Sound-stage is based on volume (not phase dependent), ear to ear timing
cues (also not phase dependent as long as shift is same for both
speakers). That also primarily happens much less than 20KHz.
I
don't think I have ever read any experiment that shows the ability to
differentiate anything but quite significant phase shifts.
@roberttdid
This is all true. However there is a reason that wide bandwidth is a sought-after thing going well back (Stewart Hegeman, designer of the original H/K Citation series, felt wide bandwidth to be quite important). That reason is simply while the ear is insensitive to phase in sine waves, triangle waves and the like, it does use phase information for echo location and so does respond to phase information when a spectrum of frequencies are involved. It also interprets phase shift as a tonal coloration; a change in FR one or two octaves outside of the audio passband can be heard as a tonality in the audio band, with little regard for the loudspeakers involved. I had this demonstrated to me in spades about 30 years ago when I was sorting out an MFA Magus preamp in my shop; the complaint was brightness in the phono. It turned out to have an extra timing constant that set the phono EQ to flat at 50KHz. This was clearly audible as a brightness. By simply removing the components responsible in the EQ network, the RIAA slope was restored (at 50KHz) and the brightness removed. When the RIAA curve was tested using an inverse RIAA network, it was perfectly flat up to 20KHz before and **after** the change. I passed this information back to MFA and they incorporated it into their later production of that preamp.
In high end audio we are always concerned about the nth degree; if you really want to get soundstage right, you need wide bandwidth in the electronics even though the speakers won't reproduce it. And we also want the tonality to be correct; especially in an amplifier with a higher output impedance like an SET its going to have FR errors; it will do it no good if it sounds dark on top due to a rolloff above 20KHz.
Alternatively you can employ enough feedback so that the amplifier can compensate for phase shift in the audio passband. But to do that you need north of 60dB of open loop with good Gain Bandwidth Product and good phase margins, since you'll need to blow off 35dB of that gain (if we are talking about a power amp) with feedback. This is next to impossible with tube amps and very difficult with traditional solid state designs (to my knowledge the Benchmark is one of the very few to do this). The most common amps that employ this much feedback are self-oscillating class D amps (mostly designed by Bruno Putzeys).
For example, one ’higher power output single tube’ amp has Bandwidth listed as 10Hz through 60kHz.
@david_ten Almost any power tube has bandwidth from DC to several MHz at least. We have 40MHz bandwidth in the output section of our amps- we intentionally limit the bandwidth in the voltage amplifier. The real issue is the bandwidth of the output transformer (which we haven't got).
The power bandwidth and the low power bandwidth are two different things BTW. I'm of the opinion that they should be nearly the same. But in many amps they are not- its common for the 1 watt bandwidth to be much wider than the full power bandwidth. So it might be a good idea to get some clarity about what is meant by
Bandwidth listed as 10Hz through 60kHz.
