Interesting results. Pretty much confirms my subjective feelings towards MDF, they sound "quiet". The thicker the MDF the more there is difference between the surface hardness and the inner part hardness (surface hard, inside soft), "natural CLD"
Good sounding cabinet can be made out of almost anything in the end. Mdf is very good for as-is lively sounding pro paper cones that transfer energy quite efficiently to the baffle (low loss and stiffer suspension). Plywood might be subjectively better choice for dead sounding rubber surround, low qms, metal/ceramic drivers adding some "life"? Generally speaking.
Good sounding cabinet can be made out of almost anything in the end. Mdf is very good for as-is lively sounding pro paper cones that transfer energy quite efficiently to the baffle (low loss and stiffer suspension). Plywood might be subjectively better choice for dead sounding rubber surround, low qms, metal/ceramic drivers adding some "life"? Generally speaking.
Last edited:
I am still struggling a bit to understand the configuration and, for example, how the loads that are vibrating the main box are being applied.
Is the small driver bolted to the small box and then the small box attached to the large box in some way(?) with the small driver poking through a hole in the baffle of the large box as suggested by the photo in #50. Or is the small driver bolted to the baffle and then the small box attached over it in some way(?).
Is there stuffing inside the main box to remove the internal box resonances or is it empty?
Is the small driver bolted to the small box and then the small box attached to the large box in some way(?) with the small driver poking through a hole in the baffle of the large box as suggested by the photo in #50. Or is the small driver bolted to the baffle and then the small box attached over it in some way(?).
Is there stuffing inside the main box to remove the internal box resonances or is it empty?
Short sidestory: Im renovating our house and installed new front door. Almost 10cm thick, one aluminium sheet layer, wood vaneering on both sides and XPS foam insulation inside. Terrible sound, sounds ugly when opening/closing. Very bad sound when music is playing inside and listening from outside. I should measure it, propably a high Q resonance. Anyway, for measuring sound of a speaker panels one should build a brick house, attach panel to an opening, play music inside, measure from outside to get transmisson through the panel.
That is good practice when measuring transmission in an acoustics lab. Two highly reverberant rooms, one standardized opening in the wall between. I won't go into the measuring details here... but the results are translated into practical rules for the calculation of sound transmission (or isolation).
This. Source driver screwed to baffle, small box over it also screwed to baffle. So it will transfer some energy to the baffle, but it is very small and therefore stiff (much higher resonance than the main box), and well damped, and is the best I can do. Small box is stuffed.
The test box is currently unfilled. Next test is cut a hole for the dummy woofer, fill or line the box, and measure radiation out of the dummy woofer. I could also use the same wood to cap the hole and do these side/rear measurements again, wouldn't be exactly the same starting condition as these though since the hole+cap will surely change response somehow, but might be interesting anyway.
There is no debate whatsoever if you possess an introductory or better knowledge about the physics of vibration. Although not widespread this knowledge is possessed by a few that post here. The problem is that the many without this knowledge have difficulty distinguishing between posts from those that genuinely know what they are talking about and those with confidence and enthusiasm but limited knowledge. The posts keep rolling in.
Do tell, oh wise one.
Remember that we will check what you say against what augerpro finds.
I disagree, the problem is the results have never been presented in a relatable form (relative SPL), across typical construction methods, with standardized box and methods. Just because we know what is better, doesn't mean we know *how much better* in the real world where decisions and compromises must be made. I aim to change that with this testing.
Also: I've been in discussions where it was hypothesized there are different "behavior regimes" that manifest at different frequencies, and would be important to know depending on, for example, the box was only for a mid range, versus something that will produce low bass. There might be a difference between the output of a flexing/vibrating panel and one where the sound is moving through the panel to the outside. I'll get into this more later.
Last edited:
Thanks for the information which I think has clarified what you are doing.
For me an experiment looking at the forced response of various boxes of similar size but different materials and methods of construction would be useful if the measurements provided information on the real-world mass, stiffness and damping. The mass is easily determined from the dimensions and weight, the stiffness will follow from the frequencies of the resonances and the damping, by far the most interesting property, will follow from the shape of the resonances in the FRF.
What are the dimensions of the box and when you change wall thickness is the inner dimension, outer dimension, or something else held constant?
Absorbing material inside the box will remove the internal resonances of the air cavity which might be showing up as small resonances in the measurements. The frequencies would be easy to determine though.
I presume the parameters for the small driver are known? A simple lumped model should be OK to determine the forcing (force vs frequency applied to the baffle) at least over the lower frequencies of the passband.
