Do high end CD players actually make a difference? (1 Viewer)

[Lee Scoggins]

Yes, a web site that makes claims that do not stand up to scientific scrutiny.

Robert,

I don't believe it is helpful to the Forum spirit to include DBT debates everytime a high end question arises. Let's leave these comments to their own thread and let people "opt-in" if they want to discuss that wholly separate topic.

 

[mike_decock]

Do high end CD players make a difference?

I think so, but not a "night and day" difference. IMO, the sound of a CD player will be determined by the quality of the DAC implementation.

I'm pretty skeptical of the "bit is not bits" argument. If all the vibration control and jitter were SO critical in getting the data off a CD without error, it would be a MAJOR challenge to get data off a CD ROM in a computer. Heck they do it at more the 50X the speed of an audio CD player with NO corruption of the data.

-Mike...

 

[RobertR]

that wholly separate topic.

Where was it "decided" that "that" topic is "wholly separate" from dubious audio claims?

 

[Saurav]

If all the vibration control and jitter were SO critical in getting the data off a CD without error, it would be a MAJOR challenge to get data off a CD ROM in a computer. Heck they do it at more the 50X the speed of an audio CD player with NO corruption of the data.

The data itself does not get corrupted, i.e. a 1 doesn't turn into a 0 or vice versa. However, that's not all there is to it. Digital audio derives timing information from the edge transitions between the 1s and 0s, that timing generates the clock which drives the DAC. Jitter is an error in the timing between these bits, which creates an irregular clock driving the DAC. Digital data has no such timing requirements, it does not matter if one sector of a CD-ROM is read a few microseconds late or one IP packet of an email arrives a few milliseconds late.
Many people seem to confuse data transmission with audio/speech transmission. Even though both are digital, the requirements are completely different. Look at telecom and datacom networks and you'll see the differences. With audio, the timing is more important than the actual data. If you lose one audio sample due to errors, the system can usually interpolate it from the previous samples and you wouldn't notice a difference. If the samples stopped arriving for a few seconds, or they arrived in the wrong sequence, the system will have no idea what to do, your audio stream will be interrupted and you'll get silence. Data is exactly the opposite - you absolutely need every bit to get to the other end intact, but it doesn't matter if there is some delay, or some packets arrive out of sequence.
All of that is network theory, and not really applicable to CD players and DACs Also, exactly how much jitter is audible is still pretty hotly debated.
Anyway, just thought I'd clear that up. CD-ROMs have little to nothing to do with CD audio. Even if you play an audio CD on a computer, AFAIK most software copies the data to RAM and then reads it out from there, so the "evenness" with which the bits are delivered does not directly depend on the CD-ROM drive.

 

[Saurav]

And in case anyone thinks this hasn't been discussed to death before, here's one 4 page thread
http://www.hometheaterforum.com/htfo...858&highlight=

 

[Lin Park]

I'm a subjectivist!

I wasn't saying you were an objectivist. I was just trying to stir the subjectivist/objectivist pot and see if I could get this thread up to 1000 posts since Yogi slipped in and stole that opportunity from me in our last mega-thread.
Ah well, back to the drawing board ...

 

[Saurav]

I'm mostly a sexist.

 

[Chu Gai]

I'm just waiting for somebody from Sweden to nominate Audioquest's 'designer' for the biggie

 

[mike_decock]

Thanks for the link to that thread, Saurav. It helped clarify jitter in terms of timing rather than data. However, a CD player is an isolated digital network so getting data "out of order" like you do over the internet should be a non-issue unless the player is very poorly designed.

As far as timing errors, this should be easily remedied by buffering/reclocking the bitstream coming off the transport before it goes to the DAC.

My comment was mainly in response to the concept of "stable platter mechanism and better case creation which helps isolate vibration, etc." This should have no effect on getting the data correctly off the CD. Also, it should have no effect on the analog output of the player unless it is VERY microphonic.

-Mike...

 

[Saurav]

Also, it should have no effect on the analog output of the player unless it is VERY microphonic.

There is some debate about the microphonics of crystal oscillators. Damping the XO in a CD player or DAC is a popular tweak, and some manufacturers put the crystal on its own PCB which is mechanically and electrically isolated. While electrical isolation is certainly theoretically beneficial (you don't want the power supply blips caused by servo motors engaging/disengaging to show up on the voltage supplied to the oscillator), the theory behind the need for mechanical isloation is a little more murky, I believe.
And of course, all of this leads right back to the audibility question. Even though you can measure jitter performance difference from one component to the next, that tells you nothing about whether that difference will be audible. Enter DBTs and 1000 post threads Personally, I would take the product/design/tweak/whatever that performed better under a scope, assuming of course that I could afford it if it was more expensive than the other option. Higher cost doesn't always guarantee better objective performance.

