Wayne, I have a typical 160hz null. In simple terms, how can I fix this?
I have tried your method of "over crossing" the woofer and midrange but it sounded strange.
I have tried your method of "over crossing" the woofer and midrange but it sounded strange.
From a musical perspective I draw the line between what sounds like "bass" and what sounds like "midrange" at around 180-200Hz. A tone of say 160Hz still sounds clearly like bass to me. By the time it reaches 200Hz it sounds like midrange.
The significance of this is that as bass in music is generally individual discrete notes with most of the power in the fundamental, holes are much more obvious when a particular pitch note happens to fall into the hole, while most other instruments whose fundamental are at 200Hz or higher tend to have much more complex harmonic structures where some attenuation of the fundamental isn't nearly as obvious. (But still important, obviously)
Individual bass notes which disappear into a black hole is one of my pet peeves with many speaker/room setups, and is not easily solved with a single pair of speakers, as moving them around just tends to move the frequencies of the notches.
All the deep modal / boundary cancellation notches in most typical sized rooms seem to occur between 60-200Hz, with problems between 100-200 very common. I would put the dividing line where it's important to smooth modal notches and where it's not important at 200Hz, which I think not coincidentally is very close the Schroeder frequency of many rooms.
I think the whole discussion got a bit derailed due to confusion over the cause of the notches, and whether the notches being discussed were "floor bounce" notches, or just boundary/room mode related notches in general.
It's fairly obvious now that the notches we're discussing between 100-200Hz at the listening position - which are by far the worst in most rooms, are not floor bounce path length delta related notches - they're (probably) caused by vertical room modes as well as boundary cancellation effects.
Some playing with simulations of my room with the room response calculator suggests that no one boundary is solely to blame and that the depth of the notch is a cumulative effect of several of the boundaries including the ceiling - if I start with all boundaries except floor set to 0, and start adding in the various walls and ceiling, the notch appears early on and becomes progressively worse as more of the room boundaries are added back in.
In any case, physically lower sources of bass in the speaker design is the only thing that provides significant reduction in 100-200Hz notches in all room sizes and speaker positioning. As long as a slight EQ change is made in the woofers crossover (relative to mid/tweeter) to return to a balanced response, the lower woofer gives a more uniform response in different room positions than one with a higher woofer.
Also, one thing maybe overlooked in the discussion but very relevant is the perception of notches vs peaks in response. It's fairly well known that notches in an otherwise flat response are far less objectionable or noticeable than an equal sized peak at the same frequency. (Especially if the notch falls at "less important" frequencies)
In other words the absence of something is far less obvious than having too much of it. What I tend to notice with notches between 100-200Hz is that unless the music draws attention to it (such as a missing bass-line note, or a thin sounding male vocal) it's not always apparent that there is even a problem when there is no reference to compare it to. It's easy to convince yourself that things sound just fine on many pieces of music.
But then switch on a physically displaced (low) woofer filtered to only operate between ~100-250Hz or so (your "flanking sub") and you immediately notice what was missing. Only when the missing range is restored is it obvious that it was missing in the first place.
I wonder if some of the people that feel notches between 100-200Hz don't matter have tried carefully controlled A/B tests which very precisely target the range where the notch occurs.
Just lowering the height of an existing speaker design (for example using a lower stand on a stand mounted 2 way) doesn't really prove anything because there is so much change in the overall response by lowering the speaker that you're doing an Apples/Oranges comparison.
Likewise comparing a 2 way stand mounted design with a 3 way design with a low woofer is unlikely to prove much as the overall characteristics of the speakers are going to be a lot different - in particular the bass below 100Hz, so you don't know how much the difference in sound is related to curing the notch, and how much is other things.
The best way to do an A/B test is to start with a 2 way system with a high midbass driver where a notch is easily measured at the listening position somewhere between 100-200Hz, then add a separate woofer below it near floor level that is bandpass filtered to only operate from about 90Hz to 250Hz, and whose level is set about 3dB or so below the main speaker.
That way, when you switch the "flanking sub" in, you know (and also verify with measurements) that it's only targeting a relatively small frequency range where notches are occurring. For example in my test which I posted measurements of a page or so back (post #1624) the change in response at the listening position was very specific to the notch regions - outside of those there was almost no change in measured response.
