• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Bass null with two speakers but not on their own?

@Farenheit so I just tried doing this and it works. It's mindboggling because my original measurements, when I use a sub, were measuring that null with both the left and right channels. I'm guessing that the sub it contributing to it even at a 95hz crossover. So the EQs are boosting the left by 10db and the right by 7.9db. But when I boost both channels the null doesn't budge a single db. So instead I inversed the left from a 10db boost to a -11db reduction and lifted the right to a 10db boost. It's cut the null from a -10db dip to a -5db dip. So it's not completely fixed it but it's made a huge difference. I did all of this inversing and changing in increments and each time the delta between -10db and -5db got smaller. I'm sure I could take it further but I'm concerned about boosting it too much as +10db is already big.

All of this happened without having any effect on the rest of the frequency range so I'm looking pretty good now.

While doing this I also experimented with having my door open or closed. Previously I had kept it closed but I'm realised that the bass roll off changes from 20hz to 10hz with the door open. A boost for me as I listen to electronic music :)
Let's see:

I tried with up to two subwoofers, but despite measuring better, the bass was, I don't know how to say... "not compact", even though my Swiss army knife (DBX Venu360) gives me the opportunity to vary the cut, phase, time, etc. of each subwoofer.

The secret is to dedicate an extra hour to equalization. My FARENHEIT method is not about lowering a specific frequency on one channel and raising a specific frequency on the other. You have to start by lowering the specific frequency on one channel with a Q16 where the null resides and see what happens without equalizing the other channel and measuring again.

From there, it may be good to also lower another frequency with Q 16 to 2 Hz or 4 Hz below or above the main null peak, or even 9 Hz! and measure again.

When you see that it has no greater effect on the null, then you go to the other channel and go up to "fill" the null in the same way, but not necessarily the same frequencies and the same Q!

It's a trial and error exercise, but I guarantee that with patience you will flatten it, you will make money selling your subwoofers... and your partner will love you even more, (great night of passion, I predict, hahaha)



I repeat:

FAST, EASY AND FOR THE WHOLE FAMILY
 
Last edited:
What would I be asking for? The ability to rephase individual speakers? Or is it more specific than that?
So indeed, the pre-requisites from the system are:
  • An Ability to EQ each channel separately (again feature may already be supported)
  • A decent(*) convolver that supports uploading *.wav (or similar files).
    (*) important, I am using a PC as a streamer, so computation power is good, and I can set up compensation filters with large number of taps (>100K-300K as "a standard") and get an accurate compensation filter.
    In standalone system (like, e.g.: minidsp, WiiM) the computation power is limited so number of taps\filter length may be smaller (e.g. <10K), this may also lead to reduced accuracy and effectiveness of the compensation filter.
    I did not try to limit the length of my filters in order to 'save' computation power, so I do not have a lower threshold for the filter length; moreover, it is markedly dependent on the behavior of the Lt & Rt phase at your room (hence very specific) so I do not want to generalize. I believe that for my room, it will take more than >50K taps to retain a solid performance.
 
First several considerations:

- My method is self-taught and unorthodox, but it measures me and sounds good.

- My room is asymmetrical and I don't know how it would work in symmetrical rooms.

- Regarding your question "Didn't that result in putting too much strain on one of the speakers by heavily boosting it?: my speakers have 15" woofers and high sensitivity (JBL S3100), but I amplify them with 20w per channel in a 30 m2 room and I have no problems. In fact, I have another YAMAHA P3500 350w professional amplifier and I don't feel the need to use it.

Therefore, I present to you...


"FARENHEIT METHOD FOR DESPERATE NULLS":


The graph with a red line corresponds to the measurement at the L+R listening point without any equalization where the null is observed with a vertex at 62 Hz.

At that frequency I tried applying a PEQ with Q 16 and -10 db on only one of the speakers (the cause could be the left or the right), observing that it largely disappeared, so I applied the inverse equalization on the other speaker and filled the null, as can be seen in the green trace graph.

However, you have to play with different +/- EQs at various points to get a flat response.

Finally, I applied equalization to L+R to flatten the response down to 200 Hz (blue trace graph).

Easy, fast and for the whole family.
Very interesting stuff. Before I try this with my S3100mk2 - how are your distortion measurements at the null position after that procedure? In my experience, boosting a null in any way balloons up the distortion into double digits.

Also - cool to see your S3100 go down to 20hz in-room. Mine only really manage 30. I have recently discovered the pleasure of low Q bass boosts as recommended by Greg Timbers for all „Japan“ tuned products, really transformed the speakers. The native bass response almost seems broken without EQ.
 
