CD Specs--What do they mean? Marketing?

[MBM135]

Is comparing specs with CD players really meaningful?What is really meant by Signal/Noise ratio and dynamic range? THD?

My $300 Denon CD changer has the following specs

S/N Ration 112dB

THD .0025%

Dynamic Range 100dB

Channel Seperation 98dB

Freq Range 2Hz to 20KHz

I've been lead to believe that bigger is better for S/N ratio and dynamic range. But, lets look at Denon's top of line DCD-1650AR retailing for $1000:

S/N ratio of 118

THD .0018

Chan Sep 110dB

Specs are only slightly better. And, many other models in the $500+ price range that people rave about (NAD, Rotel, Tjoeb) don't sport specs any better than my $300 Denon. So, should one disregard specs and let the ear make the comparison?

Hmmm...

MM

[MBM135]

Oh, and if these specs aren't what's critical, what should one be looking for for awesome sound? 20 bit DACs etc??

MM

[Ray Garrison]

Hi there.

These days, the specs on just about any decently designed, engineered and built CD player will be pretty much meaningless. They will all show S/N ratios, frequency response, dynamic range, channel separation, distortion levels, etc. etc. etc. that are so far beyond (er, better than) what is generally accepted to be the threashold of audibility that comparing them won't tell you anything.

This means that either (a) all CD players sound the same (the $ensible $ound and Consumer Reports position) or ( differences in sound between CD players are due to things that those measurements do not reveal. Sound and Vision has started doing some interesting measurements (something called excess noise at 16 and 20 bit quantization, other stuff) and Stereophile for years has been publishing detailed jitter measurements, amoung other things.

The best thing to do is read everything you can get your hands on, go listen to several of the CD players that got the better reviews, and see if what you're hearing corresponds to what the reviewers were saying. If you can find a reviewer(s) who's comments seem to be pretty much in alignment with what you hear when you listen to the same equipment, you can use that reviewer as a fairly accurate guide to what the various players sound like.

Nothing will replace listening to them yourself. This is where a good dealer comes in. Any reputable high-end dealer should let you take a CD player home so that you can try it out in your own system before you commit to a purchase.

------------------

Music is art

Audio is engineering

Ray's Music System

[Colin]

step aside, Ray and Mike Tyson are back!

The November #88 issue of the $ensible sound does mention DVD/CD players that the staff likes, but they are fairly common names: Sony, Technics, Onkyo and Apex. I would certainly expect to see others. Perhaps they feel the high prices of some uniuts are not $ensible.

Specs are only slightly better. The qualities that people like about a CD player are not easily measured or described. The higher S/N and channel seperation would be noticable on CD players as slightly better sound, but other features, not so easily measured, would be just as important.

Stereophile not only found that jitter matters to a CD player's sound, but also cable direction and player location did too!

Best thing to do is to listen to your own piano, drum, guitar, flute, saxophone and vocal solos, classical and rock arrangements on as many different players on the system as possible.

------------------

Colin's Music System Cornwall 1s & Klipsch subs; lights out & tubes glowing!

[soundog]

My experience is that the transports are not anywhere near as critical as the quality of the DAC. However, jitter reduction seems to be a significant benefit with lower priced transports.

------------------

Soundog's HT Systems

[WMcD]

I don't quite get the effect of the admittedly very good specs on CD players. It seems to me that speakers and the room create larger problems.

For example. I think that 100 dB would be very loud in a room and you might drive your system to that on the peaks output of a CD. But there has got to be about 10 dB of ambiant noise in the room. So the room set up only has 90 dB range before hitting the noise floor, roughly.

Ordinary CD's have about 96 dB of dynamic range.

So why seek a CD player going much beyond those limits?

Regarding THD, I'd think that even good speakers are introducing far more than any piece of electronics, including CD players.

Therefore, I'm happy with the run of the mill CD player.

Of course, consider the source: I grew up with LP's, 8-track, cassettes, somewhat noisy FM, and AM top 40. So any CD sounds clean to me. =8^o.

Gil

This message has been edited by William F. Gil McDermott on 01-24-2002 at 08:49 PM

[Prana-Bindu]

Thanks for pinging, dude. I've been very interested in the various options available for the digital end recently, so I'm glad I get to spew for all y'all.

Jitter is timing error. It is more present in separate digital systems (separate DAC and separate transport and separate Digital Signal Processing). Single-unit cd players exhibit less jitter, because time errors are introduced in the encoding process to S/PDIF (Sony Phillips Digital Interface?) -- the format used by today's common digital outputs. Where do these timing errors come from? The explanation still confuses me into just wagging my finger. Nevertheless, I'll give it a try. More talkie talkie...

Music is digitized at a particular digital-word length (16 bits) and sampling frequency (44,100 words per second). The length of the word is relevant insofar as the more bits used by a binary numbering system the more numbers it will be able to express (the more words that will be available to use). Each word has an association with the aural music (volume, frequency, etc.) qualities. The more words available, the more different qualities will be expressible. Remember: sound spans an infinitely divisible continuum. The more bits in a binary encoding system, the more of the continuum that will be available to express. It's like comparing a crayon box containing only two shades of what we often call red with those outrageous crayon boxes with sharpeners built in. Crazy!

Higher sampling frequency crams more differently placed (on the continuum) expressions (digital words)within the time unit used to measure the frequency. It's like the difference of plotting a 5-foot long wave with parallel slide rules and doing so with straws. Again, more of the continum is available for expression.

