REW MMM measurements

[Keith_W]

Here's the impulse response of the mains and subs:

I see that you decided to sweep both main speakers together and the microphone is poorly centred. Double peak in the main speaker impulse response. There is a 1ms discrepancy between the two peaks, meaning that the microphone is 343mm closer to one speaker than the other. About 1 foot. That is huge!

First, zoom in to t=0 to make sure that REW has correctly aligned the tweeter reference for sub (green) and mains (red). All good, so we can proceed.

We want to align the start of the tweeter impulse with the start of the subwoofer impulse. The discrepancy is either 13.3ms or 14.3ms depending on which main speaker peak we use. Let's say it is 13.3ms. Since the subwoofer impulse is to the left of the main speaker impulse, this means that the sub needs to be delayed by 13.3ms for impulse alignment.

Open up REW's alignment tool with the above settings. To start off, put a 13.3ms delay for the sub. Then move the "Fine delay adjustment" slider whilst keeping an eye on the frequency response. Your goal is to find the delay which gives you the flattest response. In this case it is 16ms.

Once the sub and mains are aligned, use EQ to chop off any remaining peaks. And that's it, you are done.

 

[JD1]

Here's the impulse response of the mains and subs:

View attachment 472940

I see that you decided to sweep both main speakers together and the microphone is poorly centred. Double peak in the main speaker impulse response. There is a 1ms discrepancy between the two peaks, meaning that the microphone is 343mm closer to one speaker than the other. About 1 foot. That is huge!

View attachment 472942

First, zoom in to t=0 to make sure that REW has correctly aligned the tweeter reference for sub (green) and mains (red). All good, so we can proceed.

View attachment 472944

We want to align the start of the tweeter impulse with the start of the subwoofer impulse. The discrepancy is either 13.3ms or 14.3ms depending on which main speaker peak we use. Let's say it is 13.3ms. Since the subwoofer impulse is to the left of the main speaker impulse, this means that the sub needs to be delayed by 13.3ms for impulse alignment.

View attachment 472945

Open up REW's alignment tool with the above settings. To start off, put a 13.3ms delay for the sub. Then move the "Fine delay adjustment" slider whilst keeping an eye on the frequency response. Your goal is to find the delay which gives you the flattest response. In this case it is 16ms.

Once the sub and mains are aligned, use EQ to chop off any remaining peaks. And that's it, you are done.

I did use the LP location for the microphone instead of in the middle of both speakers.

So should I use only left speaker to TA sub ?

Also why is the circle on the sub (graph above (green line) considered the start of the IR instead of behind the small peak? This is what I don't really understand.

Dows this mean I should invert sub polarity?

 

[JD1]

R vs L speaker I moved left speaker to left room corner to get an equidistant triangle to LP from R and L. and changed crossover to 117 hz.

No eq used.

 

[JD1]

im struggling so hard with sub + mains but id say this is correct?

 

[JD1]

What I mainly struggle with is the sweep frequency range, should I do 60 to 160hz since this is what I'm testing (given I've selected main speaker output bass) | I tried with and without, should I do 10 to 20khz, I'm struggling with what to use to TA and then if using the 100% peak or the second peak of sub with the first peak of mains, or the beggining of the impulse response of both sub and mains given I'm doing 60 to 160 hz.

Anyone care to explain?

Thanks for all the insights.

Edit: I've tried searching fir videos and Google but many conflicting info, most use acoustic reference instead of loopback, most are doing it for avr and home theater, many conflicting info, and practical zero about 2.1 or 2.2 stereo with loopback and rew.

 

[JD1]

or is this correct using sub xo bypass and wiim xo bypass , kef r5 meta is passive so obviously cant deactivate xo.

 

[JD1]

latest mmm and EQ.

 

[Keith_W]

Also why is the circle on the sub (graph above (green line) considered the start of the IR instead of behind the small peak? This is what I don't really understand.

This is because subwoofers have slow energy build-up before the main impulse. It's a minimum-phase phenomenon.

This is a subwoofer simulation. I created a minimum-phase 8th order Linkwitz-Riley bandpass. In the upper graph you can see the amplitude response, in the lower graph - the impulse response. The start of the impulse is usually taken as the zero crossing line to the main impulse.