Last edited:
Let's see:

I tried with up to two subwoofers, but despite measuring better, the bass was, I don't know how to say... "not compact", even though my Swiss army knife (DBX Venu360) gives me the opportunity to vary the cut, phase, time, etc. of each subwoofer.

The secret is to dedicate an extra hour to equalization. My FARENHEIT method is not about lowering a specific frequency on one channel and raising a specific frequency on the other. You have to start by lowering the specific frequency on one channel with a Q16 where the null resides and see what happens without equalizing the other channel and measuring again.

From there, it may be good to also lower another frequency with Q 16 to 2 Hz or 4 Hz below or above the main null peak, or even 9 Hz! and measure again.

When you see that it has no greater effect on the null, then you go to the other channel and go up to "fill" the null in the same way, but not necessarily the same frequencies and the same Q!

It's a trial and error exercise, but I guarantee that with patience you will flatten it, you will make money selling your subwoofers... and your partner will love you even more, (great night of passion, I predict, hahaha)



I repeat:

FAST, EASY AND FOR THE WHOLE FAMILY
@Farenheit thanks for this. So you're saying reduce just one channel at 115hz with a q factor of 16 while leaving the other channel alone? So I'll be reducing quite a large frequency range because a q factor of 16 is a really large range right? I'll give it a trying and see. I'm a bit confused as you did say that you boosted the other channel but I must have misinterpreted you.
 
I've been using REW and the WiiM peq slots to do some room correction and when I measure the speakers individually (post eq) they measure pretty well. However, when I measure them together it shows a big bass null between 102 and 115hz.

Open an RTA window on REW and play some music.

You'll probably find the null when the music in the range of the null is mono (mono bass) but not when that frequency range is stereo.

That's what happens here.

Using test sweeps, left right and both speakers:

1737393789678.png




Stereo bass music RTA Peak (red) using music with stereo bass, No broad dip.

1737393819917.png



Mono bass gives a result similar to the sweeps.

I don't "hear" it, and didn't know it was there until measured, and don't worry about it.
 
Open an RTA window on REW and play some music.

You'll probably find the null when the music in the range of the null is mono (mono bass) but not when that frequency range is stereo.

That's what happens here.

Using test sweeps, left right and both speakers:

View attachment 422631



Stereo bass music RTA Peak (red) using music with stereo bass, No broad dip.

View attachment 422633


Mono bass gives a result similar to the sweeps.

I don't "hear" it, and didn't know it was there until measured, and don't worry about it.
From what I understand, though, almost all bass is mixed in mono below 200hz, so fixing that null should probably still be beneficial, no?
 
Boosting the signal to cancel a true null is not something I would encourage. If the null is from a room mode or other boundary reflection (SBIR), then boosting the signal may not help (0 * 10 is still 0), and means that frequency will be very loud out of the null (anywhere else in the room). The other issue is that (e.g.) a 10 dB boost requires 10 times the power, so your amp and speakers (subs) may quickly run out of headroom trying to overpower a null. I prefer to solve it other ways, including just ignoring it if it is very narrow.

IME/IMO - Don
 
From what I understand, though, almost all bass is mixed in mono below 200hz, so fixing that null should probably still be beneficial, no?

Depends upon what to which you listen, I suppose.

Run the RTA on your music as suggested above with the mic at the listening position and see what you get.

If you see AND hear a hole, work on it.

I don't seem to, so, I leave it be, maybe I'll do something later if I'm still alive when I get around to it. Soaring is my new little waste of time that makes a life for me.

1737396564071.png
 
Depends upon what to which you listen, I suppose.

Run the RTA on your music as suggested above with the mic at the listening position and see what you get.

If you see AND hear a hole, work on it.

I don't seem to, so, I leave it be, maybe I'll do something later if I'm still alive when I get around to it. Soaring is my new little waste of time that makes a life for me.

View attachment 422649
Will do that & may I say, what you describe sounds absolutely grand. I’ve taken notice of your new hobby before - colour me green with envy, what with me living in a small European country and all
 
Can you provide a link? Or more details about the recommended equalizer.
Will screenshots suffice? I had these saved - that’s the post by GT which I was alluding to. He talks about the 4367 and mentions that most of the larger systems made for the Japanese market are lean in the bottom octave and would do well with EQ. I apologise if there was a misunderstanding; from memory, GT does not recommend any specific hardware towards that end
 

Attachments

  • E8054DBE-A7F3-4E46-8793-FB4927FDD220.png
    E8054DBE-A7F3-4E46-8793-FB4927FDD220.png
    582.7 KB · Views: 31
  • 11147859-85C9-4442-9CF0-1B10336C6D24.png
    11147859-85C9-4442-9CF0-1B10336C6D24.png
    471.3 KB · Views: 30
Thanks for this @ebslo

Is there anywhere I can read up on this a bit more. But like an "all filters" for dummies in REW because I'm a bit of a noob lol. I'm not even sure if the WiiM allows me to do it. It has 10 peq slots per channel but not much else.
I don't have a WiiM, but it looks from their howto's that it doesn't support all-pass. It's a type of PEQ filter, but they only mention types peak (PK), low-shelf (LS), and high-shelf (HS) as being available, with support for high-pass and low-pass in the future. No mention of all-pass, sorry.

