I am pleased to hear your observations.
My experience is different. I spent years recording live music, where I could step from the audience to the monitor booth quickly. To keep from being bored (four cello recitals in one day) I learned much about imaging from concert hall to speakers (and headphones). Many different mike techniques (almost always two mikes direct to tape) and auditioned on many different playback rigs.
The "best" localization I ever heard was commercial 16-track LP mixes to a cannibalized suitcase recordplayer, two 60FX6 and two Fostex FE103 in cardboard boxes, hung in specific places over my bed. Simplicity is a virtue. Of course "localization" from multi-channel overdub recordings is a fantasy.
I do not think it is possible to do much more than "water color" spatial imaging. Yes, I have heard ear-exciting "images". I have even fooled my dog. (A dancer dropped small beads on stage; my dog in the thought it sounded like food hitting the floor... the only time he reacted to speakers.) But I strongly suspect that most of this is serendipitous alignment of the brain's ear-paths.
You can learn a lot just sitting in the concert hall. The "image" can be very monophonic if you sit dead still. You get much better imaging if you turn your head. (DF96's "few mm".) Note that hunting dogs and cats twitch-big-ears and turn-head to localize prey.
"30dB NFB" is perhaps not the criteria for phase shift. As DF96 says, if the response is flat, the phase is flat. I also believe that if the response is not-flat, the ear "knows" the phase is shifted the same way. The ear becomes very smart about real-world acoustics. When it is not eat-or-starve, the ear seems to be good at finding patterns which are not there. (Probably a vestige of hunting: if there is no clear pattern, focusing on possible-patterns may improve your odds of dinner.)
30dB NFB "is" interesting because in a device with ~~5% open loop THD, 30dB NFB is about the point that high-order IM products decline (<30dB NFB increases IM on multi-tone signals). To my ear this makes a "haze" over the sound. (Interestingly, more now that I have lost much hearing.) Perhaps this relatively well correlated "haze" has components which line-up in ways which suggest localization.
Or I may be an old deaf REC-button civil servant who is full of crap.
There are many examples in the Fullrange and Multiway speaker forum where via FIR filters in DSP, and all-pass filters, phase can be arbitrarily altered but amplitude response vs frequency is flat. I have heard strange speakers with “phasey” sounds that are very fatiguing and prevent correct imaging. Just look at the frequency response of a perfectly flat LR4 XO in phase space - it is far from flat. That’s why full range speakers or first order XO sound better. This is happening in speakers - and same thing can happen with amplifiers. Depends on topology and how it is designed and implemented.
Simplicity is a virtue, but it’s more because you are listening to fullrange drivers, which are inherently flat phase. Look at phase relationship of popular multi-way speakers that use LR2 or LR4, they are not phase flat. That’s one of the reasons why so many people like fullrange single driver speakers. They sound more natural and have exceptional imaging.
Yes, there are multitude of ways speakers and filters change the phase, all of which can be measured and are well understood. There may be some very poorly designed amplifiers that change phase too, easy to find ones that don't, and again, it's measurable, no mysteries.
There is no such thing as a "fullrange" driver 🙂 Find a speaker with a flat frequency response and you may have found one with a flat phase response. 😉
Why is there focus on amplifiers in the replay chain regarding phase? Active and passive xover network effects are a separate issue. The microphone, mic preamp and loudspeaker make the biggest contribution to phase deviations. Also I am not aware of anything in audio that will record phase flat to 300KHz.
Although this is all very interesting, though some of it going over my head, I think I worded my initial question wrong.
What I wanted to know was, if I used two amps in an active system, one a chip amp with minimal components, and one a great bbig pro sound amp with 100s of components, would there be an appreciable time delay difference between them, that the simple polarity/phase adjustment on a minidsp/dcx2496 could not fix.
I.e. Could the phase of one go past 360°? And be unable to catch up with the other.
What I wanted to know was, if I used two amps in an active system, one a chip amp with minimal components, and one a great bbig pro sound amp with 100s of components, would there be an appreciable time delay difference between them, that the simple polarity/phase adjustment on a minidsp/dcx2496 could not fix.
