Lapaix, having worked for several years trying to get the ultimate jitter-free sound from a computer source, I can say that it remains better, audibly better, to play files from USB RAM rather than from hard drives. Yes, hard drives are read in packets, and yes, they are buffered extensively throughout a computer system's busses and operating system and playback software and hardware drivers, but the sound remains better from a USB RAM. And this remains worse than the sound achieved when slaving a CD player to a master dac synchronously. The ultimate point in buffering is not the size of the buffer, but the quality of the clock signal. And the quality of the clock signal depends ultimately on all of the surrounding factors including: its own power supply, the radiating ambient high frequency EM signals from other quartzes and other busses, vibrations (hard disks vibrate), and many other things knowns and possibly unknown. THere are several forums on the internet where people have been discussing computer audio for years and some have reached a very high end sound this way. However, once you put all the solutions up to the line and really compare, it remains improbable that merely enlarging a buffer will take away all the digital nasties. Another way of looking at it is that a CD already is a buffer. It is the buffer that is holding the DATA which was recorded at the finite moments of A/D sampling (during the recording). Whether CD, DVD, hard disk, ram, or anything else, the playback depends on the quality of clocking it. |
This is certainly a naive question, but why can't a CD player store the entire contents of a disc in memory, then send them to the DAC, which would be the equivalent of reading a file from a hard drive? Is that the definition of a music memory player? If it is, why is that controversial? Not spoken, of course, as a digital engineer. |
Rather than open yet another thread on MP(I've heard MP & yes it does sound good), I'll address the post by observing that the manufacturer's claims concerning MP theory of operation remain opaque & controversial. There has been too much delay bringing MP to market to consider it at present a serious threat to existing technology. Finally, there appear to be questions as to whether its technology is unique and patentable. The latest AA thread below addresses the controversy & also includes Gordon J. Rankin's recipe on how to build a "DIY memory player" in several easy steps. As regards music servers in general, in his review RH elevates the Esoteric with rubidium clock above HD systems like Sooloos & Qsonic. In any event with servers there is still the need for a clock at the DAC.
http://www.audioasylum.com/audio/digital/messages/13/136279.html |
The Memory Player, produced by the Nova Physics Corp. made clocking obsolete! Here's how it was explained:
"All clocking does is synchronize the laser with the DAC. Clocking synchronizes the beginning of the time the sampler is "open" for the laser to seek & read the bits. So clock & clock again & you only get the sampling period to start exactly when the laser begins reading a new section of the CD.
The problem is the bits can be read at any time during the sample period, and are read randomly. So the bit, a representation of a moment of music, is always late. Sometimes a little, sometimes intolerably late.
So clocking ever more precisely improves an area that is already nearly perfected in $200 DVD players at Walmart.
All Memory Players extract music bits as a mirror of the master. No more clocking to sync to anything, no jitter as nothing moves, only laser reading efficacy & ONLY reading music bits, hence, a mirror of the master.
As you have no error concealment bits to back up a missed bit, Memory Players must reread to capture all bits dropped in the first pass. If we used error correction bits, we're back to synthetics again so the goal is to reread more & more & faster & faster. In this way, very few bits will be lost & the ONLY bits you CAN hear, are MUSIC BITS.
It is not dependent upon a clock. All of the music bits are on memory & just stream off the memory in the order & TIME they were recorded.
The small community that is either copying the MP or designing their own now that the cat is out of the bag, know the CD player as we know it is going extinct. Good riddance. It was a poor compromise they chose in 1982. They had it nearly perfect (& perfect on CDs containing programs instead of music) but relaxed error correction to fit Furtwangler's Beethoven's 9th symphony on it & digital audio was never right again."
