Hmm. I've read Moncrieff's papers (very long-winded... jeez, could he be more wordy? um, *ahem* perhaps I should look in the mirror before pointing fingers, but at least my words are there to be precise, not redundant 
). Anyway, I think Moncrieff has one basic concept correct, but just doesn't have the technical knowledge to really understand what he's talking about. While he may have a valid point here or there, I have to dismiss his writings as uninformed and overly opinionated. His observations regarding the treble quality in SACD may also be influenced to a certain extent by "experimenter expectancy" (to quote the term as used by another, perhaps even more controversial audio personality, Douglas Self). This is unfortunate, as it would suit me just fine to be able to support the man, since I personally believe that DVD-A has greater potential than SACD.
In any case, I do have a rather more thorough understanding of digital signal processing and audio data formats than Mr. Moncrieff seems to have, and I can say that it is indeed true that the DSD format will have an effective "resolution" which is inversely proportional to frequency. I haven't sat down and done the hard math or looked at the specific equations which will govern this, but the basic concept is correct. I also don't buy into the 30kHz+ business, as far as it relates to human hearing. Of course, the effect of ultrasonics on other electronics and devices in the audio chain is another matter, and is almost certain to be detrimental, so this is quite possibly a major weakness for SACD. I also recall seeing some AES papers by Lipshitz and Vanderkooy, which basically state that 1-bit ddata structures are unsuitable for use in high quality audio, due to the fact that they are imperfectible. I have yet to get my hands on the full text of these papers, but maybe I'll join the AES so that I can...
Subjectively, although I haven't been listening to SACD for very long (just a few weeks), I have not found the treble to be anything special at all. The main improvements seem to be in the form and texture of midrange and low tones. In fact, I am not convinced that the cymbals do sound any better than my standard CDs of the same recordings (mostly jazz - could be something to do with the quality of the old analogue master tapes?). I am also familiar with the sound of real instruments, having played violin from the age of three to nine, then choir for a few years and a brief stint with clarinet, before finally settling on drumming and percussion for the past 11 years. It seems to me that redbook CD can be pretty darn good at reproducing cymbals, if properly recorded, whilst SACD seems to offer little if any advantage to my ears. But, perhaps more extensive listening and comparison of more recordings will reveal otherwise...
). Anyway, I think Moncrieff has one basic concept correct, but just doesn't have the technical knowledge to really understand what he's talking about. While he may have a valid point here or there, I have to dismiss his writings as uninformed and overly opinionated. His observations regarding the treble quality in SACD may also be influenced to a certain extent by "experimenter expectancy" (to quote the term as used by another, perhaps even more controversial audio personality, Douglas Self). This is unfortunate, as it would suit me just fine to be able to support the man, since I personally believe that DVD-A has greater potential than SACD.In any case, I do have a rather more thorough understanding of digital signal processing and audio data formats than Mr. Moncrieff seems to have, and I can say that it is indeed true that the DSD format will have an effective "resolution" which is inversely proportional to frequency. I haven't sat down and done the hard math or looked at the specific equations which will govern this, but the basic concept is correct. I also don't buy into the 30kHz+ business, as far as it relates to human hearing. Of course, the effect of ultrasonics on other electronics and devices in the audio chain is another matter, and is almost certain to be detrimental, so this is quite possibly a major weakness for SACD. I also recall seeing some AES papers by Lipshitz and Vanderkooy, which basically state that 1-bit ddata structures are unsuitable for use in high quality audio, due to the fact that they are imperfectible. I have yet to get my hands on the full text of these papers, but maybe I'll join the AES so that I can...
Subjectively, although I haven't been listening to SACD for very long (just a few weeks), I have not found the treble to be anything special at all. The main improvements seem to be in the form and texture of midrange and low tones. In fact, I am not convinced that the cymbals do sound any better than my standard CDs of the same recordings (mostly jazz - could be something to do with the quality of the old analogue master tapes?). I am also familiar with the sound of real instruments, having played violin from the age of three to nine, then choir for a few years and a brief stint with clarinet, before finally settling on drumming and percussion for the past 11 years. It seems to me that redbook CD can be pretty darn good at reproducing cymbals, if properly recorded, whilst SACD seems to offer little if any advantage to my ears. But, perhaps more extensive listening and comparison of more recordings will reveal otherwise...