In case you or others are still interested, read on. You can see the phase response of an all-pass filter here. Since your cancellation is around 108Hz, if you were to put an all-pass on the channel with higher phase around 93Hz and an all-pass on the channel with lower phase around 123Hz, it would bring the phases closer together in the problem area, while keeping the relative phases unchanged at lower and higher frequencies. Of course you would need to muck around with the precise break frequencies and Q's to get the best result, but that's the general idea.
 
You‘re welcome!

To answer your question: I EQ L+R together up to 200hz to fix room modes and add a 6.5db 1-2Q boost at 55hz. An additional low Q boost of 5db at ~40hz is needed to have full output down to 30 in accordance with the Harman curve. Works great, but I have two subwoofers and prefer to see less cone movement, so I currently only use the one boost. Outputting subwoofer levels of bass really makes the 15“ swing. It’s almost scary

I then EQ each speaker individually from 200-500Hz.

I have tons of measurements for these speakers, as well as detailed pictures of the drivers/insides/horn and crossovers, which I had to rebuild from scratch with a friend. Will share at a later date in a different thread:)
 
Personally I would not do anything to try to fix this (IMO) non issue. Here's why:

- main speakers are meant to be played with a signal specifically designed to come from one speaker, not two speakers combined with precisely the same signal, and as such the L+R test is somewhat meaningless unless you plan on only listening in Mono
- any pair of two speakers in a room will cancel each other out at certain frequencies and at a given mic position (a form of comb filtering), but our ears are able to ignore this filtering as we have two of them separated by about 6" with complex ear canals to boot to compensate
- research on psycho acoustics has shown that our hearing gives a significant priority on the direct sound that hits our ears first, and uses the mixed phases of sound that come after, which can result in cancellations in a FR plot (which is a sum of all the reflections typically in a 0.5s long time window), to interpret the spatial characteristics of the room or the soundstage. A frequency response plot is not able to separate this direct from indirect sound and how it might be interpreted by our hearing (and it's not meant to) - you need time domain plots for that purpose
- if you try to "solve" the issue and add any type of AP filter, or muck around with the phase of one speaker relative to the other, you are risking screwing up imaging between the speakers, both for Stereo sound and also 5.1 or Atmos/DTS mixes making the placement of sounds and objects incorrect or difficult to place in the soundstage. Even an AP filter set at 110Hz will affect the phase at a range of frequencies above and below it depending on the Q of the AP filter


That said, adding REW responses together and looking at their combined Frequency response is appropriate and highly desirable when looking to combine multiple subs into one single "virtual" sub
 
Personally I would not do anything to try to fix this (IMO) non issue. Here's why:

- main speakers are meant to be played with a signal specifically designed to come from one speaker, not two speakers combined with precisely the same signal, and as such the L+R test is somewhat meaningless unless you plan on only listening in Mono
- any pair of two speakers in a room will cancel each other out at certain frequencies and at a given mic position (a form of comb filtering), but our ears are able to ignore this filtering as we have two of them separated by about 6" with complex ear canals to boot to compensate
- research on psycho acoustics has shown that our hearing gives a significant priority on the direct sound that hits our ears first, and uses the mixed phases of sound that come after, which can result in cancellations in a FR plot (which is a sum of all the reflections typically in a 0.5s long time window), to interpret the spatial characteristics of the room or the soundstage. A frequency response plot is not able to separate this direct from indirect sound and how it might be interpreted by our hearing (and it's not meant to) - you need time domain plots for that purpose
- if you try to "solve" the issue and add any type of AP filter, or muck around with the phase of one speaker relative to the other, you are risking screwing up imaging between the speakers, both for Stereo sound and also 5.1 or Atmos/DTS mixes making the placement of sounds and objects incorrect or difficult to place in the soundstage. Even an AP filter set at 110Hz will affect the phase at a range of frequencies above and below it depending on the Q of the AP filter


That said, adding REW responses together and looking at their combined Frequency response is appropriate and highly desirable when looking to combine multiple subs into one single "virtual" sub
@DaveBoswell and @RayDunzl thanks. I think you're both saying the same thing right. That when I'm doing L+R sweeps it's playing mono and that causes the null. In stereo that shouldn't be the case. @jeffaegrim points out that bass is in mono? I will test it out using the RTA window in REW in the next day to see.

