Propeller Head Plaza

Technical and scientific discussion of amps, cables and other topics.

Return to Propeller Head Plaza

What is missing in measurements, so there's lack of correlation between them and sound quality?

129.33.19.254
Posted on October 27, 2009 at 09:49:33
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 I know it was discussed, if not beaten to death, before, but this post is prompted by recent exchange on PC Audio forum about Benchmark vs. Wavelength DACs, used with computer via USB. What measurements are there, show superiority of Benchmark in every category (which prompts "objectivists" to talk about "engineered colorations" and "deliberate colorations as design goal"). However, posters who actually listened to both, have no doubt which one sounds better - and that ain't Benchmark. Almost the same goes for Benchmark vs. Ayre QB-9 - comparable measurements, superiority of Ayre in sound quality.

Links to measurements:

http://www.stereophile.com/digitalprocessors/108bench/index4.html

http://stereophile.com/digitalprocessors/wavelength_cosecant_v3_usb_digitalanalog_converter/index5.html

From my own experience, Benchmark (silver-faced, so one of the latest generations, if not the latest) via AES/EBU was somewhat harsh in highs and forward sounding, easily outclassed by several solid-state DACs I tried in the same and higher price range - so preference for "tube" sound (Wavelength) is not the reason.


So, what gives?
Did you hear the Benchmark at it's best?, posted on October 28, 2009 at 08:21:58
Don Till
Audiophile

Posts: 675
Location: Tempe
Joined: February 16, 2009
 Did you spend time finding the best interconnects and PC or did you use stuff that was selected for some other DAC?

I don't know these things. Did you consider them?

Is it possible that what you heard through the Benchmark was actually the recording, which was "tamed" by other less revealling components?

"From my own experience, Benchmark (silver-faced, so one of the latest generations, if not the latest) via AES/EBU was somewhat harsh in highs and forward sounding, easily outclassed by several solid-state DACs I tried in the same and higher price range - so preference for "tube" sound (Wavelength) is not the reason."

Though you listened to these players the question is did you do what it takes to get the best from them before you reached your conclusions?

Your preference for "tube" sound comment is kind of ridiculus to say the least. Do people still take that comment seriously?

" What measurements are there, show superiority of Benchmark in every category (which prompts "objectivists" to talk about "engineered colorations" and "deliberate colorations as design goal"). "

How a designer choses his compromises may be characterized different ways but regardless of whatever criticism one should listen prior to considering their relative importance.

Often people will mention that any coloration detracts from the ideal. In my opinion this is a fair as well as an inarguable point. However how often do we reach such an ideal and isn't worth considering that some loss of performance is preferable given the ideal in order to better benefit the vast majority of recordings available?

I think the answer is simply yes. My position isn't based so much on "performance" as much as it is on the ability to enjoy a wide diversity of recordings regardless of genre as well as recording style. But there is no doubt increased performance plays a role in this diversity as well and colorations/compromises can only be such they allow for an expanding diversity.

My guess is that the majority, but not a vast majority, of hi end audio equipment is designed by ear and not by measurements.

Too Transport Sensitive.........., posted on October 28, 2009 at 15:47:45
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 At best, it sounds decent in the short term. The problem with this particular DAC is it uses asynchronous sample-rate conversion (24/110 kHz upsampling, a little different from 24/96 or 24/192), which I find very fatiguing to listen to long-term. There is a "mist" of HF noise that initially seems to be part of the music ("added detail"), then it "rides" on the music, then it ultimately *dominates* the music.


RE: Did you hear the Benchmark at it's best?, posted on October 28, 2009 at 15:09:14
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 yes, its nowhere nearly as good as the Monarchy M24 DAC for about the same price. Not even close, actually.

I suspect it's a modern day Adcom GFP 565., posted on October 29, 2009 at 04:56:27
bjh
Audiophile

Posts: 11876
Location: Toronto, Canada
Joined: November 22, 2003
 SF has a poor record when it comes to giant slayers.
Everything matters, don't forget to tweak your placebos!
RE: Did you hear the Benchmark at it's best?, posted on October 28, 2009 at 12:03:31
kerr
Audiophile

Posts: 3695
Location: Central Indiana
Joined: November 10, 2003
 >My guess is that the majority, but not a vast majority, of hi end audio equipment is designed by ear and not by measurements. <

I'm going to go with "designed by measurements, fine tuned by ear".

One of the best sounding integrated amplifiers I ever experienced (within a reasonable budget) gave the user the ability to change the amount of negative feedback. The best measured setting gave the worst sonic performance and the worst measured setting gave me the most glorious sound imaginable. The whole concept of giving the user the choice was because the original, perfect measuring, design didn't sell because it sounded like crap. :)
Don't forget supports (isolation footers, etc.) Then again, posted on October 28, 2009 at 11:06:31
bjh
Audiophile

Posts: 11876
Location: Toronto, Canada
Joined: November 22, 2003
 perhaps you're not quite ready for another epiphany.

:)
Everything matters, don't forget to tweak your placebos!
RE: What is missing in measurements? In this case, not much, posted on October 27, 2009 at 22:48:50
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "Almost the same goes for Benchmark vs. Ayre QB-9 - comparable measurement"

err...no. the high FR measurements are not comparable at all.

http://stereophile.com/digitalprocessors/ayre_acoustics_qb-9_usb_dac/index6.html

http://www.stereophile.com/digitalprocessors/108bench/index4.html

look at those measurements again, even in measure mode, the QB-9 is 1.7dB down at 20kHz in comparison to the Benchmark's 0.2dB. In Listen mode the rolloff starts earlier and looks likes its greater than -3.5dB @20KHz, that's a dramatic difference. Finally, in fairness to Ayre, the measured performance of it's USB input is very good.

Music making the painting, recording it the photograph
So... any idea why both of Wavelength and Ayre sound better,..., posted on October 28, 2009 at 08:30:33
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... despite "inferior" measurements? That's the topic of this thread, as you hopefully noticed.

Todd linked very good article, which actually makes a lot of sense (I'm in the beginning now).

RE: So... any idea why both of Wavelength and Ayre sound better,..., posted on November 2, 2009 at 00:15:11
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 The Benchmark uses asynchronous sample-rate conversion, the Ayre does not........ I don't think the Wavelength does either. That IMO would likely be the main "technical" reason why the Ayre and Wavelength would sound better. The Ayre also uses minimum phase filtering, which IMO widens the advantage.

Sometimes how a product works from a qualitative perspective is just as important, if not more-important, than mere measurements alone.


RE: So... any idea why both of Wavelength and Ayre sound better,..., posted on November 2, 2009 at 09:47:56
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "The Benchmark uses asynchronous sample-rate conversion, the Ayre does not........ I don't think the Wavelength does either. That IMO would likely be the main "technical" reason why the Ayre and Wavelength would sound better. The Ayre also uses minimum phase filtering, which IMO widens the advantage."

And in true audiophile style, the most significant measureable difference up to 3.75dB @20dB in respect of the Ayre is conveniently ignored. FTR, Wolfson implemented a Linear phase as well minimum phase filter with just -0.18dB @20kHz. Therefore there is lot more going on with Wavelength and Ayre DACs, the MP filter does not account for the severe roll-off @20Hz measured on both DACs.

"Sometimes how a product works from a qualitative perspective is just as important, if not more-important, than mere measurements alone."

qualitative accounts are of limited value if adequate controls are not in place.


Music making the painting, recording it the photograph
Oh boy... It's that rare case when I actually agree with Don., posted on November 2, 2009 at 10:05:35
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 From Don Till's post below:

"Indeed, and as we move on it's assumed we are all in agreement with the conclusion as forwarded - ie. that the Benchmark does indeed sound worse than the others.

Truth of the matter, for the conversation to move on with any relevance whatsoever, we should be in agreement on this point."


It seems you're inherently incapable of staying on topic of discussion...

You missed my point - I'm not in agreement with the original assumption., posted on November 2, 2009 at 16:23:52
Don Till
Audiophile

Posts: 675
Location: Tempe
Joined: February 16, 2009
 And even if I was, and I might not be surprised if I actually would be, the most I would conclude it that the most accurate is not always the most preferable.

But that fact I came to that exact conclusion decades ago.

What's objectionable is attempting to find objective evidence why? As if one really expects science to back up their own subjective preferences? That's so arrogant and self-serving it's repulsive.
I thought I made it clear what exactly I agree with, quoting your 2 sentences., posted on November 3, 2009 at 08:37:40
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 If it's of any solace - the rest of it, or any of your follow-ups, didn't make sense to me whatsoever, as usual is the case with your posts.

RE: You missed my point - I'm not in agreement with the original assumption., posted on November 2, 2009 at 22:08:49
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 "And even if I was, and I might not be surprised if I actually would be, the most I would conclude it that the most accurate is not always the most preferable."

"Accuracy" should be associated with a standard or benchmark (no pun intended). Otherwise, "accuracy" will have different meanings to different people.

(There is a standard for color accuracy of still picture cameras called Imatest. People use this to determine the color accuracy of a camera under review. The deviations from accuracy are ultimately displayed on a chart.)



I no longer use the term "accurate" in an audio assessment..... I might *think* it's accurate, but that does not mean it would perform well referenced to an established standard, if it were to exist.


RE: Accuracy, posted on November 3, 2009 at 10:43:50
rick_m
Audiophile

Posts: 2529
Location: Oregon
Joined: August 11, 2005
 I love the chart, it's really colorful...

