Can you fix cancellation by changing phase?

[daftcombo]

By the way, my dip doesn't look as bad as yours because I used a 1/3 smoothing. If I remove the smoothing, it goes down 20dB as yours.

 

[reza]

You cannot "correct" this. Your ears "correct" the asymetry themselves.

Do you think you hear something strange? if not, just don't worry about it.

Not really, I don't notice the dip when listening. In fact when I measure both speakers, the dip is mostly filled.

 

[reza]

By the way, my dip doesn't look as bad as yours because I used a 1/3 smoothing. If I remove the smoothing, it goes down 20dB as yours.

Yeah, I figured there's some smoothing going on

 

[andreasmaaan]

Just to confirm that it is the sidewall reflection primarily, could you give the following approximate measurements:

• A - distance from speaker to sidewall
• B - distance from listening position to centre-point between the front baffles of the speakers
• C - distance from listening position to sidewall

Assuming it is the sidewall reflection, it is technically correctable using DSP (see the Linkwitz article @daftcombo linked earlier), but as daftcombo pointed out, it is not psychoacoustically important. And the cure is likely worse than the disease.

 

[reza]

Just to confirm that it is the sidewall reflection primarily, could you give the following approximate measurements:

• A - distance from speaker to sidewall
• B - distance from listening position to centre-point between the front baffles of the speakers
• C - distance from listening position to sidewall

View attachment 25548

Assuming it is the sidewall reflection, it is technically correctable using DSP (see the Linkwitz article @daftcombo linked earlier), but as daftcombo pointed out, it is not psychoacoustically important. And the cure is likely worse than the disease.

Relying on memory here, but numbers should be in the correct neighborhood:
A- ~75cm
B- ~200cm
C- ~200cm

I guess I can no longer avoid reading that article carefully then.

 

[andreasmaaan]

Relying on memory here, but numbers should be in the correct neighborhood:
A- ~75cm
B- ~200cm
C- ~200cm

I guess I can no longer avoid reading that article carefully then.

Thanks When I run those numbers I get a predicted sidewall reflection notch at 165Hz, so higher in frequency than your measured notch, and most likely that second, shallower notch in your measurement:

If you're interested in going deeper into working out the cause of the deeper notch slightly lower in frequency, the next step would be to give the room dimensions.

But honestly, it looks very likely to me to be a room mode cancellation frequency, and therefore something that can only be dealt with using room treatment, and very heavy room treatment (bass traps) at that...

 

[reza]

Thanks When I run those numbers I get a predicted sidewall reflection notch at 165Hz, so higher in frequency than your measured notch, and most likely that second, shallower notch in your measurement:

View attachment 25549

If you're interested in going deeper into working out the cause of the deeper notch slightly lower in frequency, the next step would be to give the room dimensions.

But honestly, it looks very likely to me to be a room mode cancellation frequency, and therefore something that can only be dealt with using room treatment, and very heavy room treatment (bass traps) at that...

Ah, thank you. I did actually enter the room dimensions in REW's simulation and got the following result. But I wasn't quite sure how to interpret the result. To be honest, I don't think I'll be pursuing a fix any further, particularly since it doesn't seem to be damaging the sound in any perceptible way.

 

[andreasmaaan]

Ah, thank you. I did actually enter the room dimensions in REW's simulation and got the following result. But I wasn't quite sure how to interpret the result. To be honest, I don't think I'll be pursuing a fix any further, particularly since it doesn't seem to be damaging the sound in any perceptible way.

I probably wouldn't try to tackle it either. The room is already very small, especially on the right-hand side, and getting effective treatment in there looks like it would be a tall order. Your measurements are actually very good given the size and shape of the room, and it seems subjectively to work too - stick with it IMHO

 

[andreasmaaan]

@reza one last thing you should do for the sake of completeness is to take a nearfield measurement of the right speaker's woofer. Just to rule out that there's something odd going on there, which would be very unlikely of course.

 

[reza]

@reza one last thing you should do for the sake of completeness is to take a nearfield measurement of the right speaker's woofer. Just to rule out that there's something odd going on there, which would be very unlikely of course.

In fact I've already done that. Nearfield (~20cm) of the right speaker without any DSP or post processing:

Nothing fishy going on. So the dip must be due to the reflections I think, right?

 

[andreasmaaan]

In fact I've already done that. Nearfield (~20cm) of the right speaker without any DSP or post processing:
View attachment 25553

Nothing fishy going on. So the dip must be due to the reflections I think, right?

Seems almost certain, yeh. FYI, it would be better to take the nearfield measurement closer to the woofer's dustcap, say 1-2cm from it. But it shouldn't show any significant difference, you clearly have healthy output in the frequency range of the in-room dip.

 

[RayDunzl]

Not really, I don't notice the dip when listening. In fact when I measure both speakers, the dip is mostly filled.

I have the opposite "fill" result here.

Left (green) or Right (blue) speakers measure well enough in SPL, but a phase anomaly shows up and cancels around 48Hz at the listening position when both (violet) are playing a mono signal:

I don't "hear" it. Didn't know it was there till measured. Stereo bass can fill it in.

Room:

Something like this:

 

[reza]

Seems almost certain, yeh. FYI, it would be better to take the nearfield measurement closer to the woofer's dustcap, say 1-2cm from it. But it shouldn't show any significant difference, you clearly have healthy output in the frequency range of the in-room dip.

Yeah, that was a quick and dirty nearfield measurement with the mic handheld kinda sorta close to the woofer.

@andreasmaaan and @daftcombo have been most helpful. I appreciate your help.

 

[levimax]

Of course. Here you go. Red is the right speaker with the dip, purple is the left speaker (I'm afraid I don't know how to place a legend in REW's graphs). Both are single measurements, not averaged, with Var smoothing.
View attachment 25534

I have a similar shaped room and a similar measurements, one channel has a dip and the other a peak around 100 Hz... if you look at sweep with both channels the "dip" will probably be much smaller and probably not a concern. Have you run a sweep with "both" channels? While there are differences of opinion I like to look at "both" before I start to do any EQ.... for sharp dips especially I like to error on the "less EQ" side of things.

 

[reza]

I have a similar shaped room and a similar measurements, one channel has a dip and the other a peak around 100 Hz... if you look at sweep with both channels the "dip" will probably be much smaller and probably not a concern. Have you run a sweep with "both" channels? While there are differences of opinion I like to look at "both" before I start to do any EQ.... for sharp dips especially I like to error on the "less EQ" side of things.

Indeed I have. The dip isn't really there when both channels are sweeped:

 

[levimax]

Indeed I have. The dip isn't really there when both channels are sweeped:

View attachment 25559

If it was me I would use ERB or Psycoacoustic smoothing on this sweep and import it into Rephase and use some low Q EQ adjustments and concentrate on getting the "peaks" down and getting a reasonably smooth 10 db downward slope keeping in mind "less is more" and then use the same correction on both channels and be done. I think you will be pleasantly surprised with results. Of course there are many other ways to "skin this cat."

 

[reza]

If it was me I would use ERB or Psycoacoustic smoothing on this sweep and import it into Rephase and use some low Q EQ adjustments and concentrate on getting the "peaks" down and getting a reasonably smooth 10 db downward slope keeping in mind "less is more" and then use the same correction on both channels and be done. I think you will be pleasantly surprised with results. Of course there are many other ways to "skin this cat."

I have tried a bunch of different approaches. In the end what worked best was not to mess with anything above 100 Hz, and not to follow any automatic filter assignment, like the one in REW. For me, violating these heuristics always ended doing more harm than good.