Help with eq for an asymmetric room with REW

[soypeter]

Hi amigos,
I'm dealing with finding a right eq for my living room; its my main listening room. It's strongly asymmetric, and cant move the speakers so much.
I'm adding a picture of the room and REW/Umik measurements from the listening position (fixed and MMM),
from each channel individually and from both simultaneously.

The speakers are a DIY project from a friend of mine, D'Appolito configuration, so no spinorama available.
Filters will be implemented by my RME ADI 2 Dac Fs. The amp is an AUDIOPHONICS MPA-S250NC.

1. First question: below the Schroeder freq, must I eq based on both speakers sounding simultaneously?
(I've reading Dr Toole comments recommending that, but the location of the speakers make their FR so different...)
2. I've made quasi-anechoic measurements for the speaker, and in the crossover region (aprox 2,8Khz) there's a dip (visible in the measurements): it is fixable by eq (for example, a low Q)?

Thanks in advance,
Pedro

 

[janbth]

I have an asymmetric setup as well. From your picture, it looks like you have an open space to the left (so no reflections) and curtains/windows on the right. Similar for me. In my room, I try to reduce this asymmetry by toeing the right speaker inwards, in order to reduce reflections from the right boundary. Maybe you could try to experiment along such lines.

As for your two questions:
1. EQ based on L+R simultaneously. This is because mono content dominates in the bass. (Note: There are some who suggest EQing L and R separately to achieve «stereo bass». However, this is an advanced topic. Do L+R properly first.)
2. I say try it. Do listening tests with the EQ on and off. You will hopefully hear what works best.

 

[soypeter]

@janbth thanks for your advice. So I'm doing some experiments with both speakers on and a few filters. I'll keep you updated with the advances.

 

[Keith_W]

When you look at this, you have to wonder whether the right woofer (green) is outputting the same SPL as the left (red), or whether the SPL difference that we see is simply due to 8th space radiation.

1. First question: below the Schroeder freq, must I eq based on both speakers sounding simultaneously?
(I've reading Dr Toole comments recommending that, but the location of the speakers make their FR so different...)

Agree with @janbth, EQ them together. What you want to do is equalize that L+R MMM directly.

2. I've made quasi-anechoic measurements for the speaker, and in the crossover region (aprox 2,8Khz) there's a dip (visible in the measurements): it is fixable by eq (for example, a low Q)?

Please post that MDAT. You need to be sure that it's actually a quasi-anechoic measurement.

And one last thing, I recommend you get rid of that coffee table. It's causing a lot of problems in the ETC.

 

[thewas]

Agree with @janbth, EQ them together. What you want to do is equalize that L+R MMM directly.

For me and my rooms that never sounded as good as equalising below Schroeder left and right separately so I would always recommend to try both and let your ears decide.

 

[Mike123]

When you look at this, you have to wonder whether the right woofer (green) is outputting the same SPL as the left (red), or whether the SPL difference that we see is simply due to 8th space radiation.

What is surprising to you? They have pretty similar SPL. But room modes have a 6 dB greater impact on the right than on the left. This is the topic starter's problem.

 

[Mike123]

1. First question: below the Schroeder freq, must I eq based on both speakers sounding simultaneously?

I don't think it's worth it. Trying to perfect the frequency response at one point isn't the best idea. In my experience, you should be quite careful when correcting peaks below +6 dB. I'd leave the left speaker alone and slightly reduce the peaks on the right. Not to zero, but to somewhere around +5 dB. That might well be enough.

 

[soypeter]

View attachment 548025

When you look at this, you have to wonder whether the right woofer (green) is outputting the same SPL as the left (red), or whether the SPL difference that we see is simply due to 8th space radiation.

Thanks @Keith_W for your comments.
Yes, both are sounding with the same SPL.

Agree with @janbth, EQ them together. What you want to do is equalize that L+R MMM directly.

Ok, I'm working on that method.

Please post that MDAT. You need to be sure that it's actually a quasi-anechoic measurement.

Not at my pc now, but I'm pretty sure the gating and measurement is reasonably right.

And one last thing, I recommend you get rid of that coffee table. It's causing a lot of problems in the ETC.

Yeah, that table is problematic, but... is our dining table... so...

 

[soypeter]

I don't think it's worth it. Trying to perfect the frequency response at one point isn't the best idea. In my experience, you should be quite careful when correcting peaks below +6 dB. I'd leave the left speaker alone and slightly reduce the peaks on the right. Not to zero, but to somewhere around +5 dB. That might well be enough.

Thanks @Mike123 for your advice. Yes, I agree with that. I'm trying a couple of filter options and see how it sounds.

 

[soypeter]

For me and my rooms that never sounded as good as equalising below Schroeder left and right separately so I would always recommend to try both and let your ears decide.

