Amp Phase - Am I going mad?

[tubeswell]

Remember also that flux and magnetising current lag applied voltage by 90 degrees in transformers, and that induced EMF (which is proportional to the rate of change of flux) is 90 degrees out of phase with the applied voltage, therefore the secondary voltage is 180 degrees out of phase with the applied voltage. Furthermore, because of interwinding capacitance, the phase shift can be different at different frequencies in different output transformers that have different numbers of interleaving.

[Malcolm Irving]

Not sure if that’s quite right. I think the induced EMF has to be in a direction that opposes the applied voltage. Think about using one battery to ‘oppose’ another – you would have to connect positive to positive and negative to negative (in phase in other words) otherwise the two batteries would ‘assist’ each other causing a huge current.
I think the ‘rate of change of flux’ leads the flux by 90 degrees. So -90+90 = zero and the induced EMF is in phase with the applied voltage.
You can make the secondary voltage in phase or 180 degrees out according to how you connect the wires, of course.

[pdf64]

... because of interwinding capacitance, the phase shift can be different at different frequencies in different output transformers that have different numbers of interleaving.

Yes, phase shift in the OT can be a big issue, I think it may be a key reason why guitar amps can't accommodate the amount of global negative feedback used in hifi amps (thankfully), and even with the levels used, the open loop response needs shaping to keep the loop stable, eg the small snubber caps across the LTP plates of a 5F6A and its descendants.
Over on TGP, Jeff Gehring has reported on stability issues caused by the little 022848 OT used in Bandmasters etc, when the snubber caps are removed ('blackfacing' eughh)
https://www.thegearpage.net/board/index ... t-19447469
https://www.thegearpage.net/board/index ... /22725183/

[cxx]

...Wait, I think I'm getting what you have been saying. Thinking of a signal as composed of many frequencies, and phase shift being a delaying effect, it alters each of the components of the signal. They won't all be multiples of the fundamental and won't be 180 degress out when the fundamental is. I keep thinking of the signal as a single sine.

Yes, that's my perspective on things.
And even for a more complex wave eg square, where the fundamental frequency has been shifted 180 degrees, when a component of the signal is an integer multiple of the fundamental, the phase shift will be 180 + (n x 360) degrees (ie those components would be 'out of phase' but with a phase shift much bigger than 180 degrees).

I think it's the conflation of 'signal' with 'sine wave' that led to experts / textbooks of the golden age using 'out of phase' and 'polarity inversion' interchangeably, and so to the confusion that inevitably results when that conflated terminology became used by those outside of the original intended audience, eg those who haven't studied the textbooks (no offence intended ).

None taken.

Guitar is a six string instrument, so I suppose that the signal could be a pretty complex combination of fundamentals and their overtones making it difficult to find an appropriate period for a phase shift that would cancel out the original signal.

[Reeltarded]

Hat is off to the engineer who can imagine how exactly to use all this jargon and convert it into numbers that become a screaming monster. I have to think about things for a year at a time then solder it all in and play it.

Making a good amp is at first about staying away from known bad ideas and then doing as little as possible to get as close as you can to where you wanted to be. You guys are my Uber. hehe

[norburybrook]

...Wait, I think I'm getting what you have been saying. Thinking of a signal as composed of many frequencies, and phase shift being a delaying effect, it alters each of the components of the signal. They won't all be multiples of the fundamental and won't be 180 degress out when the fundamental is. I keep thinking of the signal as a single sine.

Yes, that's my perspective on things.
And even for a more complex wave eg square, where the fundamental frequency has been shifted 180 degrees, when a component of the signal is an integer multiple of the fundamental, the phase shift will be 180 + (n x 360) degrees (ie those components would be 'out of phase' but with a phase shift much bigger than 180 degrees).

I think it's the conflation of 'signal' with 'sine wave' that led to experts / textbooks of the golden age using 'out of phase' and 'polarity inversion' interchangeably, and so to the confusion that inevitably results when that conflated terminology became used by those outside of the original intended audience, eg those who haven't studied the textbooks (no offence intended ).

None taken.

Guitar is a six string instrument, so I suppose that the signal could be a pretty complex combination of fundamentals and their overtones making it difficult to find an appropriate period for a phase shift that would cancel out the original signal.

Guy's for the last time it doesn't matter how complex the signal is if they're time aligned and you flip the phase on one the combined output will be zero--SILENCE...

as i said this is how audio engineers test/prove if two files are identical or not


3:15 is a full orchestral track, they null to total silence https://www.youtube.com/watch?v=A_BuXzQ9gzg


Marcus

[martin manning]

...it doesn't matter how complex the signal is if they're time aligned and you flip the phase on one the combined output will be zero--SILENCE...

