Good Article CD vs. SACD vs. DVD-A
Does the author not seem a bit, no pun intended, biased to you? If you take note he frequently mentions how the noise created by sacd is within 40db of maximum output at "just above" 100KHZ and PCM is 144db below from dc-100 KHZ. Thus, there is no common denominator. Propaganda people?

Happy Listening,
Why is it Propaganda. Did Sony/Philips not sell us the big Lie Once Before. Perfect sound forever. WAKE up people.
Here's what my ears tell me: SACD beats normal redbook CD easily in the midrange and in the lower registers, but there is seriously something wrong with the top end. I't somehow dead, lifeless and something, hard to define unpleasant, unnatural in its rendering! At any rate - I have plasma tweeters from Acapella - I suffer from listeners fatigue after too long exposure to SACD, which I don't with vinyl nor generally with well recorded CDs and my SACD playback isn't cheapo, but pretty much SOTA. Cannot say if my experience has something to do with the findings in the above mentioned article, but John Atkinson has in his testing of an Accuphase player published results, which point in the same direction. So, sadly, I've come to doubt, that SACD is the great breakthrough to finally fullfill the promise of CD. Natalie may well be right!
SACD sounds better than most CDs because the technology used to record in SACD is the same as CDs today. In fact most of the SACD converters that I have seen are based on the same single bit Delta Sigma A/Ds as are used in most PCM converter boxes. They simply grab the stream before the DSP that converts to PCM. So the media on the CD is the same BUT there are more digital conversion steps, Each one of which has some loss (not much but some). This means that CDs sound worse than SACDs if you use the same equipment to record them. However this is not a problem of the format. There are other ways to "fill the bits" than a single bit Delta Sigma A/D converter but these all cost much more money ($500 a chip for a 18 bit A/D instead of $5 for a pretty good single bit Delta Sigma A/D). Because these chips are cheap and most people can't hear the difference on their home stereos almost all CDs are mastered on these chips. In fact most CDs are mastered on DAT machines simply using the analog inputs!! (Not many of which contain very good analog stages much less good A/D converters and they are typically run through these boxes multiple time to "MIX" them down to a CD.) At least SACDs are mastered on dedicated equipment that is much better designed. I have heard some carefully mastered recordable CDs converted from master tape (just recorded of a CD recorder with digital in) using a direct conversion A/D converter (these converters directly measure an input signal instead of generating a stream of pulses that are then digital filtered) capable of filling a CDs full resolution and played back on a Direct conversion DAC that sounded almost exactly like the tape. In contrast the commercial master of the same material made on a DAT machine sounded just like any other commercial CD.

The resolution of a digital system is easily calculated. For PCM represented with binary numbers you can figure the resolution by raising 2 to the number of bits in a single word. An 8 bit system has 256 possible levels, a 16 bit system has 65,536 possible levels, and a 24 bit system has 16,777,216 possible levels to describe the sound. This resolution holds to the nyquist frequency with no degradation (Nyquist is half the sampling rate e.g. CD is 22050Hz.) A 1 bit system like 1 bit Delta Sigma or SACD has only 2 possible levels therefor you must average some pulses to reproduce any given level. If I want a 1 bit system with a resolution the same as an 8 bit system I would need to average 256 pulses, a 16 bit system would need 65,536 pulses to be averaged. (For 8 bit resolution, If I want a "1" I would turn on 1 pulse and turn off the 255, if I wanted a "200" I would turn on 200 pulses and turn off 56, ect...) So to calculate the resolution of a 1 bit system at any frequency just calculate how many pulses it can produce at that frequency. SACD uses some tricks to increase the resolution of REPETITIVE signals by shaping an imperfect representation of the repetitive signal many times and correcting the imperfections from the previous wave. Signals that do not remain long get NO BENEFIT from this noise shaping. The noise shaping also adds lots of noise to the converted signal (the noise is the "leftover" distortion from each of imperfect conversion steps). To avoid swamping the audio band with this noise the converter makes sure that most of this noise is above 20,000Hz. Multi-order noise shaping (usually 3-7 order e.g. SACD is 7th order) pushes more of this noise above the hearing range at the expense of a super high noise floor above 20Khz and low system stability (seventh order systems must be monitored for instability continuously and reset often). This instability normally shows up as severe damage to the low frequency part of the signal! The portion that is normally very good in single bit systems. Single bit SACD attains higher resolution than CD but only for low frequency signals or repetitive signals.