Speaker enclosure vibrations - a few measurements with accelerometer

[Tangband]

I have subs that are 27x27x60 cm, 19 mm MDF, no bracing. What do you think the main modes will be?

The lowest about 283 Hz , so no trouble if you are crossing at 80 Hz , 24 dB/oct. It would be a problem with 12 db/oct filtering.
The subs I used with HYBRID had about the same dimensions , by the way. Crossing at 80 Hz in stereo was fine, and 75 Hz with 30/18 dsp crossover had the best sound.

 

[Vintagefan]

Rowan Atkinson (nephew?), driving a brand new mclaren, having misinterpreted the resonanzes of his ‘cage’ he was driving ( too fast ) took advantage of the low resonance when he hit an accelerometer! By fft transform it showed a perfect peak with high Q and low bandwidth.

 

[Thomas_A]

A little off topic, but …maybe all those resonances are small, compared to the out of phase sound coming from the inside of the box, and thru the drivers membrane.

How much delayed sound comes through the cone from the inside of the loudspeaker in a closed box ?

One of the reasons Linkwitz didnt like the sound of boxed full range speakers was because of the delayed sound coming through the cone and giving coloration to the sound . In the lxmini he used a 80 cm long pipe heavily stuffed with damping material at the bottom of the pipe , and with damping...

www.audiosciencereview.com

From what I remember the distortion products were well above audible level, mostly because of presence of higher order distortion products. Perhaps I could have some time time during the week to measure it.

 

[witwald]

About this good braced:

View attachment 237290

Does anyone have a 3D CAD model of a 2-way or 3-way loudspeaker enclosure, or one such as that above? The CAD model could be imported into a finite element analysis program and meshed, after which the vibration resonance frequencies and mode shapes could be calculated. Any pointers to such CAD models would be helpful. Thanks.

 

[DanielT]

Does anyone have a 3D CAD model of a 2-way or 3-way loudspeaker enclosure, or one such as that above? The CAD model could be imported into a finite element analysis program and meshed, after which the vibration resonance frequencies and mode shapes could be calculated. Any pointers to such CAD models would be helpful. Thanks.

There was only one picture I found. Unfortunately, I can't find now which page I found it from. In any case, here is something to read, if you or anyone else is interested:

Loudspeaker Cabinet Bracing: A Detailed Look on Do's and Dont's

This article does a detailed FEA analysis on loudspeaker cabinet bracing and its affect on reducing resonance modes and coloration of sound. The stiffer cabinet always wins as the analysis shows.

www.audioholics.com

 

[fineMen]

There was only one picture I found. Unfortunately, I can't find now which page I found it from. In any case, here is something to read, if you or anyone else

Loudspeaker Cabinet Bracing: A Detailed Look on Do's and Dont's

This article does a detailed FEA analysis on loudspeaker cabinet bracing and its affect on reducing resonance modes and coloration of sound. The stiffer cabinet always wins as the analysis shows.

www.audioholics.com

So, how come that the panel gets into a secondary mode, while the force is uniform?!

(citation from above link, only for scientific discussion ;-)

The force is mathematically uniform, point symmetrical, right?! No, that force wasn't considered atall when generating the image--it shows a possible motion of the panel, but not the concrete reaction to any force that may be applied by the internal air. Period. The picture presents an Eigen-mode, a singular building block of more complicated patterns that may (or not) comprise the 1st, 3rd etc mode also.

Such kind of a 'fundamental' analysis is always seen, but I never saw a really fundamental analysis that necessarily has to include the driving force for the (desired?!) panel movement.

Please think Yourselves. Some posts above I gave You otherwise quite costly data that I generated in private, while being, by the papers, a full fleshed scientist. My observations are true. What I didn't consider to be necessary to mention is that model I once set up using the old Akabak. Same result: forget about the eiigenmodes of the panels, but concentrate on the internal air column resonances.

You are willing to think yourselves. Just do it! How come?

 

[Tangband]

From what I remember the distortion products were well above audible level, mostly because of presence of higher order distortion products. Perhaps I could have some time time during the week to measure it.

This would be very interesting

 

[Thomas_A]

This would be very interesting

No time yet, giving my hopes to a few hours ”hobby” work in the weekend.

 

[fineMen]

This would be very interesting

I do not feel it that way. You cannot derive general conclusions from, if at all, only loosely connected observations from an unknown set-up.

Actually, the graphs on distortion given above don't make sense to me. They are unexpected in terms of amplitude and spectrum, no correlation to whatsoever, except to possible HD of the driver which to my knowledge isn't documented. And may it be rattling originating in a flimsy stand.