With a test set-up like this it's possible to evaluate the audibility of the notch vs the absence of the notch without altering the bass response below 90Hz or the midrange response above 250Hz. It also lends itself easily to ABX style testing, if one so desires...
Last edited:
The floor bounce can't be measured because it is swamped by other effects. If it could be measured it would make an audible difference.
Im one of those that has a multiple sealed subs around the room to smooth out the response below 100Hz (LMS5400 sealed designs and AV15X sealed designs). I doubt anyone is completely happy with what happens in the 100hz to 300Hz region at all but what can be done??
Room treatments already exist and if we add more bass bins to handle that mid bass region we will have localization issues.
Now i have started a a couple of new builds recently, My mains are full range now with AV15X sealed subs and i will run a double bass setup with my subwoofers (My DCX allows for proper EQ/gain structure), I have wondered about running multiple bass bins up higher into that 150Hz range. If my mains also handle that frequency range as well then I might not have a localization issue.
Last edited:
Hi,
In simple terms, build a line array ! Then forget about your troubles ! 😀 Dipole better.
- Elias
Much to my surprise, my electronically EQ'd sealed woofer boxes measure flat to 20 HZ at the listener position (room slightly on the smaller side). The EQ is just a 14dB rising amplitude from 100HZ to 20 HZ, and then a drop off below that. I was expecting some kind of mode issue notch, but none below 100HZ. I've got 2 significant notches between 100HZ and 200HZ, from my vertical array of four five inch open baffle drivers (the 4th order active electronic crossover freq. is 100HZ). My experience with floor or ceiling bounce cancellations is that they don't usually occur below about 300HZ, so I'm instead thinking "room modes", Schroeder frequency definitions, room air compliance... and then just adding a little high frequency resolution EQ to minimize mostly the peaks, and yes I dare to try and pull up the nulls a bit since I'm not really convinced that they are a product of a cancellation as much as a reactance effect of the room air compliance as dictated by the room boundaries (I wonder if there's a difference(?)).
What about different locations in the room? That's where things often get worse when you try to pull up a cancellation frequency. I think having multiple woofer boxes helps that the best. I only have room for two, so that's what I have (12 inch Peerless XLS closed cubes). Above 100HZ, my speakers in that frequency range being open-baffle or dipole - it's probably not as bad as a closed box speaker since energies at the sides cancel in the lower midrange and bass freqs, and so don't stimulate room mode issues as much.
Who would have thought that the conversation would get this interesting when someone originally asked if flat is what we want... I especially like what DBMandrake has to say. Many others too.
What about different locations in the room? That's where things often get worse when you try to pull up a cancellation frequency. I think having multiple woofer boxes helps that the best. I only have room for two, so that's what I have (12 inch Peerless XLS closed cubes). Above 100HZ, my speakers in that frequency range being open-baffle or dipole - it's probably not as bad as a closed box speaker since energies at the sides cancel in the lower midrange and bass freqs, and so don't stimulate room mode issues as much.
Who would have thought that the conversation would get this interesting when someone originally asked if flat is what we want... I especially like what DBMandrake has to say. Many others too.
I have found that a couple sound sources placed a couple feet apart, blended in the 100Hz to 150Hz (or sometimes as high as 300Hz, depending on distance), removes the notches in this region. I like having them offset in both the vertical and fore-aft, because it smooths the notch that would have existed from the wall behind the speakers as well as vertical notches and room modes in this region.
Try one of the techniques described in the High-Fidelity Uniform-Directivity Loudspeakers paper, linked above.
I'm sure there are lots of setups that don't work well, like pushing a physically too small midrange down too low, trying to blend all the way to 100Hz. It works great with prosound midwoofers, but little five-inchers probably won't cut it. Likewise, a flanking sub needs to be well behaved at moderately high frequencies (for a sub) and it has to be placed within a couple feet of the mains and symmetrical in order to image right. I won't start the whole symmetrical/asymmetrical debate on more distant subs - I almost don't care about their placement because once you have enough subs, it matters less where you put them (provided they aren't all grouped in the same spot). But as for flanking subs, they should be symmetrical to the mains and close to the nearest boundary (or boundaries).
