Enclosure resonances, not a big deal?

[Pano]

I don't buy boxes at all.

 

[gedlee]

It's beyond the Haas effect range in time for most of it's output for a speaker out in the room.

I don't buy most of what I read about boxes.

Dave

Dave

Mostly agreed, but the time lag from box resonances would be well be within the summing locaization time region and if the level were high enough then they would be an audible effet on imaging and coloration. But I seriuosly doubt that the level would be high enough in any well designed box.

The way I tested this was to start with an overkil box and measure its polar response. Then I began to remove things like CLD panels, internal damping, etc. trying to note any measureable changes in the polar response. I was not able to measure any significant changes nor hear any audible differences in any of these changes.

 

[xpert]

xpert,
... 20 dB+ is very modest - ...Resonances? I see the major deviations from a smooth curve at just above 100 Hz, above 200 Hz, 400 Hz and 800 Hz. Makes me thinking ...

The peaks and dips may be due to frequency dependent leaking of front baffle output into backside measurement. Once I posted it I thought it would improve clearity regarding dips'n peaks to measure the thing without the only partly sealing blanket.

All those "resonances" are 20dB below the direct radiation. It is common wisdom that those won't be audible even if they were resonances. It has to be understood that those resonances are not bad per se. They are NOT distortion. They have to be evaluated by their LINEAR impact on the amplitude/phase versus frequency graph. And that ain't much with more than 20dB below level.

And realize it was a box that neither I would build that way as final. In this case it was for ckecking a pro midwoofer for harmonic distortion only.
I hate woodwork. It's loud, dusty, potentially harmfull ... no, not mine. But plywood 1/2" plus some crossbracing plus that mentioned felt is my standard. Maybe I measure such a beast sometime again.

If you have a panel that has a total panel sectional output that is 40db down from the first arrival, what is the magnitude at the listener after it's reflected off of walls and anything else that disperses and/or dampens? It's in the room response region and seems mighty unlikely to me that any of that is provably audible, it if is at all.

Then of course, the two lateral side panels are vibrating roughly equivalently and in-phase, but are facing the opposite direction. How does that affect the response at the listener's spot? And another thought, we're talking a flat baffle. What do you suppose the directionality is for such a large area? The listener is pretty close to 90 degrees in the off-axis of side and top baffles. Another confounding factor having an impact on the direct response.

I don't buy most of what I read about boxes.

And woodwork makes deaf if earprotection is not applied. The more golden the ears the more woodwork has to be done to not hear all that resonances ...

 

[djarchow]

The other point that seems to get glossed over in these discussions is how optimzed is the rest of the design? Doing things like excessive bracing, constrained layer damping etc. don't make much sense if the rest of the design isn't optimal as well. Much like the discussions where people are debating which brand of $75 boutique caps to use in a XO designed with textbook formulas.

I tend to use the cabinet construction methods I learned from the old North Creek Music cabinet handbook which includes thick baffles (front and rear) along with staggered plywood bracing and have never heard (nor measured) anything that would make me think I am not building my cabinets well enough.


Regards,

Dennis

 

[dlr]

Dave

Mostly agreed, but the time lag from box resonances would be well be within the summing locaization time region and if the level were high enough then they would be an audible effet on imaging and coloration. But I seriuosly doubt that the level would be high enough in any well designed box.

My thought was on the overall room reverberant environment, but some direct reflections would be short enough in time as you point out. The first reflections from the driver off-axis would, I think, swamp the output from the box panels for any reasonable box where reflections may be an issue.

Dave

 

[gedlee]

My thought was on the overall room reverberant environment, but some direct reflections would be short enough in time as you point out. The first reflections from the driver off-axis would, I think, swamp the output from the box panels for any reasonable box where reflections may be an issue.

Dave

Dave

To me I think more in terms of a box, but it easier for me to think of a sphere, that has a major portion in motion and the rest of the sphere has small motions that are at the same frequency. IF this extraneous vibration is significant then it would have to show up in the polar response even though it may be minimal on axis. On-axis has the unique position of being immune to any vibration that "averages out" and that would be true of most box vibrations. But the off axis locations are hyper sensitive to these "average" motions. If you don't see something in the axial response OR the off-axis response then, quite honestly, there IS nothing. But there can be box motions that do not average out and these would appear on axis (and off-axis) and would have a large group delay associated with them. These, IMO, would be the most audible effects and they are not reflections but direct radiation from the enclosure.

 

[thoriated]

With my most recent speaker design, I felt that I had to replace the rear panels due to their sonic contributions. I can guarantee that I didn't go to all the extra effort of tearing down the cabinets and rebuilding them to this extent because of a hypothetical whim. It was because once I identified this problem, it proved quite annoying during listening sessions and wasn't worthy of the general quality of the design. I found that the rear panels were calling attention to themselves through empirical listening tests with easily identifiable and characteristic changes (of the presence or absence of relatively high Q lower midband resonances) verifiable when the rear panel contribution was filtered out acoustically. All but the rear panels were 13 layer Baltic birch - but the rear panel was originally a cheaper grade plywood which significantly deteriorated the overall sonic presentation. Given that room response aberrations can often approach 20db, a 20 db margin between panel and driver outputs may actually decrease to almost 0 db at certain frequencies in an actual listening environment.

