[IPodPJ]

I read the part I was able to.  Even though a master clock is mainly for studio use where you have multiple devices that need to be synchronized for recording, the article is very informative and explains why timing is critical for audio reproduction.

 

[IPodPJ]

Quote:

 

Jitter has nothing to do with the quality of the signal itself.  The signal is all still there and can be reclocked and sent about its business.

 
I'm not talking about the 1s and 0s part of the signal.  Obviously that isn't going to change.  But the timing of the samples still gets transmitted and received, and in order for the transport and DAC to work in perfect harmony their clocks should be synchronized.  True, this has to do primarily with clock timing and very little with jitter unless the levels are extremely poor -- but there are some components with very poor jitter levels in the nanoseconds.
 
Have you ever used a sound card that let you adjust the sample rates manually?  When you change the sample rate to a value that doesn't correspond with the song's sample rate, the pitch changes drastically.  Obviously this is a gross exaggeration because the timing of two devices isn't going to vary by tens of thousands of samples per second.  But small deviations in the timing can cause very small pitch shifts.  Those pitch shifts may not be apparently audible but they do cause changes to perceived soundstage and imaging.  The accuracy of the crystal (the closer to 0ppm stability the better, and the closer to 0ps jitter the better) will determine how accurate the timing is.
 
Edit to add:  According to Rosendahl who makes the "nanoclocks" audio clock, "Ground potential variations, similar to earthloop induced hum in analog audio, can cause jitter in clock signals.  A transformer isolated input eliminates this effect"

 

[Prog Rock Man]

Quote:

Quote:

I'm not advocating only testing audio gear with objective, technical measurements, because even though I think its theoretically possible, no one is probably going to spend the time to create such tests that will only be useful to a small market like us.  Some things could be measured on an absolute scale, while others could be rated on a scale with one subjective preference on each end.  It's also possible for a bunch of poor measurements to somehow add up to something that sounds good, so subjective impressions need to be included along with objective measurements.  My point is that we need both.
 
There are two main reasons for consumers to worry about measuring audio equipment.  The first is because you can't always get a chance to test something before you purchase it.  The second reason is to make sure you're not fooling yourself.  The tests we do have aren't perfect, but if you have two pieces of equipment, one $500 and one $5000, and you're don't perceive a very large difference, and they test the same, then any difference you notice is likely psychoacoustic.  Like I said, it's not perfect, but it's the best we've got.
 
I'd also argue that DACs also have a more narrowly defined purpose which make them easier to objectively test.  Even if something like the "NAIM sound" isn't a simple EQ curve, it's nothing that couldn't be replicated by a DSP.
 
Again I'm not saying that non linear DACs are necessarily bad, but it's nothing you can't more easily replicate with something more versatile that can switch between different sounds as the mood strikes you.  If you already know you like the sound of a piece of equipment than go ahead and buy it.  My only beef with people who don't disclose what they are actually selling.

Totally convincing for me 

 

[maverickronin]

Quote:

You don't seem to get it. If something measures "perfect" it often will sound like crap (sterile, analytical, non musical). Back in my days of car audio I did a ton of measuring and you know what? The amps with the lowest measured THD were often the most non musical, hyper analytical amps I ever came across. By the way I will never measure home audio with all the nasty AC power involved. DC is practically pure and is a far better foundation for measuring equipment. "A dac that doesn't transparently output an analogue version of the digital input".... I must be missing something here because that makes absolutely no sense, can you please clarify? Dsp's still operate in solid state which cannot produce even order distortion. No matter how much math is involved, it's still transistor based and that is its limit.

Saying the amp isn't your favorite 'color'  doesn't mean it sounds like crap.  It means you wish it had tubes in it.
 
It's pretty easy to test the transparency of a DAC if you've got suitable equipment.  Take a given digital music file, and open it with an appropriately sophisticated waveform analyzer.  The take the same file, play it through the DAC and record it via similarly high quality sound card.  Compare the two waveforms.  The closer they match, the closer to transparent the DAC is.
 
