Correct high frequencies "holes" in EQ

[Dox82]

Hello,

Maybe this is a noob question, but I'm new of REW
I'm trying to generate some EQ filters for my mesurements. In general, I'm able to generate a good EQ but there are some "holes" in the bass range that seem to be "ignored" by REW during the "Match Response To Target" process.

Please, look at attached picture. There is some way to "fix" che frequencies "in the pit" around 65Hz, 95Hz and 140Hz?

Many thanks in advance!
Bye

 

[ddude003]

Generally speaking those are nulls... A common thing in "small rooms"... You won't want to boost those... Without a little more information about your room or system and what your use case is (Stereo, HT, etc) there are a few things you can try... Move your sub(s) if you have some... Move your main listening position... Add bass absorbers and if you only have one sub, you could add another...

In REW you can use the Room Simulation to see what may be going on in your room with your speakers...

 

[Dox82]

Hello!

Thanks for your reply!

I have no sub. I have a pair of Neat Ekstra in an HiFi setup. Room is treated with panels and traps and they have reduced the nulls but there something yet, especially around 100Hz. 65Hz and 140Hz are "recovered" mainly by the R channel (not affected by those nulls), but both L and R have the nulls at 100Hz. The L channel (shown in the pic above) is the most problematic but is difficult to say which is the destructive element in the room: maybe the door in the opposite side, maybe the window in the same side (there is a panel in front of the window for the first reflection here). I suppose the culprit is a room element that I can't "remove" or "move".

So, in general, do you say this kind of nulls doesn't have to be corrected with convolution filters but investigating the room cause?

Generally speaking those are nulls... A common thing in "small rooms"... You won't want to boost those... Without a little more information about your room or system and what your use case is (Stereo, HT, etc) there are a few things you can try... Move your sub(s) if you have some... Move your main listening position... Add bass absorbers and if you only have one sub, you could add another...

 

[ddude003]

Is it your floor or ceiling??? A coffee table??? It does seem those speakers are down firing for bass... Hard to predict what is really going on in the room...

I believe boosting nulls just magnify the issue of nulls... Below the Schroeder frequency the room is in control... And 15db is a lot to boost... You might try moving the mic around the room and see if you can find the issue... 100Hz is about 9.8 feet...

 

[Dox82]

eh, the main problem is the room dimension: 10.5 feet x 10.8 feet x 9.51 feet (9.51 is the ceiling and this one isn't treated).

I've tried to boost the nulls and the problem seems to disappear but I exaggerated and in some cases the left speaker goes into overdrive. I can work on it but if this isn't the right methodology is better to avoid it...

Is it your floor or ceiling??? A coffee table??? It does seem those speakers are down firing for bass... Hard to predict what is really going on in the room...

I believe boosting nulls just magnify the issue of nulls... 15db is a lot to boost... You might try moving the mic around the room and see if you can find the issue... 100Hz is about 9.8 feet...

 

[ddude003]

Ah, the dreaded square room... Lol... Almost a cube... Try and visualize what is going on with https://amcoustics.com/tools/amroc... The thing with boosting is how many watts does it take to boost by 1db... Now think of what it takes for 15db... The other thing is if there is pre-ringing and if it is audible... To be clear... Boosting nulls isn't the right methodology...

And this may be a fun read for you... https://www.audiosciencereview.com/forum/index.php?threads/solving-bass-nulls-with-dsp-or-not.35473/ Especially post #5...

 

[Dox82]

Ok, thanks! Basically I have to investigate the 100Hz because the others two (pseudo-)nulls are "compensated" by the right speaker in a good way. The really bad guy is the 100Hz for both speakers and in some tunes this is really evident. !

 

[ddude003]

Your welcome!!! Let us know how it goes so others may learn...

 

[DanDan]

Dips are caused by a destructive reflection presenting the same frequency back on itself, but out of phase. Cancelling or weakening. Deep narrow dips such as yours suggest strong cancellations. Increasing power at that frequency will only increase cancellation pro rata. You can't cancel a reflection.....

 

[Dox82]

Nice, than we need a methodology to find where is this reflection in the room and try to avoid it with panels and so on...this is the problem now
Something happens boosting the dips a little in terms of "resurrection" of those frequencies, but there is a bad repercussion on others frequencies reducing the air of the sound and introducing some "boom" somewhere else.
I think moving some traps and first reflection panels I'm improving the situation, now I do a new mesurement

Dips are caused by a destructive reflection presenting the same frequency back on itself, but out of phase. Cancelling or weakening. Deep narrow dips such as yours suggest strong cancellations. Increasing power at that frequency will only increase cancellation pro rata. You can't cancel a reflection.....

 

[Dox82]

And now:

1.
I advanced the listening position of 10cm.
2.
I moved (again) the traps in the back not at 45° in the corner but more towards the center of the room (a little, the corner is ever covered).
3.
I moved the first reflection panels more close to the speakers.

