Help with 2 sub alignment using REW and AVR distances for delay

[Keith_W]

Have you applied an "octave window" to those IR's though? You can't visually/directly compare IR's that have different "windows".

I am not sure what you mean. If you are talking about octave smoothing, that does not alter the appearance of the impulse response. If you are talking about window functions to reduce sidebands, I am using the standard REW IR window function that places the windows well outside the range of the IR. If you are talking about gating, then I am not using any gating in this instance.

 

[nadrealista]

When aligning subs to mains in the impulse response, you have two goals:

1. No part of the subwoofer response should occur before the main response. If it is unavoidable, it should be low in amplitude and within the psychoacoustic threshold of inaudibility (about 20ms of pre-masking).
2. The highest peak of the subwoofer response should not occur after the main response. Ideally, it should occur together. If this is unavoidable, then it should be within the psychoacoustic threshold of inaudibility (generally about half a period of the wavelength in question).

Both goals are mutually contradictory with minimum-phase DSP. You have to choose one or the other - either the small wiggle occurs before the main impulse, or you align the small wiggle to the main impulse, then the big wiggle occurs later than ideal. One would give you "pre ringing", the other would make the subwoofer sound "slow". I prefer prioritizing the second goal because the initial deflection is small and short.

Sometimes, subwoofers have additional peaks that may be even higher in amplitude than the initial deflections (e.g. the 3rd or 4th peak). This is due to reflections and not the direct sound of the subwoofer. Some caution needs to be taken when interpreting the response.

View attachment 499913

A good example is in his MDAT. Notice how the green subwoofer response is distorted compared to the red sub? Assuming it's the same brand of subwoofer with the same response, this is due to a reflection distorting the response of the sub. It's not so clear where the main impulse is.



Think of it this way: each speaker and sub is a "subwoofer" with different frequency responses thanks to different positions in the room. If we are talking about mono bass, then you can think of the sum of L+R mains as a single subwoofer. Every time you add another subwoofer, the sum of all the speakers and subs is a single subwoofer.

Of course you could do it the way you describe. This way is how I do it. That post will get very long if I were to go through every possible permutation of how to align subs to mains. I am a bit of a collector of different sub aligning methods, every time I see something new, I go and experiment with it. Some methods work better than others, some are easier than others.

Thanks everyone for their input and thank you @Keith_W for taking the time to go through my measurements!

Will try those delays and see what kind of response they produce vs AVR calibration.

These are two different subs in asymmetric locations in the room as shown in the first post.
SW1 is big SVS PB2 Plus sub with dual 12" down firing drivers

SW2 is much less potent Velodyne CHT 10 10" front firing sub

wonder if these dual 12" SVS drivers are making the impulse response wonky (green line).

@Keith_W Looking at the subs impulse responses is SW1 ahead or behind SW2?

I will play with alignment tool and see what I can come up with. Will report back.

 

[nadrealista]

I did bunch of measurements where I looked how subs and mains behave will all DSP off and XO set to 80, 100 and 200hz.
with XO set at 100hz I first aligned subs and got -6.2ms delay on SW2 and then aligned that combined sub aligned measurement with mains and best response was with the subs inverted and mains delayed by -1.38ms. then I compared that with response I have with AVR calibration and DSP.

REW aligned subs (red) vs AVR calibration/DSP (blue):

looking at this it would seem that sub alignment can't give me better response than what I already have with my AVR calibration, but I will apply the settings and measure to confirm.

then I aligned mains to combined subs

here is REW aligned subs with mains (purple) vs AVR calibration measurement (red):

again will measure with these settings this to confirm

With XO at 200hz

REW aligned subs xo 100hz(red) vs REW aligned subs xo 200hz(green) vs AVR calibrated subs response (blue):

REW aligned to mains

REW aligned subs to main xo 100hz(purple) vs REW aligned subs to main xo 200hz(green) vs AVR calibrated response xo 100hz(red):

here looks like I get better response over AVR baseline, however not sure I can achieve this as xo will most likely be set at 100hz.

will try both set ups and see how they measure/compare.

I will use true mains distance and then adjust the subs from that reference point.