I.e. Could the phase of one go past 360°? And be unable to catch up with the other.
Your first question was fine and answered by rayma.
There has been some confusion since regarding time delays and how they may or may not relate to phase shifts that may or may not be actually happening.
There has been some confusion since regarding time delays and how they may or may not relate to phase shifts that may or may not be actually happening.
How many amplifiers contain an all-pass circuit? You are confusing "can happen" (i.e. it is physically possible to put an all-pass in an amplifier) with 'does happen' (i.e. amplifiers often contain all-pass), or maybe you are hoping that we will confuse these two?xrk971 said:
Amplifiers are minimum phase systems, so flat frequency response means flat phase response so nothing funny happens. In or near the passband the phase shift is unlikely to exceed 90 degrees. Added filters can change this: assume 90 degrees per filter pole.
Hi Hugh, I've got the same observation that's been a surprise for me when I noticed the phenomenon for the first time. My Simpelstark 1.3 with 18db loop gain via ODNF within the audio band demonstrates fantastic localization of sound sources with the high-resolution recordings. A number of highly experienced listeners have noticed the stereo base being wider than the distance between the speakers with the sources clearly placed in 3D space, including the difference in vertical coordinates.
I'm also pretty interested to dive deeper into this point.
Cheers,
Valery
It could be that PRR has a point, similar to how a correlated delayed reflection can aid with the perception of depth and localisation.
That's a good point, however, what puzzles me is why the highly-linear low-feedback amplifiers demonstrate such an advantage in comparison with some very fine low-phase-shift but high-feedback amplifiers (those oned practically don't show any vertical localization with the same recordings).
He mentioned the higher order IM products of amplifiers with lower feedback. Also one needs to define the cues that can help create a vertical image, higher frequencies, for example...
I don't observe any "higher" IMD in Simpelstark. It runs well even with no global loop at all, however, 18db ODNF further reduces the noise and lowers the higher-order components. 0.007% IMD is not the one I would classify as high. This is the live measurement. The level of detail and localization demonstrated by this amplifier is exceptional even with the most "difficult" music material like the big bands, classical music, multi-voice vocals, etc.
An amplifier with 55db global loop gain and even lower IMD - very transparent and low-distortion - does not demonstrate the level of vertical localization as high as Simpelstark.
An amplifier with 55db global loop gain and even lower IMD - very transparent and low-distortion - does not demonstrate the level of vertical localization as high as Simpelstark.
Attachments
Coming back to this, I suggest that whatever is causing the perceived expansion of the image, it is inflating it, if you see what I mean, so that it's expanding in 3D space.
Hi Valery,
Interesting...... what was that adage of designing a good amp from decades past?
'Design a very linear amp, then apply global feedback'. Is this a sign that your SimpleStark has some very good ideas? And maybe even more linearity and a little less NGFB might be the best option? I really like your SimpleStark, it's beautifully designed.
HD
Interesting...... what was that adage of designing a good amp from decades past?
'Design a very linear amp, then apply global feedback'. Is this a sign that your SimpleStark has some very good ideas? And maybe even more linearity and a little less NGFB might be the best option? I really like your SimpleStark, it's beautifully designed.
HD
It's an option, but why the best? What is the mechanism that causes it to produce a better soundscape and why?
You might recall, Scottjoplin, that first I came to this thread asking the smartest guys in the room myself. Do you have any ideas? I really have no idea.
Time delay is measured in seconds, but phase is in degrees. If you want to preserve the wave shape, you need a constant time delay with frequency, which results from a linear phase curve. That is, the phase shift must be linearly proportional to frequency at all frequencies of interest.
Amplifiers do have a small time delay due to their limited bandwidth. If the bandwidth is defined by a single time constant tau, then the time delay for signals well within the amplifier's bandwidth is -2 Pi x tau. For signals band-limited to the audio band, the time delay should be nearly constant with frequency, and the waveform will be preserved, for amplifiers with bandwidth of around 100kHz or more.
Last edited: