Need help interpreting phase measurements

[ggegan]

I understand that phase is an extremely important element of acoustics, but I am unclear how to measure and interpret it. Phase relationships are pretty straightforward to recognize when comparing input from two separate electronic signals using an oscilloscope, but I am confused about how to derive useful information from swept sine acoustic measurements. I believe I can figure out how to remediate phase issues once I recognize their source, but I am unclear how to determine the source from REW or FuzzMeasure graphs. The manuals seem to be especially vague regarding interpretation of the measurements.

I would be very grateful for any education on this subject.

[Max Dread]

Originally Posted by ggegan

I understand that phase is an extremely important element of acoustics, but I am unclear how to measure and interpret it. Phase relationships are pretty straightforward to recognize when comparing input from two separate electronic signals using an oscilloscope, but I am confused about how to derive useful information from swept sine acoustic measurements. I believe I can figure out how to remediate phase issues once I recognize their source, but I am unclear how to determine the source from REW or FuzzMeasure graphs. The manuals seem to be especially vague regarding interpretation of the measurements.

I would be very grateful for any education on this subject.

+1 and a bit! I have no idea but have often wondered..... But unlike you, I have no idea how they are recognised or remedied. Don't even really know what they are TBH!

As they are seldom discussed in these forums I'd kinda stuck them to the bottom of the list of "things to find out about". But I've done that with an apprehension that perhaps I'm glossing over something quite significant.

Look forward to hearing more......

[DanDan]

A bit beyond my pay grade also, and on a list....
However this is just another way of looking at the measurement taken.
Probably useful in designing loudspeakers or combining multiple loudspeakers in a large system. In the REW John has singled out an area of the Phase family which is practically useful. Using extractions he seeks to identify frequency areas which are minimum phase and thus amenable to Eq.
Soffit LF reinforcement is a MP system and can be readily corrected by Eq.
Front Wall SBIR, at least the peaks should be amenable.
There is an awful lot more to this though. REW Forum at Home Theater Forum and Systems - HomeTheaterShack.com
I haven't spend much time there myself but this has prompted me to start doing so.
DD

[ggegan]

The REW manual has a section called "Why Can't I Fix All My Acoustic Problems With EQ" which explains a bit about phase issues and suggests examining the ETC graph, but I couldn't find anything about how to interpret the ETC and determine the source of the problem for a particular frequency.

I'm using the REW RTA with external pink noise because my firewire interface can't seem to generate an output from REW due to Java issues with MacOS, so I can't use it for ETC. I'm using FuzzMeasure for swept sine measurements. The REW manual is more comprehensive, but I'm having trouble applying some of the info it contains to FM.

[DanDan]

ggegan, sounds like you are just starting on this measurement lark.
Perhaps Phase is not featured for good reason.
There is a Java problem. Try to convince AudioMidi setup that your interface is 2 channel only. If this doesn't work, the onboard sound I/O is good enough for measurement. Just connect your measure mic to a pre to line in.
Line out to one speaker at a time or your controller.
There is a considerable section in the REW manual on Minimum Phase with a purpose in mind as I said. Beyond that is really text book stuff, at a pretty high level too.
You can take measurements in FM, export them as Wavs, import into REW.
FM has MP, Excess Phase, and so on. Chris and John may be using slightly different terms.

DD

[ggegan]

Originally Posted by DanDan

ggegan, sounds like you are just starting on this measurement lark.
Perhaps Phase is not featured for good reason.

DD

I'm a mixer, not an acoustician - I'm not going to be able to deal with the more esoteric math required, but I can understand the basic logic behind the physics and I figure that there is still a lot I can do myself in terms of optimizing the acoustics of my personal studio.

[DanDan]

The software has very powerful features right up front. The Waterfall.
As an Acoustician I have never delved into Phase and many of the other features of FM or REW. If I were trying to use an Eq to help I would definitely study John's manual on that topic.

DD

[Ethan Winer]

Originally Posted by ggegan

I understand that phase is an extremely important element of acoustics, but I am unclear how to measure and interpret it.

This is a conceptual problem. Phase has no meaning until you have two otherwise identical sources that become combined. So first you need to answer the question, "Measure the phase of what?" A room doesn't have "phase," though reflections can have a phase relationship between direct and reflected sounds, or between two different reflections. But moving a speaker, or moving your ears or microphone, changes everything.

So what exactly are you asking? And what specific problem do you think you have or may have?

--Ethan

________________
The Acoustic Treatment Experts

[DanDan]

+1 Ethan.
DD

[ggegan]

Originally Posted by Ethan Winer

This is a conceptual problem. Phase has no meaning until you have two otherwise identical sources that become combined. So first you need to answer the question, "Measure the phase of what?" A room doesn't have "phase," though reflections can have a phase relationship between direct and reflected sounds, or between two different reflections. But moving a speaker, or moving your ears or microphone, changes everything.