It is worth noting that I do get the nulls at 115hz with both channels when measuring them individually. So even if the stereo playback does show that the null doesn't exist in stereo that would only be the case when I've boosted both channels. I'll set some time aside and play around with it.
 
Personally I would not do anything to try to fix this (IMO) non issue. Here's why:

- main speakers are meant to be played with a signal specifically designed to come from one speaker, not two speakers combined with precisely the same signal, and as such the L+R test is somewhat meaningless unless you plan on only listening in Mono
- any pair of two speakers in a room will cancel each other out at certain frequencies and at a given mic position (a form of comb filtering), but our ears are able to ignore this filtering as we have two of them separated by about 6" with complex ear canals to boot to compensate
- research on psycho acoustics has shown that our hearing gives a significant priority on the direct sound that hits our ears first, and uses the mixed phases of sound that come after, which can result in cancellations in a FR plot (which is a sum of all the reflections typically in a 0.5s long time window), to interpret the spatial characteristics of the room or the soundstage. A frequency response plot is not able to separate this direct from indirect sound and how it might be interpreted by our hearing (and it's not meant to) - you need time domain plots for that purpose
- if you try to "solve" the issue and add any type of AP filter, or muck around with the phase of one speaker relative to the other, you are risking screwing up imaging between the speakers, both for Stereo sound and also 5.1 or Atmos/DTS mixes making the placement of sounds and objects incorrect or difficult to place in the soundstage. Even an AP filter set at 110Hz will affect the phase at a range of frequencies above and below it depending on the Q of the AP filter


That said, adding REW responses together and looking at their combined Frequency response is appropriate and highly desirable when looking to combine multiple subs into one single "virtual" sub
Your contribution is very interesting. It is a point of view and that is what it is about, enriching ourselves in the space that a forum like this offers us.

But 99% of the recordings are mono approximately from 60 Hz down except extreme cases of more or less old recordings.

Let's take TAKE FIVE by Dave Brubeck & Paul Desmond as an example. The double bass is poorly panned and in the left channel it could sound bad due to the equalization applied according to my method, but it is also true that the drum kick is recorded with a single microphone IN MONO.

From that date until today, the bass drum of a drum set (as of any other instrument) would not be heard if we did not correct the 60 Hz null.

It's just a compromise solution.

Physics is very stubborn...

To your statement of:

"any pair of two speakers in a room will cancel each other out at certain frequencies and at a given microphone position (a form of comb filtering), but our ears can ignore this filtering since we have two of them about 6 apart" with complex ear canals to compensate."

I think it is correct, but my method would only be applicable to omnidirectional frequencies. I assumed it was understood.
 
Your contribution is very interesting. It is a point of view and that is what it is about, enriching ourselves in the space that a forum like this offers us.

But 99% of the recordings are mono approximately from 60 Hz down except extreme cases of more or less old recordings.

Let's take TAKE FIVE by Dave Brubeck & Paul Desmond as an example. The double bass is poorly panned and in the left channel it could sound bad due to the equalization applied according to my method, but it is also true that the drum kick is recorded with a single microphone IN MONO.

From that date until today, the bass drum of a drum set (as of any other instrument) would not be heard if we did not correct the 60 Hz null.

It's just a compromise solution.

Physics is very stubborn...

To your statement of:

"any pair of two speakers in a room will cancel each other out at certain frequencies and at a given microphone position (a form of comb filtering), but our ears can ignore this filtering since we have two of them about 6 apart" with complex ear canals to compensate."

I think it is correct, but my method would only be applicable to omnidirectional frequencies. I assumed it was understood.
Thanks for the additional information @Farenheit

However, I was clear that my issue is at 115hz. So I'm confused if your suggestion applies to that null or not?
 
I would love to read about it. At one time I wanted such asymmetrical horns, but then I stayed with the regular ones. I found an owner of 3100 nearby, maybe I'll listen to them someday, I'm curious.
You probably made the safer bet. I just wanted a pair of speakers with good dynamics and some directivity control. I figured, the effect from the asymmetrical horns was an afterthought in the „baby“ Everest - but no, it still kind of works. Only matters when you’re sitting off axis, of course. Very peculiar speakers with clear advantages/disadvantages. High DI makes everything sound „in your face“.

Will tag you when I’m done collecting all the pictures/data!

To get back on topic: @C0mbat, 115hz is considered a directional frequency - everything above 80hz is, afaik.
 
Back
Top Bottom