I never said they did. What I said was that the others don't either. Not ones that matter at all, and that are in no way different to an additional 8mm of air distance from one speaker. Care to answer the questions above, since you insist on asserting that the pure one sample delay is so deleterious, and in some crucial way different in nature to 8mm of air?
Earlier in this thread, ALP observed that the interchannel delay, which is characteristic of Audio Note-style DAC implementations, was audible and deleterious. For that heresy he was set upon by a band of diyAudio experts who informed him that the delay was one sample time, 22.7 us, and was equivalent to a free air, sound-travel distance of about 8mm.
As usual, the diyAudio experts are wrong. Apparently, a basic understanding of digital circuits and acoustics is beyond them. First, the delay is ONE-HALF of a sample period, 11.3 us. Second, a pure time delay in one channel is not the same thing as moving one loudspeaker further away. While moving one loudspeaker further away will delay the arrival time of the direct sound wave relative to the other speaker, it will also change the relative amplitudes of the direct sound waves and the interaction of the reflected sound waves with the room boundaries and each other. The entire acoustic field will be different from the case of a pure time delay in one channel. I should think any audiophile with an educated brain and experienced ears could understand and hear the difference.
As usual, the diyAudio experts are wrong. Apparently, a basic understanding of digital circuits and acoustics is beyond them. First, the delay is ONE-HALF of a sample period, 11.3 us. Second, a pure time delay in one channel is not the same thing as moving one loudspeaker further away. While moving one loudspeaker further away will delay the arrival time of the direct sound wave relative to the other speaker, it will also change the relative amplitudes of the direct sound waves and the interaction of the reflected sound waves with the room boundaries and each other. The entire acoustic field will be different from the case of a pure time delay in one channel. I should think any audiophile with an educated brain and experienced ears could understand and hear the difference.
Konnichiwa,
Ahhh, yes darn it, you are right, the problem is halve as big as we thought. My bad for not spotting this.
That is correct, in effect moving a speaker by 4mm will have a MUCH GREATER sonic impact than the 11.3uS interchannel delay.
My points remains, that whoever does not reliably position his speakers and head precisely to much tighter tolerance than 4mm will find hearing the interchannel timedelay considerably difficult.
Let us be clear, you suggest that moving your head by around 4mm has a definitively audible and negative impact?
Sayonara
Ulas said:
Ahhh, yes darn it, you are right, the problem is halve as big as we thought. My bad for not spotting this.
Ulas said:
That is correct, in effect moving a speaker by 4mm will have a MUCH GREATER sonic impact than the 11.3uS interchannel delay.
My points remains, that whoever does not reliably position his speakers and head precisely to much tighter tolerance than 4mm will find hearing the interchannel timedelay considerably difficult.
Ulas said:
Let us be clear, you suggest that moving your head by around 4mm has a definitively audible and negative impact?
Sayonara
Francis_Vaughan said:
Francis, - makes alot of sense to me.
Some years ago we tried squaring up a 1Vp-p sinewave master clock for a DAC and transport link as most people say you need a clock with fast clean edges to reduce jitter. We tried no end of squaring circuits, comparators, analogue amps, gates etc. In the end I thought the DAC chip input squares it anyway so why put another stage in front of it. We biased the waveform around the DAC circuits input switching point using a resistive divider decoupled from the power supply rails. It proved to be clearly the most sonically pure configuration.
The other point about there being so many other areas of compromise that are so much more significant in that other commercial design mentioned also supports our experience.
rgds, Craig
Kuei Yang Wang said:
I guess you still don’t get it. As I explained, moving one speaker does not even approximate interchannel delay. Moving one speaker changes the way the entire sound field is created not just the arrival time of the sound from one channel. Most of the audiophiles I know, when fine-tuning an audio system, move the speakers in increments as small as 1 mm. Are you saying that repositioning one speaker by 4 or, in your case, 8mm is inaudible?
Of course repositioning my head by 4mm is audible. When I move my head I am not just changing the arrival time of the direct sound waves but I am changing my location in the entire acoustic field. While one head position may or may not be preferred to another position that is just 4mm away, the two positions each present different aural pictures of the acoustic space I am in. It is those tiny changes that let me track the location of different sound sources and track my location as I move around in the aural landscape.
You and the other diyAudio experts claim to be able to hear minute amounts of jitter. Guido Tent says he can hear 1ps of jitter. In one picosecond, sound travels about 1/3 nanometer. Using your speaker/head position analogy, 1ps of jitter is the same as randomly moving your hear back and forth thousands of times per second an average distance of 1/3nm. To put that into perspective, the wavelength of visible light is 400-700nm.
Konnichiwa,
I DO get it. However in testing myself I found the interchannel delay inaudible, but the added circuitry to correct for it VERY audible.
No, it does not approximate it, it goes considerably BEYOND the simple and less damaging effects of interchannel delay.