The only issue of note that I can currently see and mentioned earlier is the resolution of the FRFs failing to resolve the high Q resonances in order to calculate the damping. The reason the damping is interesting for a full box is because it varies significantly with frequency unlike mass and stiffness, it tends to be only vaguely known as a material property for typical speaker materials and, unlike mass and stiffness, is also affected by what goes on at surfaces like joints and mountings.
andy19191> 1) I can't hear mass, stiffness, or damping. I only hear SPL. Which the mic is measuring. 2) outer dimensions are held constant. Not sure this was the correct choice, but differences should be tiny anyway. Might have to revisit this later. 3) I've switched to 1/24 octave smoothing, do you think higher would be better showing Q in a meaningful way? I mean our ear/brain smooths this anyway, for some reason I remember that being one reason Harman used 1/20 or 1/24 smoothing versus the typical 1/3 or 1/6 used in the industry.
This illustrates my point beautifully. You have no idea whether I have relevant knowledge or not but you can see and value enthusiasm and confidence.Do tell, oh wise one.
Remember that we will check what you say against what augerpro finds.
I don't know about you but I don't design my speakers with my ears. I design them using mass, stiffness and damping. I do check them with my ears though.
The mic is measuring SPL at a location that cannot be immediately related to something more solid like the SPL at the listening position or the velocity at a known point on the surface of the speaker. With the information from a simulation of the experiment what is being measured can be quantified. It will likely be close to some form of spatial average of the velocities around the nearest point on the surface. It is likely at the frequencies of interest the spatial averaging will be unimportant but it would be wise to check even if only to have evidence to deflect future criticism given what you are doing is unusual.
Consistency across the builds is likely most important.
Ideally you want high frequency resolution with ensemble averaging rather than bin averaging to reduce the grass/wiggles.
It is common to present the same information in different ways to serve different purposes (e.g. time or frequency space, unweighted and weighted, waterfall,...). What tends to be important is keeping the raw unprocessed data in order to present in different ways when the need arises in the future. Ensemble averaging might get built in depending on your software but you are likely to be able to back out the bin averaging.
The perceived level of discrete tones will follow a highly resolved FRF. The perceived level of noise will more closely follow some form of smoothed FRF. What is "right" depends on what information you want.
I'm starting to get the feeling that my effort to read each of your plentiful words returns less and less value.
Do you care what the exact thickness of a speaker cone is? Why or why not?
Firstly, even if what augerpro is doing isn't perfect, it's still a lot better than what we currently have, and second, "future criticism" is kind of rich. You really mean "present criticism".
andy19191> "I don't know about you but I don't design my speakers with my ears. I design them using mass, stiffness and damping. I do check them with my ears though." So you trust your sims more than your own lying ears, eh?
"With the information from a simulation of the experiment what is being measured can be quantified." This illustrates an issue I'm having with your posts. You seem to be claiming precisely quantifiable numbers from a theory/simulation is more valid than an actual physical measure. This is anti-science. Might as well be metaphysics.
"The mic is measuring SPL at a location that cannot be immediately related to something more solid like the SPL at the listening position" Another issue I have with your posts, always pointing out problems never solutions. I've previously provided measurements at several distances and you have made no comment, you just bring it up again later like I never made the post.
"Consistency across the builds is likely most important." No ****. We've already established consistency is important, yet we must choose between two inconsistent realities. Give me an answer by choosing one, not repeating a problem we're all already aware of.
"With the information from a simulation of the experiment what is being measured can be quantified." This illustrates an issue I'm having with your posts. You seem to be claiming precisely quantifiable numbers from a theory/simulation is more valid than an actual physical measure. This is anti-science. Might as well be metaphysics.
"The mic is measuring SPL at a location that cannot be immediately related to something more solid like the SPL at the listening position" Another issue I have with your posts, always pointing out problems never solutions. I've previously provided measurements at several distances and you have made no comment, you just bring it up again later like I never made the post.
"Consistency across the builds is likely most important." No ****. We've already established consistency is important, yet we must choose between two inconsistent realities. Give me an answer by choosing one, not repeating a problem we're all already aware of.
The transmission measurement method in acoustic labs I mentioned before uses reverberant rooms. This has a reason: creating sound fields that are as diffuse as practically possible. The sound pressure level in both rooms is measured at various positions in order to acquire an adequate value of the sound intensity in both rooms. Then and only then one can claim some reasonable isolation or transmission value of the sample.
So measuring transmission isn’t that simple. Measuring at one position isn’t very adequate and only gives relative results. Mainly because you don’t have a clue about what direction the radiated sound has, given the flexing of the panels an their relatively large size. It likely is pretty directional though.
So measuring transmission isn’t that simple. Measuring at one position isn’t very adequate and only gives relative results. Mainly because you don’t have a clue about what direction the radiated sound has, given the flexing of the panels an their relatively large size. It likely is pretty directional though.
Measured baseline for the fill and lining testing. Empty box, no lining, no bracing. Dummy woofer is an SB Acoustics SB15, terminals shorted. I had both the poly MFC and aluminum NAC on hand so measured both. I would have guessed the poly would pass less sound, at least at higher frequencies, but...
