Ultrasonics and DSD

Ultrasonics and DSD....
Actually read something on the web a few weeks ago in which a Scandanavian engineer (Can't remember his name, especially with all those bizarre syllabic combinations) was attacking the DSD process and claimed SACDs were inferior to DVD-As due to the ultrasonic distortion issue. Funny, ultrasonics and their possible effects on the listening process have long been an issue of debate. I can't claim to have ever heard things I shouldn't have been hearing due to ultrasonics, but I do remember when I was younger hearing (ar at least imagining I was hearing) the "dog whistle" on the choruses of "I Know There's an Answer" on Pet Sounds. Now with my hearing greatly diminished from middle age (and playing in too many loud bands), I certainly cannot hear the whistle anymore (Doubt if I can hear anything much above 10 K anyway!) Yet I still hear a kind of strange presence during that musical passage - and at other times - which is difficult to explain. While I haven't heard anything really weird on most SACD's, there are a few tracks on some of the Stones' hybrid-layered stuff which seem to contain things that sound strangely like overtones or barely definable sounds which shouldn't be there. Perhaps I'm being paranoid (the self-fulfilling prophesy, etc), but I'm often right about things like this (even the producer and engineer in the studio I am working in usually rely on my judgment when it comes to sounds like this). The effect of ultrasonics and overtones with regard to the DSD process has been mentioned before, but in this case, there were no conclusions offered, other than a sort of debate between this engineer and people at Phillips defending DSD. Is it possible that we could actually be hearing things due to the DSD process which relate to ultrasonics? Wonder if Steve (or anyone else) has some answers?

 

"was attacking the DSD process and claimed SACDs were inferior to DVD-As due to the ultrasonic distortion issue"

I have studied this in detail. There have been white papers put back and forth from Lipshitz and Vanderkooy and an opposing team, Dirk Reefman and James Angus. Summaries of the debate can be viewed on the AES website and technical papers can be purchased there as well. After much debate, they concluded that newer implementations of DSD and adding "dither", reduce the noise floor to such an extent that DSD artifacts are practically impossible to hear. The audibility of the ultrasonic noise is actually disputed by many audiophile engineers like John Atkinson who claims one cannot hear it.

From an equipment standpoint, on some SACD players you have to be careful what amps and preamps you use as some ultrasonic sound can cause damage. For most of the high end industry, this ultrasonic sound is being exploited to add more "air" to instruments on playback. You are starting to see "supertweeters", etc. as part of the landscape.

In short, this is added detail from the newer hirez technology and a very good thing.

 

White papers must be taken with a grain of salt.

Consumers will ultimately decide which format sounds better.

 

If you want to feel stupid just take a glance at those papers. I had to read them during college. Still trying to figure some of them out.

Seriously, if you're interested in this topic you need to read these articles. Thanks for pointing this out Lee.

 

I wonder if the scandinavian engineer wasn't, in fact, complaining about the noise shaping DSD requires. While this is added intentionally and is not an unintended byproduct of the process; it could still, I suppose, be termed "distortion." As a 1-bit coding system, DSD has a theoretical S/N of only about 6dB. Several methods are employed to remedy this, the most drastic of which is noise shaping. This noise shaping moves enough quantization noise out of the audible band to make DSD a viable format; but since this noise can only be moved, not removed, it ends up in the ultrasonic region. Furthermore, since the total power of this noise doesn't change )only its spectrum changes), there ends up being quite a bit of it up there.

I can't be entirely sure without reading the article myself, but I'd suspect that this is he is complaining about. I can think of no other reason that DSD (or hi-res PCM, for that matter) should behave differently simply because they've moved beyond 20kHz.

 

Technically speaking, ANY sound not present during the original recording session is distortion. But whether or not it sounds good is very subjective.

[T]

 

For quite a while I left my SACD player on 'Standard' setting which attenuated the hyper-ultrasonic frequencies. I finally decided a while ago to switch to 'Custom' and have never gone back. I do feel there is a presence of air in presentation that was missing when the attenuating filter was engaged.
Sony was concerned that some (very rare cases) amps could be sent into "oscillation" thus causing damage but I don't think that this would be common.

 

I have read some of the papers, but I'm not a rocket scientist. I understand that the noise shaping is supposed to move the noise to the ultrasonic range, where it can be dealt with by relatively simple analog filtering. So why not just filter out most of it? How much ultrasonic garbage actually comes out of an SACD player, and how much of a real worry is it, in terms of possible equipment damage?

 

"it ends up in the ultrasonic region"

Yes, but the Angus/Reefman papers show that the noise floor drops to minus 160 db with dithering applied. It's not going to effect even the most high end of systems.

"I finally decided a while ago to switch to 'Custom' and have never gone back. I do feel there is a presence of air in presentation that was missing when the attenuating filter was engaged."

Same here with my SCD777ES. Never looked back. No blown Maggies yet.

 

"how much of a real worry is it, in terms of possible equipment damage?"

I'd suspect that there really isn't that much danger. Most amplifiers are band limited in order to keep them from oscillating when driving tricky loads. The rising inductance of dynamic tweeters tends to help keep them out of trouble as well. I can't speak for large planar ribbons like Maggies, but electrostats and ribbon tweeters are often (nearly always, in my experience) driven via transformers which also act as a low-pass filter yielding further protection. So, barring a confluence of particularly poor designs, I doubt there is need for great deal of concern.

 

Probably not a big worry Gerry as you suspect...there are some reports of older tube amps getting fouled but even that depends on brand, circuit topology, etc.

 

Lee,
Don't forget, the"added detail from the new hi rez technology" you mention on the top end is also ultrasonic distortion. I'm not complaining about the sound of SACD's, which I find to be generally more natural than DVD-A's. But ultimately there must be some distortion that gets into the lower audio spectrum as a result of the ultrasonic problem...this is inevitable since any kind of overtone affects the listenable spectrum, even if only minutely. i know this has been debated up the wazoo, but I have yet to find any scientist or engineer who has been able to refute this point.

 

"i know this has been debated up the wazoo, but I have yet to find any scientist or engineer who has been able to refute this point."

Perhaps you are right, but I am not so sure at this point. I have yet to find a recording engineer including myself that can hear a real audible amount of lower spectrum distortion. Me thinks this is one of those things that looks like a problem in theory but is not really audible.

 

And you may also be right, Lee....Still I can't help thinking I'm hearing a few just slightly strange things on SACD now and again. Overall, DSD is the way to go, though, I think.

 

"Overall, DSD is the way to go, though, I think."

Agreed. I have really gotten back into music with this format. I was suffering on all but the best of redbook.

DSD just pulls me in somehow. I like DVDA also but we need more 24/192 recordings.