe they don't. No one can say for sure but you guys are making pretty bold statements that transports absolutely don't matter, especially since it doesn't sound like you have any experience with actually listening to good transports. The whole audiophile thing makes no sense if you try to boil it down to science. A non audiophile would say, "my stereo will sound just as good as yours because look my home theater receiver has 150 watts per channel with 20-20khz frequency response, my bose speakers have the same data sheet specs as yours, how could they possibly sound different." The least you guys could do is back up your arguments with actual extended in home experience.

You guys are geniuses, you should enlighten the dimwitted engineers at the leading audio manufacturers. Im actually dissapointed in you guys, you could have saved so many companies time and money by simply telling them "transports don't matter, all transports are bit perfect." Thier only response would be "wow, I guess i never thought of that, god it makes perfect sense, we are such morons, well I guess we have a lot to learn." While you are at it your going to have to enlighten all the reviewers at all the reputable review houses. Now that I think about it they are pretty dimwitted as well, why do they keep insisting that transports do matter when obviously transports don't matter, all transports are bit perfect.



"In my experience I've found that one of the biggest differences between transports is the ability to create a natural soundstage." - Lucio Caddedu Tnt-audio, reviewing the northstar model 192 transport
http://www.tnt-audio.com/sorgenti/northstarcd_e.html

I urge you to read this review and the countless other transport reviews here: http://search.ecoustics.com/Editoria...lpha-Name/All/

Seriously folks we have to educate these people that transports don't matter, they are all bit perfect.

I guess after we have all the companies and reviewers on our side we can start informing the millions of people with expensive transports that transports don't matter they are all bit perfect.

I don't mean to sound melodramatic, but you guys have saved my life.