You gotta love Stereophile. Or not. I used to subscribe and read it as entertainment, then at some point, realizing how serious they were, just didn't find it as entertaining anymore.
The article misses a few key points regarding soundstage, and the fundamental problems with getting it to work. It focuses on several miking techniques and theories on how to capture a soundstage, but pretty much ignores the reproduction side. Taken as a whole system, though, the issues start to pop out.
Consider the scenario of listening to a live event, as an audience member. Take a single sound source an consider how sound gets to your ears, the direct path, and all the reflections, the angle it hits you at, both azimuth and elevation, and how that's different for each ear. Consider the difference in all of that for each sound source on the stage, and you begin to see what it actually takes to hear the positions of each sound source, the size and shape of the room, and your seat position.
Now take a recording of that event, it's not all that important what mic technique is used, and play it on a two channel stereo system. Now plot the direct and reflected paths for each sound source. You'll see they are completely different from the original performance, and mostly, there are really two sources, the speakers. The recorded direct and reflected sound arrivals have now had their relationships changed by virtue of being played through only two speakers in a different, probably smaller, room. There's no hope of recreating the original that way, so all recording engineers try to do is present a representation that's artistically acceptable, and asks for only a bit of the suspension of disbelief. They don't even attempting faithful capture of direct and reflected sound by a pair of mics that mimic ears in position and directional characteristics, simply because if you squeeze all of that through a pair of speakers in a room, all the special spacial hearing cues are scrambled by the speaker and room configuration. So what's done is by using multiple mics, or multiple stereo pairs of mics, and perhaps, depending on music type and venue acoustics, several "spot" mic on individual instruments, plus real and artificial reverb, they capture enough sound information to create a new stereo soundstage that works on a pair of speakers, but is its own reality, not a faithful re-creation of the original.
We can translate this problem to headphone listening quite easily. Since most recordings are made to be played on a pair of speakers in a room, placed in front of us at a 45 degree angle or so, squirting those same two channels into headphones presents an entirely different reality, bearing little resemblance to the original, or the same mix on speakers in a room. Each ear now hears only its own channel, and pretty much all the HRTF stuff has been shorted out. Center images are now in the middle of the head, etc.
So how do you record and reproduce a soundstage? The answer came first from Harvey Fletcher of Bell Labs back in 1934. He suggested that if you put an orchestra on stage and set up an acoustically transparent curtain in front of the stage and covered it with an infinite number of infinitesimal microphones which would then feed a similar curtain in another similar hall, but this time populated with "sound projectors" that would project sound with the identical intensity and frequency of the original sound, you'd have the perfect soundstage (though the term wasn't used). But Fletcher wasn't really that impractical. He referred to a paper by Steinberg and Snow that suggested the minimum number of microphones, channels and speakers required to reproduce a believable soundstage, including depth, was three. Not two, three. So why did we get stuck with two channels when we really need three? It was the practical matter of distributing sound on disc, specifically the record. Early attempts at two-channels were difficult, requiring two styli in two grooves, but the final 45/45 groove idea was the only practical way to get it done. So, we have two channels, not because it's good, pure and right, but because that's the best they could do at the time.
And that, basically, is why we can't seem to produce a believable and palpable, reliable soundstage with speakers in a room. Not enough channels. Of course, multi-channel music on DVD Audio and SACD should have solved this, but the idea had marketing issues that were just hard enough to surmount that the idea, still around, never really got going. I dearly love the 5.1 music DVDs and SACDs I own, though.
But, this is Head-fi. So we only have two ears, and pretty much only two earphones fit. Binarual to the rescue, right? To follow the Stereophile format, yes and no. Binarual recording, where mics are placed in either real ears on a person or artificial ears on a dummy head, work well, but there are just enough differences between individuals that the don't work equally well for everyone. When they work, they do present a dimensionality, and soundstage more real and palpable that any other method. So why don't we have more binaural recordings? Two reasons: They sound just terrible on speakers, and the afore mentioned inconsistent results. Everyone's ears, head and chest are just different enough that total realism is not available to everyone with every recording. They pretty much have to be done just for us. Stuff tiny mics in your years and record something, then play it back on headphones and you are darn close to reality. But give the cans to your buddy, and he may not quite believe he an reach out and tap the conductor on the shoulder.
Now, all of those points relate to soundstage, and ways to make it or break it. The are all huge things, number of speakers and mics, placement, binaural recording, nothing minute about those being different from stereo, or each other. But when we are talking about a DAC or amp affecting soundstage, the actual differences between units is very tiny. So where do these differences come from?
Mostly, expectation bias. Sorry to say, it explains pretty much all of it. If you were to do a real ABX test on a pair of amps (actually not hard if you have the right switching gear), gain match them and then attempt to match A and B to a randomly chosen X, the vast soundstage differences simply vanish. Sort of sad, really, because hearing a soundstage improvement from an amp is, well, fun. ABX tests of DACs is hard to do because a computer will generally only talk to one DAC at a time, and you have to be cautious of and time delay in a DAC giving itself away. But if we could do the ABX test on DACs, it's my strong suspicion that the results would be similar to amp tests.
So if you hear a great soundstage with a particular DAC/Amp combo, just enjoy it. Don't try to figure it out, you'll only burst the bubble.
As to triode tube amps simulating 3d sound, well... we'd have to do an ABX test to see if there was a difference or if just knowing there are hot bottles at work does the trick. There are differences in performance of tube amps, but it's been proven many times that once characterized, the same results can be had from SS designs. This is a long other topic, and we started with soundstage. So perhaps the SS/Tube thread?