On to DSPs, also know as Digital signal/sound processors.  A DSP may be hardware, or software.  A hardware DSP is a specialized microprocessor designed to process digital sound, and is usually embedded in a specialized electronic device.  A software DSP is a computer program designed to process digital sound, that runs on more generalized hardware, such as a PC or Mac.  Note that digital sound is just and arrangement of numbers, and that numbers are good for math.  To simulate the sound of a tube, a DSP would first analyze the waveform, second create the even order harmonics that a tube would, and third, add them to the original signal.  And that is how solid state components can create even order harmonics.  Programmaticly, not electrically.
 
Please note that no magic transistors which produce even, instead of odd order harmonics are used.  Also note that solid state components, such as opamps and passive crossovers, perform analog signal processing, and that it is possible for vacuum tube based digital computers to perform digital signal processing.  Such a machine could also programaticly produce odd order harmonics if you were in to that kind of thing.

 

[maverickronin]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif

But the timing of the samples still gets transmitted and received, and in order for the transport and DAC to work in perfect harmony their clocks should be synchronized.  True, this has to do primarily with clock timing and very little with jitter unless the levels are extremely poor -- but there are some components with very poor jitter levels in the nanoseconds.

 
Timing information is not sent, only the raw bits.  The relative timing is what's important.  Each clock does not need to agree that right now is exactly is exactly 5:15PM GMT plus 2 nanoseconds.  The important part is that each clocks idea of a nano or pico or whatever second it the same, and consistent from tick to tick.
 
Imagine I'm sending you a message in morse code.  It's easy if we do it nice an slow and can take some time to think about each, dot, dash, and pause in relation to what came before it and what came after it.  That's a form of error correction.  Its used by asynchronous transfer protocols, such as TCP/IP, but not used in bitstream protocols like S/PDIF and TOSLINK.  Now imagine that I'm sending you another message in morse code, but this time, by divine fiat, you must transcribe each dot or dash immediately, without using later ones for reference, and it is impossible for you to remember anything about the length of a dot or dash after you have written it down.
 
In such a convoluted system, the only way to get a message across is to have ahead of time agreed on very precise timings for what constitutes a dot or a dash.  If we say that 1ms is a dot ans 1.5ms is a dash then we both need to know exactly how long a millisecond is, but it will not matter if you think it 5AM and I think its 6PM.  This is the way bitstream hardware works, because it would be more expensive to give it the brains to correct errors.
 
The clocks don't really have to be synchronized.  A long as they tick at the same rate there will be no jitter.  If transmitter A sends a bit on its count A 0 and receiver B receives it on its count B .7 then it will expect the next bit at count B 1.7 which will be the same as count A 1, provided both clocks think "1" is the same period of time.  Like I said before, it's relative, not absolute.

 

[computerparts]

 
Quote:

Saying the amp isn't your favorite 'color'  doesn't mean it sounds like crap.  It means you wish it had tubes in it.
 
It's pretty easy to test the transparency of a DAC if you've got suitable equipment.  Take a given digital music file, and open it with an appropriately sophisticated waveform analyzer.  The take the same file, play it through the DAC and record it via similarly high quality sound card.  Compare the two waveforms.  The closer they match, the closer to transparent the DAC is.
 
On to DSPs, also know as Digital signal/sound processors.  A DSP may be hardware, or software.  A hardware DSP is a specialized microprocessor designed to process digital sound, and is usually embedded in a specialized electronic device.  A software DSP is a computer program designed to process digital sound, that runs on more generalized hardware, such as a PC or Mac.  Note that digital sound is just and arrangement of numbers, and that numbers are good for math.  To simulate the sound of a tube, a DSP would first analyze the waveform, second create the even order harmonics that a tube would, and third, add them to the original signal.  And that is how solid state components can create even order harmonics.  Programmaticly, not electrically.
 
Please note that no magic transistors which produce even, instead of odd order harmonics are used.  Also note that solid state components, such as opamps and passive crossovers, perform analog signal processing, and that it is possible for vacuum tube based digital computers to perform digital signal processing.  Such a machine could also programaticly produce odd order harmonics if you were in to that kind of thing.