The situation has changed! I think in a better direction! Basically the nulls are decreased and, magically, there isn't the 100Hz nulls in both speakers, but now the left one has the major nulls at 65Hz and the right one has the major nulls at 135Hz. Because one speaker balance the other one, now the situation is audibly better listening to the most critical tunes.

 

[Dox82]

Another update: I found some pre-ringing in some tunes (I think this is a pre-ringing what I'm listening). This happens in a couple of tunes but there is. Basically I think this is do to the "Individual Max Boost" at 18db which is applied to some frequency. Lowering the "Individual Max Boost" to 12db the problem seems gone.

In any case I did some additional mesurements and I'm not able to find a specific guilty in the room for the null at 90Hz and 135Hz, I think they are a "limit" related to the room and maybe correlated to the speakers also.

 

[ddude003]

Some thoughts... Have you tried REW's Room Simulation to model your room and speaker placement... I am wondering if moving your speakers forward or back a few inches and/or side to side would make much of a difference... Are they perfectly symmetrical? And are you using REWs timing to see if they are time/distance alined? Are you toeing them at all?
This is a little unorthodox... And... Have you tried moving a bass free standing absorber around in the room to see if you can "absorb" the null(s) a bit... Have you tried a house curve other than flat? And are you using REWs EQ filters or trying REW's new math inversion features?

And again... Boosting the nulls is not the way to go... And at some point you will have to compromise for what sounds best to you with the system and room you have... Its not about a pretty flat graph...

 

[Dox82]

Hello,

I'm not boosting the nulls anymore. Unfortunately I don't know many things you talking about like house curve, math inversion, the time/distance and I'm not sure to really understand the room simulation info. Sorry, I have to study, many things are new for me. Probably I have to experiment more speakers placement...

To say the truth, nulls are only a part of the problem, as you can see in the graphs there is some "boom" until 130Hz, but calm down those peaks is more easy via EQs filters.

Some thoughts... Have you tried REW's Room Simulation to model your room and speaker placement... I am wondering if moving your speakers forward or back a few inches and/or side to side would make much of a difference... Are they perfectly symmetrical? And are you using REWs timing to see if they are time/distance alined? Are you toeing them at all?
This is a little unorthodox... And... Have you tried moving a bass free standing absorber around in the room to see if you can "absorb" the null(s) a bit... Have you tried a house curve other than flat? And are you using REWs EQ filters or trying REW's new math inversion features?

And again... Boosting the nulls is not the way to go... And at some point you will have to compromise for what sounds best to you with the system and room you have... Its not about a pretty flat graph...

 

[ddude003]

No worries @Dox82... Take your time... You have done well so far...
It would really help if you filed out your profile and some more info about your kit... You said convolution... What do you use?
House curve = https://www.hometheatershack.com/threads/house-curve-what-it-is-why-you-need-it-how-to-do-it.96/
and https://accuratesound.ca/standards.html see room response curves
Math inversion = https://www.you tube.com/watch?v=5YcH7j2-L1Y&t=14s take out the space between you and tube...
Time/distance = https://www.roomeqwizard.com/help/help_en-GB/html/makingmeasurements.html see timing reference
Room sim = https://www.roomeqwizard.com/help/help_en-GB/html/modalsim.html

You may be able to find a copy of Master Handbook of Acoustics by F. Alton Everest & Ken C. Pohlmann on the web... Google is your frenemy... Chapter 12 has lots of information about building various types of absorbers... Some that can be tuned to specific frequencies...

 

[DanDan]

Dip are caused by reflections, which can be resonant or non resonant. The resonant ones are visible in the Waterfall as Ridges and Valleys. The non resonant single reflections from boundaries are visible in the ETC graph or EchoGraph as I like to think of it. Some of these concepts are best understood by poking at them IMO. Move the microphone towards a suspected reflection...... The line on the ETC should move a little towards 0 time delay. The frequency of the dip should rise. And vice versa. Modal, resonant, reflections are set up by boundaries bouncing energy back and forth. Their frequency cannot be changed by moving the mic or speaker or listener.

 

[Dox82]

As I experimenting, the dip at 60-80Hz is unavoidable, indipendently to the speakers/listener position.The RT under 80Hz increases from 0.8s to 1.5s at 60Hz and this surely the cause of the "boom" and the main "caos" in the bass range. I think this one can be cancelled using specific traps only, maybe on the ceiling also. Conversely, the dip at 80-100Hz and 130-150Hz are "position dependent" and can be tamed down with speakers and/or listener position. Effectively, I've cut out the last ones with a new position, placing the speakers really close to the rear walls where there are the bass traps. Obviously, we have to deal with the soundstage and imaging response but this is another story. Basically I have 2 ways to go: one is to insist in the room treatment with the help of a technician, the other one is to better understand the acoustic concept and the REW potentiality in order to do some correction at "software level". I'm using Roon to play digital content and I can use the convolution filters generated by REW (this is what I'm doing now) to correct some aspect.