And color is only a small part of the picture, pun intended. How about MTF, pincushion and barrel distortions, off-axis glare resistance, speed, noise, handling, size, weight, cost, auto focus speed and error, flexibility vs ease of use? And that's just a camera, how about the mapping of the gamut through your computer display error and service's printer?

But when push comes to shove, who cares? It's just like stereos, is accuracy in the look (or sound) or is accuracy in the emotional impact? If the latter then you're almost always better off with black and white prints.

Rick

ASRC not necessarily bad., posted on November 2, 2009 at 07:47:48
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "That IMO would likely be the main "technical" reason why the Ayre and Wavelength would sound better."

Sorry Todd. While your old post may be applicable to some implementations of ASRC, it is not necessarily applicable to all implementations. Perhaps the ASRC in the Benchmark isn't good enough. I don't know, as I haven't heard the product or seen the necessary technical details (except that the output sample rate of 110 kHz seems inadequate on the face). IMO it is inappropriate to reject a design approach qualitatively, if there is a possible good implementation of that approach. However, one needs to carefully select any listening tests and/or measurements based on the technical approach in a product, if one wants to do a good job of evaluation.

I will now address each of your three points in your original post.

1. In particular, the whole point of using ASRC is to allow the use of a fixed output clock, something which can be built at much lower intrinsic jitter than a variable clock as needed for an analog PLL. It would be stupid to use this mechanism and not use a high quality oscillator, except for a cheap implementation where ASRC is used to eliminate the need for separate clocks for the 44.1 x n kHz and 48 x n kHz families of sampling rates.

2. In a proper implementation of ASRC it may be necessary to interpolate the output signal, since the input signal may appear between two output samples. If one does not do this, then there will be, as you suggest, jitter introduced. However, on can interpolate the output sample if one knows the time factor involved (see next paragraph).

Jitter in the input timing is eliminated in a proper ASRC by using a digital phase lock loop. This can be designed to have an extremely narrow lock bandwidth, a minute fraction of one second if desired. The limit is given solely by the long term stability (drift) of the oscillators and the acquisition delay (e.g. when the input source goes from one nominal sampling rate to a new one). There is nothing to preclude (in a sophisticated design) a wider initial bandwidth for rapid lock coupled with an increasingly narrow bandwidth once the input oscillator has been observed to be stable in frequency. A very small FIFO is used (perhaps only a latch) to accommodate the input jitter, and the output comes out at known points. (The ideal jitter free sample times computed by the digital PLL are referenced as fractional values measured according to the output clock which is the stable oscillator.) The resolution of these virtual times can be made arbitrarily large by representing them as binary fractions. The net result is that the uncertainty in timing can be reduced to a tiny fraction of an output clock sample. The known timing between samples is then input to the interpolation algorithm (next step).

3. The final stage of the process is to calculate an output sample. This occurs at integer clock ticks according to the output sampling rate. The input samples appear at constant periods not necessarily related to the output sample times. If one uses a zero order hold for the interpolation, one will get your jitter. But one can use as good interpolation as one wishes. For example the ASRC in the 24 bit SABRE chip uses 1st order interpolation at an output sample rate around 40 MHz. If this isn't good enough, one could use even higher order interpolation.

If one wants, one can built a system where input jitter has no effect on output jitter, measured down to the least significant bit (or even lower if desired) of the output samples. In other words, there would be no discernible jitter on the output of the sample rate conversion, as could be determined by looking at the digital samples. (There would still be jitter on the output analog signal due to the jitter of the output master clock, and if this weren't completely isolated electrically from the processing done by the ASRC function then there could still be coupling introduced this way, but this would be a criticism of the implementation of the ASRC, not its architecture.)

In short, there is no theoretical reason why an ASRC need corrupt the signal in any way shape or form. In practice, of course, that's another matter. Many ASRCs are used primarily as a cheap digital way of eliminating a more expensive crystal oscillator. The new Wavelength option is going to use the 32 bit SABRE chip, which has an ASRC as an (optional) feature. Gordon will have to comment on whether he will be using this feature or not. Presumably he will make this determination based on sound quality, as the cost of an extra oscillator isn't likely to be terribly important at his price point.



Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: ASRC not necessarily bad., posted on November 2, 2009 at 18:04:19
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 "1. In particular, the whole point of using ASRC is to allow the use of a fixed output clock, something which can be built at much lower intrinsic jitter than a variable clock as needed for an analog PLL. It would be stupid to use this mechanism and not use a high quality oscillator, except for a cheap implementation where ASRC is used to eliminate the need for separate clocks for the 44.1 x n kHz and 48 x n kHz families of sampling rates."

But the whole problem is **how** this jitter is "eliminated"....... What happens is the time shifted data is re-sampled by the precise output clock, but because the data being sampled was time-shifted (due to the input jitter), the sampled data isn't at the correct amplitude. (The time shift means the output grabs the amplitude from the signal, interpolated at an extremely high rate, slightly before or after where it should have taken place.) This is where the whole problem is.

"2. In a proper implementation of ASRC it may be necessary to interpolate the output signal,"

It is not only necessary, but it wouldn't otherwise function in this particular implementation...... (I think you meant to say "input signal.")

"since the input signal may appear between two output samples. If one does not do this, then there will be, as you suggest, jitter introduced."

That's not where jitter is introduced, at face value. What happens is if the input signal is time-shifted from jitter, which the intermediate interpolated signal is correlated to, what's sent to the output will not be of correct amplitude, and these amplitude errors, which is a function of the input jitter, introduce noise.

"However, on can interpolate the output sample if one knows the time factor involved (see next paragraph)."

I'm not sure what you mean by "time factor"...........

"Jitter in the input timing is eliminated in a proper ASRC by using a digital phase lock loop."

The PLL supposedly reduces jitter, whether it's ASRC or not. The PLL is not unique to ASRC playback. Many older synchronous conversion DACs use PLL for jitter reduction.

"This can be designed to have an extremely narrow lock bandwidth, a minute fraction of one second if desired. The limit is given solely by the long term stability (drift) of the oscillators and the acquisition delay (e.g. when the input source goes from one nominal sampling rate to a new one). There is nothing to preclude (in a sophisticated design) a wider initial bandwidth for rapid lock coupled with an increasingly narrow bandwidth once the input oscillator has been observed to be stable in frequency. A very small FIFO is used (perhaps only a latch) to accommodate the input jitter, and the output comes out at known points. (The ideal jitter free sample times computed by the digital PLL are referenced as fractional values measured according to the output clock which is the stable oscillator.) The resolution of these virtual times can be made arbitrarily large by representing them as binary fractions. The net result is that the uncertainty in timing can be reduced to a tiny fraction of an output clock sample. The known timing between samples is then input to the interpolation algorithm (next step)."

Useful for those who want an idea how a PLL works.............

"3. The final stage of the process is to calculate an output sample. This occurs at integer clock ticks according to the output sampling rate. The input samples appear at constant periods not necessarily related to the output sample times. If one uses a zero order hold for the interpolation, one will get your jitter. But one can use as good interpolation as one wishes. For example the ASRC in the 24 bit SABRE chip uses 1st order interpolation at an output sample rate around 40 MHz. If this isn't good enough, one could use even higher order interpolation."

What gets lost here is the intermediate interpolated signal is still correlated to the *input*....... (The PLL might reduce it, but it's still there.) Once again, if there is a shift from jitter, what's actually sampled to the output, which is from that interpolated signal, will contain an amplitude error, due to the sample being grabbed slightly before or after it should have been.


RE: ASRC not necessarily bad., posted on November 2, 2009 at 19:49:14
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "What happens is if the input signal is time-shifted from jitter, which the intermediate interpolated signal is correlated to, what's sent to the output will not be of correct amplitude, and these amplitude errors, which is a function of the input jitter, introduce noise."

This noise comes from three sources, all of which can be addressed.

1. Error in estimating the time at which the input signal should have arrived if it were jitter free. (Reduce this error by using a better digital PLL, one that averages over a longer time period and which using more bits of virtual time resolution.)

2. Error in performing the interpolation between output clock points due to inaccurate algorithm. (Reduce this error by using a higher order interpolator or a higher output sample rate or a combination of the two.)

3. Error in calculating the interpolated values. (Use a longer word length in the interpolation calculations.)

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: ASRC not necessarily bad., posted on November 2, 2009 at 21:43:50
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 "1. Error in estimating the time at which the input signal should have arrived if it were jitter free."

Since PLLs have been common prior to ASRC (my Prism DA-2 DAC uses it), if it completely eliminated jitter (it doesn't in the real world), then using ASRC to eliminate something that was *already* eliminated would be pointless. (Unless variable input or output rates were to be utilized, which is the one thing ASRC has over synchronous conversion. A true jitter-free input signal would eliminate the noise issue. But no ideal exists in the real world. And still, the ASRC isn't what would even be actually eliminating the jitter. It would only benefit from the PLL doing so.)

"2. Error in performing the interpolation between output clock points due to inaccurate algorithm. (Reduce this error by using a higher order interpolator or a higher output sample rate or a combination of the two.)"

I'm not sure what you mean by "between output clock points"...... The interpolation is actually performed *prior* to the output clock sending samples to the output. The samples sent to the output are *already* interpolated. (This is why if there's jitter on that interpolated signal, which is correlated to the input signal/PLL, the errors at the output would be that of amplitude/noise.)

"3. Error in calculating the interpolated values. (Use a longer word length in the interpolation calculations.)"