 

[janbth]

- Why did you choose to align sub after sub with l+r instead of all subs together first and then the aligned subs in total to l+r?

Some of us do it that way. I tried a number of «methods» for several years in order to integrate my two subs with my mains, including the one-sub-at-a-time method. I never got good results until I tried the multisub method and aligned the subs together first.

Strictly speaking, though, I wouldn’t call it «alignment», because the time domain (impulse response) is not useful for sub integration. You need to look at the frequency domain. Possibly also decay or waterfall.

Here is a minimal description of the method I’m currently using:

1. Place the subs where they are useful. Ideally, learn how the standing wave pattern is in your listening room, and put the subs in opposite polarity regions of the offending modes.
2. Play the subs in mono - the same signal, no relative delay. Measure, and try tweaking level and delay for better frequency response. In my case, I need to play one sub 6 dB louder than the other, because the former is located at mid-wall, the latter is in a corner. No relative delay.
3. Measure the sub combo and L+R separately, inspect the frequency responses, and decide on a crossover.
4. Measure all together - subs combo and L+R - with the crossover applied. Is there a dip in the response at the crossover frequency? If yes, add or subtract a delay for the combined subs until it is smooth.
5. The integration is now complete, but it may be necessary to add filters to knock down remaining peaks and/or smooth out the response.

Of course, I have left out most of the details, but this is the general idea.

 

[Keith_W]

@Keith_W Looking at the subs impulse responses is SW1 ahead or behind SW2?

In that particular graph, I aligned the start of the IR so that it is easier for your eyes to compare the two measurements. Your actual measurement looks like this:

Once the polarity of both subs is inverted and compared to the L+R vector sum, it looks like this:

Green sub: delayed by 7.4ms; red sub: delayed by 15.8ms.

 

[nadrealista]

I also took a look at the phase alignment between L, R, SW1 and SW2 with XO at 80, 100 and 200 measurements:

XO 80hz:

XO 100hz:

XO 200hz:

 

[nadrealista]

Quick follow up on this. I found this document on how my AVR Pioneer MCACC Pro calibration system works and it was very informative on many aspects of setting up and integrating speakers for HT. I think it will be very good read for most members here, especially last section after page 11.

Here is section on how it handles phase control:

Also adding phase measurements with AVR DSP/phase correction on and XO set to 100hz, not bad at all:

here is AVR impulse alignment of sub1 and 2

vs no DSP and equal distance setting for both subs in AVR and XO at 100hz:

 

[nadrealista]

@Keith_W and others folks following this in your opinion how well has my AVR calibration and DSP aligned subs between themselves and then with mains?

How much more can I squeeze out of my current set up over what AVR is doing?

Measurements attached.

 

[Keith_W]

Please label your graphs properly. I don't know what "M1 baseline" is. It looks like L+R+both subs, but when I sum "Sub 1" and "Sub 2" with "L+R" I get a different looking curve. Without knowing what those measurements are, I can't comment.

 

[nadrealista]

M1 baseline is the measurement of both subs with mains. I am guessing when they all play together you also get some cancelations happening between them hence sum of sub1 + sub2 + "L+R" is not equal of M1 baseline which tells me something is not optimal?

 

[Keith_W]

This is your L+R speakers (blue) compared to the measurement of all speakers and subs (yellow). You expect the two measurements to be the same above the XO point. They are not. Something has changed between these two measurements, maybe it's a new mic position, maybe another DSP setting. I don't know.

This is the step response of your current measurement of "L+R M1" compared to your previous measurement of "L+R no sw" taken from this post. You expect them to be the same, but you can clearly see they are different. The current measurement has a low frequency impulse appearing before the main impulse which made me suspect that your so-called measurement of "two main speakers with no subs" actually has a subwoofer included.

Anyway if we ignore all those time domain measurements which are clearly nonsense, and ASSUME that this measurement of your entire system playing together has been taken properly, this result does not look too bad.