I really wonder again, how strictly the discussion of (presumably) science driven results remains in the riverbed of hearsay and magazine-fed rumours ...

 

[Thomas_A]

I do not feel it that way. You cannot derive general conclusions from, if at all, only loosely connected observations from an unknown set-up.

Actually, the graphs on distortion given above don't make sense to me. They are unexpected in terms of amplitude and spectrum, no correlation to whatsoever, except to possible HD of the driver which to my knowledge isn't documented. And may it be rattling originating in a flimsy stand.

I really wonder again, how strictly the discussion of (presumably) science driven results remains in the riverbed of hearsay and magazine-fed rumours ...

I am not up for writing peer-review articles here, but as noted in other threads linked to Purify drivers and cabinets, you may or may not end up with non-linear effects.

 

[fineMen]

I am not up for writing peer-review articles here, but as noted in other threads linked to Purify drivers and cabinets, you may or may not end up with non-linear effects.

Frankly spoken, I'm not quite convinced that kind of a model-free measurement attempt would yield valuable results. Of course, science begins with just fumbling around with some new material--except it is known dangerous.

You've got some loudspeaker box, driver included, an accelerometer and a spectrum analyser.

Now, after Your first trials fumbling around with the new toys, how to organize some real data, fit it to a model, which model? One needs a model to just talk about things. Re-build, in the head, the chunky bulky real thing from imagninative elementary parts.

I strongly advertize to redo the measurements but with a more analytical understanding of the questions thrown at the device under test. As a teaser I give You a riddle, reiterated:

If the panel resonates from whatever driving force, what about the phase of the acceleration measured at different points distributed over the panel, and what about the relative amplitude of the respective accelerations? If the plane would go into a higher order mode that should be clearly visible. If no higher modes are visible, the model shall be adjusted to that finding!

Another:

What about the spectrum of accelerations when knocking at the wall versus driving it (still: somehow) by the speaker, same or different? If different, the model shall be adjusted to that finding.

In case You are able and willing to really elaborate on answers to those two, quite basic by the way, questions the results could encourage a productive dispute. Otherwise, what do You think?

People might help You to go through the problem once the language is clear in terms of--terms. I'm afraid that the current terminology in this thread once again is still too much tied to questionable blurp from magazines. Of course not all contributors here have a real scientific background, I have by paper at least. Why don't You trust me? I wonder. Too hard to follow that simplistic stuff given above? Sorry for being less diplomatic.

More frankly, some guys hang around waiting, as freeloaders, for a free lunch. Don't let them push You. I've got all the worst experience with such lazy, headless folks especially with regards to stereo. Sorry.

 

[Thomas_A]

Frankly spoken, I'm not quite convinced that kind of a model-free measurement attempt would yield valuable results. Of course, science begins with just fumbling around with some new material--except it is known dangerous.

You've got some loudspeaker box, driver included, an accelerometer and a spectrum analyser.

Now, after Your first trials fumbling around with the new toys, how to organize some real data, fit it to a model, which model? One needs a model to just talk about things. Re-build, in the head, the chunky bulky real thing from imagninative elementary parts.

I strongly advertize to redo the measurements but with a more analytical understanding of the questions thrown at the device under test. As a teaser I give You a riddle, reiterated:

If the panel resonates from whatever driving force, what about the phase of the acceleration measured at different points distributed over the panel, and what about the relative amplitude of the respective accelerations? If the plane would go into a higher order mode that should be clearly visible. If no higher modes are visible, the model shall be adjusted to that finding!

Another:

What about the spectrum of accelerations when knocking at the wall versus driving it (still: somehow) by the speaker, same or different? If different, the model shall be adjusted to that finding.

In case You are able and willing to really elaborate on answers to those two, quite basic by the way, questions the results could encourage a productive dispute. Otherwise, what do You think?

People might help You to go through the problem once the language is clear in terms of--terms. I'm afraid that the current terminology in this thread once again is still too much tied to questionable blurp from magazines. Of course not all contributors here have a real scientific background, I have by paper at least. Why don't You trust me? I wonder. Too hard to follow that simplistic stuff given above? Sorry for being less diplomatic.

More frankly, some guys hang around waiting, as freeloaders, for a free lunch. Don't let them push You. I've got all the worst experience with such lazy, headless folks especially with regards to stereo. Sorry.

Not sure how far I would take this further other than acoustic measurements. This is because the end result is satisfactory to me. Building cabinets and testing various damping materials just for the sake of modelling and measuring resonances is out of scope for me. Others may chime in with their own accelerometer measurements and modal analyses with respect to driver and cabinet resonances.

 

[DanielT]

So, how come that the panel gets into a secondary mode, while the force is uniform?!