Dows this mean I should invert sub polarity?

Subwoofer polarity is surprisingly difficult to determine from in-room measurements. There are many reasons for this which I won't go into. In any case, it does not matter whether your subs have correct polarity or not. What matters is whether they have correct polarity relative to each other, and relative to your main speakers. Fortunately, that is much easier to determine.

Use REW's alignment tool, then check/uncheck the "invert polarity" button. One will look better than the other. In your case, the answer is "no".

 

[JD1]

This is because subwoofers have slow energy build-up before the main impulse. It's a minimum-phase phenomenon.

View attachment 473181

This is a subwoofer simulation. I created a minimum-phase 8th order Linkwitz-Riley bandpass. In the upper graph you can see the amplitude response, in the lower graph - the impulse response. The start of the impulse is usually taken as the zero crossing line to the main impulse.



Subwoofer polarity is surprisingly difficult to determine from in-room measurements. There are many reasons for this which I won't go into. In any case, it does not matter whether your subs have correct polarity or not. What matters is whether they have correct polarity relative to each other, and relative to your main speakers. Fortunately, that is much easier to determine.

Use REW's alignment tool, then check/uncheck the "invert polarity" button. One will look better than the other. In your case, the answer is "no".

So in the above graph I posted, "sub vs mains IR.jpg" would the main impulse be where the cursor is or the big bump at 85 Ms? (Blue line 12:sub) or the bump at 80ms?

 

[JD1]

One descent xover method is to apply Butterworth 2nd order hpf to the speaker and Linkwitz 4th order lpf to the sub at the xo frequency. Most speakers are considered have natural BW2 roll off and when this adds up with your applied BW2, it's equal to LR4 of the sub and will sum flat. The exact xo frequency is dependent on the target curve used but yours seem around 120hz. So you apply a 120hz BU2 hpf to each speaker, vector average these two. Apply 120hz lpf LR4 to sub and use alignment tool to align this sub response with the lpf to vector average of speaker (use impulse alignment at 120hz). Rew will give you exact timing and inversion requirement of the sub. Then use vector sums of aligned sub + speaker (with lpf and hpf applied) for eq filtering.

Ok so basically Wiim PEQ can set HPF and LPF in PEQ page, and depending on the Q given I could create a BU2 I'd do a HPF with a Q of 0.70 or 0.71 which is as close as possible to a BU2 (wish Wiim ultra let the hpf and lpf filters Q to be adjusted by 3 decimals but they don't) then I'd use the subwoofer LPF which has -6db to -24 db slopes (in 6db increments) so a -24db slope = LR4, so I can try to do what you suggested.

In wiim app sub out page I'd select main speakers output bass (since HPF is coming from L/R/Stereo PEQ), and subwoofer bypass mode (since sub is using it's LPF with 24db slope). which basically ignores the wiim crossover from sub out page.

edit:
hmm maybe it doesn't work as the PEQ with HP at crossover frequency will affect sub aswell.

 

[JD1]

mock attempt at PEQ since I found out using wiim L+R peq which gives 20 bands (10 bands per L+R) also allow to set PEQ for subwoofer by using twice the value of DB gain meaning if I have to cut 6db at sub I can use -12 at either R or L PEQ band, giving me more bands to play around with. If I have to cut -10 I do need to both use a PEQ band at both left and right however anything at 6db or below I can use only one band. didint sweep and try to even phase or TA which I hope improves crossover region at 115 hz.

 

[OCA]

Ok so basically Wiim PEQ can set HPF and LPF in PEQ page, and depending on the Q given I could create a BU2 I'd do a HPF with a Q of 0.70 or 0.71 which is as close as possible to a BU2 (wish Wiim ultra let the hpf and lpf filters Q to be adjusted by 3 decimals but they don't) then I'd use the subwoofer LPF which has -6db to -24 db slopes (in 6db increments) so a -24db slope = LR4, so I can try to do what you suggested.

In wiim app sub out page I'd select main speakers output bass (since HPF is coming from L/R/Stereo PEQ), and subwoofer bypass mode (since sub is using it's LPF with 24db slope). which basically ignores the wiim crossover from sub out page.

edit:
hmm maybe it doesn't work as the PEQ with HP at crossover frequency will affect sub aswell.

