It has been reasonably well established for half a century if not longer. For example Figure 12 here shows an estimate of the audibility threshold. There are few details given of the procedure suggesting they may been in a separate publication. Anyone?
I seem to recall seeing a similar plot from a different source about 10-20 years with the threshold in a slightly different place and without such a strong decline at low frequencies. Is this ringing any bells with anyone? Would be grateful for a link since I am doing a bit of work in this area at the moment.
But the answer is pretty much what one might expect. The threshold for hearing resonances is in the same ballpark as other forms of distortion. We are most sensitive to it in frequency ranges where it occurs naturally and where the ear/brain has developed to listen for it in order to distinguish sounds. Design your speaker cabinets accordingly.
Annoying though it was at the time I think it has turned out to be a good thing by prodding me into assessing the value of such a post both to myself and the forum. For example, I have made many posts like the first part of this one on aspects of speaker design but they have had close to zero impact despite them correcting matters of fact rather than opinion. Does my background and outlook mean the forum is out of step with me or that I am out of step with the forum?
The lost post was mainly about the practicalities of performing FEM/BEM simulations of speakers, reviewing the B&W paper and to some extent the design of the B&W speaker. The comparison between simulations and measurements clearly shows a significant systematic error almost certainly in the simulations. The results also show one two rather doubtful design choices. This is all of substantial interest to me covering both the day job and a hobby interest in speaker design. But would it have been of any interest to others on the forum? The answer I am pretty sure is no for the vast majority and perhaps a little bit for a few. My effort would very likely have been better direct elsewhere and in a more appropriate format.
I have looked at the article, blog and wiki sections here but the level of activity looks to be well below critical mass. Any suggestions for more active locations for reasonably lengthy and reasonably technical articles and/or interaction on the engineering of speakers? I do not intend to leave completely or flounce off in huff just looking for a more appropriate place to interact and blog about the technical side.
I don't recall those materials specifically, but there's the idea of using sand. There might have been an article in Audioxpress or Speaker Builder.
Google finds many diyaudio threads on sand-filled speaker walls:
sand speaker site:diyaudio.com - Google Search
To use the threshold of audibility data we need know how loud the cab walls are at the listener's chair for a specified speaker loudness (and measured with no leakage from the driver) or conversely how loud the speaker needs to be to cause cab wall sound that crosses the threshold. That gives you a near-useless figure for detection of wall vibration in the absence of a speaker playing (which of course isn't what we are talking about at all).
Curiously, we need to know how the sound of the cab sounds to listeners because it is singing in harmony (mostly) with the speaker. That really is a masking paradigm, not a threshold test. And even then, "detection" isn't quite the same thing as "annoyance", but good enough for most of us.
I suppose you could take the cab wall data and create a filter. That can be used to power a second speaker, Speaker 2, during the following test. Then play something on a speaker while testing listeners for the point at which they can detect the wall sound from Speaker 2. That test would be analogous to a far less interpretable test in which you build a concrete cab and a whole series of wood cabs and could switch listeners between them blindly.
B.
Last edited:
In industry this is easy enough to simulate with FEM/BEM or to measure with a laser vibrometer, a Kirchoff integral and some perseverance. DIYers could perform the simulations after some study but I am not aware of anyone apart from myself with a current interest. Laser vibrometers aren't expensive and could be hired if a DIYer was keen but again I am not aware of anyone taking an interest.
Not sure this makes sense. If you increase the sound from the speaker drivers the sound from the cabinet will increase similarly.
Don't understand this.
If the sound from the cabinet is below audibility why should what it sounds like be of any relevance?
Don't understand what is being suggested or why.
I wondered whether the lower frequency rise was representative of the nature of the panels, or of our ears. Then I realised that the lower frequencies could perhaps respond to EQ in either case. With basic bracing and damping in addition, now I'm disappointed that the audibility chart doesn't seem helpful anymore. I'd also go for a link that anyone knows of..
Do you mean the LF rise of the Fletcher-Munson-like curves? Remember, those are perceived equal loudness curves except for the bottom one which is threshold (minimum you can detect). In addition, you have to be careful about conclusions looking at one curve (as everybody does); you really need to consider how the curves change next to one another. That's a failure of the chart makers to present this information properly (now doing on 90 yrs).
The issue with cab audibility is masking in the presence of the very same (or usually the very same) notes coming out far louder from the speaker. Now, as it turns out, there is a voluminous literature by psychologists on masking. But the topic is so very very complicated, I can't hardly understand it.
Earl has some cred in this topic; but unless you've read his work on annoyance from harmonic distortion, you'll be surprised.
B.
The issue with cab audibility is masking in the presence of the very same (or usually the very same) notes coming out far louder from the speaker. Now, as it turns out, there is a voluminous literature by psychologists on masking. But the topic is so very very complicated, I can't hardly understand it.
Earl has some cred in this topic; but unless you've read his work on annoyance from harmonic distortion, you'll be surprised.
B.
Last edited:
It's not masking. The loudspeaker is a system and the way it vibrates creates a sound field - the cabinet included. Masking involves a second source, unless you are talking about masking in our ears, which is another thing, which I am not sure comes into play here. If the cabinet has substantial radiation from vibrations then this has to affect the far field radiation pattern, or it is insignificant. Along those lines I have never been able to detect anything. A sharp resonance might ring - making it audible - but then that is simply poor construction.
But the simple fact is that this discussion is like arguing about how long the soul of a fly lasts after it's dead. I don't really care, it's not important.
Gosh, I never realized your talent as a lawyer. Seems to me that is rather nit-picky to say masking only happens when the sound comes out of two different boxes not one. Anyway, the test paradigm I outlined involves (1) identifying the sound coming from the walls and in relation to speaker loudness levels and (2) using TWO speakers, crank up a simulation of the wall sound until it is no longer masked by the primary speaker (AKA masking).
I'm sure Earl is too modest to go into it, but he wrote a very good study showing that HD is best understood if you took into account the human audibility of the distortion components not just their acoustic loudness. See how that relates directly to this topic?
I think this is an important topic because I see great consternation on this website about bracing, finding Baltic Birch (esp if you have a Russian cousin to ship it to you), and esoteric physics of panel vibrations.
But if it ain't audible, waste of time to fuss over extreme bracing.
B.
I'm sure Earl is too modest to go into it, but he wrote a very good study showing that HD is best understood if you took into account the human audibility of the distortion components not just their acoustic loudness. See how that relates directly to this topic?
I think this is an important topic because I see great consternation on this website about bracing, finding Baltic Birch (esp if you have a Russian cousin to ship it to you), and esoteric physics of panel vibrations.
But if it ain't audible, waste of time to fuss over extreme bracing.
B.
Last edited:
Andy's link, page 19. A reduction in audibility toward lower frequencies.
Dynamic variable loudness processor?
Not sure what you mean by nature of the panels but as the frequency falls a number of factors kick in. Firstly the ear/brain becomes less sensitive to distortion (e.g. subs can have relatively high levels of distortion which don't hear but would if they were at midrange frequencies). Secondly the motion of the walls and the driver become increasingly aligned as the wavelength gets longer with the result the addition of the cabinet radiation tends towards making the signal a touch quieter/louder rather than being an interfering/different sound. The air pressure inside the cabinet becomes a significant factor (or dominant for small cabinet volumes) in driving the walls at the lowest frequencies and this will track the motion of the cone closely. Thirdly, most low frequency musical sounds are not click-like and tend to have a tail.
Perhaps to some extent but if you consider the extreme case of the cabinet motion cancelling the driver output then equalisation obviously cannot fix this. Not sure it would be an effective tool in practise.
Don't understand. It seems to me to be exactly what is needed to guide the design in terms of bracing and damping the cabinet in order to the push resonances low enough in magnitude to be inaudible. For example, given the audibility levels, which of these two types of cabinet construction is likely to be superior? (Obviously the other 5 panels need measuring and a surface integral evaluated to allow a direct evaluation but there is enough in the plot to get a feel for the reasoning that would be used if a speaker cabinet was to be designed following standard engineering principles conventionally used in higher tech industries.)
I guess I didn't intend to sound so dismissive.
For example the conjecture that the sloped region might benefit most from the bracing and a little EQ, and the flat region of the audibility result might benefit most from panel damping and it's especially important that it works sufficiently on its own as EQ will not.
Mentioning EQability was meant as a commentary on the importance of that part of the problem.andy19191 said:
For example the conjecture that the sloped region might benefit most from the bracing and a little EQ, and the flat region of the audibility result might benefit most from panel damping and it's especially important that it works sufficiently on its own as EQ will not.
In the bass region I find it harder to discern less than a dB of variation for a typical adjustment. The -10dB at the left of the chart, broadly speaking has the potential to change the 0dB by a couple of dB or so. The curve is in the same direction as, but appears perhaps way off my personal idea of sensitivity threshholdsandy19191 said:
No problem but I think I may be seeing less of an issue than you or Ben. After playing with a few structural designs for a cabinet I would look at the FRF at the listening position in the light of an estimated audibility threshold like the BBC one in order to get a feel for the level of damping that needs to be added if any. Subwoofer cabinets and some woofer cabinets will generally require no additional damping. Midrange and midwoofer cabinets generally will. Once the simulated peaks are 30 dB and preferably 40 dB below the signal they are of little interest. Don't care what they sound like or about equalisation because they are going to be effectively inaudible in use (although possibly not with every test tone). Job done.
On the other hand an interest in quality is the motivation for performing the simulations in the first place (or measurements if you are that way inclined and have the time and resources). Proper quantitative evidence that cabinet sound radiation isn't an issue is a part of a high quality speaker design. Relying on faith does not lead to high quality designs particularly if that faith is misplaced by, for example, relying on high stiffness when there are cabinet resonances within the passband of the drivers.
But as a number of people have pointed out earlier, audible cabinet sound radiation is likely to be a minor issue compared with the main factors governing the quality of sound unless there is something significantly abnormal about the design.
I have worked on numerical simulation R&D with aerospace, power generation, automotive and one or two others over the decades. Never worked with any speaker companies although I did once apply to work for KEF in the 70s (position already taken) and instead went to work for an aerospace company. Career could have been rather different had I not left applying to the last minute.
No, I don't see how it relates because box vibrations are linear and my above work was on perceptions of nonlinear effects. Nothing in common as I see it.
My point is simply that I have stated my position, which isn't that far from yours and beyond that I just don't have any interest in the detailed arguments of what is audible and what is not in this context. There simply are no good data on the topic, so it is all just conjecture, but again, my investigations indicate that it is not a major effect. So I don't worry about it.
As you identify and deal with acoustic issues, new ones come to light. Ones that you may not have noticed before. Cabinet issues too. At this time I'm not going to be thinking it's minor. Cabinets are capable of contributing to bad sound.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.