I agree. I don't discount the importance of smoothing the deeper range, below 100Hz. But I do argue the point that the 100-200Hz transition range is at least as important. More often than not, what's going on above 100Hz (or maybe 90Hz) is as ugly or uglier than what's happening below 100Hz. To me, the 100Hz to 200Hz range needs attention.
Absolutely. I noticed this five or six years ago, and have been suggesting the flanking sub arrangement ever since. It can be used in conjunction with more distant subs used to smooth lower modes, if needed. But I think to use distant subs and omit flanking subs is a mistake.
It always bothered me when I would see discussions of multiple subwoofers that overlooked this midbass/midrange smoothing aspect. Distant subs are fine for smoothing bass below 100Hz, but should not be used to smooth the transition range above that.
If you try to blend a distant sub too high, it begins to sound unnatural. Voices sound like a muffled man is talking in the sub, localization clues clearly start to emerge. But there's no problem blending up through the 100Hz to 200Hz range in closer spaced flanking subs. You can also use a 2.5-way arrangement or a three way with a large midrange, if it's capable of clean output down to 100Hz or so.
Good idea, definitely, I agree. This is, in my opinion, the best way to deal with the 100Hz to 200Hz range. It's notoriously hard to treat this range, as absorbent materials have to be impractically large. Distant multisubs would betray their location, if blended in this range. But flanking subs or blended midwoofers do a splended job of smoothing the upper modal range, just under the Schroeder frequency.
Try one of the techniques described in the High-Fidelity Uniform-Directivity Loudspeakers paper, linked above.
I'm sure there are lots of setups that don't work well, like pushing a physically too small midrange down too low, trying to blend all the way to 100Hz. It works great with prosound midwoofers, but little five-inchers probably won't cut it. Likewise, a flanking sub needs to be well behaved at moderately high frequencies (for a sub) and it has to be placed within a couple feet of the mains and symmetrical in order to image right. I won't start the whole symmetrical/asymmetrical debate on more distant subs - I almost don't care about their placement because once you have enough subs, it matters less where you put them (provided they aren't all grouped in the same spot). But as for flanking subs, they should be symmetrical to the mains and close to the nearest boundary (or boundaries).
I agree. I don't discount the importance of smoothing the deeper range, below 100Hz. But I do argue the point that the 100-200Hz transition range is at least as important. More often than not, what's going on above 100Hz (or maybe 90Hz) is as ugly or uglier than what's happening below 100Hz. To me, the 100Hz to 200Hz range needs attention.
Absolutely. I noticed this five or six years ago, and have been suggesting the flanking sub arrangement ever since. It can be used in conjunction with more distant subs used to smooth lower modes, if needed. But I think to use distant subs and omit flanking subs is a mistake.
It always bothered me when I would see discussions of multiple subwoofers that overlooked this midbass/midrange smoothing aspect. Distant subs are fine for smoothing bass below 100Hz, but should not be used to smooth the transition range above that.
If you try to blend a distant sub too high, it begins to sound unnatural. Voices sound like a muffled man is talking in the sub, localization clues clearly start to emerge. But there's no problem blending up through the 100Hz to 200Hz range in closer spaced flanking subs. You can also use a 2.5-way arrangement or a three way with a large midrange, if it's capable of clean output down to 100Hz or so.
Good idea, definitely, I agree. This is, in my opinion, the best way to deal with the 100Hz to 200Hz range. It's notoriously hard to treat this range, as absorbent materials have to be impractically large. Distant multisubs would betray their location, if blended in this range. But flanking subs or blended midwoofers do a splended job of smoothing the upper modal range, just under the Schroeder frequency.
In one of my previous living rooms a notch at 75Hz was a huge problem at the listening position - there was a very deep notch such that a 75Hz tone was almost inaudible whilst sitting, but if I just lifted my listening position by a few inches it started to return. Forwards or backwards movement made very little difference, (at least less than a couple of feet) neither did left/right. Up and down movement was the sensitive axis - when standing up in the same location there was no notch.