 

[Yogi]

Robert,
I don't believe it is helpful to the Forum spirit to include DBT debates everytime a high end question arises. Let's leave these comments to their own thread and let people "opt-in" if they want to discuss that wholly separate topic.

I dont have anything to add to this thread. I was just surfin the threads and stopped by. Sorry for the interruption. Carry along guys.
Over and out.

 

[Craig_Kg]

Bullsh*t. The dielectrics do adjust over time and that is what accounts for breaking in. I have worked with hundreds of recording engineers over dozens of project and maybe all except four or less have used good quality cable that was broken in.
And they do it in three major ways:
1. Beefier power supplies. Having good clean power is critically important.
2. Better transports which lower audible jitter (time-based distortion) to the 50 picosecond realm. Good transports often use a "stable platter" mechanism and better case creation which helps isolate vibration, etc.
3. Better DAC implementation. Here is where things get interesting...you can get better sound from both better DACs and the circuit path implementation of those DACs. Output filters in particular are very important.

Gee, thanks for the nice, measured response Lee! :frowning: Did those 4 or less exhibit significantly lower fidelity in their recordings (and were their unbroken cables of otherwise good quality)?
I will agree with you on all the points about the CD players and add that the analog stages past the DAC can also greatly affect the output.
As far as buffering and reclocking the output goes, the Chord64 DAC (well and truly in 4 figures) can be switched to 1 or 4 second buffering yet Stereophile found that it STILL had measurable (albeit very low) levels of jitter. It seems normal crystal oscillators aren't sufficient to totally remove jitter although the audible threshold is still up for debate (like most audio subjects).

 

[Manuel Delaflor]

A place in which one can note that higher end CDP's do make a difference are the Stereophile reviews. On their measurements they show (among other things) a graph of the response of DAC's dealing with 1Khz sine wave at very low volume.

Lower end CDP's show, as expected, a very distorted (with higher harmonics) signal. Higher end CDP's can do a much better job.

 

[RobertR]

On their measurements they show (among other things) a graph of the response

Of course, if a subjectivist were to use such evidence, he would be in violation of one of the primary subjectivist commandments: Thou shalt not cite measurements as proof of anything, because thine ears are the only thing that matters (I've seen it claimed many times that objectivists never bother to listen to anything, but instead rely solely on measurements). Check out the Stereophile review of the Cary Audio amp I cited for evidence of this commandment.

 

[Saurav]

I've seen it claimed many times that objectivists never bother to listen to anything, but instead rely solely on measurements

Except when it comes to digital gear, where "bits is bits" is the guiding commandment that rules over everything else, and hence it overrides test measurements
I think all philosophies and mantras should be appended with "except where inconvenient"

 

[Manuel Delaflor]

Robert, you really need to give up in your personal crusade over subjectivists. They have what they want, you have what you want. Where is the problem?
Regarding the "subjectivists and objectivists commandments" as you call them, I seriously doubt anyone here have them, or follow them to death. Most of us just use common sense and our personal beliefs when buying equipments.
Back to the topic. If CD's do make a difference or not, well, that question should be answered in a general way in the forum, in a personal way by the interested poster.
Sure it helps to know that sometimes a DBT will show that there are no differences among equipments. But there will be other times in which the differences will be there even in a DBT.

 

[RobertR]

I seriously doubt anyone here have them

I've seriously seen the "who cares about measurements, ears are all that matter" attitude expressed a number of times.

 

[Ian Montgomerie]

The data itself does not get corrupted, i.e. a 1 doesn't turn into a 0 or vice versa. However, that's not all there is to it. Digital audio derives timing information from the edge transitions between the 1s and 0s, that timing generates the clock which drives the DAC. Jitter is an error in the timing between these bits, which creates an irregular clock driving the DAC. Digital data has no such timing requirements, it does not matter if one sector of a CD-ROM is read a few microseconds late or one IP packet of an email arrives a few milliseconds late.

A crucial misunderstanding here seems to be about exactly what is hooked into "the DAC". You are talking as if the CD drive or a digital cable from a CD player is hooked directly into an audio DAC. But this isn't how it works.