I have to ask. Have you ever heard an amp that measures .0006% THD? I have yet to hear anyone tell me that kind of amp is their "favorite color" as you put it. I measured more amps as well. The amps that measured .01%+ THD sounded significantly better. Both were solid state amps. Both amps were placed in the same system of neutral toned electronics. My point is just because something measures good doesn't always mean it will sound good. This has been proven time and time again. I know what a dsp is and it's based on transistors, that is its limit. Dsps's don't just exist in computers and they are not only designed to process sound. They exist in receivers, car stereos, also calculators, and even microwaves. A dsp can try to emulate but it will never create true even order distortion unless it contains tubes. I know a lot of people say harmonics but that term is wrongfully used as they are two different things. Also remember there is not just one type of tube, but rather several and each tube has its own signature. Anyway If you wish to argue this further, please give me a link to a published white paper that states otherwise.
 
Again the transparency thing makes no sense. I don't know if it's the term "transparency" you're using or what. Can you provide a link so that I may read up on whatever it is you're talking about? I'm not sure why you provided reference to ENIAC of all things I find this a bit funny. Yes it contained tubes but no dsp of any kind was present. There were no transistors in ENIAC, which is one reason why it was so massive. I suggest you do some further reading on ENIAC in order to have a better understanding before you give reference of it. Also, regarding the whole digital signal thing. I have done some experiments regarding this. Me and a buddy of mine were curious one day so we decided to see if digital signals have any weakness. Our first experiment consisted of having two of the exact same digital setups.
 
Each setup consisted of a PS Audio Lambda transport connected directly to a Sony digital receiver via optical. There were no dacs involved. The only difference between the two setups was that one setup contained a PS Audio power re-generator. The only thing plugged into the power re-generator was the transport. It turns out there was an audible difference. The setup with the power re-generator sounded better. Why was this? Digital signals should not contain noise we thought. We then packed up our equipment and set out to his electronic lab where he works. We analyzed the digital signal and we then were able to confirm, the digital signal is only as clean as the power supply which produces it. It seems you have done some reading on the subject but very little if any experimentation. Anyway this thread has gone way off topic so this will be my last post. If you insist on having only equipment which measures good, then you're missing out on some really good music.   

 

[Fidelity Castro]

Here a very good read wich helped me to understand what jitter is  :
 
http://www.jitter.de/english/engc_navfr.html
 
Here a good read to understand the Sense of oversampling / upsampling and how pcm works  :
 
http://www.mother-of-tone.com/cd.htm
 
I recommend reading the whole page, very interesting approach to Hifi in general
 
 

 

[nick_charles]

Quote:

Here a very good read wich helped me to understand what jitter is  :
 
http://www.jitter.de/english/engc_navfr.html
 
 

Actually it is mostly self-serving (they are trying to sell you something) wafle with a couple of stray facts thrown in, a good read on jitter is: E. Benjamin and B. Gannon, ‘‘Theoretical and audible effects of jitter on digital audio quality, Preprint of the 105th AES Convention (1998). You can get this from the AES and it us a must read for anyone with a serious interest in Jitter. The Jisco people very carefully do not mention that Dunn was re-evaluating his position on jitter audiblity in the light of B and G study, nor do they provide any jitter levels or correlated controlled listening tests.

 

[SP Wild]

Thanks nightrhyme for such an insightful look at these hi-end dacs.  I for one appreciate how the maturity of the old-timers is able to prevent such controversial threads from spilling into immature name calling.

 

[maverickronin]

Quote:

 
I have to ask. Have you ever heard an amp that measures .0006% THD? I have yet to hear anyone tell me that kind of amp is their "favorite color" as you put it. I measured more amps as well. The amps that measured .01%+ THD sounded significantly better. Both were solid state amps. Both amps were placed in the same system of neutral toned electronics. My point is just because something measures good doesn't always mean it will sound good. This has been proven time and time again. I know what a dsp is and it's based on transistors, that is its limit. Dsps's don't just exist in computers and they are not only designed to process sound. They exist in receivers, car stereos, also calculators, and even microwaves. A dsp can try to emulate but it will never create true even order distortion unless it contains tubes. I know a lot of people say harmonics but that term is wrongfully used as they are two different things. Also remember there is not just one type of tube, but rather several and each tube has its own signature. Anyway If you wish to argue this further, please give me a link to a published white paper that states otherwise.
 