(citation from above link, only for scientific discussion ;-)

The force is mathematically uniform, point symmetrical, right?! No, that force wasn't considered atall when generating the image--it shows a possible motion of the panel, but not the concrete reaction to any force that may be applied by the internal air. Period. The picture presents an Eigen-mode, a singular building block of more complicated patterns that may (or not) comprise the 1st, 3rd etc mode also.

Such kind of a 'fundamental' analysis is always seen, but I never saw a really fundamental analysis that necessarily has to include the driving force for the (desired?!) panel movement.

Please think Yourselves. Some posts above I gave You otherwise quite costly data that I generated in private, while being, by the papers, a full fleshed scientist. My observations are true. What I didn't consider to be necessary to mention is that model I once set up using the old Akabak. Same result: forget about the eiigenmodes of the panels, but concentrate on the internal air column resonances.

You are willing to think yourselves. Just do it! How come?

They are quite vague in their conclusions. They measure, but the practical significance of those measurements is another matter.

I agree with you that studies conducted on a strictly scientific basis should be done.

 

[Thomas_A]

I strongly advertize to redo the measurements but with a more analytical understanding of the questions thrown at the device under test. As a teaser I give You a riddle, reiterated:

If the panel resonates from whatever driving force, what about the phase of the acceleration measured at different points distributed over the panel, and what about the relative amplitude of the respective accelerations? If the plane would go into a higher order mode that should be clearly visible. If no higher modes are visible, the model shall be adjusted to that finding!

Just a comment. The fundamental modes and the higher order modes are there in the graph. A glued MDF cabinet should not have those non-linearities (while using screws and bolts may introduce those things) - putative explanations are driver, cage and and transition from frame to cabinet resonances causing non-linearities, as well as standing waves (cf. #5 ). Key findings were that these disappeared in the constrained layer cabinet (three previous cabinets with 19 mm MDF had the same "problem") and pushed to lower frequencies. Are the findings suprising? I can't say. If they are real-world (=audible) problems, can they be dealt with? Certainly, I would say.

 

[Thomas_A]

Today I made some acoustic measurements of the two speakers, about 2.9V signal, frequencies 510 Hz, 620 Hz, and 790 Hz, and RTA using REW. Speaker A is the old one with 19 mm MDF and speaker B with constrained layer damping. Note that the two 5 inch drivers are different Peerless drivers, but as mentioned, I do not have the previous conventional cabinets that I made for the newer CSX drivers. Matching levels were difficult, but THD was consistently lower in speaker B, and especially around 600 Hz.

At 510 Hz, speaker A

Speaker B:

620 Hz, speaker A

Speaker B:

790 Hz, speaker A:

Speaker B:

 

[Tangband]

Very interesting Thomas .
-40 dB is about 1 % distortion ?
Many modern drivers ( SB NBAC 15/17 or purifi ) have then less distortion than some of the cabinet resonances.
The distortion from the cabinet walls ( regardless of which ) seems to be lower in this case than the sound potentially coming through the cone of the boomer ? Can you find a way to measure this also ?
( Linkwitz mini have -40 dB through the cone using a damped 80 cm long pipe heavily stuffed, and Bosse Bengtsson measured -20 dB worst case in his experiments with a conventional loudspeaker box )

 

[Thomas_A]

Very interesting Thomas .
-40 dB is about 1 % distortion ?
Many modern drivers ( SB NBAC 15/17 or purifi ) have less distortion than the cabinet resonances.
The distortion from the cabinet walls ( regardless of which ) seems to be lower in this case than sound potentially coming through the cone of the boomer ? Can you find a way to measure this also ?

Thanks Tangband!
I am not sure if continue these kind of measurements any more. I became satisfied when I made the constrained layer cabinets. The constrained layer idea was based on things I heard and vibrations I felt on the cabinet, at the time when I had my main DIY period. I may pursue some other measurements later on, but not on this.

 

[Tangband]

Thanks Tangband!
I am not sure if continue these kind of measurements any more. I became satisfied when I made the constrained layer cabinets. The constrained layer idea was based on things I heard and vibrations I felt on the cabinet, at the time when I had my main DIY period. I may pursue some other measurements later on, but not on this.

Ok I understand.

 

[Thomas_A]

I also tested the 250 Hz since it appeared that the 2nd order vibration was higher on the constrained layer cabinet. Distortion was similar/slightly higher for second order component, but the 19 mm cabinet had all higher distortion for the higher order distortion components.

Speaker A, 250 Hz:

Speaker B:

 

[Thomas_A]

Stupid of me though, I should have measured the 510, 620, and 690 Hz bands (not 790 Hz).