Let's take the typical example where you hook a CD player up to a receiver via a digital cable. This creates a data path with, at a minimum, the following distinct stages:

1. Data is read off the CD by the drive controller.

2. Data is sent over a digital cable to the receiver.

3. The receiver's DSP gets the digital input, and processes the data to run effects on it.

4. The DSP sends the data to the audio DAC.

In this process there are two distinct system clocks and two memory buffers (one of each on the CD drive controller, one of each hooked to the receiver's DSP).

The CD drive has a master clock which is designed to create a 44.1 KHz output rate of data from the drive. This clock controls the rate at which data is sent out over the digital output. Data is sent out from a large memory buffer which the drive keeps as full as possible. So for example, if the drive encounters dirt, scratches, or excess vibration that give it a problem reading the data, the audio already in the buffer will give it some time to catch up its reading, before it runs out of data to send out. Drives in portable CD players and car audio players have larger buffers to deal with the additional vibration.

The DSP has a SEPERATE master clock, and its own input memory buffer. The digital input feeds into the DSP's input buffer. In order to avoid interruptions in the audio, the DSP must keep its buffer from overflowing or underflowing. This means that on AVERAGE, its clock must be running at the same speed as the CD drive's clock. If the CD drive is sending out audio at an average of 44.1002 KHz, then the DSP must also be sending out data to the DAC at an average of 44.1002 KHz, to avoid getting so far ahead or behind that there is an interruption in the audio.

BUT, this is only an average. The DSP's input memory buffer gives it some headroom. It doesn't have to match its output timing to the timing of the input, on a bit for bit level. It just has to match the average over a longer period. So it doesn't "slave" its clock directly to incoming bits, it synchronizes it using a hardware device called a PLL (phase locked loop). What the PLL does is, make relatively "slow" tweaks to the receiver's master clock rate, to keep the receiver clock at the same average speed as the CD drive clock.

These PLL changes are basically on a much longer time frame than "clock jitter" on the input signal. The PLL has the effect of drastically reducing jitter on the digital input. A truly bad CD drive clock may produce so much jitter than the PLL doesn't eliminate it all, but even then, it will dramatically smooth it out. With any normal, decent quality source and receiver, basically the source/drive jitter is so drastically reduced, that it is insignificant compared to whatever clock jitter exists in the DSP's own master clock.

Also note that in all the receiver designs I'm familiar with (not including soooooper high end audio DACs), the receiver's master clock directly drives the DACs, which have no clock and no buffers of their own. I guess ridiculously expensive DACs could have their own internal buffer and PLL and reclock the data... but that begs the question of why you don't just get a receiver with a more stable clock of its own. (This doesn't require an expensive clock, so much as a stable power source - variations in the power supply are a big cause of jitter).

Basically in conclusion, essentially all clock jitter is going to come from the last PLL + clock in the digital chain, right before the signal is converted to analog. You will not reduce jitter by buying a 3000 dollar CD player rather than a 300 dollar CD player, if you plan on hooking the player to a receiver. You will not reduce jitter by choosing different digital cables.

 

[mike_decock]

ears are all that matter" attitude expressed a number of times.

As a subjectivist, I have to admit that even if you choose to ignore measurements, a DBT is simply an unbiased method of "using your ears". You simply isolate the subjective experience of what you are hearing from what you expect or believe you should hear.

Have there been any DBTs on CD players?

-Mike...

 

[Saurav]

Also note that in all the receiver designs I'm familiar with (not including soooooper high end audio DACs), the receiver's master clock directly drives the DACs, which have no clock and no buffers of their own.

What about a situation where there is no DSP? Or is there always a DSP? None of this applies to DD or DTS audio, my whole post was related to PCM audio being sent from a transport to a DAC. Receivers are a different story, they'll probably handle PCM differently because they also need to handle DD/DTS which needs memory buffers to handle all the framing information.
So... my understanding is that most DACs use a clock that is a PLL-smoothed version of the clock derived from the incoming bitstream. Here, the term DAC refers to a component that's a standalone DAC (with the analog section, PLL, and other stuff), not just the IC by itself.
Anyway... it all comes down to a question of audibility. Some will say that the numbers are insignificant, I say that it is possible that they could be audible (which is different from saying it will be audible ).
Anyway.. thanks for an informative post, that was very interesting. Do you work with digital audio components for a living, by any chance?