hi chad,
thanks for the interesting perspective. it's really puzzling how people have so many different takes on the treble quality of SACD. some find them harsh, some find them about the same as CD, some find them vastly superior to anything they've heard. i personally think they are excellent and far better than cd, but maybe your CD setup is better than mine (modified MSB Link DAC). but then, my editor, who has extensive experience with tons of different gear, has a MSB Platinum DAC (one of the best DACs out there) to compare to a Sony XA777ES SACD, and after some initial disparaging comments about SACD treble he's come around to feel that it's among the best he's ever heard. here's an excerpt from his review of the Sony, which was just published:
"If there should be anything to criticize with the 777 and
its DSD processing it should be the highs - the noise
shaping thing. As it turned out, however, I ended up liking
the highs with the 777 very much. Cymbals were uncannily
reproduced with a natural ringing that's seldom heard
anywhere other than in the hall - shimmering, truly
shimmering is the only way I can describe them under certain
conditions. Flute and recorder were equally airy, having a
lilting, delicate, lofty, floating persona. Nicely done,
and that in spite of my overly critical, even derogatory
reference to them before at an earlier date. I was wrong."
my impression is similar, i was a classical musician for many years and i find the representation of treble to be fantastic... BUT it varies tremendously from recording to recording. i think the quality of the particlar DSD mastering process used in a recording is definitely a huge variable here. also, if it is a remaster of an old analog recording, deterioration of the tape over the years could make the treble unbearably hissy (e.g. the Philadelphia Orch. Carmina Burana on Sony SACD - i can barely listen to it).
so far my only theory is the one i mentioned before, maybe the extended HF response of SACD is disturbing some people's systems. it would be interesting to survey what sort of equipment people are using w/their SACDs, and seeing if there is any correlation to their perception of the format - e.g. amps, speakers. getting the best out of SACD treble may be as simple as installing a little RC filter at the input of the amp...
thanks for the interesting perspective. it's really puzzling how people have so many different takes on the treble quality of SACD. some find them harsh, some find them about the same as CD, some find them vastly superior to anything they've heard. i personally think they are excellent and far better than cd, but maybe your CD setup is better than mine (modified MSB Link DAC). but then, my editor, who has extensive experience with tons of different gear, has a MSB Platinum DAC (one of the best DACs out there) to compare to a Sony XA777ES SACD, and after some initial disparaging comments about SACD treble he's come around to feel that it's among the best he's ever heard. here's an excerpt from his review of the Sony, which was just published:
"If there should be anything to criticize with the 777 and
its DSD processing it should be the highs - the noise
shaping thing. As it turned out, however, I ended up liking
the highs with the 777 very much. Cymbals were uncannily
reproduced with a natural ringing that's seldom heard
anywhere other than in the hall - shimmering, truly
shimmering is the only way I can describe them under certain
conditions. Flute and recorder were equally airy, having a
lilting, delicate, lofty, floating persona. Nicely done,
and that in spite of my overly critical, even derogatory
reference to them before at an earlier date. I was wrong."
my impression is similar, i was a classical musician for many years and i find the representation of treble to be fantastic... BUT it varies tremendously from recording to recording. i think the quality of the particlar DSD mastering process used in a recording is definitely a huge variable here. also, if it is a remaster of an old analog recording, deterioration of the tape over the years could make the treble unbearably hissy (e.g. the Philadelphia Orch. Carmina Burana on Sony SACD - i can barely listen to it).
so far my only theory is the one i mentioned before, maybe the extended HF response of SACD is disturbing some people's systems. it would be interesting to survey what sort of equipment people are using w/their SACDs, and seeing if there is any correlation to their perception of the format - e.g. amps, speakers. getting the best out of SACD treble may be as simple as installing a little RC filter at the input of the amp...
Another factor that shouldn't be overlooked is the overall quality of the implementation.
I have an original SCD777ES, and while it was a significant step up on redbood CD as well as bringing some interesting SACD recordings, limitations in it's performance (probably due to jitter in digital section, and a bazillion op amps and coupling caps in analog section) render it ultimate performance for CD and SACD mid-fi- perhaps upper end mid-fi, but still...
I have a DAC constructed with sound techniques, and a NLFB analog output stage, using the CD4390, a "low end" Crystal 24 bit chip. It is clearly better than the 777 on CD, and is good enough to reveal subtle shortcomings of the SCD777ES as a transport compared to a better unit. I'm working on one now with an 8420 for SRC (to 88.2 kHz) and reclocking, with a CS8397 converter, and a more sophisticaed NLFB output stage. SACD disks have some nice qualities, but are still have a signature sound through the 777ES that has become somewhat annoying to me.
My only real point is that I think SACD has a lot of potential, but the available players don't realize it- and it's not just a question of throwing gobs of money at the problem (Sony certainly did that with the contruction of the 777ES), but of better design, more attention to some of the finer points.