I challenge anyone to demonstrate the ability to psotion even modest size speakers (never mind serious behemoths) and himself with an accuracy of better than 4mm and to see this positioning accuracy retained for any reasonable amount of time.
Not neccesarily, however I find that a placement accuracy of around 1cm (10mm) is sufficient to minimise any issues, plus I find it rather difficult to attain any realistic precision to higher degrees for positioning speakers, as for positioning myself, I'd guess at best 10cm accuracy with any reliability.
And hence you keep it in a vice that makes sure that you do not move at all, or if by less than 4mm?
I do not claim to be able to hear any specific amounts of jitter, past that, the differences between a static displacement and a dynamic modulation are quite drastic and obvious.
Yup, but it is the frequency of the movement that matters here, not the amount.
I suggest to all to try for themselves and make their own jugement. I found the simple inverter (ideally actually a pair of very fast switching transistors set up to keep the current flow constant and not a logic gate) subjectively preferable to "stopped clock" and "shift register" aligning channels. As usual, YMMV.
Sayonara
Ulas said:
I DO get it. However in testing myself I found the interchannel delay inaudible, but the added circuitry to correct for it VERY audible.
Ulas said:
No, it does not approximate it, it goes considerably BEYOND the simple and less damaging effects of interchannel delay.
Ulas said:
I challenge anyone to demonstrate the ability to psotion even modest size speakers (never mind serious behemoths) and himself with an accuracy of better than 4mm and to see this positioning accuracy retained for any reasonable amount of time.
Ulas said:
Not neccesarily, however I find that a placement accuracy of around 1cm (10mm) is sufficient to minimise any issues, plus I find it rather difficult to attain any realistic precision to higher degrees for positioning speakers, as for positioning myself, I'd guess at best 10cm accuracy with any reliability.
Ulas said:
And hence you keep it in a vice that makes sure that you do not move at all, or if by less than 4mm?
Ulas said:
I do not claim to be able to hear any specific amounts of jitter, past that, the differences between a static displacement and a dynamic modulation are quite drastic and obvious.
Ulas said:
Yup, but it is the frequency of the movement that matters here, not the amount.
I suggest to all to try for themselves and make their own jugement. I found the simple inverter (ideally actually a pair of very fast switching transistors set up to keep the current flow constant and not a logic gate) subjectively preferable to "stopped clock" and "shift register" aligning channels. As usual, YMMV.
Sayonara
First you say interchannel delay is inaudible. Then, at the end of your post, you say using a pair of fast transistors as an inverter is preferable to a logic gate when aligning the channels. If interchannel delay is inaudible, why would you want to use anything to align the channels? And how does an inverter, be it a pair of fast transistors or a logic gate, align the channels?
If you found adding a shift register damaging to the sound then you did something wrong. Either you did not observe proper digital design technique or you miscalculated the length of the shift register needed to compensate. It seems to be a common mistake among the diyAudio experts. The shift register should be 32-bits long because that’s the number of SCLKs in ONE-HALF of a sample time.
Earlier in this thread you were browbeating ALP with the claim the two were equivalent. Don’t you remember?
What made you change your tune?
Most speakers, especially behemoths, stay were they are put and accurate positioning can be easily retained until the speakers are moved.
Clock jitter is the random deviation in the timing of a clock edge and it is measured by the magnitude of that deviation. Random events do not have a frequency component. When a clock is influenced by a non-random, outside source, it is called frequency modulation or correlated phase noise; but it is not jitter.
OK, a delay is not jitter. Pure and simple. Jitter is changing time interval, a delay is a constant change in time. There is no equivalent to jitter induced distortion products from a constant delay. Maybe people should read Hawksford on the jitter equivalence of distortion.
If you place a constant delay in one channel the sound field is rotated about the axis normal to the plane that the speakers and listener define. If we take a reasonably typical domestic setting, speakers 1.5 metres apart, the rotation is arctan(4mm/750mm) which is 0.3 degrees. That is it.
Moving one of the speakers by 4mm will result in exactly the same rotation of the sound field. Of course, this may also result in a change in the interaction of the speaker with the room, so moving the speaker may result in other changes, but the rotation of the sound field will be the basic result.
Some idea that the delay is equivalent to a rapid perturbation of the location of the speaker driver results from a trivial mistake in understanding. Jitter is the equivalent of such a changing perturbation - but a delay isn't jitter.
The jitter resultant from a pure delay is exactly zero. There is no resultant creation of distortion products. Moreover other static timing errors are also benign. An asymmetry in the the sample clock, so long as it remains static, results in a constant gain change - i.e. the same as a microscopic change in volume, or in a DC offset.
Another thought experiment. Draw a waveform on a piece of paper. Now delay it. The most trivial way of doing that is to move the paper to the right a few mm. Doesn't seem to be much changed does it? Does the moment you hit play on your CD player matter? That clearly results in an arbitrary delay.