So who would pay $15k on a clocker that merely reduce jitter? |
Audiofeil: "I think the manufacturer in this discussion is suffering from what Freud called 'Product Envy'" I choose to dissemintate what I hold to be knowledge at opporunities when my conscience is troubled, for I have a general disposition to side with defending any truth when I suspect disinformation being spread. This reflects a type of world view which I attempt to consciously carry. You may choose to examine further any subconscience aspects of this behavior, but please be aware that in the human science of psychology, it is by now generally known that one tends to notice upfront those aspects of others which are in truth deeply embedded in our own subconscious selves. Liudas |
Dazzdax, The mods I've made to my CDP that have obtained the most audible improvement are refinements to DC power into the digital section: power to master clock, but also separate discrete power circuits into DSPs and motor/servo control throughout the entire digital section. It's conceivable that some manufacturers do a better job than Esoteric to improve performance of the digital section in ways unrelated to accuracy of the master clock. But the benefits that one hears through these improvements to the digital section are still most likely related to a reduction in jitter. If not, how do I account for the audible improvements that I hear in my modded unit? I can rule out upsampling technologies as a consideration, as no changes were made in this area.
Mods to the digital section make digital sound far from awful and in most respects RBCD has surpassed vinyl in my system. My experience confirms JH's finding that short strokes related to banishing jitter really open up the true potential of digital. Problem is this may require heroic efforts like $15K clocks or 300lb. battery supplies. |
$15,000 for a clock? With that kind of moola one could buy one hell of a turntable set-up and still have $ left over for software. Oh, and jitter removal is on the house... |
I think the manufacturer in this discussion is suffering from what Freud called "Product Envy" |
The P-01 transport has very low jitter but still benefits from the Rubidium clock. I wonder whether some jitter in clock frequency is reintroduced in the clock-link cables that connect the G-ORb to the transport & DAC. In digital, distance is very important. If a P-01 transport sounds better using an outboard clock, it speaks volumes about the low quality of the internal clock of that unit. The best scenario is when the low jitter clock is running the whole show at the very DAC chip. This way, data reclocking can be done right before conversion. If ever there is a better sound from using an external clock vs. an internal clock, it proves that the internal clock is of poor quality. Think about all the RCA and BNC connectors on the market. Think about all the digital cables. Why do they exist? Think about Eichmann, WBT, small metal contacts, 75 Ohm impedance plugs, balanced AES/EBU format, etc. etc. All of these differences are audible because Jitter is the result of a pulsating high frequency clock signal interacting with reflections as well as intermodulation with other induced ambient electromagnetic fields. The aether is a chaotic sea of electromagnetic fluctuations going in every possible direction and at all possible frequencies. Just try to introduce a single frequency through this electromagnetic chaos without it even touching another airborne frequency -- and you will see that this is nearly impossible due to inductance and intermodulation. Therefore, shielding is of utmost importance. Liudas |
Do you really think jitter is the only reason for awful digital sound? The P-01 transport has very low jitter but still benefits from the Rubidium clock. There are other transport that have more jitter but sound as musical and with the same detail resolution as the P-01.
Chris |
I would love to hear for myself the effect jitter has on sound. |
JH's point regarding frequency is that a crystal clock is subject to transient fluctuations in output frequency caused by power supply variations & ripple, whereas the output frequency of a rubidium clock is inherently stable. He states that a transient variation in clock output frequency is "the very definition of jitter." Timing errors in the clock translate directly into a misshapen waveform and amplitude errors in the reconstructed analog signal, as well as introduction of spurious sideband frequencies unrelated harmonically to the original signal.
While I accept the benefits of ultra low-jitter clocks, I wonder whether some jitter in clock frequency is reintroduced in the clock-link cables that connect the G-ORb to the transport & DAC. |
Jitter occurs all over the place. It is of no significance except at the real-time A/D process or the real-time D/A process, which is when the distortion takes place.
Drubin: The point that Serus and I are both making is that frequency accuracy means nothing regarding audio quality. It is used by some marketers of digital audio technologies to introduce new numbers with flying colors (they look really good on the ads). However, the point in Jitter is not whether the frequency is accurate, it is however the point whether the SAMPLES are accurate. And that's what I am trying to show in my example: samples can be way off and cause massive distortion at the D/A or A/D process but the frequency can be right on target. Frequency is a total amount of oscillations per unit time. Jitter is how much each sample is off time target each and every time. And with clocks, this can be 33 million times a second. So potentially, a clock can make 33 million little mistakes a second and still be accurate to a fraction of a second within years and years of running.