Again the transparency thing makes no sense. I don't know if it's the term "transparency" you're using or what. Can you provide a link so that I may read up on whatever it is you're talking about? I'm not sure why you provided reference to ENIAC of all things I find this a bit funny. Yes it contained tubes but no dsp of any kind was present. There were no transistors in ENIAC, which is one reason why it was so massive. I suggest you do some further reading on ENIAC in order to have a better understanding before you give reference of it. Also, regarding the whole digital signal thing. I have done some experiments regarding this. Me and a buddy of mine were curious one day so we decided to see if digital signals have any weakness. Our first experiment consisted of having two of the exact same digital setups.
 
Each setup consisted of a PS Audio Lambda transport connected directly to a Sony digital receiver via optical. There were no dacs involved. The only difference between the two setups was that one setup contained a PS Audio power re-generator. The only thing plugged into the power re-generator was the transport. It turns out there was an audible difference. The setup with the power re-generator sounded better. Why was this? Digital signals should not contain noise we thought. We then packed up our equipment and set out to his electronic lab where he works. We analyzed the digital signal and we then were able to confirm, the digital signal is only as clean as the power supply which produces it. It seems you have done some reading on the subject but very little if any experimentation. Anyway this thread has gone way off topic so this will be my last post. If you insist on having only equipment which measures good, then you're missing out on some really good music.   

I can't say I've listened to an amp with distortion that low, but I can say depending on my mood, type of music, system synergy, etc that sometimes I would prefer something with ultra low noise and distortion, and sometimes I would prefer a thick syrupy tube.  I did not say that one single measurement is how we should judge a piece of audio equipment. I did say that measurements are useful because you will likely never have a chance to hear everything you would consider buying before you actually buy it.  I did say that although it would be possible to objectively measure and rank audio equipment according to both preferences  and absolute standards, no one has devised such a system, and no one likely ever will, because the audiophile market is so small.  I also just realized that the fact that that many audiophiles are violently opposed to measurements of any kind will also be an impediment.
 
You are correct that transparency is is not a technical term.  I would say it is more of a Platonic ideal and would define it as faithfulness to the original file.  (We were talking about DACs here so I'm disregarding vinyl, and yes the tracks on your CDs,and SACDs are files.)  It is reproducing the waveform exactly as it is stored in the file, without any coloration or alteration from the equipment used to reproduce it.  Obviously, as you have stated before, this may not always be desirable for the whole system.  I am not advocating perfect transparency for the whole system however.  Only for the DAC, which was the original topic of this thread.  My reasoning is that since there are already enough variables in the audio chain, and even relatively cheap DACs can achieve near perfect transparency, that you should simplify things and remove the extra variable.  Amps (especially tube amps) and your output transducers color the sound enough already, and EQ's and DSPs can replicate anything a DSP might add or subtract, and are easily changeable.  It may not always be easy to replicate a specific sound, so I do not say it is wrong, or bad, to use such a 'colored' DAC, but because you can tweak free and cheap DSPs to your heart's content it is a better option for most people who have more time than money.  In fact with a sufficiently advanced DSP, an ultra low distortion amp and phones could be desirable because a sufficiently advanced DSP could be tweaked to allow any type of sound, without the amp or 'phones interfering.  Note that such a DSP probably doesn't even exist, but is theoretically possible.
 
Now on to ENIAC.  I mention it because it is a digital computer based on vacuum tubes.  A DSP is also a digital computer, but most today are based on transistors.  Because they are both programmable digital computers they are capable of similar things.  They manipulate data.  Your digital music is data.  If I felt like it could calculate even order tube distortion on a digital computer based on wooden beads.  I could do it by hand.  One could make an electrical or mechanical analog computer to do it.  Obviously these would not be in real time, but if you felt like punching the data back into a computer you could do it.  DSPs do not generate effects by happenstance of their electrical properties, they do it by manipulating data.
 