With the sheer speed of the calculations in such high oversample rates utilized in ASRC, this *might* be an issue. But to be honest, I think the jitter in the signal would likely result in larger errors, unless some calculations were corrupted outright. (Such errors would render a DAC unusable.) The output is good enough (in this context) to quench such doubt.

I think the ultra-high oversample rate performed in ASRC in itself is a technological achievement. (The processing power has got to be downright immense.) But since there is no readily explicit information on how these values are actually generated, I'd raise the question in regard to short-cuts possibly being done in order to achieve such high oversample rates.

At the end of the day, I think asynchronous conversion should have never seen the light of day in consumer CD playback. And also consumer CD recordings. I think CDs cut from a high-rez 24/192 master is a similar compromise.


RE: ASRC not necessarily bad., posted on November 3, 2009 at 06:13:51
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 I'm sorry that my explanation was difficult to understand. It's a difficult subject, and no doubt the inherent difficulty has resulted in many designs that don't quite work correctly. For a description of some of the compromises applied in practical designs (which have been reported by audiophiles to be good sounding) look at the white paper for the ESS SABRE chip and the patent filing on ASRC technology by the authors of the paper.

There is no point in arguing the case against using analog PLLs to derive sample clocks. This has been discussed numerous times in numerous threads. If you have one that has low enough jitter that it doesn't bother you, then I'd just enjoy it.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: ASRC not necessarily bad., posted on November 3, 2009 at 10:44:18
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 I listened to the SABRE chip at the L.A. Can Jam several months ago. It was actually one of the better performers there. Although it was with music that I was unfamiliar with, so I couldn't really compare it to anything.


RE: ASRC not necessarily bad., posted on November 3, 2009 at 12:40:06
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 As I understand it, it is possible to bypass the ASRC on the SABRE chip if one runs the chip with a master clock that is an appropriate multiple of the sample rate. That would be up to the DAC, i.e. if it didn't use the built-in ASRC it would probably need to have two different master clocks to accommodate the 44.1 and 48 kHz families of sampling rates.

Do you recall the DAC that was running at the demo?

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Your words, not mine., posted on October 29, 2009 at 06:38:34
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "Almost the same goes for Benchmark vs. Ayre QB-9 - comparable measurements"

Your words, not mine. Anyway moving on

"So... any idea why both of Wavelength and Ayre sound better ... despite "inferior" measurements."

Sounds better is a preference issue. the published independent measurements provides many clues as to why the DACs may sound different. For example, it is certain that a component (Ayre) that measures about -3.8dB @20kHz will sound different from one that -0.2dB @ 20kHz.

Tone controls were once common place on domestic audio equipment for good reason. In addition, the mechanics of typical vinyl playback systems readily permit topend FR adjustment. IMO, this is is just another case of looking for zebras when there is a herd of horses nearby.

Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 1, 2009 at 08:09:56
Don Till
Audiophile

Posts: 675
Location: Tempe
Joined: February 16, 2009
 "Your words, not mine. Anyway moving on"

Indeed, and as we move on it's assumed we are all in agreement with the conclusion as forwarded - ie. that the Benchmark does indeed sound worse than the others.

Truth of the matter, for the conversation to move on with any relevance whatsoever, we should be in agreement on this point.

But alas we move on anyways and even worse than that the assumption seems to be that if there is a preference for equipment with more colorations and less measureable performance there must be some measureable reason why? Well if we all had a common frame of reference there might be some truth to this reasoning, however stereo by it's very nature is ill-defined, and getting such a reference is no small task if outright impossible.

Geez seems reasonable to me to conclude that the best measureable audio performance isn't always the best thing because it reveals the limitations of stereo itself. And beyond that there's an infinite amount of flexibility allowed to the recordist and an infinite number of speaker placements / room combinations.

To think audiophiles haven't figured out the subjectiveness of all of this after 50 years is mind boogling. Of course much of the marketing and advertising that fills our hobbyist magazines (ie. marketing vehicles) appeals to our objective nature (ie. in general our lack of smarts).

The old Exposure distributor coined a term for this kind of psuedo-scientifical kind of thinking - he called it "gooberism" and I think it's a fitting term.
RE: Your words, not mine., posted on November 2, 2009 at 14:17:27
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "Indeed, and as we move on it's assumed we are all in agreement with the conclusion as forwarded - ie. that the Benchmark does indeed sound worse than the others. "

I do not agree, preferring a rolled-off (read equalised response) ala Ayre and the Wavelength is a matter of preference and subjective judgement rather than an objective qualitative assessment. The issue dovetails into a point I have made repeatedly on this board, why do folks presume that their preference would always be for accurate sound? And when they come to the realisation that they prefer their music with a bit of engineered distortion give flimsy reasons for why they think measurements do not tell the whole story? The reasons audiophiles give for preferring various forms of distortion are tortuous at best. Tone controls and equalisers were once ubiquitous for very valid reasons. These tools have largely disappeared yet audiophiles go to ridiculous lengths to achieve what's effectively an equalised response in all but name.


"Truth of the matter, for the conversation to move on with any relevance whatsoever, we should be in agreement on this point."

The conversation has already moved on and folks are already chiming in with their opinions.





Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 2, 2009 at 23:46:42
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 "I do not agree, preferring a rolled-off (read equalised response) ala Ayre and the Wavelength is a matter of preference and subjective judgement rather than an objective qualitative assessment."

I've mentioned why the "rolloff" exists several times. It's not a typical FR rolloff like what exists in band-limited analog products. It's due to the beat/modulation, caused in most part by a sample rate that is too low. Since the signal modulates, peaks are sampled might be close to flat or maybe even lower than what the FR curve would suggest. That curve only suggest the "average" output at HF. But subjectively, the HF does not seem "rolled off", relative to a digital filtered signal that's flat to 20 kHz....... Since the time response is improved, the HF even seems greater in some cases.

When it comes to digital audio playback, the measured "HF rolloff" only tells half the story.


He still won't even attempt to explain how is that rollof responsible..., posted on November 3, 2009 at 08:56:19
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... for those DACs sounding better, nevermind how is it SOLELY responsible for them sounding better.

Awkward attempt at obfuscation, nothing more.

RE: Your words, not mine., posted on November 3, 2009 at 05:14:21
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "I've mentioned why the "rolloff" exists several times. It's not a typical FR rolloff like what exists in band-limited analog products. It's due to the beat/modulation, caused in most part by a sample rate that is too low. Since the signal modulates, peaks are sampled might be close to flat or maybe even lower than what the FR curve would suggest. That curve only suggest the "average" output at HF. But subjectively, the HF does not seem "rolled off", relative to a digital filtered signal that's flat to 20 kHz....... Since the time response is improved, the HF even seems greater in some cases."

Your comments are somewhat long-winded and per Werner post (and many others besides) not particularly correct.

Secondly, per my original post all the filters employed by Ayre has deeper roll-off than the alternative implementations. The LP filter -1.7dB@20Hz is fully 1.5dB off the Wolfson and Benchmark implementation, MP filter is worse as it's fully 3.5dB off the Wolfson implementation, Those are dramatic differences in any sense.

And no time response is not improved with MP filters as it sacrifices phase linearity for amplitude linearity and that makes the deep roll-off even harder to explain. The other factor that's supposedly improved is pre-ringing (in decent implementations) but that's at the expense of increased post-ringing.

Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 3, 2009 at 07:23:15
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 I don't consider myself any kind of expert in filter design, but I do believe that the minimum phase / linear phase tradeoff dimension is independent of the high frequency roll off dimension. The duration of significant impulse response will depend heavily on the steepness of the cut off slope, and if aliasing/imaging is to be avoided this implies there is a tradeoff between impulse response and high frequency amplitude response. The phase response is a separate issue. One can shift the needed transient response to be after the impulse (minimum phase) or centered on the impulse (linear phase). One can even, for amusement, shift the needed transient response to entirely precede the impulse (maximum phase).

One can see these different dimensions with some file based SRCs, such as iZotope 64 bit SRC where both dimensions are exposed separately via the user interface. If one wants to experiment with maximum phase effects one can use minimum phase filters in conjunction with an audio editor that allows time reversal of files.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Your words, not mine., posted on November 3, 2009 at 08:28:14
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 You know a lot more about filter design than me your post address the issues more lucidly than I can manage. I certainly agree that high frequency roll-off is independent of the minimum phase/linear divide which was essentially one of the main points I attempted to put across in the previous post. Thanks for jogging my memory on the trade-off between impulse response and high frequency amplitude response.

The only issue I have is wrt time response which is what I believe Todd mentioned as opposed to transient response. Time response and phase response are obviously related terms and an essential difference between the Linear phase and minimum phase is that one exhibits constant group delay and the other doesn't, therefore by definition the time response of both filter is very different.

Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 2, 2009 at 16:12:31
Don Till
Audiophile

Posts: 675
Location: Tempe
Joined: February 16, 2009
 "I do not agree, preferring a rolled-off (read equalised response) ala Ayre and the Wavelength is a matter of preference and subjective judgement rather than an objective qualitative assessment. The issue dovetails into a point I have made repeatedly on this board, why do folks presume that their preference would always be for accurate sound? And when they come to the realisation that they prefer their music with a bit of engineered distortion give flimsy reasons for why they think measurements do not tell the whole story? The reasons audiophiles give for preferring various forms of distortion are tortuous at best. Tone controls and equalisers were once ubiquitous for very valid reasons. These tools have largely disappeared yet audiophiles go to ridiculous lengths to achieve what's effectively an equalised response in all but name."