Best regards,
Jon
I have an original SCD777ES, and while it was a significant step up on redbood CD as well as bringing some interesting SACD recordings, limitations in it's performance (probably due to jitter in digital section, and a bazillion op amps and coupling caps in analog section) render it ultimate performance for CD and SACD mid-fi- perhaps upper end mid-fi, but still...
I have a DAC constructed with sound techniques, and a NLFB analog output stage, using the CD4390, a "low end" Crystal 24 bit chip. It is clearly better than the 777 on CD, and is good enough to reveal subtle shortcomings of the SCD777ES as a transport compared to a better unit. I'm working on one now with an 8420 for SRC (to 88.2 kHz) and reclocking, with a CS8397 converter, and a more sophisticaed NLFB output stage. SACD disks have some nice qualities, but are still have a signature sound through the 777ES that has become somewhat annoying to me.
My only real point is that I think SACD has a lot of potential, but the available players don't realize it- and it's not just a question of throwing gobs of money at the problem (Sony certainly did that with the contruction of the 777ES), but of better design, more attention to some of the finer points.
Best regards,
Jon
You've got a good point there Jon. Implementation can make all the difference. This is one reason I've been restricting my CD/SACD comparisons to a level playing field (or, at least as level as I can make it) - I am doing playback of both formats on my modified NS500V. This way, I'm listening to the same transport, same clock jitter, same DACs, and same analogue circuitry. Granted, the NS500 may not exaclty be the greatest thing going, but at least that part of the playing field is roughly level.
My homebuilt preamp, which is flat to 5MHz, should be quite indifferent to the presence of ultrasonics, but who knows about the Sennheiser HD-600s. the phones themselves are extremely revealing... far more so than most combinations of amp + loudspeaker + room acoustics. I feel that with this very simple signal chain, I should be extracting pretty much the greatest possible performance out of the NS500, although I may be missing some of the more ethereal qualities of a recording such as the soundstage and general ambiance.
In any case, dorkus has another valid point, and that is the extreme variations in recording quality. Perhaps the only conclusion I feel secure in stating at this point is that the possible variations in mixing, mastering, ADC processes, subsequent DSP operations, etc etc etc are so complex (and totally uncontrolled by the end user) that direct comparison, even between different versions of the same album are exceedingly difficult to call valid in terms of evaluating one data format vs. the other. In this respect, perhaps Mr. Moncrieff had one advantage: some measure of control over the technical details of the implementation on the recording end. Of course, this still doesn't mean that I'm going to take his word for it...
Considering all this, I'm starting to wonder if the better approach is to evaluate on more pragmatic terms: what can I acheive with existing recordings in my listening environment. That is, give each format the benefit of the best possible reproduction chain I can, and just ignore the uneven playing field?
In the end, I'm really very happy with the sound of my SACDs, and can't wait till DVD-As become more prevalent so I can spend more money on equipment and software. Curiously, I've recently purchased the DVD-A release of the Eagles' Hotel California. This album has a DTS track on it (despite the conspicuous absence of DTS labelling on the package), which actually sounds quite good on the NS500 (that DTS decode feature did come in useful, after all! ) With a bitrate of 1500Mbps, the DTS track is apparently mastered at 20 or 24bit, 48kHz, which is better than CD, but then compressed (yuk) using the DTS algorithm. I need to sit down and do further comparison to the original redbook release, but my initial impression is that the DTS is better. I'm tempted to go get the DTS release of another favorite of mine, Sting's Nothing Like the Sun. Anyone else have any experiences to share regarding the quality of DTS music?
Regards,
My homebuilt preamp, which is flat to 5MHz, should be quite indifferent to the presence of ultrasonics, but who knows about the Sennheiser HD-600s. the phones themselves are extremely revealing... far more so than most combinations of amp + loudspeaker + room acoustics. I feel that with this very simple signal chain, I should be extracting pretty much the greatest possible performance out of the NS500, although I may be missing some of the more ethereal qualities of a recording such as the soundstage and general ambiance.
In any case, dorkus has another valid point, and that is the extreme variations in recording quality. Perhaps the only conclusion I feel secure in stating at this point is that the possible variations in mixing, mastering, ADC processes, subsequent DSP operations, etc etc etc are so complex (and totally uncontrolled by the end user) that direct comparison, even between different versions of the same album are exceedingly difficult to call valid in terms of evaluating one data format vs. the other. In this respect, perhaps Mr. Moncrieff had one advantage: some measure of control over the technical details of the implementation on the recording end. Of course, this still doesn't mean that I'm going to take his word for it...