 

[xpert]

But there can be box motions that do not average out and these would appear on axis (and off-axis) and would have a large group delay associated with them.

I agree in the result, I disagree with Your - ever more well known - argumentation. Speaking of frequency response You have to take the whole radiation into account. To pick one specific part is invalid mathematically. That for the first. The second is Your reference to psychoacoustic data. This data has never been taken with loudspeakers in mind. That for they don't apply exactly.

Example given: temporal masking. Got some signal first. Then add this signal to itself but with a temporal delay and with lower level. With what combinations of delay and levels - one and two - the adding becomes audible? Reverted, how far does the masking from the first signal hold?

For that situation some investigations have been performed. But obviously they do not apply to any real world situation! What is the structure of the first signal? Is it a burst, a steady sine, discrete spectrum, what?! What is that for the second, as You mentioned some "group delay" with it - from what?! All that fancy numbers don't apply because the real situation ain't defined!

If we know something from scientific work on the human hearing: it's complex as mad. You can't extrapolate some specific results to any different. I would wish You could for the sake of scientific honorability be more sensible with Your references to psychoacoustics.

Regarding that box zound issues. Define the worst case and analyse it thoroughly. My bet was to get the group delay / amplitude response of the whole radiation and relate that to well known thresholds. Since the upcomming of digital signal processing that should be very simple.

It is pure mathematics that if amplitude response and group delay of a given linear transmission line are the same, these transmissionlines are equivalent. The experiment mentioned above - the delay/amplitude adding - can be described in a amplitude/group delay response of ONE simple signal. There is no need to come up with some "temporal masking" wording. That might be useful with contruction of concert halls, but it is not related to loudspeakers.

Given that room response aberrations can often approach 20db, a 20 db margin between panel and driver outputs may actually decrease to almost 0 db at certain frequencies in an actual listening environment. [/B]

You WANTED to have the backsides changed because they were cheep. But Your one dimensional calculation ain't right, sorry.


cheers

 

[thoriated]

You WANTED to have the backsides changed because they were cheep. But Your one dimensional calculation ain't right, sorry.

Since I will not accept a conflation of optimized measurement conditions with real world listening environments I have to disagree with your effective contention that the listening environment cannot make up to 20db difference to the speaker response at the listening position at any frequency. For example, diffraction/reflection modeling shows a lower midrange null of several db resulting from front panel cancellation against the back wall as compared to energy from the back panel at the same frequency for the speaker in question when the speaker is placed fairly close to the back wall, as is the case here. This is not taking into account any of the several significant room reflections that also have significant but different effects on both the front and back panel contributions to the overall sound field.

It's silly to assert that I 'WANTED' to go to a great deal of extra effort to tear the two speakers down and reassemble them. That begs the question of why I used the inferior material for the rear panel in the first place, for one thing. I used the cheap stuff first, found it inadequate strictly based on listening quality considerations, and learned from my mistake. That's the only reasonable interpretation of the events I described. The speaker in question is not intended to represent an 'ultimate' construction approach - but the cabinet had to avoid incompetent choices in material and construction not to compromise or obscure the effects of certain design principles I incorporated into the speaker crossover.

 

[xpert]

Since I will not accept a conflation of optimized measurement conditions with real world listening environments I have to disagree with your effective contention that the listening environment effectively cannot make up to 20db difference to the speaker response at the listening position at any frequency.

The leakage of sound through the cabinet walls is a linear effect. It sums up to the soundfield as any other contribution example given (a) the speaker chassis or (b) reflections of (a)s sound by walls. The key here is that leakage due to the walls acoustical transparency is linear. If that leakage can be emphasized by room acoustics +20dB any other contribution could too. You won't seperate them, neither by measurement nor by listening.

cheers

 

[gedlee]

I agree in the result, I disagree with Your - ever more well known - argumentation. This data has never been taken with loudspeakers in mind. That for they don't apply exactly.

Well as usual you are ill-informed. My wife and I did and published such a psychoacoustic study - with loudspeakers in mind. You should read more, and criticize less.

 

[thoriated]

<i>

If that leakage can be emphasized by room acoustics +20dB any other contribution could too. You won't seperate them, neither by measurement nor by listening.

Sure I can. A high Q resonance can often be detected and isolated. It manifests itself as 'coloration' or a smearing of detail in its milder forms, far beyond what its steady state amplitude contribution might indicate, because of its delayed and much longer lasting effect. And guess what, these are all nominally 'linear' effects. This kind of thing will not be an issue only if you listen to continuous waveforms instead of music.

I've distinctly heard sheetrock, ductwork, port resonances and panel resonances that are all fairly easy to separate from the original incident waveform from the driver. Not to mention reflections through the cone itself from within the cabinet which can also be detected and suppressed. Suppression of secondary resonant contributions is, for instance, much of the rationale for feedback control at live concerts.