They don't seem to understand the hypocrisy of their position. I mean if you got rely on some guys masters thesis to make some point you're beyond a laughing stock. But alas these people do these things as they deny what's are clearly leaps of faith, wishful thinking and hoping against all hope.

I am a minimalist kind of audiophiles, no tone controls, no balance controls and a minimum of frill in the circuitry. I don't think the most accurate equipment sounds the best, though potentially it might in some situations, but I think that's just the obvious nature of this beast. What sounds the best - whether it is the most accurate or not hardly makes a difference and certainly has little to do with what sounds the best in another situation.

"The conversation has already moved on and folks are already chiming in with their opinions."

It's hilarious they are offering objective evidence as proof to support their subjective opinions. Now isn't that really hilarious as these as the same people who are continually bash RBNG - their positions are as incredulous as his but they are completely oblivious to that fact!!!!

RE: Your words, not mine., posted on November 3, 2009 at 02:51:43
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 "I don't think the most accurate equipment sounds the best,"

And yet Keith Howard showed with his distortion adding software (I tried this at home by the way with very similar results to him) that NO added distortion to a music track sounded the best. Even so called "euphonic" additions degraded the sound.

HOWEVER; he also found that even order and progressively decreasing amplitude with increasing order led to LESS degraded sound than distortion packets with mostly odd order and/or high order harmonics (i.e. most commercial SS PP amp designs have patterns like this). So while ZERO distortion was the best, in the real world where ALL amps have some distortion then what is defined often as "euphonic" distortion patterns sound the least annoying. It is not a question of annoying or not annoying but of more or less annoying.

Amplifier distortion is a subtle but long term pleasure affecting tone control. Some amps have enough high order harmonic distortion that they sound "loud" at nearly any volume. Others suffer from sounding "dead" where you always feel the need to turn it up to get some life out of the sound. Many so called "accurate" amplifiers fall into this category. They simply turn music into boredom. It may not be distortion per se, but the techniques used to eliminate that distortion also seem to flatten dynamic contrast in a very audible way. I have not seen a good measurement yet to explain this very audible phenomenon.
RE: Your words, not mine., posted on November 3, 2009 at 06:52:25
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 There is a difference between non-linear distortion, such as Keith Howard was adding, and linear distortion, which amounts to an equalization change. The former is difficult/impossible to undo and the latter is easily corrected by passing through inverse transformations, some of which even happen naturally. One might not find that a rolled off DAC was "inaccurate" if one played hot recordings and/or one had bright speakers and a bright listening room.

The whole question of what it means for a component to be accurate is not meaningful in the subjective realm, because human subjects don't listen to components. They don't even listen to playback systems. They listen to record-playback systems. In the objective realm, one can call a component accurate if it passes all the measurements, but then there remains the question of whether the measurements are complete, relevant and what constitutes a passing grade on the "transparency test".

Some signal modifications, such as equalization changes (amplitude and phase) are relatively simple to understand, whereas others, such as non linear effects and jitter effects, are much more complex. IMO the only way to make progress without seemingly endless wheel spinning is through a well chosen combination of objective and subjective methods.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Your words, not mine., posted on November 3, 2009 at 09:33:51
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 "Some signal modifications, such as equalization changes (amplitude and phase) are relatively simple to understand, whereas others, such as non linear effects and jitter effects, are much more complex. IMO the only way to make progress without seemingly endless wheel spinning is through a well chosen combination of objective and subjective methods."

I am in full agreement with what you say here. I would like to point out though that I am not really interested in linear distortions, such as frequency response/amplitude curves because with a good digital equalizer you can eliminate many of these things very easily. Dispersion can even be compensated for to some extent.

What you cannot easily correct for are things like driver/cabinet resonances, box/port tuning errors, box/driver energy storage (related to resonances but not quite the same), thermal and mechanical compression, harmonic distortion, cone breakup etc.

It turns out that these things and not on-axis, off-axis or even averaged frequency response go a long way towards defining the characteristic sound quality of a speaker. In many cases the non-linearity IS the characteristic of the speaker in question.

For digital, where distortions are low in magnitude, the characteristics of the sound are made up largely by analog non-linear effects (harmonic distortion, phase shifts) and digital non-linear effects (jitter). This is why the linear distortion of a slightly rolled off high frequency (who is going to hear -3db at 20Khz anyway?? Not many I suspect) is likely to be inaudible and unimportant but how the two deal with jitter and phase-shift possibly very important.

For amplifiers you have phase shifts, harmonic distortions, intermodulation distortions, shifting noise floors (due to multitudes of harmonics) etc. The complexity of these distortions is a big reason, IMO, why they are so audible even though they measure very low. "Simple" distortions are not such a big deal and may even be largely inaudible. Complex ones are not only audible they can make a product sound bright or dead and lifeless, thin, fat it just depends on what and where the distortions are centered. Tube amps tend to sound warm and wooly because they have a lot of distortion centered at lower frequencies (transformer saturation being a prime culprit). SS amps with ample negative feedback tend to sound "bright" or "thin" or "sterile" because their distortions tend to be centered at upper frequencies (negative feedback?). I have also heard tube amps that could sound both warm and wooly AND bright and hard at the same time!

I haven't figured out yet what could be the cause of amps that sound "dead" or "lifeless". I does an amp with seemingly huge power deliver a sound that sounds like it has no power?? I have heard this attributed mostly to amps with a lot of negative feedback but it is not clear to me whether that is a cause or incidental. It sounds like a kind of compression where you feel the need to always turn it up to get it going but it never does and by the time you have it loud you want to turn it back down because it still doesn't sound good...just annoying. Have you experienced this? I had an amp, the Sumo Polaris III (not designed by Bongiorno), that was just this kind of amp. No transparency and just dead sounding. Why??? It measured great, sounded really awful.

I even demonstrated this convincingly to non-audiophile friends but who happen to be good listeners anyway (lots of concerts). They wanted to buy a system so I said I would help them. We found a nice pair of floor standing 2-way speakers (AudioPlan Kontrast IIIsi), which cost more than they had alloted for a budget. This didn't leave much for a cd player and amp. However; I had a very nice Transcendent sound grounded grid preamp and a pair of Cary 572SE Single ended monoblocks that I thought would work well. We tried it and they loved the sound, it was rich, detailed, transparent and powerful sounding. What they didn't love was the my asking price, which was very fair btw.

So I decided to help them find a more affordable alternative. We tried several SS amps, a hybrid and a couple of other tube amps. All were inferior, some shockingly so (a Denon Integrated that they had, which sounded as if it might really be broken) and they were disheartened because they could hear so clearly what some on this forum would try to deny. The differences were obvious to them!! Finally, they called me and said, "Could you please bring your amps and preamp back over again". I did and they bought them...that was three years ago and they haven't changed a thing. They only asked me where to get replacement tubes.

They have a decidedly esoteric system by many audiophile's standards but it sounds really good and as they didn't know anything but how it sounds to them they were not influenced by prestige, pedigree and brand names. THey had never heard of any of the brands they currently own previously. However, they know more about good sound than most of my so-called "audiophile" friends.

The problem I think is that in nature we evolved to deal mostly with using non-linear distortions to distinguish between two different sounding things. For example, two woodwind instruments playing the same note (can even be the same instrument) are characterized not by the fundamental frequency of that note but by everything else that is surrounding that note. This means we are sensitive to things sounding similar but different and able to differentiate these things.

I once measured the harmonic structure of a Stradivarius violin as compared to a Guarneri Del Gèsu as compared to an Guadinini using an RTA and a very skilled professional violinist. She would play the same note at about the same amplitude (important as the harmonic structure changes with volume) and the patterns were different but not night and day different.

Now you take distortions that we were not evolved to hear and the filtering mechanisms of our brain were not designed to address and you have a result of being able to hear very minute quantities of such distortions.

I tried Keith Howard's software and I looked at the harmonic patterns that were available from literature (i.e. measurements from Soundstage taken at 1Khz, which I think is more representative than at 50Hz...especially for SS amps where their distortion vs. frequency starts to rise in that range) and put the harmonics and amplitudes into the program thus simulating to some extent real world amps. The result?? Clearly audible differences between the same track encoded differently. Try it yourself, its pretty cool in a geeky way ;-).

RE: Your words, not mine., posted on November 3, 2009 at 11:51:44
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 "For digital, where distortions are low in magnitude, the characteristics of the sound are made up largely by analog non-linear effects (harmonic distortion, phase shifts) and digital non-linear effects (jitter). This is why the linear distortion of a slightly rolled off high frequency (who is going to hear -3db at 20Khz anyway?? Not many I suspect) is likely to be inaudible and unimportant but how the two deal with jitter and phase-shift possibly very important."

Did you notice in the Stereophile review of the Ayre QB-9, in the measurements section, the effect of having the digital LPF in "Listen" mode on intermodulation distortion using 19kHz and 20kHz signals? Check out Figure 13 on this page. At first, I was baffled about why a change to the digital filter, which is nominally a linear circuit, would affect measured distortion that much (though it's not really that bad when you look at the numbers). Then I realized that in "Listen" mode, it's letting through quite a bit of signal at the image frequencies of 44100-20000=24100Hz and 44100-19000=25100Hz. Then these two images at 24100Hz and 25100Hz are causing the analog circuitry some trouble.

I don't think I like these "Listen" filters. I'll take a linear-phase brick wall filter any time (for Redbook anyway).

RE: Your words, not mine., posted on November 4, 2009 at 04:17:14
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 I agree with your assessment of the IM distortion test. The questions then are this, "which sounds better and why?"