Considering all this, I'm starting to wonder if the better approach is to evaluate on more pragmatic terms: what can I acheive with existing recordings in my listening environment. That is, give each format the benefit of the best possible reproduction chain I can, and just ignore the uneven playing field?
In the end, I'm really very happy with the sound of my SACDs, and can't wait till DVD-As become more prevalent so I can spend more money on equipment and software
. Curiously, I've recently purchased the DVD-A release of the Eagles' Hotel California. This album has a DTS track on it (despite the conspicuous absence of DTS labelling on the package), which actually sounds quite good on the NS500 (that DTS decode feature did come in useful, after all! ) With a bitrate of 1500Mbps, the DTS track is apparently mastered at 20 or 24bit, 48kHz, which is better than CD, but then compressed (yuk) using the DTS algorithm. I need to sit down and do further comparison to the original redbook release, but my initial impression is that the DTS is better. I'm tempted to go get the DTS release of another favorite of mine, Sting's Nothing Like the Sun. Anyone else have any experiences to share regarding the quality of DTS music?Regards,
Well, I think you've hit the nail on the head with this thought- though obviously future developments may change our perceptions (hopefully) about what is possible and how to benefit best from it.
Don't take my remarks the wrong way- I don't think of my self as a detractor of SACD- I'm even considering getting another player just for the purpose of digging into it. I'm also hopeful re DVD A, and have a few 24/192 disks, but there the situation for trying to find a first class player is even worse, it seems, than for SACD!
My reference standard for CD is a unit designed by an old friend of mine, the Ayre D1. Even he admits that he thinks the SACD format has much sonic potential, but he's pretty negative about the commercial posibilities for either SACD or DVD-A, and so they won't invest in the development effort for a player. Because of the issues around encryption, and the inability to get a digital bit stream out of these players (by normal means- but abnormal means is another story!
a high performance outboard DAC is not a ready potential solution.At this time, I'm confining my SACD purchases to Hybrid disks for the most part, then I can choose whether I want the additional openess and detail along with some extra grunge using SACD playback, or playback the CD layer on one of my own DACs. These disks, which use a newer form of SBM to downconvert the DSD to CD, are the closest to being possibly valid tools for evaluation as well as listening, and at least they seem to eliminate some of the other mastering and production variables which even exist between different CD versions of the same album.
Best regards,
Jon
Jon,
As for first class DVD-audio players, I very happy with my modified Pioneer AX10 (I posted the details on another thread in this forum), using a transformer off of the I/V stage that consists of one AD825. The DACs are PCM1704. I wouldn't be surprised if the same mods could be done to their latest DVD player the Elite DV47-A. This unit plays DVD-a, SACD, CDs. It goes for less than $800 or so on Audiogon.
Regards, Robert
As for first class DVD-audio players, I very happy with my modified Pioneer AX10 (I posted the details on another thread in this forum), using a transformer off of the I/V stage that consists of one AD825. The DACs are PCM1704. I wouldn't be surprised if the same mods could be done to their latest DVD player the Elite DV47-A. This unit plays DVD-a, SACD, CDs. It goes for less than $800 or so on Audiogon.
Regards, Robert
I use a pair of Jensen JT-11SSP-8MA line transformers; they're shielded, very well built, and have a natural bessel top end roll off at about 220 kHz. The low end extends to about 0.25 Hz at -3 dB. Their only real drawback is the size (relatively large) and the cost.
I/V design approaches will vary depending on philosophy; I preferr a non-feedback discrete circuit, either complementary common base or common source. But I've never AB'd them against the 825; I hear good things about it from a lot of people.
Regards,
Jon
I/V design approaches will vary depending on philosophy; I preferr a non-feedback discrete circuit, either complementary common base or common source. But I've never AB'd them against the 825; I hear good things about it from a lot of people.
Regards,
Jon
I used a Sowter 3603 (http://www.sowter.co.uk/specs/3603.htm). This worked well for interfacing between the I/V stage and the input to a solid stage line stage. Sowter makes another transformer that is better for a tubed line stage interface (http://www.sowter.co.uk/specs/8347.htm). Jensen and Lundahl also make transformers that can work, depending upon the DAC.
As for the I/V stage, this is typically found with most BB/TI DACs like the 1702/1704 series which output current, not a higher voltage. To get the voltage up you need a I/V stage, which is basically an op-amp. (Look at the PCM1704 data sheet and evaluation board sheet for examples.) The signal is usually then filtered with a second order Butterworth, followed by perhaps another (or several) op-amp used as an output buffer. My Pioneer AX10 had 4 dual op-amps, which I reduced to one single AD825 (for the I/V stage). Some DACs produce more voltage and do not need the I/V op-amp. In this case, you can use the transformer directly between the DAC and your line stage.
