I split out some stuff about this from the Blowtorch thread and merged it here. So if the conversation seems choppy, it's because we were having two conversations simultaneously.
SY said:
Not quite. Two $250 players and one Sony. Apologies, I've been too busy with work to properly pursue. The Sony might be the real deal, though Stereophile's measurements cast doubt on the reported noise difference to CD. The Yamaha was the unit found to be wanting in independent testing. As to the Pioneer, some of what I've read online suggests it does an internal conversion of SACD to PCM, making it questionable as well. The best person to weight in on the latter would be Charles Hansen if he'ld be so kind, he reviewed it in depth.
SY, you don't even one duff unit casts the summary into doubt? At the end of the day Meyer and Moran skipped the most basic precursor to any scientific experiment purporting to wider validity, confirming the performance of their instruments.
rdf, no, I don't think that having three units available, including the very expensive and highly rated Sony, somehow degrades the test.
One could argue that maybe THEIR Sony was a bad one, which they may be able to confirm or refute. But jeez, if I can't go out and spend $3000 for a unit that was well-reviewed by the Holy Priesthood and expect it to be up to snuff, what the hell good is SACD anyway?
From the BAS writeup:
One could argue that maybe THEIR Sony was a bad one, which they may be able to confirm or refute. But jeez, if I can't go out and spend $3000 for a unit that was well-reviewed by the Holy Priesthood and expect it to be up to snuff, what the hell good is SACD anyway?
From the BAS writeup:
My good friend has a Sony 777ES SACD player that has been heavily modified by Vacuum State Systems. It has both CD and SACD capability, it is quite difficult to find recordings that are first class in each (both) format(s). (The mods were not cheap).PMA said:
What I hear is an improvement in the upper registers less blending of instruments and a cleaner fundamental of cymbals. Unfortunately these are the exact frequency bands where many systems and speakers have problems. Speaker diffraction issues for example could quite easily mask the cleaner high end I hear with SACD.
The difference is not spectacular, but real enough that I am quite reluctant to purchase an inexpensive SACD player for my home system. To fully realize the improvement requires a system capable of some degree of subtlety and an already quite clean upper register.
rdf said:
Stereophile's noise measurements of digital systems are somewhere between problematic and outright invalid depending on the gear being measured. They make use of measurement gear that uses analog bandpass filters, and looks at the RMS value at the bandpass filter output to compute the noise spectral density at the filter center frequency. In cases for which the noise spectral density varies significantly over the frequency band for which the filter's response is non-negligible, this technique gives erroneous results. This is exactly what happens when measuring SACD players.
For example, many good SACD players have a noise spectral density that's almost constant with frequency up to 20 kHz, but rises quite a bit above 20 kHz. A bandpass filter whose center frequency is 20 kHz will catch significant energy at frequencies above 20 kHz and erroneously report that the noise spectral density of such a player at 20 kHz is higher than it is at, say, 10 kHz. Measurement gear that uses FFT techniques for this will correctly indicate constant noise spectral density up to 20 kHz for such a player.
Atkinson has been told numerous times about this problem, and has said he wants to keep using this technique to retain consistency with earlier measurements. So I guess it's better to be consistently wrong than to have been wrong in the past and correct in the present and future.
Sorry SY, not what I intended. No issue with multiple players though I was obviously mistaken on that account and unable to find the original account to foist blame. The Sony appears to be the one unit of arguably good pedigree, the Yamaha and Pioneer require a closer look. Money and status have nothing to do with it, checking what I could find online regarding their measured performance has everything. Charles Hansen did a very in depth technical analysis of the latter which I haven't had the time to review, maybe he could weigh in. One report mentioned it internally converted to PCM before output. The Yamaha was covered elsewhere and found to be lacking.
It's a blind spot, the assumption high end manufacturers soak in snake oil but commodity mass producers are never prone to badge engineering. Have we already forgotten mandated federal regulation for as basic a parameter as amplifier output power, a result of rampant dishonesty? $250 per short lived model doesn't leave much headroom for engineering below the 20th bit. Maybe it's having been inside this stuff for 20 years that makes me as cynical of cheap mass manufactured hifi as others are of the zoot stuff. One unit if proven duff invalidates all the published statistics and M&M did not state they confirmed the players were capable of advertised performance. It was sloppy.
Interesting insight andy_c. Would this impact the difference measured between CD and SACD or just the absolute scaling? Edit: the noise measurements appear to show the correct signature +20kHz rise in SACD noise with a trough from 10-20kHz.
It's a blind spot, the assumption high end manufacturers soak in snake oil but commodity mass producers are never prone to badge engineering. Have we already forgotten mandated federal regulation for as basic a parameter as amplifier output power, a result of rampant dishonesty? $250 per short lived model doesn't leave much headroom for engineering below the 20th bit. Maybe it's having been inside this stuff for 20 years that makes me as cynical of cheap mass manufactured hifi as others are of the zoot stuff. One unit if proven duff invalidates all the published statistics and M&M did not state they confirmed the players were capable of advertised performance. It was sloppy.
Interesting insight andy_c. Would this impact the difference measured between CD and SACD or just the absolute scaling? Edit: the noise measurements appear to show the correct signature +20kHz rise in SACD noise with a trough from 10-20kHz.