"I don't think I like these "Listen" filters. I'll take a linear-phase brick wall filter any time (for Redbook anyway). "

Hmmm...have you played around with a DAC with switchable filters??

I have some experience with switchable filters as a friend of mine and I have done extensive listening tests with his DCS Purcell/Delius upsampler/dac combo. It has 4 different filters, one of which is the measure mode and the others are varieties of listen mode. The result?? The "measure" mode, which has pre and post ringing sounds significantly worse than 2 of the other modes. The 4th mode is slightly better than "measure" but loses impact in the music. The "measure" mode sounds much more artificial and "digital" while the two other modes are more analog sounding and have only post ringing I believe.

As mentioned in the review, DCS's more extreme filters have even worse IM performance than the Ayre's but still sound better than the "measure" mode.

So in this case it seems that the wholly unnatural behavior of the standard linear phase filter that gives pre and post ringing (something impossible in the "real" world of acoustic sounds) outweighs the increase in IM distortion, which while possibly audible seems not to be as damaging to the sound quality. Now, I haven't heard the Ayre, but I would let my ears decide and not my eyes on the graph because with digital errors there is nothing in your ear/brain to filter this kind of stuff out and if it is audible then it is likely to sound very unnatural and therefore unappealing.

Of course if the sampling rate is 88.2Khz or 176.4Khz then a filter of this type should give a good result without a lot of IM products in the listening window. Something to keep in mind.

RE: Your words, not mine., posted on November 4, 2009 at 19:30:57
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 "So in this case it seems that the wholly unnatural behavior of the standard linear phase filter that gives pre and post ringing (something impossible in the "real" world of acoustic sounds) outweighs the increase in IM distortion, which while possibly audible seems not to be as damaging to the sound quality."

There's an interesting thing that comes into play though. If the input of the brick wall filter is bandlimited to less than its cutoff frequency, the brick wall filter doesn't pre-ring if there is no pre-ringing at its input. The filter, being linear, won't ring if there is no energy at its input at the frequency where it tends to ring. That is, it won't generate frequency components not present at its input. There's a great post by Werner on this subject here.

I haven't heard any of these new DACs though, as I'm not in a position to be auditioning them.

RE: Your words, not mine., posted on November 5, 2009 at 08:59:41
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "If the input of the brick wall filter is bandlimited to less than its cutoff frequency, the brick wall filter doesn't pre-ring if there is no pre-ringing at its input."


Exactly. This is easily demonstrated doing sample rate conversions using the 64 bit iZotope, which has sliders that control various parameters.

I have been doing my cassette tape transfers using this type of roll off, with energy at 22050 Hz at least 100 dB down. As a result, when played back with a brick wall filter they have only the ringing that I chose when I downsampled to 44.1 kHz so as to sound good with the particular recording. And since I used a minimum phase anti-alias filter, any residual ringing will appear after transients, despite a linear phase filter in a player.

I think it is incorrect to talk about the unnatural behavior of a playback filter. It does make sense to discuss the unnatural behavior of the combination of the record filter and the playback filter. Thanks to the lack of standardization, we are left with a problem. If recordings are made so they sound best with brick wall linear phase anti-aliasing player filters and players use slow rolloff and linear phase anti-imaging filters so they sound best with brickwalled recordings, then when a recording made according to these principles is played on a player that also follows these principles there will be unnecessary high frequency roll off (and possibly extra smearing of transients, but at least no unnatural pre ringing).

Mathematically, the combination of the anti-alias filter in the recorder and the anti-image filter in the player produce a single filter that is the convolution of their two impulse responses, and convolution is a commutative operation. Hence the order of filters makes no difference. (This last paragraph for the benefit of any lurkers, I have no doubt that you know this, Andy...) As Werner showed, these facts are easily demonstrated by using existing sample rate converter programs, so there is no need to understand the necessary mathematics or depend on outside authorities one may not trust. You must use a high quality SRC in these tests, because if the SRC uses limited word length in its calculations or fails to properly dither output values non-linear distortion will be introduced, violating the assumptions of linear systems theory used to design digital filters. In addition, if resampling causes inter-sample peaks to clip the non-linear distortion will invalidate these comments, but this will be obvious distortion. (And will render the recording unlistenable, as with much recent pop music.) Some poor quality SRCs have significant non-linear distortion in them, for example the old SRC in Soundforge 9. (Soundforge 10 now uses the excellent 64 bit iZotope SRC).






Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Your words, not mine., posted on November 4, 2009 at 07:52:16
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "Hmmm...have you played around with a DAC with switchable filters??"

Have a look at the measurements of this 10-year old player. Note John Atkinson's commentary on image leak-through.

"...The 4th mode is slightly better than "measure" but loses impact in the music. The "measure" mode sounds much more artificial and "digital" while the two other modes are more analog sounding and have only post ringing I believe."

Nothing stops someone else coming along and preferring a different set of filters and giving precisely the same reasons you've stated here. IMO, this is where controlled testing comes into it's own. Does the Ayre QB-9 sound the way it sounds because of the Minimum phase filter, I have my doubts. Or is it because of artefact's associated with it's slow roll-off, it would interesting to test Linear Phase filter with an equivalent slow roll-off and reached to some informed conclusions.

In conclusion, the OP question has been answered the Ayre's measurements are not comparable to the Benchmark. The measurements give some clues as to why the Ayre might sound different from the Benchmark. When a benchmark, such as the CD standard, is reached by juggling a set of conflicting performance parameters it's no surprise that someone else might prefer to the draw the line at a different point.


Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 4, 2009 at 15:58:23
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 "Or is it because of artefact's associated with it's slow roll-off,"

As you said, "I have my doubts".

RE: Your words, not mine., posted on November 4, 2009 at 16:45:19
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "As you said, "I have my doubts".

Without a Linear Phase filter with near-identical slow roll-off to compare against the Minimum phase filter, both us are merely speculating on the reasons behind the audible differences.

Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 5, 2009 at 02:28:06
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 "both us are merely speculating on the reasons behind the audible differences."

I never said I had firm evidence that it was the case but from my thinking my proposal is the more logical of the two.

Remember - you're dealing with someone who thinks the last post in the thread "wins",..., posted on November 5, 2009 at 08:19:03
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... no matter how irrelevant.

Besides, there's more in common between him and Pat D. than I thought - they both conveniently start having English comprehension problems, when it becomes clear that their position loses ground.

Like, what could be so difficult about the question "Why the DACs that measure worse sound better"?

What he's pushing so hard is "Those DACs measure worse" and "They sound different because of that", neither of which has any relation to the original topic.

RE: Your words, not mine., posted on November 5, 2009 at 02:56:23
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 but from my thinking my proposal is the more logical of the two.

sigh...obviously I disagree but that does not move the discussion forward and the posts that move the discussion forward do not substantiate your position.


Music making the painting, recording it the photograph
Oh and I forgot to add..., posted on November 6, 2009 at 04:20:13
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 That the analog output circuitry used by Ayre is likely much better sounding due to a discrete zero negative feedback circuit as opposed to a gaggle of op amps. This most likely makes a more ear friendly distortion pattern as well.

It must be realized that the IM distortion in the measurement is not really IM distortion but Aliasing. If you look at the 1Khz distortion pattern there is nothing there for the Ayre, essentially. Can this aliasing be heard?? Its at a pretty high frequency so I would doubt it.

RE: Oh and I forgot to add..., posted on November 6, 2009 at 10:23:49
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 "It must be realized that the IM distortion in the measurement is not really IM distortion but Aliasing."

Actually, it is both (assuming you change the word "aliasing" to "imaging"). One ends up with two desired frequency components f1 = 19kHz and f2 = 20kHz. But then there are the undesired images at f3 = 24.1kHz and f4 = 25.1kHz. This aspect is of course imaging and not IM distortion. However, the signals at frequencies f3 and f4 are slightly less than 10dB down from those at the desired frequencies of f1 and f2, and are not themselves harmonics or IM products of f1 and f2. This situation has turned what should have been a 2-tone IM distortion measurement into a 4-tone IM distortion measurement, in which the third and fourth tones don't even belong there to begin with. For the two-tone case, the frequency components at the output will be at |m * f1 +/- n * f2|, where m and n are 0, 1, 2, 3,... and so on. But with four signals, we get frequency components of |m * f1 +/- n * f2 +/- p * f3 +/- q * f4|, where m, n, p and q are again 0, 1, 2, 3,... and so forth. Many of these, as can be seen in Figure 13 here are in-band. And of course all of these for which either p or q (or both) are not zero will not show up in the simpler two-tone IM test.
RE: Oh and I forgot to add..., posted on November 10, 2009 at 05:00:04
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 I am not so sure about this analysis because if you look at the simple harmonic distortion plot there is essentially no distortion to speak of at all from a 1Khz input. I would find it unlikely then that this analog circuit would generate appreciable IM distortion when there is no distortion at 1Khz above the noise level. I could be wrong but when the one looks that good it is highly unlikely for it to have appreciable IM distortion.

RE: Oh and I forgot to add..., posted on November 10, 2009 at 09:42:07
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 "I am not so sure about this analysis because if you look at the simple harmonic distortion plot there is essentially no distortion to speak of at all from a 1Khz input."

Okay, let's look at the 1 kHz harmonic distortion data. I've linked them below to make the discussion easier.