So what exactly are you asking? And what specific problem do you think you have or may have?

--Ethan

________________
The Acoustic Treatment Experts

Yes, I understand that the human ear is not phase sensitive and that phase is only relevant for this situation when two or more waveforms interact.

I am seeing a number of notches and peaks in the low and low-mid frequencies that do not appear to be a result of room modes. I would like to reduce them as surgically as possible without getting too anechoic. I suppose what I am looking to be able to identify is which surfaces are generating reflections that cause phase induced nonlinearities at which specific frequencies. Also, I have read that EQ is not suggested for correcting phase induced nonlinearity, so I would like to be able to clearly identify those areas not suitable for EQ. Perhaps measuring phase relationships is not the way to do this, and if so I would appreciate being steered in the right direction. I'm just trying to understand how to recognize and deal with these types of issues in a precise way.

[DanDan]

The REW manual is excellent on the topic of Minimum Phase, leading on to using excess group delay plots to identifying regions where Eq would be useable.
The same section and plot identifies anomalies which are not amenable to Eq.
Unless you intend using Eq this seems like a very long and difficult way around the issue.
Similarly phase induced nonlinearities? Do you mean Peaks and nulls caused by boundary reflections, i.e. SBIR?
Treating the first biggest peak or null is often enough if the trap is tuned. Tune wider or use thick absorbent to get more of them.
To identify particular ones, unless you have the mathematical skills of Hannes, it seems simplest to move the mic or speaker, observe the change in measurement. e.g. Front Wall SBIR frequency does not change as the mic is moved directly back or forward, while a side or floor reflected SBIR's frequency will change. It gets a lot better than that though, 3D and such. Check out the SBIR thread. Also T Barefoots Wall Bounce Calculator 2D
DD

[ggegan]

Originally Posted by DanDan

The REW manual is excellent on the topic of Minimum Phase, leading on to using excess group delay plots to identifying regions where Eq would be useable.
The same section and plot identifies anomalies which are not amenable to Eq.
Unless you intend using Eq this seems like a very long and difficult way around the issue.
Similarly phase induced nonlinearities? Do you mean Peaks and nulls caused by boundary reflections, i.e. SBIR?
Treating the first biggest peak or null is often enough if the trap is tuned. Tune wider or use thick absorbent to get more of them.
To identify particular ones, unless you have the mathematical skills of Hannes, it seems simplest to move the mic or speaker, observe the change in measurement. e.g. Front Wall SBIR frequency does not change as the mic is moved directly back or forward, while a side or floor reflected SBIR's frequency will change. It gets a lot better than that though, 3D and such. Check out the SBIR thread. Also T Barefoots Wall Bounce Calculator 2D
DD

I do use EQ, and not just for low frequencies, because I tune my small room to the modified X-curve so that it translates properly to large dub stages that use the X-curve. Many small rooms around Hollywood also use the X-curve, even though it is not strictly recommended below a certain room volume. I don't know of any way to conform to either curve without using EQ. Since I am using EQ to conform to industry standard curves, I want to make sure that I am not using it inappropriately.

The REW manual does discuss minimum phase somewhat, but it doesn't seem to go into much depth. I'm also a bit confused about things like phase difference and group delay. I don't think the concepts are too difficult for me to grasp, I just need some help getting to that "ahah!" realization. I'll re-read that section and see what I'm missing. I'm sure I can probably find some internet tutorials as well, but if anyone can direct me to some well written ones, I would appreciate it.

While I have a basic understanding of how phase relationships can affect waveforms to do things like create a square wave from the interaction of sine waves, my math skills are pretty rudimentary, so maybe the trial and error approach is more appropriate for me, as you suggest, although a computer should be able to handle the math, even if I can't.

[DanDan]

The X curve or any other HF roll of is absolutely no problem. Similar to just turning down the tweeter, but an Eq or speaker processor will be just fine.
I have read that REW section twice. It seems pretty straightforward to me.
You can use Eq to cut peaks that lie within an MP area, as shown on the Excess Group Delay graph. Furthermore the effect of Eq filters parameters created by REW can be applied and REW will show if they would work or not.
Room Eq Wizard, kinda says it all.
Perhaps the Eureka understanding is that SBIR creates peaks of up to 6dB max, while the nulls may be -20dB or more. Increasing the energy at a null will increase the destructive reflection. More-more=no change. Broadly speaking you cannot eq nulls.

DD

[jonessy]

I'm not really sure what you're trying to study here, but in room acoustics phase is an issue especially at LF.