 

[badman]

<i>

Sure I can. A high Q resonance can often be detected and isolated. It manifests itself as 'coloration' or a smearing of detail in its milder forms, far beyond what its steady state amplitude contribution might indicate, because of its delayed and much longer lasting effect. And guess what, these are all nominally 'linear' effects. This kind of thing will not be an issue only if you listen to continuous waveforms instead of music.

I've distinctly heard sheetrock, ductwork, port resonances and panel resonances that are all fairly easy to separate from the original incident waveform from the driver. Not to mention reflections through the cone itself from within the cabinet which can also be detected and suppressed. Suppression of secondary resonant contributions is, for instance, much of the rationale for feedback control at live concerts.

Well put.

This argument is like tasting a bell pepper and a habanero, and then someone telling you that there's no reason to believe the habanero is hotter without measuring the capsacin (sp?) levels. These colorations, once identified, are easy for an experienced listener to hear. There are bigger problems around, but an undersized engine in a car doesn't invalidate the usefulness of an improved suspension.

 

[xpert]

Well as usual you are ill-informed. My wife and I did and published such a psychoacoustic study - with loudspeakers in mind. You should read more, and criticize less.

'Must have been very private studies indeed. Once again You're bold in repeating Your very claims and nothing.

 

[xpert]

<i>

Sure I can. A high Q resonance can often be detected and isolated. It manifests itself as 'coloration' or a smearing of detail in its milder forms, far beyond what its steady state amplitude contribution might indicate, because of its delayed and much longer lasting effect. And guess what, these are all nominally 'linear' effects. This kind of thing will not be an issue only if you listen to continuous waveforms instead of music.

I've distinctly heard sheetrock, ductwork, port resonances and panel resonances that are all fairly easy to separate from the original incident waveform from the driver. Not to mention reflections through the cone itself from within the cabinet which can also be detected and suppressed. Suppression of secondary resonant contributions is, for instance, much of the rationale for feedback control at live concerts.

My point is to measure it. What I am after is to describe what adverse the speaker does to the transmission of a signal. That is the very basic function of the device in consideration; signal, transmission. If there is any effect of something it can be measured today down 24bit below full level. By first principles a linear deviation of ideal/identical transmission would ever show as a deviation in group delay or amplitude response. There is no exception - period.

To investigate the sonic impact of a possible linear disturbance/distortion the one and only thing to do is to look up its trace in the amplitude/group delay vs. frequency plot. With cabinet "resonances" that are down by 20dB or more and that are LINEAR in their relation to the original signal no degradation is expected. The human hearing is by for more forgiving than todays measurement tools.

 

[planet10]

.... look up its trace in the amplitude/group delay vs. frequency plot. With cabinet "resonances" that are down by 20dB or more and that are LINEAR in their relation to the original signal no degradation is expected

The differences between merely good hifi & really great hifi are happening 30-40-50 dB down... A gross frequency response is not likely to show the absence or presence of something 40 dB down let alone give you enuff information to know what it is.

20 dB is not good enuff for me...

dave

 

[thoriated]

With cabinet "resonances" that are down by 20dB or more and that are LINEAR in their relation to the original signal no degradation is expected.

I doubt that any serious cites can be found that will consider acceptable as low a standard of quality as described in the above quote except possibly for incidental exposure to music where low cost is a prime consideration such as background music in commercial venues.

 

[xpert]

I doubt that any serious cites can be found that will consider acceptable as low a standard of quality as described in the above quote except possibly for incidental exposure to music where low cost is a prime consideration such as background music in commercial venues.

U may simply calculate it: a contrubution that is 20dB down and adds linear - as cabinet walls do - will show up as:


10*log10 ( 1 + (10^20/10) ) = 10*log10 (1.01) = fetch Your calculator!

in the amplitude response. The impact on group delay will show the same orders of magnitude. If You can hear the difference, You may call for the scientists to investigate Your extraterrestial skills further.

no offnece

 

[planet10]

10*log10 ( 1 + (10^20/10) ) = 10*log10 (1.01) = fetch Your calculator!

in the amplitude response. The impact on group delay will show the same orders of magnitude. If You can hear the difference, You may call for the scientists to investigate Your extraterrestial skills further.

That was implied in my post... the low level information has negligable impact on the gross frequency response, but if you can't hear them, then woe to you if you get tossed out in the jungle and need to avoid the lions & dingos.

The ability to hear that fine detail has been evolved into us for millions of years, the ones who couldn't hear that stuff never lived to breed,

dave

 

[xpert]

That was implied in my post... the low level information has negligable impact on the gross frequency response, but if you can't hear them, then woe to you if you get tossed out in the jungle and need to avoid the lions & dingos.

The ability to hear that fine detail has been evolved into us for millions of years, the ones who couldn't hear that stuff never lived to breed,

dave

I'm really impressed how far we can discuss without understanding each other. I'm talking not of a soley signal "in the jungle" but about a deviation of amplitude response. That is that the amplitude vs frequency plot shows some +/- 0.3dB ripples - one or two of them.

U should think it over