When you say "flip the phase," I think what you mean is "invert," which is flipping the polarity. For a perfect sine wave a 180-degree phase shift is the same as an inversion for all practical purposes, and as pointed out sine waves are used for analysis because they are manageable mathematically. The original query was about the polarity of the output signal with respect to the input, which is determined by the number of inverting gain stages in the signal path. In the analog world those inversions are imperfect, including phase shift and distortion. It's true that the terminology is confused with terms like "phase inverter" being in common use, but phase and polarity are two different things.

[norburybrook]

yes, it has drifted off topic from the original question. I thought it was interesting though to talk about phase generally as it's something I work with everyday in the studio.

flipping the phase is a common Audio engineering term used in recording studios.

It's dragged on because a few seem not to believe that it doesn't matter how complex a signal is if the phase is reversed 180 deg on any identical source it will null, it doesn't matter if it's a sine wave or a full orchestral mix as demonstrated quite clearly in the video I posted, pdf64 said a complex signal would cause comb filtering in that scenario which is wrong.


I think polarity and Phase are terms that are interchanged quite often in many parlances and that's where some things get confusing.

Marcus

[martin manning]

flipping the phase is a common Audio engineering term used in recording studios... ...I think polarity and Phase are terms that are interchanged quite often in many parlances and that's where some things get confusing.

Yes, and that's unfortunate. A quick google will turn up many attempts to clear the misunderstanding, and we are right in the middle of another!

It's dragged on because a few seem not to believe that it doesn't matter how complex a signal is if the phase is reversed 180 deg on any identical source it will null, it doesn't matter if it's a sine wave or a full orchestral mix as demonstrated quite clearly in the video I posted

I'm with you 100% here, as long as you replace "the phase is reversed 180 deg" with "inverted." That's the terminology getting in the way again.

pdf64 said a complex signal would cause comb filtering in that scenario which is wrong.

This?

But in this context (pro audio equipment, analogue or digital), my understanding is that 'flipping the phase' actually means inverting signal polarity, the 'phase' switch just inserts an inverting unity gain stage into the signal path; nothing actually to do with the phase of the signal. To actually affect phase, just nudge one of the tracks in time one way or the other, the result (EDIT when the signals are mixed equally together) being a comb filter.

Pete is correct here, I believe. In a complex signal an inversion is not at all like a shift of 180 degrees, in fact any phase shift is nonsense unless a particular frequency is identified. In a comb filter the amount of time shift determines which frequencies are nulled.

Not trying to bust anyone's chops as we say here, just trying to clarify.

[norburybrook]

yes and with digital audio there is no time shift at all so identical sound sources regardless of their complexity will null to zero when their Phase/polarity is reversed 180deg.

phase.png

you can see it doesn't say polarity on the mixer it says phase 180 deg so that's where I'm coming from , two identical sources on each of those tracks will null to silence when the phase is switched to 180deg.


We need to go back to building amps now martin. I've a feeling I might need to ask you a question about my bête noire----- relays


and I have a new album out soon.



marcus

[pdf64]

Guy's for the last time it doesn't matter how complex the signal is if they're time aligned and you flip the phase on one the combined output will be zero--SILENCE...

as i said this is how audio engineers test/prove if two files are identical or not


3:15 is a full orchestral track, they null to total silence https://www.youtube.com/watch?v=A_BuXzQ9gzg

As is common in engineering, you, equipment manufacturers, and the speaker in that vid are using the the term 'out of phase' when what is really meant is 'polarity inversion'. It is very unfortunate but it's just an example of the many misapprehensions we're stuck with.

Please see schematic of how a 'phase' switch is implemented https://imgur.com/a/WyIGl
IC5a is the relevant stage, and as I hope is readily apparent, it is configured as a unity gain, inverting amplifier.
Due to HPF and LPF, at ~6Hz and ~340kHz, the phase will indeed lag and lead 45 degrees, extending to 90 degrees the further out we go. But over the audio passband, its phase shift will be 0 degrees (as near as dammit!).
Nothing in the IC5a circuit will change the phase of any frequency by 180 degrees.
Yes, for audio equipment, almost certainly every 'phase' switch is mislabelled; they actually invert signal 'polarity'.