 

[nadrealista]

ok. so I think I got the subs aligned as good as possible with their current locations. I made several alignment options that I stored in calibration memory so I can switch between them. Two of them were with subs inverted and one with subs in phase with mains:

I have two configurations with inverted subs: subs + mains as large (red) vs subs + mains crossed at 100hz (green)

vs non inverted subs options

AVR Calibration (blue) vs my REW align (orange) with subs not inverted, XO 100hz mains set to small:

now If I could find a way to position my mains better to avoid all these nulls in the 20-200hz region I am having now. mains are 31" from the front wall and 55" from the side walls. there is open hallway in the left side wall about 5ft from the front wall.

any suggestions?

L is blue and R is red trace:

when I was placing mains I did some measurements (unfortunately these are with subs on) starting against the front wall (purple) and 32" (green) from the front wall. here is delta: front wall driven SBIR dips at 117 and 175 hz?

then looked at what different distances to side wall do: 47" to side wall (blue), 52" (orange) and 58" (red) which is the most I can do since I have projector screen in between mains. does this mean 85hz and 150hz dips are side wall SBIR driven?

thoughts and suggestions on how to combat these?

I also looked at moving my MLP backwards but that looks worse:

 

[thcdru2k]

I like to measure both subs and look for summation specifically around the crossover. If two waves are in phase at the crossover region you should basically only see constructive interference (summation). Basically it should always be louder with two subs in the crossover region vs one sub alone. REW alignment tool is helpful to get you into the ballpark but measuring both subs helps give you final confirmation. For two subs that are level matched I aim for around 6db of summation at crossover regions.

 

[thcdru2k]

When drivers are time/phase aligned you will actually reveal room nulls. Room nulls are inherent properties of the room and will be activated by any driver that are positioned to excite them. The benefit of using multiple subwoofers is one sub will partially fill in the null created by another sub. Im attaching two pictures one is left sub, right sub , and both subs. The other picture is both subs, both mains, and both subs and both mains. Timing acoustic reference to get mains aligned. REW alignment tool to lightly get subs align. You want to key in around the 50-80hz region where the subs and mains cross. I basically do not want any cancellation in that region at all. You can see the left side of the room has a null it is dealing with at 54 hz. You know its a room null because when you inject energy into a room null you actually make it worst. You can see clearly that is the crossover frequency that you do not see summation. There is some cancellation after crossover but that is less of a concern because we are naturally rolling off and the mains are meant to fill in that area. The biggest goal is from 50-80hz we only see summation, no cancellation. Some room nulls I decide to live with and some I will aggressively try to target by sub crawl. It just depends.

 

[thcdru2k]

Another thing is I do not try to chase or boost nulls that are room nulls. The human ear doesn't hear ultra narrow wiggles in REW. I've never seen a human listen to something and say hey did you catch that sharp null at 76 hz. If you want to see what is worth chasing I would click psychoacoustic smoothing; if something disappears by 1/3-oct smoothing its extremely unlikely to be heard by the ear as an issue. In general if I see a dip more than 4db with psychoacoustic smoothing or 1/3 oct-smoothing I'm going to try to see what I can reasonably do to solve it.

 

[janbth]

When drivers are time/phase aligned you will actually reveal room nulls. Room nulls are inherent properties of the room and will be activated by any driver that are positioned to excite them. The benefit of using multiple subwoofers is one sub will partially fill in the null created by another sub. Im attaching two pictures one is left sub, right sub , and both subs. The other picture is both subs, both mains, and both subs and both mains. Timing acoustic reference to get mains aligned. REW alignment tool to lightly get subs align. You want to key in around the 50-80hz region where the subs and mains cross. I basically do not want any cancellation in that region at all. You can see the left side of the room has a null it is dealing with at 54 hz. You know its a room null because when you inject energy into a room null you actually make it worst. You can see clearly that is the crossover frequency that you do not see summation. There is some cancellation after crossover but that is less of a concern because we are naturally rolling off and the mains are meant to fill in that area. The biggest goal is from 50-80hz we only see summation, no cancellation. Some room nulls I decide to live with and some I will aggressively try to target by sub crawl. It just depends.

I think your terminology is unconventional. There is no such thing as «room nulls». There are «room modes», and these have «nulls» at certain locations in the room.