Thanks @thewas for your advice. I'll try both. A doubt: eq'ing separately doesnt introduce some kind of interference, timing problems between the channels?

 

[Flaesh]

some kind of interference, timing problems between the channels?

The existing problems of LR non-identity may be reduced with proper equalization. If L and R APhR become more similar.

 

[thewas]

Thanks @thewas for your advice. I'll try both. A doubt: eq'ing separately doesnt introduce some kind of interference, timing problems between the channels?

You are welcome, not if you do it correctly, namely only correct modal peaks but don't try to fill narrow dips.

 

[RayDunzl]

I have an asymmetric room. The wall behind the couch has the usual corner on the right, and is open for another 10 feet or so on the left rear.

Frequency response, left, right, and both speakers (playing a mono sweep). Got a big hole there. Around 48Hz the bass goes like180 degrees out of phase between ledt and right.

With music, if the bass is "stereo". the hole usually disappears.

Red is the peak RTA frequency response with music, likely with stereo bass. Ignore the black, as it is the "instant" reading.

Mono bass tracks still give a hole, but I still don't "hear" it.


Conclusion:

It isn't something I can reasonably fix, isn't something I hear (didn't know a hole was there until measured), and doesn't seem to be a problem with what I listen to.

 

[janbth]

@soypeter A comment on computing PEQ filters to cut down peaks. If you are using REW to compute the parameters, in particular Q, you need to remove the smoothing from the plots, or use Var smoothing (no smoothing in the bass). The reason is that REW bases its EQ computation on the smoothed curve, and a smoothed peak will appear to have lower Q than the room mode actually has.

 

[soypeter]

Based on measurement with the two speakers on, I've chosen manually these filters, trying to not correct everything,
(the RME has 5 filters available, plus 2 tone controls).

38hz Q=5 -6db
72hz Q=5 -9db
97hz Q=5 -5db
2800hz Q=1.5 +3.5db (crossover region)

The measurements are below.
To my ears, it sounds nice most of the time. Some records sound more bassy, some more thinny.
In those cases, I'm using the bass control (LSQ, 80hz Q=0.5) +-2db.

I'll try to eq each channel separately, and compare with this.

 

[Mike123]

Some records sound more bassy, some more thinny.

Room correction is far from a perfect solution. It's the simplest and most cost-effective way to improve a poor sound, and it won't be perfect. If a song has notes that you've weakened, they'll play quietly but still sustained. The room's quality factor remains intact. If bass notes are outside the filters, the sound may be bassier, because without resonance, the volume is typically set higher. But on average, reducing resonance allows the music to play more smoothly, and you'll hear a lot of things your ears and brain simply didn't notice before, including in the midrange.

And i agree with janbth. It's advisable to set the smoothing to at least 1/12. Then the filters' quality factor will be higher than 5.

 

[thewas]

For room correction the REW manual recommends the variable smoothing.

 

[janbth]

And i agree with janbth. It's advisable to set the smoothing to at least 1/12. Then the filters' quality factor will be higher than 5.

I would say 1/12 is too smooth. No smoothing/Var is probably the best, but 1/24 could possibly work.

Another way to do it is to use REWs RTA and do a MMM measurement. Then, when the mic is moved around the listening position it will create a little spatial averaging, and smooth the peaks in a more «natural» way (in lack of a better term). Of course, it is necessary then to use max 1/48 smoothing of the measured curve before you send it to the EQ computation, so that no additional smearing occurs.

 

[soypeter]

Room correction is far from a perfect solution. It's the simplest and most cost-effective way to improve a poor sound, and it won't be perfect. If a song has notes that you've weakened, they'll play quietly but still sustained. The room's quality factor remains intact. If bass notes are outside the filters, the sound may be bassier, because without resonance, the volume is typically set higher. But on average, reducing resonance allows the music to play more smoothly, and you'll hear a lot of things your ears and brain simply didn't notice before, including in the midrange.

And i agree with janbth. It's advisable to set the smoothing to at least 1/12. Then the filters' quality factor will be higher than 5.

Agree. Room correction isn´t perfect but allow me to listen to music in a room where naturally is almost impossible.
Most of the records sounds fine. The rest, I accept that as a consequence of the "circle of confusion" in recording, mastering, etc.
About the filters: it is advisable to use Q>5? I've read about phase problems, etc

 

[Mike123]

About the filters: it is advisable to use Q>5?

It's recommended to use what's actually there. If you have pronounced resonances in your room, avoid attenuating adjacent frequencies that aren't resonant. When you use strong smoothing, the calculated filter attenuation and Q factor decrease. Regarding phase, minimal phase filters only improve the phase that's being affected by resonances.