Frequency, phase and time are inextricably intertwined.
Frequency (f) is how many time per second (t) a simple continuous signal repeats.
Hence f=1/t
Each complete repetition = a movement of phase of 360 degrees
so at 20 Hz, each 360 degrees takes 50ms, hence 180 degrees 25ms

But at eg 750Hz, each 180 degrees requires a different shange in time, 0.0666ms in this case.
If we add the signals together to create a slightly complex continuous signal, I hope you can see how utterly impossible it is to simultaneously apply a 180 degree phase shift to both components of that complex signal?
We can either apply a 25ms (or an integer multiple thereof) change to get an 'apparent inverse' of the 20Hz component, or a 0.0666ms (or integer multiple thereof) change for the 750 Hz component.
The only way to null this (and any complex) signal is to invert its polarity and mix it equally with its non-inverted self.

I can kinda grasp how DSP could implement frequency shaping with 0 phase shift.
But how DSP could apply a 180 degree phase shift to a complex signal, and what the result would be, is more than my rather limited understanding can conceive.
For the moment I'm sceptical; prove to me that it's not just inverting the signal

When we move to using 2 mics, both their relative polarity and proximity matter when the outputs from them are combined.
Due to the speed of a sound wave in air (~340m/s), any distance between them will cause a time delay.
I'm a bit hazy on the numbers here, but if the mics are in the same polarity and are 1m apart, as 1m represents 180 degrees they will have some cancellation at 680Hz and integer multiples thereof.
If they are in opposing polarity then the combined signal would broadly be nulled, apart from peaks around 680Hz and integer multiples thereof.
Hope that all makes sense

...you can see it doesn't say polarity on the mixer it says phase 180 deg so that's where I'm coming from , two identical sources on each of those tracks will null to silence when the phase is switched to 180deg....

I strongly suspect that the software of the digital mixer channel strip has been coded to work in the same way as a regular analogue mixer, ie the 'phase' button is actually inverting signal polarity.

[norburybrook]

Pete,

I think the term 'out of phase' and polarity swap get used or misused in the audio world and I'm probably guilty of that too just because that's the term 'we' use most of the time.

technically I suppose polarity has no time domain so you shouldn't perhaps use the word 'phase' in these contexts, however audio engineers all over the world still use the term


As I said previously to Martin, I'm guilty here of moving this post off topic ,purely from the point that I find it interesting, and it's an area I have more expertise in than amp building; where I wouldn't challenge any of you over your knowledge.



Marcus

[Reeltarded]

lol me and you against the world.. a bunch of engineers.. and then a bunch of engineers..

Which engineers get 'inverted' commas?

Here are two labels I have seen. A circle with a line straight up and down marked PHASE and a circle with a line through it about 20° off axis labeled POL..

The correct term for flipping is POL and that's fine.

[xtian]

technically I suppose polarity has no time domain so you shouldn't perhaps use the word 'phase' in these contexts, however audio engineers all over the world still use the term

Thank God we finally got here!

[pdf64]

...I think the term 'out of phase' and polarity swap get used or misused in the audio world and I'm probably guilty of that too just because that's the term 'we' use most of the time...

Hey Marcus, as much as it makes me cringe, it's hard to not to use the vernacular when talking to colleagues, and as that's the label on the equipment, and it's what everyone has got used to, sometimes cruel circumstance forces my hand and I too misuse the word 'phase'.
If it was ok with Langford Smith, who am I to criticise?

...technically I suppose polarity has no time domain so you shouldn't perhaps use the word 'phase' in these contexts, however audio engineers all over the world still use the term ...

Dunno how much if any basic electro-magnetic theory features in audio engineering courses; it's fearsomely difficult and much of it isn't readily applicable, so probably very little
I suspect that tube amps will stop featuring a HT 'standby' switch before studio folk use 'phase' and 'polarity' appropriately.
The important thing is to understand the difference, and then, what is really meant.

As I hoped I demonstrated with my 2 mics' example, in audio engineering both 'phase' and 'polarity' are important, and grasping the difference might help with identifying the most suitable fix to be used when there's an issue.
However, often I've noticed that due to eq and phase (ie distance) differences between 2 mics, combining them can sound fine, different but both ways valid, with the 'phase' (ie polarity) button of one channel in either position; it's not like one way has that hollow sound and the other sounds full.

...As I said previously to Martin, I'm guilty here of moving this post off topic ,purely from the point that I find it interesting, and it's an area I have more expertise in than amp building; where I wouldn't challenge any of you over your knowledge...

But it's been interesting and has made me at least think about this stuff for the 1st time in decades, so this (slight) digression has been useful.
And perhaps we can blame Phil, as he (inadvertently) started it off when he mentioned the Vox trem [vib] network, which uses a brain haemorrhagingly complex combination of phase shift AND polarity inversion
Cheers, Pete