Wavebourn said:
I don't have the article at hand, but if memory serves me well, one suggestion was that SACD generally is thought to sound better than CD is because they are better produced and recorded. Mainstream CD's suffer from compression and strong clipping to win the loudness wars. SACD'd are higher priced, not for the main public that just wants loud music, and therefore are better recording-wise. More dynamic, less clipping.
Of course it is a generalization but there may be something to it.
Jan Didden
rdf said:
Well, it's not so much a CD vs. SACD issue, but one of the technology used in the DAC chip. For example, the old-style R-2R ladder DACs have a noise spectral density that is quite constant with frequency, so the bandpass-filter measurement technique will give good data in that case.
But many (most?) DACs nowadays use some variant of multibit delta-sigma, which does oversampling (above and beyond the typical 8x used in digital filtering) and noise shaping. Noise shaping typically causes the noise spectral density outside the the audio band to rise to a level greater than its value within the audio band. The effect may not be as pronounced with multibit delta-sigma as it is with SACD, but it's still there.
So the effect of using the bandpass filter technique for noise spectral density measurement is to make any DAC that uses oversampling and noise shaping, whose noise spectral density is constant in the audio band, look like that spectral density is rising within the audio band. For an example of this, see the TI/BB PCM1794A datasheet. See figure 16 on page 12. Note the noise level is roughly constant up to 20 kHz but rises above that. A Stereophile measurement of a player using that DAC would likely show increasing noise within the audio band.
Bottom line is that the Stereophile noise measurement technique penalizes players that use noise shaping techniques. This includes CD and SACD players, but the effect is probably more pronounced with SACD.
Thanks andy_c. From Figure 5 here
http://www.stereophile.com/hirezplayers/491/index6.html
it was apparently still an issue. I don't know enough about the details to translate from chipset to finished circuit. Edit2: is the DAC likely identical for both CD and SACD since it's the same player? It all goes back to the significant difference reported in audible noise levels.
http://www.stereophile.com/hirezplayers/491/index6.html
it was apparently still an issue. I don't know enough about the details to translate from chipset to finished circuit. Edit2: is the DAC likely identical for both CD and SACD since it's the same player? It all goes back to the significant difference reported in audible noise levels.
On SACD. I don't have a player but I did have the use of a cheap one for a short time. Playing a dual layer disk I thought the SACD layer sounded better in the treble. On first hearing my reaction was, "where has the treble gone?" There seemed to be less treble but what was there was of better quality, it seemed less 'splashy' or 'emphasised'.
A golden eared friend hearing the same unit did not like the SADC sound. He is non technical and completely mis-trusting of any measurements, whereas I am technical and do trust measurements. The listening was done sighted.
Could it be that because SACD 'measures' better than CD he assumes the measurements don't indicate better sound, whilst I would assume the opposite?
On these tests I don't doubt the methodology because I don't have a problem with the results. To me the results do not say 'there is no difference in sound' they just say 'people can't tell the difference blind'.
All this means is that the actual differences are very small, not that they don't exist.
A golden eared friend hearing the same unit did not like the SADC sound. He is non technical and completely mis-trusting of any measurements, whereas I am technical and do trust measurements. The listening was done sighted.
Could it be that because SACD 'measures' better than CD he assumes the measurements don't indicate better sound, whilst I would assume the opposite?
On these tests I don't doubt the methodology because I don't have a problem with the results. To me the results do not say 'there is no difference in sound' they just say 'people can't tell the difference blind'.
All this means is that the actual differences are very small, not that they don't exist.
Science and Subjectivism by Douglas Self
.
This is an expanded version of an article
that appeared in the UK journal Wireless World for July 1988.
Science and Subjectivism in Audio, by Douglas Self
.
This is an expanded version of an article
that appeared in the UK journal Wireless World for July 1988.
Science and Subjectivism in Audio, by Douglas Self
Lineup
Good article, and difficult I would think to dismiss out of hand.
Most subjectivists when confronted with scientific evidence demand more.
They believe in the limitations of measurements but on the other hand they believe there is no limits to their own hearing abilities, they are a wonderful bunch of people to study.
Good article, and difficult I would think to dismiss out of hand.
Most subjectivists when confronted with scientific evidence demand more.
They believe in the limitations of measurements but on the other hand they believe there is no limits to their own hearing abilities, they are a wonderful bunch of people to study.
fredex said:
What any scientist should do when presented an experimental protocol is examine it's assumptions and conditions for error and omission. This is the ground zero of scientific process. Yet, when someone does precisely that to experimental results with which you agree, they're branded subjective. Alice through the looking glass.
I'm amused by the implication that SACD could possibly be "ruined" by passing thru negative feedback amplifiers
don't SACD fanboys know how noise shaping modulators work? (hint noise shaping BitStream/DSD ADCs use lots and lots of negative feedback from low bit resolution internal DACs to multiple stages of integrators - probably 5 or more internal D/A feedack loops in SACD noise shaping ADC in addition to the local integrating feedback loops around each stage)
don't SACD fanboys know how noise shaping modulators work? (hint noise shaping BitStream/DSD ADCs use lots and lots of negative feedback from low bit resolution internal DACs to multiple stages of integrators - probably 5 or more internal D/A feedack loops in SACD noise shaping ADC in addition to the local integrating feedback loops around each stage)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.