As for your test, I'll assume you mean no external DAC since that bitstream in a beam of light has to turn into an analog waveform somehow.  Assuming you're not fooling yourself about the audible difference, the only way you could notice a difference was if the receiver was being powered by your line filter.  Assuming that your Sony receiver has a substandard power supply and can  therefore benefit from the cleaner power you may hear a difference.  Note that this is the analog amplifier benefiting from the cleaner power.  Not the DAC or the transport.  To properly control this experiment you need to switch only the transport between normal power and the line filter.
 
The reason I find your claim unlikely and assume your experiment is flawed, is because digital equipment does not, or cannot have noise for some magical reason, but because it absolutely rejects all noise below a certain threshold.  A signal is either a one or a zero.  On or off.  High or low.  Bright or dim.  There is almost no in between.  If the amplitude of the noise is not enough to turn a zero into a one then the noise is rejected.  So even though you measured lower noise with the fancy line filer, it doesn't matter!  If the sound is not very obviously distorted, breaking up, dropping out, or just not there, then the noise is being completely rejected.
 
How would modern computers work otherwise?  Your computer, from which you type your response is probably running from what is considered to be a low quality power supply with a processor running at least 10,000x audio frequency, and relies on every single bit being perfect so it doesn't crash.  It deals with many clock generators running at many different frequencies all putting out their own RFI and trying to talk over each other, yet it still works.  If digital signals were as susceptible to noise as you claim, then the average computer would hang 5 clocks after you pressed the power button.

 

[IPodPJ]

I have to subjectively disagree with part of your above statement.  The biggest improvement to my digital front end was gained when I put my best power cord (Virtual Dynamics Master LE 2.0) on my Logitech Transporter.  I am using a stock cord on my DAC because I prefer it with this piece of gear.  Transparency increased substantially when the VD cord is used, compared to stock or other high-end cords.

 

[Fidelity Castro]

 
Quote:

Actually it is mostly self-serving (they are trying to sell you something) wafle with a couple of stray facts thrown in, a good read on jitter is: E. Benjamin and B. Gannon, ‘‘Theoretical and audible effects of jitter on digital audio quality, Preprint of the 105th AES Convention (1998). You can get this from the AES and it us a must read for anyone with a serious interest in Jitter. The Jisco people very carefully do not mention that Dunn was re-evaluating his position on jitter audiblity in the light of B and G study, nor do they provide any jitter levels or correlated controlled listening tests.

I understand your point of view, the whole hifi live would be easier if jitter would be inaudible. From my experience i can tell you it´s not. The big question is, if it´s not the jitter wich counts, what is it then ? Placebo is not Placebo if 2 people hear the same difference at the same time.
 
Blindtesting is ok and i agree with people who after failing in such test sell their whole highend audio eqipment and stick with standard consumer level gear or even a dvd player for a few bucks from walmart. But i would only trust my ears and the impression i get in my home, in my hifi chain and relaxed listening. It´s not 100% sure how the human brain works or even your listening when you have to prove something like in a blindtest. I don´t think that there´s a conspiracy going on regarding jitter to rip off people, i asure you that the real conspiracy´s wich exists are far more dangerous

 

[haloxt]

Actually it is certainly possible for two people to fall for placebo, it just means two people are affected by placebo . Relaxed listening is important, because when nervous you can induce all sorts of faulty sensations or interpretations, but we must keep in mind so long as your experiences stay in the subjective realm with no attempt at externalizing your perspective, it will never be accepted as fact in objective science.
 
And I agree with you on cmp, it's nice. But still, you must be very picky about computer parts, even then, they may not be anywhere near ideal like the components found in a hifi cd transport.

 

[Prog Rock Man]

If the PC is just a means of storing a music file or streaming the same off the internet and then transmitting it to an external DAC, how can it really effect the sound? Surely that is all going to happen in the DAC, amp and headphones/speakers? Even if a PC could influence sound quality, its role will be far less than the other components.

 

[haloxt]

Most pc's are a noisy environment, if you want a brief look at what could possibly go wrong look at this list of components recommended by cics, author of cmp^2 http://cicsmemoryplayer.com/index.php?n=CMP.05Components