Now, when music is digitized (the wave form is turned into lined-up straws), the original analog waveform signal is pecked at during the intervals and for the length of such intervals defined by the frequency rate. In other words, the music (waveform) expressed in cd's is pecked at (during the digitization process) 44,100 times per second. Each peck can be analyzed for relative volume, frequency, etc. and assigned that peck's (word's) value accoding to the table of values used by CD's (isn't that called Red Book or something like that?). Obviously, cycles per time unit is a measurement of time: the digitization process involves a clock (hardware).

So what are the effects of timing error? The playback of the limited (finite) snapshots at times relative to each other that don't match up the times, again relative to each other, at which the snapshots were taken. In other words, a note played at the wrong time is the wrong note. I'm not a practiced enough listener, and I've never heard a less jittery playback next to a more jittery playback to even try to express the effects in the vocabulary of high-end audio (decay times, transient response, detail, soundstage, dynamics, etc.), but I can understand the nature of the error, and it frightens me.

So how and where does this timing error occur? Why isn't the timing information built into the digital stream as it's stamped on the cd? I believe the answer to the latter question may be a weakness of the limited storage on a cd, but I don't know enough agout DSP to know whether that information (time) can even be written and reproduced in a more accurate and retrievable manner. Maybe it's a function of storage space, and we're still at the infancy of such technology (only a few years ago from paper and telephone)?

Back to the first question: we'll, there's no such thing as a perfect clock; even atomic clocks are subject to measurable error (noticeable error -- heh heh, dirty move). Furthermore, look at the fractions of the units these clocks in cd players and analog-to-digital converters have to tick off! CD's demand at least 44,100 cycles per second -- doesn't that seem like an error-prone task? Maybe not, but even if the tranport mechanism and the digital-to-analog converter (DAC) used the same clock information, the very fact that you have at least two physically distinct clocks(one is at the studio and the other in your listening room) means that one is not going to time the same bits at the same rate. Now we're entering fuzzy territory for me.... You see, the information the transport unit picks up is a series of flat surface and indented surface lengths along a single-track (to hide their numbers, as it were) that wraps around the cd in a spiral around itself (groovy...). The troughs and flat areas don't represent 1's and 0's themselves. No! The change from flat to dug-in and vice versa stands for a change from a 1 to a 0 or vise versa (but not respectively - heh heh). There are other rules to this encoding scheme, such as the maximum number of 1's and 0's that can be in a row, but I'm not very familiar with them.

I am under the impression that there is another level of decoding to occur before the DAC has words to conver to specific straws that will make up the reproduced waveform, but I'm not sure. My knowledge of the issue takes a sharp turn to conjecture at this point, so please extend some patience. No matter what, you have a clock that was encoding the analog waveform at one end and a totally phisically separate clock at the other end trying to play the notes at the right time relative to each other.

Now, in a single-unit cd-player, this timing error is supposed to be inaudible. The use of separate DAC's and other DSP devices brings the issue of jitter to a supposedly audible level, mostly because the S/PDIF format is another level of encoding of the digital signal that utilizes a clock. S/PDIF signals, though digital, have time information about the signal playback burned into them. In other words, S/PDIF encoding scrambles the timing error inherent in one clock in addition to the digitizing clock right into the signal. Then when the separate DAC applies its clock to the signal, more timing error is introduced. That's at least three clocks (physically separate from each other, messing with the time. For some reason, though the S/PDIF encoding is particularly guilty of time error introduction. The two levels of timing error can be cumulative as well as consecutive or even cross-correcting, much like constructive and destructive interference in waveform interference. In other words, the error of one process may play a note later than it's supposed to be played, and the error of the other clock can make it play even later or sooner or make them cancel each other out, you get the picture....

Why use separates? To upsample, add correcting noise, upgrade current cd-player's DAC process, and other DSP tricks. For example, I run an old (old!) Onkyo cd player into a Perpetual Technologies DAC. After I upgrade my speakers (RF-3's), I'll want to also use the Perpetual Technologies P-1A DSP computer. It does some neat tricks with the digital signal! Anyhow, Perp.Tech. addressed the jitter issue by using a I2S (eye-squared-esssss...) interface that carries the time info separately from the signal so no time-based encoding has to occur to get the info from the transport to the DAC. The P-1A doesn't take an optical input (bastards!), so I'll probably get a transport at the same time that outputs in a Perp. Tech. version of the I2S interface. That expense is gonna suck s**t.

Of course, one of the cutting-edge companies out there has come out with a I2S interface that can run the transport at the rate presented by the DAC clock, further reducing the chance of jitter occurring. I don't think they make transports or DAC's that take advantage of this so-called "Level 1" performance. As you can see, this technology is quite new and it's expression in the market is still a matter of handicapped speculation.

Why bother when I can just get damn well-timed sound from a solid Arcam cd-player? Because the P-1A is really neat!!! I'm too tired to go into it, but I'm sure if you do a seach on it you'll find more of my drivel about it.

What were we talkig about? Did I just fart?

------------------

May the bridges we burn light our way....

[Prana-Bindu]

Jitter is just timing error: playing a note at the wrong time. There are supposed to be several sources of jitter. As I understand it, the S/PDIF encoding is supposed to be the biggest source of them all. The various sources may be cumulative or may cancel each other out, much like wave constructive and destructive interference.

I don't know what a EFM demodulator is, dude, but it sounds sexy.

There are several schemes out there to get over the S/PDIF jitter in a separates situation, but they always entail getting the transport and DAC from the same manufacturer. This is because these schemes essentially try to run one clock as a slave to the other or do a better job of encoding and decoding the S/PDIF signal, because the two pieces are designed to work with each other.

As I understand it, most (if not all) jitter sources other than the S/PDIF encoding process add inaudible levels of jitter. Not sure about that, though.

If you can find Spider, he seems to know quite a bit about this stuff.

------------------

May the bridges we burn light our way....