Dip are caused by reflections, which can be resonant or non resonant. The resonant ones are visible in the Waterfall as Ridges and Valleys. The non resonant single reflections from boundaries are visible in the ETC graph or EchoGraph as I like to think of it. Some of these concepts are best understood by poking at them IMO. Move the microphone towards a suspected reflection...... The line on the ETC should move a little towards 0 time delay. The frequency of the dip should rise. And vice versa. Modal, resonant, reflections are set up by boundaries bouncing energy back and forth. Their frequency cannot be changed by moving the mic or speaker or listener.

 

[ddude003]

Hey @Dox82... I have been using various EQ methods for a few years now, because my Martin Logan ESLs are pretty unorthodox being hybrid electrostatics... I have found that Serkan's method of using inverse math, that John Mulcahy has recently added to REW, has delivered the best, so far, room correction, in my room with my kit... I think you should try it and see how it compares with the more traditional EQ method that you are using.. I use the FIR filters that this method creates in both Roon and Audirvana+LiquidSonics with great results... I have also been testing the CamillaDSP convolver using the same FIR filters with the same success...

Food for thought... If you cut the offensive frequencies, instead of boosting them, do you decrease the null(s)...

 

[DanDan]

@Dox82 Dirac Live works very well. Especially at getting to grips with the lowest mode.

 

[hemiutut]

Hello,

Maybe this is a noob question, but I'm new of REW
I'm trying to generate some EQ filters for my mesurements. In general, I'm able to generate a good EQ but there are some "holes" in the bass range that seem to be "ignored" by REW during the "Match Response To Target" process.

Please, look at attached picture. There is some way to "fix" che frequencies "in the pit" around 65Hz, 95Hz and 140Hz?
View attachment 56560 View attachment 56561
Many thanks in advance!
Bye

Can you put the REW .mdat file?
written with translator

Greetings

 

[Dox82]

Hello,

Yes this is my mesurement file in the current situation.
If someone is able to generate some better filters impulse response in wav (I have to feed Roon) is really appreciated.
I'm using the HarmanCurve attached as House Curve.
In the current configuration there are no manual boost of nulls (and they are but with less intensity than previous setup) and some minor hollow has been corrected automatically by REW not exceeding 12db for the single correction (there are a couple of them). I've corrected the range 20-280Hz only because I like the rest.
Now, I'm studying something about math inverse and timing reference. The current setup was driven by the REW room sim and I have to say the prediction is pretty accurate.

Can you put the REW .mdat file?
written with translator

Greetings

 

[ddude003]

I am pretty sure that this is the latest version of Serkan's inversion method... https://drive.google.com/file/d/17BwTGv933w4id3B3PFHy2zjyijtgZLDX/view

Can/will output room correction FIR in .wav format to be imported in any convolution engine that supports .wav FIR input including Roon, Audirvana+LiquidSonics and CamillaDSP which I have tested... I imagine there are many others like EqualizerAPO and HangLoose to name a few... There are also convolution reverb plugins that might work... Which is what LiquidSonics does...

 

[Dox82]

Hello!

After a lot of mesurements and movement of acoustic/not-acoustic elements in the room, I totally recovered the "hole" in the 130-150Hz!
Apart the "boom" in the bass range 40-60Hz (I'm planning to add some additional traps for that in the ceiling), the main "holes" are in the ranges 60-75Hz and 80-100Hz (the bad one).

The REW Room Sim shows a problem in the room in those ranges and maybe the current acoustic treatment isn't able to recover the situation and I need to add some specific traps.
But...looking at the phase graph there is something "weird" in the left channel (the one most affected by the "holes" problem). Can you confirm there is a problem here or I'm misinterpreting the graph?

 

[BenToronto]

Given all the efforts here in many directions, a missing starting point is running an impedance curve on the speakers. That immediately provides an X-ray of what the cone of the woofer is doing. Needs nothing but a 150-ohm resistor and some alligator clips, no need to calibrate.

B.

 

[Dox82]

Hello,

I don't know what you means but I can try to swap the speakers it the doubt is in the speaker functionality itself.
Basically, I think I'm not able to interpret correctly the phase graph. I see there is a "huge" difference just around 100Hz but I'm not able to understand what this means, neither which "element" can be the guilty.
Basically the 100Hz hole afflict both speakers, so I don't know if this "phase beahviour" is related to the hole problem.
Can someone clarify this graph difference with some hypotesis and tell me if it can be a real problem or not?

Given all the efforts here in many directions, a missing starting point is running an impedance curve on the speakers. That immediately provides an X-ray of what the cone of the woofer is doing. Needs nothing but a 150-ohm resistor and some alligator clips, no need to calibrate.

B.