Hi all,
I was thinking about this yesterday and couldn't find any other threads that answers my question.
Given that each element in a crossover shifts the phase, as does each op amp in an amp, do different amps have different group phases when all elements are taken in to account?
Different amps will have different amount of components in them, and if the crossover is passive/after the amp this doesn't really matter.
But if you have an active system with a different amp for each band, surely that could cause big differences in the phase of each section?
Thanks, Will
I was thinking about this yesterday and couldn't find any other threads that answers my question.
Given that each element in a crossover shifts the phase, as does each op amp in an amp, do different amps have different group phases when all elements are taken in to account?
Different amps will have different amount of components in them, and if the crossover is passive/after the amp this doesn't really matter.
But if you have an active system with a different amp for each band, surely that could cause big differences in the phase of each section?
Thanks, Will
There's enough feedback in most amplifiers to keep their phase deviation minimal,
except at low frequencies in some cases.
Last edited:
Thanks, I understand chip amps and their use of feedback, are you saying that discrete amplifiers will have feedback that runs from the output stage, and feed back to the input stage? Thereby "balancing" the signal?
Yes, most ss amplifiers have global negative feedback from the output back to the input stage.
This will level out deviations in phase and frequency in the audio band pretty well.
However, the large capacitor (going to ground on the lower end of the nfb resistor) will cause
both bass roll off and phase deviation, if it is not large enough (compared to that resistor).
Most ss amplifiers use this capacitor, to minimize output DC offset. That pole should be set
to around 2Hz for best results.
Last edited:
Hmmm... for me, amps with zero global negative feedback can have great flat phase (within a few degrees) out to 300kHz or more, where phase shifts equivalent to a few microseconds (which is audible from psychoacoustic standpoint) are important for maintaining soundstage and imaging. We hear frequency to 20kHz but discern phase out to several hundred kHz.
See Fig. 5
https://pdfs.semanticscholar.org/f458/2b8bde4c805b416c8fd120adefa69f426377.pdf
See Fig. 5
https://pdfs.semanticscholar.org/f458/2b8bde4c805b416c8fd120adefa69f426377.pdf
Last edited:
So long as the two amps have the same phase shift the interaural time difference will remain the same
The phase shifts are important even for spatialization via pinna-induced shifts for mono localization. Even if same on both ears, it’s not the same as no phase shift.
I'm sorry, I can't get my head around that sentence 🙂 Firstly, what do you mean by mono localization?
A mono ear can localize up down front back via phase shifts induced by pinna (earlobe). Those phase shifts are of order 2-10usec. Note absence of symmetry of earlobe.
I see, yes, I'm aware of that. So, correct me if I'm wrong, a recording with that information embedded would have the spectral content effectively repeated a number of times with required phase shifts?
Usually, the slope of the frequency response indicates the amount of phase shift. For amplifiers that means that phase shift happens below 20Hz and above 20KHz. DC amplifiers may have no phase lead at low frequencies but you can't do that at the high end. Digital filters can overcome this rule by using negative time, ie memory and overall delay.
Here is a test for whether your system can localise height etc Online LEDR™ Sound Test | Listening Environment Diagnostic Recording Test
NFB is a tool, not a result.
If an amplifier is "flat over the audio band", and has no wacky parts(*) inside, the phase response will be flat.
So a tube amp with low/no NFB can still be flat phase. True, it is hard to make a Power amp LOW phase error over the extended 20-20KHz band. And output impedance, even small, works with driver impedance bumps to make amplitude slopes which means phase error. But in basic audio (not radar-location as some image-seekers want) the errors of amplitude response are the bigger problem.
One thing to watch: a "box" can be Inverting or Non-Inverting. The majority are NI. But a mix of Inv and NI will want careful connections of the individual drivers.
(*)Primarily significant time delays, which don't make sense in amplifiers.
If an amplifier is "flat over the audio band", and has no wacky parts(*) inside, the phase response will be flat.
So a tube amp with low/no NFB can still be flat phase. True, it is hard to make a Power amp LOW phase error over the extended 20-20KHz band. And output impedance, even small, works with driver impedance bumps to make amplitude slopes which means phase error. But in basic audio (not radar-location as some image-seekers want) the errors of amplitude response are the bigger problem.
One thing to watch: a "box" can be Inverting or Non-Inverting. The majority are NI. But a mix of Inv and NI will want careful connections of the individual drivers.
(*)Primarily significant time delays, which don't make sense in amplifiers.
The next question would be, how many recordings have the necessary information?
>> A mono ear can localize up down front back via phase shifts induced by pinna (earlobe).
> ....a recording with that information embedded ....
Why would a recording have YOUR pinna response in it?
(Yes, there are people who put midget mikes in their ears at concerts; I doubt you are one.)
> ....a recording with that information embedded ....
Why would a recording have YOUR pinna response in it?
(Yes, there are people who put midget mikes in their ears at concerts; I doubt you are one.)
I am just saying that the ear can discern microsecond phase shifts in recorded material that were recorded with open air stereo mics (no pinna). So if an amp can preserve that phase relationship, it’s preferable to one that can’t. There are binaural recordings made with mics inside a mannequin head with 3D printed “generic” pinna.
PRR,
XRK and I have noticed that reducing the global fb in any amp to a level of no more than 30dB brings out the spatial cues in the recorded music. We notice depth of image, particularly with large sound fields like orchestra where the positioning of instruments are generally known. Of course, we need good recordings. This is a significant phenomenon for us and quite real. XRK mentions very little phase shift in amplifiers might be related. I do not know why, but you, an august, experienced fellow, might know something about. But we know what we hear, and perhaps you might be able to explain this.... I'm sure you know quite a bit about this, and some of your posts here over many years have been very helpful to me, and many other people. Any thoughts?
Cheers,
Hugh
XRK and I have noticed that reducing the global fb in any amp to a level of no more than 30dB brings out the spatial cues in the recorded music. We notice depth of image, particularly with large sound fields like orchestra where the positioning of instruments are generally known. Of course, we need good recordings. This is a significant phenomenon for us and quite real. XRK mentions very little phase shift in amplifiers might be related. I do not know why, but you, an august, experienced fellow, might know something about. But we know what we hear, and perhaps you might be able to explain this.... I'm sure you know quite a bit about this, and some of your posts here over many years have been very helpful to me, and many other people. Any thoughts?
Cheers,
Hugh
An amp with a flat frequency response over the audio band can preserve that phase relationship quite easily, you've yet to show it can't.
Significant phase shift starts about a decade in frequency away from a first-order rolloff (e.g. 200Hz, 2kHz for your examples). Fortunately this doesn't matter too much.steveu said:
For an amp to preserve the soundstage it needs the two channels to be the same, which is almost always true. Simply moving a mike by a few mm will add much more phase shift than almost any amp.