If you place a constant delay in one channel the sound field is rotated about the axis normal to the plane that the speakers and listener define. If we take a reasonably typical domestic setting, speakers 1.5 metres apart, the rotation is arctan(4mm/750mm) which is 0.3 degrees. That is it.
Moving one of the speakers by 4mm will result in exactly the same rotation of the sound field. Of course, this may also result in a change in the interaction of the speaker with the room, so moving the speaker may result in other changes, but the rotation of the sound field will be the basic result.
Some idea that the delay is equivalent to a rapid perturbation of the location of the speaker driver results from a trivial mistake in understanding. Jitter is the equivalent of such a changing perturbation - but a delay isn't jitter.
The jitter resultant from a pure delay is exactly zero. There is no resultant creation of distortion products. Moreover other static timing errors are also benign. An asymmetry in the the sample clock, so long as it remains static, results in a constant gain change - i.e. the same as a microscopic change in volume, or in a DC offset.
Another thought experiment. Draw a waveform on a piece of paper. Now delay it. The most trivial way of doing that is to move the paper to the right a few mm. Doesn't seem to be much changed does it? Does the moment you hit play on your CD player matter? That clearly results in an arbitrary delay.
Konnichiwa,
My, my. Back to EE101 methinks.
We where in this thread discussing the relative merits and dismerits of simply in verting the wordclock to one DAC for a non-oversampling DAC using two single channel DAC Chips (or two channel DAC Chips with seperate wordclock per channel) which causes a minimal interchannel delay vs. using a variety of solutions to make both channels latch at the time.
I found the best sounding way to for the inverted wordclock was a discrete 1-Transistor (fast switching type) inverter with a second "drone" transistor to make sure that the supply current drawn both static and dynamic remains the same, no matter of the state of the inverter. This replaces the common logic gate.
OF COURSE this does not align the channels. So I am NOT TRYING to align the channels, I am merely trying to get a good inverter.
I that clear? Thank you.
Neither, I did not spend my school fee on drinks, I did pay attention in class.
Compared to the single inverter solution both stopped clock and shift registers impacted negatively on the sound. Most likely due to supply/ground contamination which even the most competent digital design and circuit layout can at best minimise.
Nope, I don't remeber. Mainly because I made no such claims. I did note that UNTILL the placement of speakers and head was made reliably to a much tighter tolerance than 8mm the presence of a 22.7uS delay (which of course is only 11.35uS anyway) between channels would be inaudible (as it it would be swamped by much larger delays in the physical setup).
There is no claim of direct equivalence here, merely a pointing out of conditions to be fulfilled before we even need to consider audibility.
I could also present charts that relate interchannel delay to localisation of sound sources vs. amplitude differences inbetween channels and you would find that unless the speakers (and amplifiers) are also matched to a fairly small fraction of a decibel the presence of a 11.35uS static interchannel delay would be swamped out as well.
Note that there is still no claim of equivalence, merely one of precedence.
It was the next on the LP, why?
As for 8mm vs. 4mm, my bad for simply re-peating numbers others had given and which did appear reasonable without actually checking.
Hmm, have you ever made measurements by exactly what amounts of distance building subsections shift with differing ground moisture and temperature? You will find it easily several mm between summer & winter, no matter what material was used and equally, the displacement is not neccesarily uniform.
Accurate positioning within the 1mm range is invariably and unavoidably an illusion and cannot be maintained for any length of time, if achieved at all.
It is most of the time actually NOT RANDOM, which is is what makes it sonically damaging.
Well, your interpretation does not semm to be shared by the majority of the engineering community. From:
http://www.maxim-ic.com/glossary/index.cfm/Ac/V/ID/58/Tm/CLOCK_JITTER/ln/en
"Glossary Term: Clock Jitter
Definition
A periodic waveform (especially a clock) is expected to cross certain thresholds at precisely timed moments. Variations from this ideal are called jitter."
As you can see, ANY VARIATION, without any qualification of it being deterministic or stochastic or a mix is classed as jitter, so phasenoise IS clock-jitter.
Please do me a favour, read what I actually write and consider making sure you actually can substantiate any claims made as factual. I get edgy if people make utterly ridiculous claims without providing any backup (eg Speaker positioning precision).
Sayonara
Ulas said:
My, my. Back to EE101 methinks.
We where in this thread discussing the relative merits and dismerits of simply in verting the wordclock to one DAC for a non-oversampling DAC using two single channel DAC Chips (or two channel DAC Chips with seperate wordclock per channel) which causes a minimal interchannel delay vs. using a variety of solutions to make both channels latch at the time.
I found the best sounding way to for the inverted wordclock was a discrete 1-Transistor (fast switching type) inverter with a second "drone" transistor to make sure that the supply current drawn both static and dynamic remains the same, no matter of the state of the inverter. This replaces the common logic gate.