0.707 - sqrt(2)/2 is the typical Q value of a 2nd order filter.

 

[JD1]

Did everything from scratch final result. the 115 which is crossover point is also a room mode which seems pretty intense and I'm struggling to improve it 324 seems some kind of L+R cancellation.

 

[Keith_W]

Did everything from scratch final result. the 115 which is crossover point is also a room mode which seems pretty intense and I'm struggling to improve it 324 seems some kind of L+R cancellation.

Timing and gain adjustments can only get you so far. Sometimes you need to do more - find better placement for speakers and subs, or buy more subs. More sophisticated DSP techniques, e.g. run your speakers full range and overlap freqs with your subs, and then adjust the phase of small freq bands to remove cancellation caused by the interaction, might buy you a bit more perfection ... but I wouldn't count on it.

Many years ago I came across a forum post that went like this: you could design a 12 sided tyre and use sophisticated computerised suspension to smooth out the ride. But you would have been better off designing a ROUND tyre in the first place. So it's not as if DSP adjustments do not have a role. But DSP should be thought of as the icing on the cake. If your speaker/room setup is fundamentally flawed, the solution is to fix those problems and fine tune the rest with DSP.

 

[boxerfan88]

IME its quite challenging to deal with the 100-200Hz area. A small shift in speaker position or listening position can change the FR in that area quite a bit. Eventually I learned to leave that area alone unless the peaking is too high and/or it resonates.

The peaking around 550 can be trimmed using a wide-Q gentle cut, or left alone if it already sounds right (ie. not boxy nor nasal)

 

[JD1]

I wish Wiim would give us Wiim folks 10 channels dedicated to sub out ='( so basically today I redid everything, yes I know, can only get so far with EQ, but I think today I'm happier than yesterday, also did buy ear protection, and a laser distanciometer, so I did it at -10 dbfs and 90% wiim volume (limited at 95%), speakers are perfectly at same distance to LP, and impulse response confirms this, so I bypassed mains sub out crossover so now I have sub plus mains doing bass. which filled some weak spots. results below.

sub only eq is applied directly at sub internal amplifier as is the 120 hz crossover.
the first 3 left and right channel peq bands are actually for the sub but wiim doesnt have seperate sub PEQ ;'( .

 

[JD1]

I feel it's much better, now if Wiim gives me seperate sub out PEQ I think it can become as good as it can ever get without room treatment.

Bass is much more controlled, music sounds much more clearer.

 

[JD1]

Can anyone explain why periodic stereo uncorrelated pink noise gets summed wrong by subwoofer???? is it cancellations by summing??? but its fine after 50 hz, also correlated is working well.

 

[JD1]

anyone can check ta the subwoofer seems IR matched to L and R individually but I get phase cancellation in xover (99 hz) although room mode at 100hz - 110hz makes it worse. which leaves me wondering if I should even delay L+R

 

[NTK]

Can anyone explain why periodic stereo uncorrelated pink noise gets summed wrong by subwoofer???? is it cancellations by summing??? but its fine after 50 hz, also correlated is working well.

Because uncorrelated signals sum "differently", and the summing is usually treated as a statistical (random) process. The resulting sum follows a probability distribution that can be described with a mean and standard deviation.

For two correlated signals (A & B) that are time varying, the relationship between signal A and signal B does not change with time. You can predict (determine) signal B from signal A when you know enough about the system.

For uncorrelated signals, you can't determine one from the other. Their relationship varies from moment to moment. And when they are perfectly uncorrelated, the relationship is random. That's why you get different sums from correlated and uncorrelated signals.

Below is from the Wikipedia page for correlation. The plots represents two signals X and Y. Each point (its coordinates x, y), represents the X and Y signals sampled at the same moment in time. The numbers above the plots are the Pearson correlation coefficients. When it is 1 or -1, the signals are perfectly positively or perfectly negatively correlated, and the points lie on a line. Once you have enough points to determine the formula of the line, you can predict any future Y from X, and vice versa. When the correlation coefficient is 0, they are perfectly uncorrelated and there is no relationship between X and Y. (The signals are often said to be independent of each other.) That's why correlated signals and uncorrelated signals sum differently.