Ayre QB-9, 16 bit data, 1 kHz signal at -90 dBc




Ayre QB-9, 24-bit data, 1 kHz signal at -90 dBc



You've concluded from the data above that there is no measurable harmonic distortion of a 1 kHz signal. But looking at the first plot, the signal is 90 dB below full scale, and the noise floor of the measurement is around 125 dB below full scale. So this measurement really can't resolve harmonic distortion components that are more than about 35 dB below the fundamental. A similar situation exists for the second plot, though the noise floor is a bit better. What this shows is that the harmonic distortion is very good for a -90 dBc signal. This is a difficult measurement to make because of noise.

A better way to look at the harmonic distortion is to use the 50 Hz harmonic distortion data into a 100k load. This is shown below.



Ayre QB-9, 50 Hz harmonic distortion with signal of 0 dBc



From this plot it can be seen that the 2nd harmonic is about -90 to -95 dBc, while the third is about -84 dBc.

Now let's look at intermods. For the case of two carriers, one at 19 kHz and the other at 20 kHz, the distortion components are as follows:

18 kHz: third order
17 kHz: fifth order
16 kHz: seventh order
15 kHz: ninth order
...and so on

1 kHz: second order
2 kHz: fourth order
3 kHz: sixth order
4 kHz: eighth order
...and so on

Using the guide above, you can determine the order of each IM product for the simple two-tone test below:



Ayre QB-9, two-tone IM into 600 Ohms, Measure mode



Now look at the case where the images are allowed to leak through at a much higher level in "Listen" mode. The new distortion components are due to the presence of signals at the two image frequencies.



Ayre QB-9, two-tone IM into 600 Ohms, Listen mode



Considering that the load is 600 Ohms, this is pretty respectable performance. But the only difference between the last two plots is Measure mode vs. Listen mode. Clearly the measured data are different. Since both filters are flat in the middle of the audio band (say 5k to 10k), it isn't just a frequency response issue. It must be nonlinearity.

RE: Oh and I forgot to add..., posted on November 12, 2009 at 06:18:24
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 Oh yeah, you are of course right about the 1Khz plots, I didn't notice that the results were taken at -90db. That is an unusual plot as normally it would be shown at full scale. The 50Hz makes it quite clear that there are some distortion components above the noise.

RE: Oh and I forgot to add..., posted on November 12, 2009 at 07:53:13
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 That -90dB harmonic distortion test is a good one to perform though. In the past at least, DACs have had a problem analogous to a badly underbiased class B amplifier - increasing harmonic distortion percentage as the signal drops in level. The Ayre does very well in this test.

RE: Oh and I forgot to add..., posted on November 12, 2009 at 17:25:45
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 Probably because the Ayre's output stage is most likely Class A so there is no crossover distortion.

RE: Oh and I forgot to add..., posted on November 12, 2009 at 18:13:40
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 Well, I wasn't referring to crossover distortion in the analog stage, but distortion of the DAC chip itself. An ideal DAC chip will have uniform quantization steps (change the input bit count by one, and the output current or voltage should ideally change by the exact same amount for each such change over the full range of bit counts). In real DACs, these step sizes are somewhat non-uniform. As the signal variation gets smaller and smaller, the DAC error due to the non-uniform step size becomes a greater fraction of the desired signal, making distortion increase as signal level decreases. This happens even with a theoretically perfect analog stage following the DAC. The old ladder-style DACs were particularly susceptible to this, but modern DACs use a technique called dynamic element matching (large PDF file), which improves the measured distortion performance of the DAC chip a lot.

RE: Oh and I forgot to add..., posted on November 13, 2009 at 11:29:30
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 After skimming through the Thesis, I see that the technique used to in the ESS Sabre chip that converts the multi-bit sigma-delta signal to analog is "dynamic element mapping". As is the distribution technique used in the DcS "Ring DAC". So I was familiar with the technique, but not the name or the (highly detailed) analysis. Thanks.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Oh and I forgot to add..., posted on November 13, 2009 at 13:39:19
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 You're welcome Tony. I have only skimmed it so far, so I can't comment intelligently at the moment.

RE: Oh and I forgot to add..., posted on November 13, 2009 at 08:02:40
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 Thanks for the link to the thesis. I've not looked at dynamic element matching, but will do so now.

If a DAC uses a single bit sigma-delta modulator, then there won't be any need for matching, as there is only a single output element. Chips that use multi-bit modulators can use other techniques, such as randomizing the use of individual switches, which changes distortion that might be caused by device mismatches into wideband noise. Take a look at the white paper for the ESS SABRE DAC or their patents if you are interested.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Oh and I forgot to add..., posted on November 13, 2009 at 04:22:54
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 Hmmm...I think that the analog output stage is much more likely to be a cause of this than the DAC chip itself. Afterall, the opamps are not running class A so they might produce some zero crossing distortion. Ayre uses only discrete Class A circuits so there should be none of that. IMO, the analog output stage does much more for defining the overall sound quality of the DAC than the digital filter or DAC chip itself, not that these don't affect sound quality as I have heard them most certainly do just that. However, I have heard otherwise very good digital designs ruined by a bog standard multi-op amp low pass filter/buffer arrangement.

RE: Oh and I forgot to add..., posted on November 13, 2009 at 09:29:32
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 Let's set subjective issues aside and just talk about measurements. Any device whose gain varies with signal level will produce distortion. Some of the Stereophile DAC reviews include plots of the gain error vs. signal level. Unfortunately, the Ayre review doesn't include these data, but the Benchmark DAC1 USB review does (on this page). The relevant graph is linked below:



Benchmark DAC1 USB linearity error with 24-bit data

The left axis is for the blue trace (output in dB vs. input in dB), while the right axis is for the red trace (output error in dB vs. input in dB). The blue trace should ideally look like a perfect straight line, because every 1 dB increase in the signal level in the digital domain should give a corresponding 1 dB increase of the analog signal. The red trace shows the error from the ideal. As the signal level gets less than -100 dB relative to full scale, the error begins to increase, reaching about -2 dB, then swinging up to about +4 dB.

This error is orders of magnitude worse than a class AB analog circuit without any feedback at all, and completely dominated by the DAC chip.

Also notice that the measured distortion of the Benchmark is better than the Ayre, including high-order distortion components.
RE: Oh and I forgot to add..., posted on November 15, 2009 at 07:17:53
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 "This error is orders of magnitude worse than a class AB analog circuit without any feedback at all, and completely dominated by the DAC chip."

Is 1 or 2 db error at -110db worse than high order harmonic distortions?? Whether that signal is -110 or -112 or -108 probably doesn't matter very much. Afterall, loudspeakers often vary much more but I have found that linear distortion errors are much more forgivable sonically than non-linear distortions.

So, I am not at all sure that this error is worse than a Class AB analog circuit without feedback (if such a thing was even possible with op amps, which it is not actually).

RE: Oh and I forgot to add..., posted on November 15, 2009 at 10:51:26
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 "Is 1 or 2 db error at -110db worse than high order harmonic distortions?? Whether that signal is -110 or -112 or -108 probably doesn't matter very much."

I'd agree that this level error, in and of itself, is probably meaningless subjectively. But such a level error cannot occur without distortion, so it cannot be considered in isolation. For a nonlinear input-output characteristic, I'll refer you to the "Crunching the Numbers" section, of Keith Howard's article, which I believe you're familiar with. In the first equation, there's a second-order term involving the constant "b", a third-order term involving "c", and so on. Let's assume x=x(t)=V*sin(omega*t), where omega=2*pi*freq. To find the harmonics, we'll need trig identities for sin2(omega*t), sin3(omega*t), sin4(omega*t), sin5(omega*t) and so on. There are a whole bunch of trig identities that relate the sin or cos raised to integer powers to the harmonics and other undesired components one ends up with. Keith Howard mentions a few. I won't get into the details of the math, but here is a summary of the qualitative results.

2nd-order term: creates second harmonic plus an undesired DC offset term.
3rd-order term: creates third harmonic plus an undesired term at the fundamental frequency.
4th-order term: creates fourth harmonic, second harmonic, plus an undesired DC offset term.
5th-order term: creates fifth harmonic, third harmonic, plus an undesired term at the fundamental frequency.
6th-order term: creates sixth harmonic, fourth harmonic, second harmonic, plus an undesired DC offset term.
7th-order term: creates seventh harmonic, fifth harmonic, third harmonic, plus an undesired term at the fundamental frequency.
...and so on.

So you can see from the above that all the odd-order distortion terms create an undesired signal component at the fundamental frequency. These correspond to the gain error, and, depending on their polarity, they may subtract from or add to the fundamental. Another way of looking at this is graphically. If there is a gain error that depends on signal level, this says the plot of output vs. input is not a straight line. This is equivalent to saying that distortion is present. In the RF world, there's a thing called the "1 dB compression point", which is related to how much output voltage or power a circuit can produce. It's found as the point where increasing the input signal by 10dB causes the output to increase by only 9dB. Although this is only a 1dB gain error, it is usually associated with gross distortion, with distortion components in the neighborhood of 10 to 20 dB down from the fundamental. The bottom line is that harmonic distortion measurements, though difficult or impossible to relate to the subjective experience, are nonetheless very sensitive. Very small gain errors with signal level show up as significant harmonic distortion components.

It just so happens there's a great example of gross DAC distortion at low levels in a review that Stereophile just added to their web site. It's on this page. Below are plots of the linearity error vs. signal level, and the distortion of a sine wave at a fixed level of 90dB below full scale.



HRT music streamer linearity error



HRT music streamer harmonic distortion for -90dB signal

You can see that the desired signal is around 87dB below full scale instead of 90dB as it should be, and gross levels of distortion are present.


