Home Theater Forum and Systems banner

1 - 6 of 6 Posts

·
Registered
Joined
·
3 Posts
Discussion Starter #1
Hi there I'm new to this forum..
...and let me start by saying rew is some jawdropping technology. Much respect.

One thing that is not immediately clear from flicking through the pdf is whether or not the software automatically considers the phase for you before it generates the eq settings. It seems that eq points should only be added to areas where the phase is minimum, and by looking at the excess group delay graph you can judge the best and worst spots(?). It follows that rew is analysing the phase so you don't have to - is it though?

Surfing the web and trying to learn about room accoustics there are many folks saying: "don't eq against a room because phase". My understanding is that if you get the phase wrong in the eq you will have worse resonances in the time domain(?). So it will sound good for a few milliseconds and then resonate badly further on? So the waterfall plots will look good initially and then much worse at the bottom of the plot?

Am I correct in my rough summary of "the phase problem"?

Some more questions:

*Is there any way to limit then number of eq points that rew will generate? For example if you want an eq with only 10 points?

*If you use a digital linear eq on your room does this completely remove the phase problem?

*It seems like you can also attempt to "fix" your room by using impulse response convolution. This appears to be a completely separate approach(?). Would you consider that a better or worse option than eq, or does it depend on your room?

Hi and thanks once again for the great software.
 

·
Registered
Joined
·
3 Posts
Discussion Starter #2
ps: one more question...

*Why can rew take an average spl response but not an average phase response? Surely this is desirable if you are taking multiple measurements around a room?
 

·
Premium Member
Joined
·
1,562 Posts
One thing that is not immediately clear from flicking through the pdf is whether or not the software automatically considers the phase for you before it generates the eq settings. It seems that eq points should only be added to areas where the phase is minimum, and by looking at the excess group delay graph you can judge the best and worst spots(?). It follows that rew is analysing the phase so you don't have to - is it though?

Surfing the web and trying to learn about room accoustics there are many folks saying: "don't eq against a room because phase". My understanding is that if you get the phase wrong in the eq you will have worse resonances in the time domain(?). So it will sound good for a few milliseconds and then resonate badly further on? So the waterfall plots will look good initially and then much worse at the bottom of the plot?

Am I correct in my rough summary of "the phase problem"?
> Yes - Roughly correct.
> All the REW EQ devices listed in the EQ window are minimum phase (IIR) devices. Individual speaker drivers are minimum phase devices. IIR correction is thus appropriate to correct a speaker driver output. The XO range in multi-driver speakers is not min phase due to the XOs. A typical passive XO in a speaker will create some excess phase rotation and that impact is different depending upon the measurement axis.
> Room effects create SPL and phase chaos at some freqs at the LP.
> It is true that the room acoustics is a major player and EQ can only help some types of those issues. If done poorly EQ can introduce other issues. The situation rules and there is no best approach that fits all the various situations.
> Good results in overall LP EQ correction is reported and documented using various methods. It is difficult to say what will work best for a given situation.

*Is there any way to limit then number of eq points that rew will generate? For example if you want an eq with only 10 points?
> The choice of equalizer in the EQ window properly sets the max bands for that equalizer and accounts for the filter shapes of that equalizer. It's possible set some of those filters To 'manual' in the 'EQ Filters' popup. The REW auto EQ will then not use those filters. The Generic and rePhase choices are the exception and have use for those using FIR filters.

> The smoothing that is applied to the response curves will impact the number of filters needed and the aggressiveness of the filters applied. Thus impact on ringing is limited as the smoothing is increased and the filter Q is reduced.

*If you use a digital linear eq on your room does this completely remove the phase problem?
No - not normally. It may be possible to get very near that at the LP area in some well dimensions, well treated rooms with ideal speaker and LP positions. Most all rooms will have at least some major room modes and major reflections that EQ cannot completely address.

*It seems like you can also attempt to "fix" your room by using impulse response convolution. This appears to be a completely separate approach(?). Would you consider that a better or worse option than eq, or does it depend on your room?
> Yep - we could debate this forever. The room setup and speaker design impacts this and the only final analysis of which is better is subjective and may differ for each person.

ps: one more question...

*Why can rew take an average spl response but not an average phase response? Surely this is desirable if you are taking multiple measurements around a room?
We can in REW - see 'Trace Arithmetic'. It is it is sometimes helpful when the objective is to help clarify the overall phase rotation of the direct sound. I do that sometimes to create a phase correction FIR filter.
 

·
Registered
Joined
·
3 Posts
Discussion Starter #4
Wow. Thanks for a great response. I was worried my questions were maybe too ambitious.

One thing I'm still not sure of, does REW consider the phase before generating EQ? Actually, as far as I can tell it does not do this, because it works with or without phase data.

It's going to take me a long while before I can properly grasp the concepts here. It's a real pandora's box.

:)
 

·
Premium Member
Joined
·
1,562 Posts
The REW filters calculated and plotted are all based on IIR filters. The direct sound phase follows the speaker phase response plus the phase response of the applied IIR EQ filters.
I am not sure if there is a special consideration of the Excess GD in the algorithm. Depending on settings chosen, REW auto EQ calculation sometimes leaves nulls untouched and other times partially fills them in. My experience is that the predicted SPL response is very accurate assuming IIR filters are used. The SPL fill that occurs in a null is most often significantly delayed as it depends on late arriving sound that may be many cycles delayed.
We can set the filters manually instead of using the auto EQ function if we like.
We can use auto filter function and then change (remove or adjust) filters as we feel appropriate.

It is up to the user to set reasonable smoothing and boost limits for the auto filter function. Doing that will avoid/minimize the artifacts that some users are concerned with. Some users are very aggressive with filters and get a very smooth SPL response as a result. They do not seem to notice the greater artifacts. Both styles are popular and have their proponents. I am not qualified to predict what will work best for a given situation.

My sense is that you may be overly concerned with phase aberrations due to EQ. I suggest you try several EQ options and pick the one that sounds the best to you. I found that selecting a house curve that suited my preference was much more important.

What is your setup?
 

·
Registered
Joined
·
2,261 Posts
Wow. Thanks for a great response. I was worried my questions were maybe too ambitious.
One thing I'm still not sure of, does REW consider the phase before generating EQ?
jtalden does have the knack, doesn't he! I hope he corrects me if I'm wrong in trying to answer your last question... The following excerpt from page 59 of the latest REW Help Manual seems like it applies:

"Note that the predicted EQ results REW shows in its EQ window are obtained by applying the chosen filters to the measured impulse response, and include the effects of non-minimum phase behaviour, so they accurately portray the results that would be obtained at the point the measurement was taken."

I would guess that if predicted EQ is performed only in minimum-phase regions, then REW will generate filters only in those regions as well. But.. caveat emptor! As jtalden mentioned, he's seen the auto-EQ function try to fill nulls (usually minimum-phase) on occasion. Of course, all this seems to be a moot point if you're comfortable with choosing auto-EQ or manual smoothing. Hope I got that right!
 
1 - 6 of 6 Posts
Top