• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

REW phase interpretation and understanding

D

Deleted member 48726

Guest
Hi clever nerds :)

In a continuos search for better sound, I wonder how to interpret the phase diagram in REW.

Can you tell me how a correct phase should look like?

This is a recording from my room from en experiment. I used the in-line RCA crossover (built in) fixed 80 Hz 12dB/oct. in my SVS SB2000.
REW Phase - 07-11-2022.JPG
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
It’s much better if you attach the individual channel and summed (sub+mains) sweeps into a compressed zip file so it can be looked at much closer.
 
OP
D

Deleted member 48726

Guest
It’s much better if you attach the individual channel and summed (sub+mains) sweeps into a compressed zip file so it can be looked at much closer.
Maybe. But I'm just curious how a correct phase curve should look like and what the phase curves themselves represent signal wise.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Maybe. But I'm just curious how a correct phase curve should look like and what the phase curves themselves represent signal wise.

Well, it depends on the crossovers and bandwidth of your speakers... It could be that many of the multiple phase rotations between 100-200Hz is simply because of the your room. There is a way to window some of that to some degree, but outdoor or quasi-anechoic measurement methods can help make things clearer so one can separate the room effects from actual phase.
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
Just a few tips:

Don't smooth measurements below 200 Hz.

Present your measurement graphs with a vertical SPL scale of 50 dB. The vertical SPL scale in the graph presented is 230 dB.

The phase scaling for the graph should be set to +/- 360˚. The phase vertical scaling as presented is set to +/- 720˚. Phase, when presented wrapped, will not exceed +/- 180˚.

When taking screenshots, don't crop off the scaling as it makes it difficult to interpret the graph with the scaling cut off. Use the camera icon on the top left of the graph to generate a screen shot.

Examples of the same measurement with different presentations:

Unsmoothed, unwrapped phase, 50 dB vertical SPL scaling, 1800˚ to 0˚ phase scaling:
unsmunwr.png


1/6 octave smoothing, wrapped phase, 230 dB vertical SPL scaling, +/- 360˚ phase scaling:
smwr.png


The second one looks pretty good, like the cute blonde at the end of the bar after you've had a few shots. Reality is much different as you get closer.

The vertical dashed lines in the phase plot indicate a complete 360˚ rotation of phase when the phase plot is presented wrapped. Unwrapping the phase—as is plotted in the top image—indicates its 'true' nature. The scaling of the phase plot when unwrapped needs to be changed to view the entire plot.

What you are seeing is 'normal' for this frequency range. In the absence of the dips in response caused by what is probably modal behavior of the room, the phase plot would actually be a fairly smooth line. Example below:

smunph.png
 
Last edited:
OP
D

Deleted member 48726

Guest
Just a few tips:

Don't smooth measurements below 200 Hz.

Present your measurement graphs with a vertical SPL scale of 50 dB. The vertical SPL scale in the graph presented is 230 dB.

The phase scaling for the graph should be set to +/- 360˚. The phase vertical scaling as presented is set to +/- 720˚. Phase, when presented wrapped, will not exceed +/- 180˚.

When taking screenshots, don't crop off the scaling as it makes it difficult to interpret the graph with the scaling cut off. Use the camera icon on the top left of the graph to generate a screen shot.

Examples of the same measurement with different presentations:

Unsmoothed, unwrapped phase, 50 dB vertical SPL scaling, 1800˚ to 0˚ phase scaling:
View attachment 242203

1/6 octave smoothing, wrapped phase, 230 dB vertical SPL scaling, +/- 360˚ phase scaling:
View attachment 242204

The second one looks pretty good, like the cute blonde at the end of the bar after you've had a few shots. Reality is much different as you get closer.

The vertical dashed lines in the phase plot indicate a complete 360˚ rotation of phase when the phase plot is presented wrapped. Unwrapping the phase—as is plotted in the top image—indicates its 'true' nature. The scaling of the phase plot when unwrapped needs to be changed to view the entire plot.

What you are seeing is 'normal' for this frequency range. In the absence of the dips in response caused by what is probably modal behavior of the room, the phase plot would actually be a fairly smooth line. Example below:

View attachment 242205
Thank you. I will investigate a bit more this evening. But can you explain what you mean by "phase rotation" what does that mean?
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
One more example in a small "treated room" with some equalization:

1668009634727.png 1668009654487.png 1668009660535.png

Individual channels:
1668010545228.png 1668010549890.png
 
OP
D

Deleted member 48726

Guest
Here is a new pic with corrected scalings.

Good / no good?

REW Phase - 07-11-2022_unwrapped.JPG
 
OP
D

Deleted member 48726

Guest
Yes, much better. As @ernestcarl has said, these are clearly modal issues caused by the room. Think of smoothing as beer goggles. It (they) hides a multitude of sins.
Yes, room issues. :) The 30 Hz boost is annoying! Don't have them without sub of course..

Wrapped, sub @ 0 °-->
Phase sub0_wrapped.JPG


Wrapped, sub @ 180 °-->
Phase sub180_wrapped.JPG


Wrapped, sub @ 160 °-->
Phase sub160_wrapped.JPG



Now, I think it sounds better with the sub phase at 160 ° but I want to know why, so I can measure different positions and settings, look at the data and set it, so I don't have to listen to music in between to figure it out.

I want to buy a minidsp at a point in time. But in the mean time I want to know more. :)
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
But can you explain what you mean by "phase rotation" what does that mean?

Every 360˚ change in phase is one 'rotation'. The frequency content is preserved but the amplitude changes.
 
OP
D

Deleted member 48726

Guest
Every 360˚ change in phase is one 'rotation'. The frequency content is preserved but the amplitude changes.
By rotation, do you mean the test signal sine wave full cycle?
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
I want to buy a minidsp at a point in time. But in the mean time I want to know more.

Before spending money, try positioning in different places in the room to see if you can smooth out the response some. Concentrate on eliminating the dips in response.
 
OP
D

Deleted member 48726

Guest
Before spending money, try positioning in different places in the room to see if you can smooth out the response some. Concentrate on eliminating the dips in response.
I did that a year ago while measuring with umik and looking at frequency response but then bought different speakers and haven't really used the sub much since. Now I want to experiment a bit more. The current position is the one that gives the flattest response.
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
By rotation, do you mean the test signal sine wave full cycle?

It's not the test signal that is producing the change in phase. It is the measured response of speaker/room interaction—mostly the room. The last image in my post above was done nearfield after Keele to eliminate as much as possible the room's effect on the response.
 
OP
D

Deleted member 48726

Guest
It's not the test signal that is producing the change in phase. It is the measured response of speaker/room interaction—mostly the room. The last image in my post above was done nearfield after Keele to eliminate as much as possible the room's effect on the response.
So.. The phase should ideally be flat sloping, crossing 0 deg. at the middle? If it doesn't, it's disturbed by reflected waves causing the ideal line to curve differently?
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
So.. The phase should ideally be flat sloping, crossing 0 deg. at the middle? If it doesn't, it's disturbed by reflected waves causing the ideal line to curve differently?

In a measured system where all frequencies of the stimulus pass through the system with the same delay the phase plot will be flat.

Loopback measurement (wired output to input), 96 Khz SR, 48 Khz BW, 20 Hz - 24 KHz displayed, ≈ -12 dBFS, phase highlighted:

flph.png
 
Top Bottom