OF COURSE this does not align the channels. So I am NOT TRYING to align the channels, I am merely trying to get a good inverter.
I that clear? Thank you.
Ulas said:
Neither, I did not spend my school fee on drinks, I did pay attention in class.
Compared to the single inverter solution both stopped clock and shift registers impacted negatively on the sound. Most likely due to supply/ground contamination which even the most competent digital design and circuit layout can at best minimise.
Ulas said:
Nope, I don't remeber. Mainly because I made no such claims. I did note that UNTILL the placement of speakers and head was made reliably to a much tighter tolerance than 8mm the presence of a 22.7uS delay (which of course is only 11.35uS anyway) between channels would be inaudible (as it it would be swamped by much larger delays in the physical setup).
There is no claim of direct equivalence here, merely a pointing out of conditions to be fulfilled before we even need to consider audibility.
I could also present charts that relate interchannel delay to localisation of sound sources vs. amplitude differences inbetween channels and you would find that unless the speakers (and amplifiers) are also matched to a fairly small fraction of a decibel the presence of a 11.35uS static interchannel delay would be swamped out as well.
Note that there is still no claim of equivalence, merely one of precedence.
Ulas said:
It was the next on the LP, why?
As for 8mm vs. 4mm, my bad for simply re-peating numbers others had given and which did appear reasonable without actually checking.
Ulas said:
Hmm, have you ever made measurements by exactly what amounts of distance building subsections shift with differing ground moisture and temperature? You will find it easily several mm between summer & winter, no matter what material was used and equally, the displacement is not neccesarily uniform.
Ulas said:
Accurate positioning within the 1mm range is invariably and unavoidably an illusion and cannot be maintained for any length of time, if achieved at all.
Ulas said:
It is most of the time actually NOT RANDOM, which is is what makes it sonically damaging.
Ulas said:
Well, your interpretation does not semm to be shared by the majority of the engineering community. From:
http://www.maxim-ic.com/glossary/index.cfm/Ac/V/ID/58/Tm/CLOCK_JITTER/ln/en
"Glossary Term: Clock Jitter
Definition
A periodic waveform (especially a clock) is expected to cross certain thresholds at precisely timed moments. Variations from this ideal are called jitter."
As you can see, ANY VARIATION, without any qualification of it being deterministic or stochastic or a mix is classed as jitter, so phasenoise IS clock-jitter.
Please do me a favour, read what I actually write and consider making sure you actually can substantiate any claims made as factual. I get edgy if people make utterly ridiculous claims without providing any backup (eg Speaker positioning precision).
Sayonara
Kuei Yang Wang said:
The PCM56, if MJ is anything to go by, is indeed very common in Japan but the TDA based nos dacs, commercially at any rate, outnumber the AD1865 ones, but I digress. The point is that removing the delay is not inherently a bad thing or as deleterious as portrayed not to mention the fact the stopped clock method has considerably less data jitter than any other method.
Hi,
Doubtful.
AN has used Non-OS AD1865 in their CD Players and DAC's since the late 90's and they put quite a bit of volume, especially of the cheaper stuff.
I can only observe that to my ears and in my circuits both approaches (shift register & stopped clock) made the sound worse, not hugely so, but audibly and that I note no negative impact from the interchannel delay.
So perhaps my point is that NOT removing the delay is not inherently a bad thing or as deleterious as portrayed by some.
So, if it bothers you philosophically that there is a delay, remove it, if it doesn't, then don't. Audible it is not, under normal conditions.
Sayonara
rfbrw said:
Doubtful.
AN has used Non-OS AD1865 in their CD Players and DAC's since the late 90's and they put quite a bit of volume, especially of the cheaper stuff.
rfbrw said:
I can only observe that to my ears and in my circuits both approaches (shift register & stopped clock) made the sound worse, not hugely so, but audibly and that I note no negative impact from the interchannel delay.
So perhaps my point is that NOT removing the delay is not inherently a bad thing or as deleterious as portrayed by some.
So, if it bothers you philosophically that there is a delay, remove it, if it doesn't, then don't. Audible it is not, under normal conditions.
Sayonara
Re: little bit off topic
Konnichiwa,
The Voltage, AFAIK. However you may find the PCM56 to do fine even with a higher I/V conversion resistors, especially with more than +/-5V supplies, worth trying, you have the gear to measure the results.
Sayonara
Konnichiwa,
Bernhard said:The PCM56 ( to my knowledge the best DAC in the world) may operate with I/V resistors up to 1 kohm.
Is I/V resistance the limit, or the resulting voltage on the Iout of the chip ?
The Voltage, AFAIK. However you may find the PCM56 to do fine even with a higher I/V conversion resistors, especially with more than +/-5V supplies, worth trying, you have the gear to measure the results.
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.