It was a large 4m wide by 8m long, 2.4m ceiling room where the floor was carpet over a poured concrete foundation - so the floor had almost 100% bass reflectivity.
The front wall and right hand side wall whilst normal plaster board on the inside were brick on the outside so were probably very reflective in the bass too, all of which contrived to cause very large standing wave ratios and several modal problems.
The listening position was about 3 metres from the speakers, leaving about 4 metres to the back of the room behind the listener. There was about 0.8m from the front of the speakers to the front wall, and about 1 metre to the side walls. Any further than this and more serious holes started to occur in the bass response.
I recently found an old steady state listening position measurement that I took of it around 2004, which I've attached. The holes at 75Hz, 110Hz, and 200Hz are quite obvious, and that's after the speakers have been positioned for best response.
Interestingly the hole at 110Hz was present despite bass being produced by a 12" woofer whose centre was about 30cm off the floor. On the other hand a higher woofer had a far more severe hole at 110Hz so it was helping a lot, the remaining cancellation was boundary cancellation from the front wall and side walls - moving the speakers further out from the corners moved that notch down in frequency but never eliminated it.
I tried a lot of things to get rid of the hole at 75Hz, including moving the speakers, (obviously) flanking woofers, and while those helped a little bit, especially for the dips around 100-200Hz they never really solved the problem at 75Hz when they were facing forwards beside the main speakers.
Even placing them right in the corners facing forwards didn't really seem to work, it helped a little bit, but there was still a very height sensitive notch at 75Hz at the listening position. I put it down to a destructive reflection from the ceiling - ceiling bounce I guess you could call it.
For no good reason that I can remember other than experimentation I one day tried positioning the flanking woofers as shown in the second image attachment - against the front wall, right up beside the main cabinets (the gap between the main cabinets and front wall was almost exactly the width of the flanking woofer cabinets) but pointing behind the other speakers. The main cabinets were about 50cm wide by 80cm high, with the flanking woofer cabinet being about 50cm high.
Much to my surprise with this configuration the height sensitive notch at 75Hz at the listening position was almost completely gone, and the 110Hz notch was largely gone as well. Improvement in the character of the bass with those two spatial holes filled in was amazing.
Keeping the flanking woofers in the same position but turning them to face forwards or towards each other still improved the notch at 110Hz somewhat but did very little if anything for the notch at 75Hz.
To this day I'm still a little bit puzzled by this, and at the time my working theory was that the large tall cabinets were together with the corner of the room forming a very crude corner horn for the helper woofer at mid/upper bass frequencies. I never found any other configuration of main speakers or helper woofers in that room that so completely solved that hole - sitting or standing the level at 75Hz hardly changed.
Below 300Hz I've found you can get away with the woofer being at a different vertical location to the other drivers without localizing it separately from the midrange/tweeter, but left right displacements can be noticeable and start to "stretch" the image horizontally, unless you keep it below 200Hz.
I would put this down to the fact that vertical image localization is largely provided by the HRTF - which is only effective at high frequencies, so a driver that is producing only low frequencies can't have it's height accurately determined, so you tend to localize the sound of a well crossed over 3 way system to somewhere between the midrange and tweeter.
On the other hand left/right displacements could be detected by left/right ear arrival time differences, even down to low midrange frequencies.
I wasn't necessarily saying that's the best way to handle it if you're starting a speaker design from scratch rather than trying to make do with an existing design, but it's a good way to do A/B testing whilst eliminating most of the other confounding variables to get a better idea of what filling in that hole actually sounds like, and it's something relatively easy for a DIY'er to try with an existing speaker design.
Attachments
Last edited:
Thats another solution, but don't forget about the ceiling - we haven't even begun to deal with that one.
I recently found an old steady state listening position measurement that I took of it around 2004, which I've attached. The holes at 75Hz, 110Hz, and 200Hz are quite obvious, and that's after the speakers have been positioned for best response.
I see that kind of graph a lot, having a notch around 120Hz. It's pretty wide really, definitely audible. Especially when followed by a peak then another notch at around 200Hz. This makes vocals and stringed instruments sound unnatural to me, sort of thin and thrummy.