"Afterall, loudspeakers often vary much more but I have found that linear distortion errors are much more forgivable sonically than non-linear distortions."

These are actually nonlinear distortions, as explained above. With speakers, this happens at high signal levels. With class AB amplifiers and DACs, one ends up with distortions that can increase as signal levels decrease. So what I've been trying to do, maybe with mixed success, is to point out the similarities of DAC distortion and distortion of class AB amplifiers at low level.

"So, I am not at all sure that this error is worse than a Class AB analog circuit without feedback (if such a thing was even possible with op amps, which it is not actually)."

I hope I've demonstrated, at least from a measurement point of view, that the distortion of a poor DAC such as the one above can be orders of magnitude worse than a class AB circuit without feedback. You're right that such a thing can't be implemented using op-amps alone, but it might be implemented with an op-amp having an external class AB buffer such as a BUF634 at its output, outside the feedback loop. A very interesting example of a class AB circuit without global feedback is the Ayre MX-R power amp. Its distortion, shown below, is surprisingly low for a circuit of this type.



Ayre MX-R 1kHz THD+N for various load impedances
Have you looked at the MSB DAC?, posted on November 19, 2009 at 10:16:19
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 If you want the ultimate in low level linearity then you have to look at MSB technology www.msbtech.com. They are not using an off-the-shelf solution from BB, AD or Crystal but making their own discrete DACs that are capable of 32 bit 384Khz resolution. The linearity is probably the best in the business and would minimize the errors you are describing.

There is an older review in stereophile but as it is 6 years old now I see now that they have at least two higher grades of DAC (but at a huge price!!) than the ones found in the review.
http://stereophile.com/digitalprocessors/799/index5.html
http://stereophile.com/digitalprocessors/799/index6.html

Also, some interesting comments by the manufacturer
http://stereophile.com/digitalprocessors/799/index7.html

RE: Have you looked at the MSB DAC?, posted on November 19, 2009 at 18:21:14
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 "If you want the ultimate in low level linearity then you have to look at MSB technology www.msbtech.com. They are not using an off-the-shelf solution from BB, AD or Crystal but making their own discrete DACs that are capable of 32 bit 384Khz resolution."

Yikes! That's quite a challenge. I'll have to look at those articles - thanks.

RE: Oh and I forgot to add..., posted on November 13, 2009 at 11:56:01
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "Also notice that the measured distortion of the Benchmark is better than the Ayre, including high-order distortion components."

This may be true, but it is hard to tell from the graphs, since there aren't comparable scales on the -90 dBfs 1 kHz tone plots and since different frequencies and loadings are used on the high level sine wave tests. The noise level on the Benchmark does appear to be lower, but we don't know if it is really comparable without knowing that the FFS parameters were the same. Perhaps a careful study of the article text would shed more light, but I haven't done this, just looked at the graphs. I would be suspicious of comparing measurements of very low level noise taken over a long time interval, without specific evidence that the measurement gear and test procedures are stable and comparable.

I don't think any of these measurements are going to explain the audible difference between the two converters. However, one thing I can guarantee you, the response to a 75 kHz tone is going to be much better on the Ayre (when running at 192 kHz) due to the Benchmark's ASRC conversion to 110 kHz. So if you want a spec that shows the Ayre better than the Benchmark, there's one, not necessarily relevant when listening to RBCD's, to be sure.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Oh and I forgot to add..., posted on November 13, 2009 at 13:43:01
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 One thing I noticed from the graphs is the different noise floors in the measurements between the two units among other things. I'm guessing that the averaging of the measurements may be different. It does make it hard to compare.

RE: Oh and I forgot to add..., posted on November 12, 2009 at 06:16:21
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 Just so I see if I understand you correctly, you are saying then that the extra peaks in the listen mode are there because of IM between the images that are leaking through and the 19-20Khz IM test signal?

RE: Oh and I forgot to add..., posted on November 12, 2009 at 07:50:37
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 That's exactly right, yes. Identifying which are which can be problematic and may entail changing the test frequencies to be slightly different from 19kHz and 20kHz. I haven't sorted through that completely, as one's head can spin because of all the different frequency combinations possible. IOW, some of the IM products involving all four frequencies might fall right on top of IM products that could be generated by considering only the 19kHz and 20kHz signals. For example, there's something fishy about that 18kHz component in Listen mode - it's lower in level than the 18kHz signal in Measure mode.

RE: Oh and I forgot to add..., posted on November 10, 2009 at 09:02:09
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "I am not so sure about this analysis because if you look at the simple harmonic distortion plot there is essentially no distortion to speak of at all from a 1Khz input."

Any distortion wouldn't have been above the noise because the -90 dBfs signal itself was barely above the noise. If you look at figures 10 and 11 which show the response to full scale tones, you will see the presence of non-linearities that are above the noise.


Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Oh and I forgot to add..., posted on November 10, 2009 at 09:43:01
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 Looks like we posted at the same time :-).

RE: Oh and I forgot to add..., posted on November 10, 2009 at 14:40:48
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "Looks like we posted at the same time :-)."

:-)

I would add that the audible effect of the images will depend on the equipment downstream of the DAC. So while in some systems the IM caused by the images might be inaudible, in others it might be audible due to increased non-linearity. The choice of best filter is likely to be system dependent, IMO.


Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Oh and I forgot to add..., posted on November 10, 2009 at 15:13:40
andy_c
Audiophile

Posts: 963
Joined: June 2, 2007
 Good point. I was going to mention something like that earlier. The images don't just have the potential to affect the DAC itself, but also the preamp, power amp and tweeter too.

RE: Oh and I forgot to add..., posted on November 6, 2009 at 07:20:09
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 "It must be realized that the IM distortion in the measurement is not really IM distortion but Aliasing."

Could you be more specific as to which measurement and which portions of the measurement?

The reason why I ask is because "aliasing" is a mechanism at which high frequency components get mapped into lower frequency components when sampling. However, this is a function that takes place during analog to digital conversion or downsampling. In the case of digital to analog conversion or upsampling, there is no aliasing. Instead, there is imaging, which is an operation that adds spurious components at frequencies above the Nyquist rate. Assuming linear operation, there shouldn't be any mechanism in a DAC that adds spurious components at lower frequencies. In other words, the existence of such components is proof of non-linear operation somewhere in the system. If it is at frequencies and levels that are "inaudible" this may be OK. Or it may be a clue as to what additional test signals need to be generated that may show up at other frequencies or levels that are not OK.

Or it could be that the spurious signals aren't in the Ayre at all, but are in the test equipment.

It is very hard to draw conclusions as to the audibility of low level effects, even those at high frequencies. For example, the "ringing" of the traditional brick wall filters used in CD players takes place at an ultrasonic frequency that is inaudible for human adults. None the less, it appears to degrade the audible sound when it occurs, especially when it precedes transients. Similarly, changes in dither algorithms occur at such a low level that they are supposed to be inaudible, but yet the different algorithms seem to make an audible difference, at least to the careful listeners. I have certainly heard these effects at 16 bits, but there are people who have noticed them at 24 bits, which are extremely low SPL levels when playing recordings of music. (But not perhaps a space shuttle launch on a very high powered system.)



Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Your words, not mine., posted on November 6, 2009 at 03:53:57
morricab
Dealer

Posts: 4456
Location: switzerland
Joined: April 1, 2005
 Sigh right back at you...but at least I got you to state a position, "obviously I disagree", clearly for once in your posting life.

However, as usual you have done nothing to try to state WHY you think your position is more important to the sound and while you are waiting for me to push the discussion forward by my telling you why I think the other way you could have already put forward something of a hypothesis. You didn't and I would speculate that you have no reason for why you think that way. Prove me wrong (or right).

"do not substantiate your position"

They don't substantiate yours either. Just for clarities sake please restate your position. I think I know what you want to say but as usual you don't ever come right out and say it.

My position is that the Ayre likely sounds better (I haven't heard it so no comment on that aspect) despite the "worse" measurements in terms of HF roll off and IM distortion BECAUSE these are not as important to the sound quality as the things that it does better with regard to digital filtering. No one has disproven that hypothesis in this thread at least.

RE: Your words, not mine., posted on November 6, 2009 at 05:01:25
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 Sigh right back at you...but at least I got you to state a position, "obviously I disagree", clearly for once in your posting life.

Were you sleeping when you wrote this? Here's an excerpt from the OP

However, posters who actually listened to both, have no doubt which one sounds better - and that ain't Benchmark. Almost the same goes for Benchmark vs. Ayre QB-9 - comparable measurements, superiority of Ayre in sound quality.

Here's my original post to you

"In conclusion, the OP question has been answered the Ayre's measurements are not comparable to the Benchmark."

Here's what you've just written

My position is that the Ayre likely sounds better (I haven't heard it so no comment on that aspect) despite the "worse" measurements in terms of HF roll off and IM distortion

Seems you agree with me that the measurements are not comparable. That remains the case even if you think that the measurements do not give any clues as to why they sound different.

In true Morricab style you are engaged in yet another imaginary battle.


NB: I read your followup post, my previous comments still apply it's speculation and not very good speculation IMO because of the leap of logic with respect to the measured differences.

Music making the painting, recording it the photograph
RE: Your words, not mine., posted on November 4, 2009 at 18:50:50
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 I can't speak of the Ayre. However, I have used various software SRCs to up sample from 44.1 to 176.4 kHz, including the 64 bit iZotope which allows independent control over aliasing, steepness, and minimum/linear phase. So I can do the experiment you suggested. Indeed, I more or less have done it, but it was done casually and wasn't a proper experiment.

