Help understanding Alignment Tool

[guitarplayr82]

I am trying to use REW’s alignment tool to align my studio monitors with a single sub. In the screenshot below you can see my monitors in green and my sub in brown, both of which have already been corrected with EQ:

To my knowledge the black dotted line should represent the “ideal” summing if the alignment was perfect. However, when I click “average the responses”, I end up with the teal line in the next screenshot:

The teal line is what I want/expect my response to look like, while the dotted line seems to create a weird boost around the crossover region. Why does the RMS average look so different from the dotted line? If everything was aligned perfectly, shouldn’t the response more closely resemble the RMS average since the RMS average ignores phase in its calculation?

 

[sm52]

If everything was aligned perfectly, shouldn’t the response more closely resemble the RMS average

No. And yes. Ideal alignment cannot be obtained by one action. The ideal alignment in your case is shown by a black dashed line on the upper screenshot. So that it does not look like something superfluous relative to the straight line, you need to move the frequencies of the cut of the sub and monitors from each other. Then there is a chance that the hat will become lower, closer to the main level, but the perfect phase coordination will remain. The phase cannot be ignored. With equally high -profile levels of two sources emitting the same frequency, but with opposite phases, what happens? Cancellation. You will not hear anything. Where the resulting alignment differs from the ideal, occur incomplete phase addition, which can be acceptable, or subtraction of phases, which is unacceptable.

 

[John Mulcahy]

Why does the RMS average look so different from the dotted line?

RMS average is an incoherent summation (assumption of random phases, essentially), the dotted line is a coherent summation.

 

[guitarplayr82]

Thank you for the info. So if I understand correctly, I should change my EQ or crossovers to counteract the boost that is created by the proper phase alignment. I feel like I might end up going in circles, as won’t a change in EQ shift the phase again and cause a different phase summation? I feel like I may be missing something in my process. How is this usually approached?

 

[JStewart]

Edit: Deleted

 

[John Mulcahy]

So if I understand correctly, I should change my EQ or crossovers to counteract the boost that is created by the proper phase alignment.

No, the dotted line just shows what would happen if both were perfectly in phase everywhere. It doesn't mean that's what you want, it represents a limit of what could occur. The summed line (sold black) on your image looks great.

 

[JStewart]

@John Mulcahy , would having the two in phase also make an audible improvement, assuming they close in time as well?

 

[John Mulcahy]

Not sure I follow. Phase should be aligned around the crossover frequency.

 

[JStewart]

That’s what I was referring to. With only the OP’s black summation line to look at, I was assuming, perhaps incorrectly, that because the dotted line is above it a better summation and therefore phase match could be achieved. I suppose without the phase traces or the .mdat to go with it, it’s not 100% possible to say.

 

[guitarplayr82]

JStewart I believe you’re correct on your assumption. I can post the phase traces later today when I get home, but I do believe that the phase traces on the current black line are not 100% perfectly aligned. There is some cancellation happening there which is making the black line look relatively flat. When I tell the alignment tool to align at the crossover frequency, the black line matches the dotted line and the phase traces look perfectly aligned at the crossover frequency. I am wondering if it would be better practice to go with that perfect alignment and then implement a subtractive EQ filter around the crossover to flatten it out again.

EDIT: I was able to access the phase traces just now. Here is what I'm looking at:

 

[sm52]

the phase traces on the current black line are not 100% perfectly aligned. There is some cancellation happening there which is making the black line look relatively flat

Yes.

I am wondering if it would be better practice to go with that perfect alignment and then implement a subtractive EQ filter around the crossover to flatten it out again.

No. Better to do, as I described in the post #2. If there is a way not to use an equalizer, you do not need to use an equalizer. If natural sound matters to you.

I am also interested the delay that turned out to be in the alignment tool for the result from your last post.

 

[JStewart]

I am wondering if it would be better practice to go with that perfect alignment and then implement a subtractive EQ filter around the crossover to flatten it out again.

As @John Mulcahy said in his post above, it’s always been my understanding that sound quality is improved when the subs and mains are also in phase at the XO. That understanding goes back some years so I don’t have a reference to the source. Chat GPT agrees, which probably isn’t conclusive.

A second guideline I’ve been given to understand to follow is to try to keep the relative timing to achieve best phase match within a cycle or ~12.5ms at 80Hz. Going too far off will add group delay which may have an audible effect.

So yes, if the goal is to flatten it then some EQ will probably be necessary. Since the summation peak is at the XO, if your hardware allows you might experiment with lowering only the sub xo or raising only the mains. You can also experiment with lowering the sub gain, although it might cost some extension if there no headroom to boost the lowest frequencies.

Because it hasn’t been explicitly stated, just to be sure, where the measurements taken with a timing reference?

There’s smarter folks around here than me, so hopefully they’ll chime in if I’ve misstated anything or with better ideas for achieving your goal.

 

[guitarplayr82]

No. Better to do, as I described in the post #2. If there is a way not to use an equalizer, you do not need to use an equalizer. If natural sound matters to you.

I am also interested the delay that turned out to be in the alignment tool for the result from your last post.

Just so I understand what you said in post #2, are you saying that if my sub and mains are both crossed over at 80Hz, I should instead try something like 70Hz LP on sub and 90Hz HP on mains to fix my problem?

I should say that my sub is already delayed by 4.74ms in all of the measurements I have posted so far. The alignment tool suggested an additional 3.41ms plus inverting the polarity in order to obtain the phase alignment in my last post.

Thank you all for the guidance you have given me so far as I work my way through learning this. I am understanding this much better already!

 

[guitarplayr82]

A second guideline I’ve been given to understand to follow is to try to keep the relative timing to achieve best phase match within a cycle or ~12.5ms at 80Hz. Going too far off will add group delay which may have an audible effect.

That's a great rule and makes a lot of sense to me, thank you for that!

So yes, if the goal is to flatten it then some EQ will probably be necessary. Since the summation peak is at the XO, if your hardware allows you might experiment with lowering only the sub xo or raising only the mains. You can also experiment with lowering the sub gain, although it might cost some extension if there no headroom to boost the lowest frequencies.

Because it hasn’t been explicitly stated, just to be sure, where the measurements taken with a timing reference?

Thank you for the good ideas! Yes, I am using loopback as a timing reference on all these measurements.

 

[sm52]

I should instead try something like 70Hz LP on sub and 90Hz HP on mains to fix my problem?

Yes. Thus, you maintain the best approval of the phases and reduce the overall level at the frequency of the crossover. This was your discontent - too high the general level at the frequency of the crossover.

sub is already delayed by 4.74ms in all of the measurements I have posted so far. The alignment tool suggested an additional 3.41ms

Up to 13 ms - a normal delay for a sub.

 

[JStewart]

Just so I understand what you said in post #2, are you saying that if my sub and mains are both crossed over at 80Hz, I should instead try something like 70Hz LP on sub and 90Hz HP on mains to fix my problem?

I had somehow missed that good suggestion in @sm52 ‘s post and later thought the same. Edit: Don’t know about the exact numbers, but worth experimenting.

 

[guitarplayr82]

Thank you all! I think I've got a good solution figured out for my room now. The help is much appreciated!