When kids use those nasty graphic equalizers, the most annoying bands to fiddle with are the 125Hz and 250Hz bands. Room modes and self-interference in this region are like fiddling the 125Hz and 250Hz knobs.
Yes, I've found this general configuration works very well. Of course, my constant directivity cornerhorns have this sort of built-in, and the geometric relationships between sound sources are fixed. It is nestled tightly back into the corner, so there is no boundary interference. Speakers with dual midwoofers in a 2.5-way configuration have pretty close fixed spacing too, so they can be blended fairly high, I imagine, probably 300Hz or so.
But if you use an external flanking sub, you can place it wherever works best and suits your decor. That's kind of nice. I like putting them on the ground and against the wall behind the speakers, just a little bit outside. This makes them offset a couple feet in all three axis, and puts them in-between the mains and the closest boundaries. Seems to work well in most situations. But I don't run them up very high, usually only to 120Hz or 150Hz.
That all may very well be true. I'll have to admit to being more empirical, but I have sort of reached a similar conclusion. The constant directivity cornerhorns have the woofer run higher than I would allow the flanking subs to go, largely because they're more closely spaced. There is only vertical offset in this design, because the woofer is already very close to the walls. No need for offset in any other axes. I'd expect the low-pass of the bottom woofer in a 2.5-way design could be run higher too, since it's just vertical offset. But the flanking sub approach doesn't strike me as being something I'd want to run as high. I suppose you could, in the right conditions (spacing, woofer design) but my approach has always been to put them a couple feet away in each dimension, on the ground (below the mains), near the wall (behind the mains) and just outside. I usually run these up into the low 100's, like 120Hz, and use a fairly gradual low-pass slope like second order.
Last edited:
My listening room ceiling is suspended gyprock (aka drywall, I think), with 8 inches of fibreglass on top, then a 2 foot cavity, then the same again above it. Anything that goes through the gyprock probably won't come back.
It would be nice to have that kind of ceiling height to work with. The biggest problem with my room was the height, it was less than 8 feet and that meant solutions like yours were not viable. I built a ceiling difuser which attempts to spread the reflection more than absorb it because absortion was not in the cards because of the height limitations. Measurements indicate that this was fairly effective.
My ceiling and floor both "float" to a certain extent so at LF they are probably fairly effective dampers. When we talk about how "thick" an absorber has to be to be effective at LFs that assumes that we are atlking about wave proagation through the material. When the whole structure moves due to pressure, then this thickness concept does not apply and one can get substantial LF absorption with minimal thickness. When thinking about LF absorption you have to change your paradigm.
That's what most of us in North America have too. Most homes here have framed drywall construction, with drywall ceilings and insulation above that. There is a dead air space above the insulation, usually several feet, then deck and roof surface (shingles or shake).
I think this kind of construction does damp room modes somewhat. It is definitely not as rigid as concrete, brick or stucco.
In fact, I've been in some rooms that seemed to have pretty well damped room modes down low, even without any special treatment. The drywall must have enough give to do a decent job of acting somewhat as a panel damper.
The key here is how is the drywall attached to the structure (you said "suspended", but now I am not sure what you meant by that). If it is a rigid attachement (nailed or screwed) then the surface will be mostly reflective at all frequencies, but if it is atached flexibly, such as with a flexible suspension system with a well damped mounting (mine are 1/4" nylon cables about 6" long), then it can move and will be mostly absorptive at LF, but reflective at HFs. Standard 1/2 drywall is almost perfectly reflective at HF, but it depends on the paint used.
Ah, it's a false ceiling (for thermal reasons) on a standard pine frame. Ceiling sheets are 3/8" IIRC.
Drywall constructin definitely helps a lot. While living in the States I've never seen those high Q modal effects like in conventional European constructions.
The weaker construction of a typical US home certainly does create some structural loss for the room at LF, but this can be vastly increased with "floating" wall construction such as I use. In my room, I only see a single mode at about 35 Hz and then it appears to become non-modal and no other discrete modes are evident. I certainly believe that this is a key factor in the high quality bass that I get in my system.
Thats standard construction for the US as well, but some local codes require 1/2". It varies.
Gedlee do you know what your room gain is? Is it just a function of the room dimensions?
- Status
- Not open for further replies.