Unless you changing your DAC, I'd base the analog output design on what DAC is already in your machine. Here are some other sites that discuss these issues: http://digilander.libero.it/paeng/magnetics_frame.htm , http://www.dddac.de/ , http://www.quadesl.com/dac.shtml , http://hjem.get2net.dk/aaholm_audio/projects/DAC/dac.htm
Hope this helps, Robert
As for the I/V stage, this is typically found with most BB/TI DACs like the 1702/1704 series which output current, not a higher voltage. To get the voltage up you need a I/V stage, which is basically an op-amp. (Look at the PCM1704 data sheet and evaluation board sheet for examples.) The signal is usually then filtered with a second order Butterworth, followed by perhaps another (or several) op-amp used as an output buffer. My Pioneer AX10 had 4 dual op-amps, which I reduced to one single AD825 (for the I/V stage). Some DACs produce more voltage and do not need the I/V op-amp. In this case, you can use the transformer directly between the DAC and your line stage.
Unless you changing your DAC, I'd base the analog output design on what DAC is already in your machine. Here are some other sites that discuss these issues: http://digilander.libero.it/paeng/magnetics_frame.htm , http://www.dddac.de/ , http://www.quadesl.com/dac.shtml , http://hjem.get2net.dk/aaholm_audio/projects/DAC/dac.htm
Hope this helps, Robert
promitheus said:
No, harmonics are tones, if they are over 20 kHz and you can't hear over 20 kHz, then you can't hear them.
Here's a quote of Oohashi & al by James Boyk at http://www.cco.caltech.edu/~boyk/spectra/spectra.htm (sorry, I didn't find a closer source).
This EEG is weird, since it persists after the sound is gone, and the people say they didn't hear a difference. I don't understand how they could rate the sound quality differently while claiming that the sound wasn't affected.
When I think the sound isn't affected, I don't rate it differently.
I can't comment without having access to the original paper.
dorkus said:
Could you privide some backup ? I've heard that audio CD could reproduce phase information with better accuracy than needed for human hearing (from Nika Aldrich, I think).
By the way, what is the phase accuracy of audio CD ? I've got no idea. Couldn't it be into several kilohertz too ?
For Example....
1. A few dB down at 50kHz will cause phase errors down into the audible range.
2. Addition of various ultrasonic waves can and often does create sound in the audible range.
3. Because standard CD's have a 20kHz filter, phase problems will extend way down into the audible range.
1. A few dB down at 50kHz will cause phase errors down into the audible range.
2. Addition of various ultrasonic waves can and often does create sound in the audible range.
3. Because standard CD's have a 20kHz filter, phase problems will extend way down into the audible range.
1. This will depend very strongly on the implementation of the filter. Also, it should be remembered that a perfectly linear phase shift is equivalent to a simple delay, and is therefore irrelevant. This is the case with a Bessel filter, which will have (if properly designed) a very nearly linear phase shift across the audible frequency band. If you normalize the phase plot to remove this linear component from DC - 20 kHz, you will find that most well designed filters produce negligible phase shift of only a degree or two in the audible range, despite having something like 20deg or more absolute phase shift at 20kHz... remove the delay component, and you'll find very little phase shift residual.
2. I can't comment directly on this, as I've never done any experiments myself, and have not read a large enough volume of papers on the matter. I will say this: those ultrasonic motion sensors sometimes used in security systems, along with the h-sync frequency of television sets drive me nuts. I hate the sound, and believe me, some TV's are very LOUD. Some of these security motion sesors are so loud, they can cause a sensation akin to a stabbing pain in my ears if I stand in just the wrong spot. I wonder how much damage is occurring to people's hearing, even though they can't hear these high amplitude ultrasonics?
3. Again, this depends on the implementation. If the original material was sampled at a higher rate, or using sigma-delta converters before reducing down to 16/44.1, then this filtering was done in the digital domain, most likey using symmetric FIR filters. Symmetric FIR filters have perfectly linear phase, and thus produce precisely zero phase shift at all frequencies. The only phase shift that would be contributed to the recorded sound at that point would be from any analogue low-pass placed before the ADC input. FIR filters are not, however, entirely immune to amplitude effects such as ripple in the passband (corresponding to pre- and post-sample impulses). But, in general, these amplitude effects are extremely low in level, and a well designed FIR filter can essentially eliminate them or otherwise mitigate their sonic effect. In any case, the point is, just because a "brick-wall" filter was used to cut off CD audio material at just over 20kHz does not necessarily mean that there will be significant, or even any, phase shifts in the audible band.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.