For example, whether you get a 'peak' or 'dip' at your room's LF response depends on how room modes coincide at the point of observation, in phase or out of phase (or in between-phase).

A good exercise would be to compare the LF magnitude response plot to the LF phase response plot. It shows up clearly.

Otherwise, please be more specific about your question...

Cheers,
Jon

[spice house]

Originally Posted by ggegan

Yes, I understand that the human ear is not phase sensitive and that phase is only relevant for this situation when two or more waveforms interact.

It's not that you can't hear phase, its that you aren't very sensitive to absolute phase, especially at bass frequencies. If things are phasing (interfering) along the absolute phase graph you can certainly hear it.

The only way I can see phase being useful in mixing rooms is to chart interfering reflections at various distances from the speakers in order to improve clarity in reproduction, and to ensure the phase coherence of your monitoring system.

[ggegan]

I don't need a tutorial on how phase interactions work, I understand that very clearly. What I am trying to determine is whether applications such as FuzzMeasure, REW, or SMAART, etc. have modules that can break down a swept sine IR into its contributory wave forms in order to help me visualize the various specific phase relationships that are occuring between the direct sound the different reflections. I understand that is is a very complex proposition, but it would be very helpful to me in steamlining the trial and error approach, which, to be honest, I don't have a lot of time for, since I make my living as a rerecording mixer, not an acoustician. The more info I can extrapolate in advance, the more efficient I can be when experimenting. From reading the manuals, I haven't been able to determine whether that is possible or not, so I suspect that it may not be, or perhaps some of the aps can do this in a form I'm not recognizing.

[JohnPM]

Originally Posted by ggegan

What I am trying to determine is whether applications such as FuzzMeasure, REW, or SMAART, etc. have modules that can break down a swept sine IR into its contributory wave forms in order to help me visualize the various specific phase relationships that are occuring between the direct sound the different reflections.

Your original question was not well related to what you are interested in. The plot to use for studying the reflections that are affecting the response is the impulse response envelope or ETC. The spikes in the ETC that occur after the initial peak are due to reflections from the room's surfaces - yes, there is a frequency at which the time delay due to the longer path travelled by the reflection corresponds to 180 degrees of phase shift, but it is the reflection that is important, not the measured phase of the response.

Ideally a Polar ETC is needed, to provide information about the direction from which any particular reflection arrived and hence make it easy to locate the offending surface. There are other ways to work that out, however.

[ggegan]

Originally Posted by JohnPM

Your original question was not well related to what you are interested in. The plot to use for studying the reflections that are affecting the response is the impulse response envelope or ETC. The spikes in the ETC that occur after the initial peak are due to reflections from the room's surfaces - yes, there is a frequency at which the time delay due to the longer path travelled by the reflection corresponds to 180 degrees of phase shift, but it is the reflection that is important, not the measured phase of the response.

Ideally a Polar ETC is needed, to provide information about the direction from which any particular reflection arrived and hence make it easy to locate the offending surface. There are other ways to work that out, however.

Thank you.

[Ethan Winer]

Originally Posted by ggegan

I am seeing a number of notches and peaks in the low and low-mid frequencies that do not appear to be a result of room modes.

Yes, that's called SBIR and LBIR, and those non-modal peaks and nulls are typical in small rooms. Large rooms too when you or the sound source are near to a boundary.

I would like to reduce them as surgically as possible without getting too anechoic. I suppose what I am looking to be able to identify is which surfaces are generating reflections that cause phase induced nonlinearities at which specific frequencies.

As I said in my first post, when you move even an inch or two everything changes. That's why they're non-modal.

So there really is no way to do what you want. The solution, as always, is absorption that can target the peak and null frequencies based on distance. This might help:

Frequency-Distance Calculator

--Ethan

[ritelec]

Originally Posted by DanDan

You can take measurements in FM, export them as Wavs, import into REW.
DD

Thanks Dan.....never even considered it........good to know.

-rich

[ggegan]

Originally Posted by Ethan Winer

Yes, that's called SBIR and LBIR, and those non-modal peaks and nulls are typical in small rooms. Large rooms too when you or the sound source are near to a boundary.



As I said in my first post, when you move even an inch or two everything changes. That's why they're non-modal.

So there really is no way to do what you want. The solution, as always, is absorption that can target the peak and null frequencies based on distance. This might help:

Frequency-Distance Calculator

--Ethan

Thank you, Ethan.

[ritelec]

LBIR

L=? please.


Don't think my search as Low Background Infrared Radiometry is it.???

[ggegan]

Originally Posted by ritelec

LBIR

L=? please.


Don't think my search as Low Background Infrared Radiometry is it.???

I believe that would stand for "Listener".

[Mala]

But I believe L would stand for "Listener"

[ritelec]

Thank you and thank you.