I ended up with a moderate amount of roll off (less than Ayre) at 20 kHz, complete (100 dB +) cutoff by 22,050, and minimum phase. This sounds good on a wide range of CD rips that I have. But I make no claims as to this being optimum. I was happy to get settings that made my better CDs sound decent to very good. I don't miss a few dBs of highs around 20 kHz—when I have a problem with high frequencies it is still that things sound too bright, rather than too dull, but that speaks more about my room than anything else (and my 65 year old ears).

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

"OP question has been answered" - you haven't understood the question, let alone..., posted on November 4, 2009 at 09:08:18
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... answering it.

However, other posters in this thread made some useful contributions.

RE: Your words, not mine., posted on November 3, 2009 at 17:35:51
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 I have been using minimum phase anti-aliasing filters that roll off moderately smoothly starting at 20 kHz in the cassette tape transfers that I make to 44/16 downloads. This allows the recordings to sound nice on players that have brick wall filters. However, my filters are set to keep all the aliases at least 100 dB down. I don't want any spurious tones in my recordings. (The ADC runs at 88/24, this format is used for editing and any mastering adjustments. The final result is downsampled using the 64 bit iZotope SRC and dithered down to 16 bits.)

Of course, these recordings will sound best when played on brick wall players. They will be rolled off and smeared slightly when played on present "filter du jour" equipment such as the QB-9.


Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

RE: Your words, not mine., posted on November 3, 2009 at 14:13:08
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "Then I realized that in "Listen" mode, it's letting through quite a bit of signal at the image frequencies of 44100-20000=24100Hz and 44100-19000=25100Hz."

Not very surprising when one considers that @22.05KHz, the filter is only -6dB down, the filter roll-off in "Listen mode" is VERY leisurely, therefore quite a few image frequencies will slip through.

"I don't think I like these "Listen" filters. I'll take a linear-phase brick wall filter any time (for Redbook anyway)."

Some folk are bound to feel differently ss their choice is based solely on listening experience which is perfectly valid by the way. There's nothing to stop anyone from preferring the output of a filter that lets a few image frequencies slip through. And surely a tad more distortion a'int that bad if one's goal is to get closer to the absolute sound. :-)

Music making the painting, recording it the photograph
RE: Thanks Don, posted on November 2, 2009 at 17:13:48
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 You've hit the nail on the head when you wrote

"I don't think the most accurate equipment sounds the best, though potentially it might in some situations, but I think that's just the obvious nature of this beast. What sounds the best - whether it is the most accurate or not hardly makes a difference and certainly has little to do with what sounds the best in another situation."

What's accurate may not necessarily sound the best in every situation, sometimes some subtle 'enhancements' are just the right ticket to musical bliss . It's time folks put aside that tired worn out excuse "My ears are the most accurate measuring tools, why don't the measurements correlate to what I am hearing". It's pathetic and as you wrote earlier hypocritical. Worse still, purveyors of snake oil and various audio myths thrive where such thinking is prevalent.

Music making the painting, recording it the photograph
You're still unable to grasp the topic of this discussion - and no, it's not "accurate sound",..., posted on November 2, 2009 at 14:28:10
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... whatever that means to you.

You'll have to try again, however at this point it's pretty clear that all further attempts will be as fruitless as this latest one.

I like the version in the article that Todd linked better., posted on October 29, 2009 at 09:03:03
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 Your persistence is noted, but the issue at hand is "Better", not just "Different".

The linked article contributes to explaining why.

RE: I like the version in the article that Todd linked better., posted on October 30, 2009 at 01:31:02
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "I like the version in the article that Todd linked better."

Good luck in your search for zebras then.


"The linked article contributes to explaining why."

It doesn't. Of course, you can persist searching for answers in the wrong places, it's your choice.

Music making the painting, recording it the photograph
If you say so, oh Enlightened One., posted on October 30, 2009 at 07:54:57
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 Apparently, the idea that you're trying to drive home is that you have nothing of substance to contribute to this thread.

I get it - you can give it a rest now.

RE: Paraphrased, posted on October 31, 2009 at 01:11:14
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 I'll paraphrase your comments as "Tell me what I want to hear"

To quote Newton restatement of Occam's razor "we are to admit no more causes of natural things than such as are both true and sufficient to explain their appearances. Therefore, to the same natural effects we must, so far as possible, assign the same causes"

To seem content to chase a mirage while a thoroughly valid explanation sits right in front of you.




Music making the painting, recording it the photograph
No, you're wrong again. It was not "Tell me what I want to hear", but rather..., posted on October 31, 2009 at 04:59:40
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... "Don't bother to tell me anything", or, simplifying it for you, "Just go away".
RE: LOL! Good luck in your search. -nt, posted on November 1, 2009 at 01:25:53
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 *

Music making the painting, recording it the photograph
RE: What is missing in measurements? In this case, not much, posted on October 28, 2009 at 00:19:59
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 "look at those measurements again, even in measure mode, the QB-9 is 1.7dB down at 20kHz in comparison to the Benchmark's 0.2dB. In Listen mode the rolloff starts earlier and looks likes its greater than -3.5dB @20KHz, that's a dramatic difference."

Do have any idea why the FR is worse in "listen" mode?

The HF extension/flatness in CD playback is a technical hot potato..... The raw signal on the CD is inherently rolled off due to modulation. The digital filter "fills in the nulls" in the modulation, restoring the FR to flat. But time response suffers (ringing) in the process. There is a time response/frequency response tradeoff. You either have flat response and ringing or "rolled off" HF and superior time response. I personally think the most-satisfying CD playback goes somewhere in between, where FR is somewhat rolled-off with damped ringing.


RE: What is missing in measurements? In this case, not much, posted on November 2, 2009 at 02:42:19
Werner
Audiophile

Posts: 1829
Joined: September 30, 1999
 "The raw signal on the CD is inherently rolled off due to modulation."

That is somewhat of a misrepresentation.

The signal on the CD, i.e. the sequence of sample values, is not rolled off.
But it comes with the requirement that the sample values should be interpreted as being the energies of a sequence of dirac impulses.

Upon playback, or even upon numerical analysis, the samples are often
interpreted as attached to levels that are invariant during one sample period (hence the staircases in graphical representations or the typical NONOS DAC output signal). This interpretation, named zero-order hold, is a linear operation implicitly added to the signal's processing, and the frequency response of this operation happens to be a treble rolloff (which follows Sinc(f) by the way).

If the samples were to be replayed through a hypothetical dirac impulse DAC (*) then this treble rolloff would be absent.


(* Which does not exist but can be approximated by oversampling with a filter kernel [0 0 0 0 0 0 0 1 0 0 0 0 0 0 0], as demonstrated in the link below)


bring back dynamic range
RE: What is missing in measurements? In this case, not much, posted on October 28, 2009 at 01:38:17
theaudiohobby
Audiophile

Posts: 4438
Joined: January 16, 2003
 "Do have any idea why the FR is worse in "listen" mode?"

Not willing to speculate. But it's pretty clear that the two measurements are not comparable though :-)

" The raw signal on the CD is inherently rolled off due to modulation. The digital filter "fills in the nulls" in the modulation, restoring the FR to flat."

hmmm....that's a very idiosyncratic description of the D/A process.

Music making the painting, recording it the photograph
RE: What is missing in measurements? In this case, not much, posted on October 28, 2009 at 15:59:59
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 " The raw signal on the CD is inherently rolled off due to modulation. The digital filter "fills in the nulls" in the modulation, restoring the FR to flat."

hmmm....that's a very idiosyncratic description of the D/A process.

That is true...... The problem is nobody has been able to build a better mousetrap for this...... The latest idea is minimum-phase (asymmetrical impulse response) filtering, which IMO is a step in the right direction, for it better preserves leading transients, yet fills in the "nulls" on the decay side, better depicting the pre-digitized music signal.

I personally think "waveform fidelity" (a buzzphrase made famous by Technics/Panasonic) would be the best objective arbiter of digital audio record and playback. The output waveform of the digital playback should be compared to the pre-digitized waveform. The less the overall deviation, the better the A/D and D/A. (This is also an issue with recording.)


Daniel H. Cheever Wrote an Article on the Subject.............., posted on October 27, 2009 at 17:53:43
Todd Krieger
Audiophile

Posts: 22046
Location: SW United States
Joined: November 2, 2000
 His paper could provide some valuable information..........


Very interesting - should keep me occupied on the train for some time., posted on October 28, 2009 at 09:04:06
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 Looks like I'm late to the table, as usual.

Or maybe someone forgot to ask for your personal opinion?nt, posted on October 27, 2009 at 16:00:07
Dan Banquer
Manufacturer

Posts: 7440
Joined: November 13, 2002
 nt

Just in that case, I stated it in my post. The thing is - ..., posted on October 27, 2009 at 19:08:02
carcass93
Audiophile

Posts: 2634
Location: NJ
Joined: September 20, 2006
 ... it's not just my personal opinion, but rather a pattern that I see. Benchmark is the best representative of that contradiction, where exceptional measurements go along with mediocre sound.

Just to be more specific - better DACs in my experience are PS Audio DL III (somewhat - better highs, weaker bass), Apogee MiniDAC, Bryston BDA-1, and DAC in Simaudio Supernova CDP (these three in all respects). First two are in the same price range, latter two - more expensive.

Page processed in 0.311 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2009, Audio Asylum, all rights reserved ] [ Go Green ] [ Take5 ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]