These two things must be differentiated. And it is important to understand that Bach sounds great, whether the music is tuned to A=440 Hz or A=440.2 Hz. Nobody, not even Bach himself, would ever notice the difference. But I think it's safe to say he wouldn't have liked Jitter.
Liudas |
I'm having a little trouble following this, but I wonder if "frequency accuracy" is not being used in the same way by all of the posters above. Just to be clear, we are not talking about audio frequency response, at least as I understand what RH is talking about. Dgarretson's summary is excellent. |
Rubidium is not necessary for timing accuracy, but it makes sense that a commercial rubidium reference oscillator would pay special attention to close-in phase noise in the vicinity of the main carrier, which is what's improtant in regard to jitter.
To put the frequency accuracy issue to rest, here are the facts. Standard (i.e. cheap) crystals are guaranteed within 100 PPM (parts per million) in the proper circuit. Add another 100 PPM for extreme temperature/humidity and circuit voltage variations. The total is 200 PPM or 0.02% of frequency accuracy. For a 1 KHz signal it will be off by 0.2 Hz. Even an acute dog wouldn't be able to tell the difference...
It's a joke when some manufacturers offer an "upgrade" to a TCXO option. How about upgrade to a low-noise oscillator? I'll take that option any day! |
Jitter occurs mostly at the DAC, whether with music server or CDP. |
WARNING: There is a lot of marketing going on here. Please consider the fact that frequency stability is not the issue of Jitter! The issue of Jitter and how that distorts audio at the A/D process is a problem of each sample coming at slightly the wrong time. Frequency stability has nothing to do with this. A small example:
The numbers 3,3,3,3,3,3,3,3 average out as the number 3.
The numbers 2,4,2,4,2,4,2,4 average out as the number 3. (This is as simple as 24 / 8 = 3).
The numbers 1,5,1,5,1,5,1,5 also average out as the number. (This is as simple as 24 / 8 = 3).
However, were these Jitter deviations from perfect timing (fluctuations from the ideal value, which was in this example, 3) you would have had worse and worse audio going down the line.
The marketers of all three pieces of equipment in the above example could easily all have said the same thing without lying: "Our Clock is so stable that you can play it for 3 million years and it will never be off by more than a fraction of a second".
Please notice the significance of this. Be aware of the logical trap that this "frequency stability" terminology is putting people into!
Liudas |
Will jitter be an issue in the future if music is primarily read from music servers? |
It's a fascinating product review & survey of the jitter problem. If as Joe Harley observes, the ear can easily discern improvements in word clock accuracy down to +/-0.05 parts per billion (translating into jitter reductions on the order of tens of pico seconds), then it would appear that there's still a long way to go toward true SOTA in digital sound. Better clocks are far from an exercise in diminishing returns.
JH makes interesting observations. Unlike the crystal oscillators found in conventional CDP clocks, the rubidium-based clock in the Esoteric does not exhibit frequency fluctuations as a result of instability in the power supply. This correlates with my own experience that improving the quality of DC power into a VCXO Superclock greatly improves its sound. He recounts the luddite position of the Audio Engineering Society in the early 90's (opposing the very notion of jitter), and the gradual acceptance of the concept in the general community. He provides a nice working definition of the signature sound of jitter as "loss of space & depth; softening of the bass; hardening of timbre; a glassy sound on initial transients (most noticeably on the leading edge of upper-register piano attacks); a metallic sheen overlaying the treble; and an overall flattening of the soundstage & homogenization of instrumental images with the stage." Finally, JH notes that the rubidium clock option elevated the Esoteric P-03/D-03 well above the music servers that he had formerly favored over the Esoteric. Perhaps there is hope after all for ye olde compact disk. |
>I read some of the article but the price of the trio left me speechless.
Sure, but that's sort of beside the point. The technology, if worthwhile, should trickle down. What's important, I think, is what this product tells us about the problems with digital today, the merits of various claims being made about vanishingly low jitter, and what we may be able to look forward to in the future.
Rja, that's interesting. Which label? Are they good-sounding CDs? |
I read some of the article but the price of the trio left me speechless. |
I own several Japanese CDs that state they were mastered using a Rubidium atomic clock. |
Let me give this one a bump. Must have chosen the title poorly because I thought this would stimulate some discussion. |
