Measuring sound output from speaker cabinet walls

Status
Not open for further replies.
Thanks to y'all who have provided info & links.

DDF: I will check over that infodump in a day or three, and update post one to be a convenient link repository. Maybe I should put an edited version (no questions; just links and snippets) in the construction forum.

Today: I'll pick up some stuff - I'll buy some damping material and pull a light, empty ply box out of storage - and begin some tests (like my proposal in post one) to see if I can get good results (with reasonable weight, and basic materials).
Charlie has suggested an interesting test, which seems pretty easy to run with existing cabinets:

You might instead can put a mini speaker in a small cube and point it AWAY from the cone, e.g. at the back of the box. I think that is a fair test. The cone leaks way more sound that people realize, especially those lightweight "pro audio driver" cones... I heard this exact problem with a "high scoring DIY loudspeaker" very recently. It used two 8" or 10" pro audio woofers in a large cabinet, possibly vented. Wow, you could hear the internal reflections from the cabinet walls very clearly. I think most of it was coming straight at me right through the woofer cones.

If, in the midrange, the levels that are leaked through the cone are vastly larger than the levels radiated through the walls (when I run the same test, but with the unloaded cabinet face-down), that'd imply I should spend more effort & weight on treating the former.

I mostly use lightweight pro drivers, so I think this test will favour large cabinets, relative to driver size, since they'll have more stuffing to eat those internal reflections.

At LFs the rear sound just acts as a spring and the energy gets absorbed mostly in the speaker itself

By 'speaker', do you mean the drive unit itself? Because this?

because the speaker would be playing and cone movement that is not in the signal would be strongly opposed by the amplifier. This is what the impedance, i.e. back emf is all about.
 

They get a bit fancier than an accelerometer:

"The rest of this tutorial will focus on the use of the Klippel Scanning Laser Vibrometer (SCN)."

Case study 2 shows how good selective bracing can be "The 350Hz resonance has been reduced 25dB", but it's a picked example of an expensive measurement tool improving a cheap speaker, the opposite of the usual DIY scenario.

Case study 1, their ABS versus composite comparison is also impressive, and probably easier to apply. The two enclosures look exactly alike, and are presumably of comparable weight, but the composite "side panel has a sound power that measures 10-15 dB lower".

I suspect someone probably already posted these measures:
http://downloads.bbc.co.uk/rd/pubs/reports/1977-03.pdf

That was my first ref in post one 🙂

In the BBC report, the closest thing to AcoustaComp they measured was ye olde fibreglass. They commented, "the figures of merit obtained for the glass reinforced plastic materials clearly exceed that for the wood based types".

...so maybe I should change up the idea I put forward earlier, by making the fibreglass proportion larger.

The shell would be ply (4mm F27)* onto which I'd laminate an aesthetic layer (5mm strand woven bamboo). I'd line it internally with a thin layer of fibreglass. For a light ~10mm shell, that's about as strong as I think I could make it.

AcoustaComp* seems to be kitchen-sink version of fibreglass, with lots of different components. I can't duplicate that, but I can make my own mess that might work almost as well: I might see what fillers I can get locally, for a decent price, possibly just from the supermarket: this discussion has a champ who found cornflour to be a good additive to polyester resin to make it less brittle (as a low cost alternative to Renshape).

Renshape alternative

*(AcoustaComp polyester BMC, polyester SMC, vinylester SMC, Epoxy SMC, phenolic SMC, granular phenolic, granular diallyl phthalate, granular melamine formaldehyde, chopped silaca phenolic prepreg, chopped quartz phenolic prepreg, and chopped glass epoxy prepreg).
 
The high Q/low Q Harman work was applied to the signal. It is correct. Cabinet resonances are low level distortion. High Q means loud and close to the signal in level and continuing to resonate after the music stops. Low Q means tens of dB below the signal and stopping when the music stops. There is not much doubt about which is preferable.

A high Q resonance is over a very narrow bandwidth while a low Q resonance has a much wider bandwidth. The latter often means that it is more often perceived. More importantly a big Q resoance needs to have large amounts of energy pumped into it over a very narrow bandwidth, something that does not easily happen with music as the energy driver, particularily if the resoance is at HFs. The resonance potential is there, but rarely excited.

dave
 
Fraud alert... for all you folks who think Klippel is above reproach and think it is always OK to link commercial websites as if they were objective.

In that otherwise interesting plug for Kilppel's expensive approach, they mention that accelerometer testing is faulty because of 60 HZ pick-up.

Ho, ho, ho, shown in their picture is the ACH-01-4 which is the accelerometer model sold with unshielded leads. DAMHIK*.

The shielded model, ACH-01-3, would be great for testing boxes and save you prolly $10,000 over hiring Klippel.

BTW, that paint from Parts Express sounds like the latex stuff I mentioned earlier. I bet latex with cork pieces is just as good.

B.
*OK, I bought an ACH-01-4 on eBay and having been testing motional feedback with it this week... not an issue if you run RTAs in which case that peak is just one spot in the tone compass
 
Last edited:
You can try this.

I may use something like that.

That particular product = glue + fillers, and is quite bulky. Since I can mix my own with locally available materials, and not pay for shipping, that's probably how I'll roll.

I've got a well caulked but otherwise bare 9mm ply box ready for testing. I'll start off simple, and try the surface damping that was good enough for the BBC. I want to see if this alone will perform better than my existing (thick, heavy plywood) boxes, at much less weight.

The resonance potential is there, but rarely excited.

I've seen you state this before, but I don't fully understand the reasoning.

"shifting the panels' resonance to higher frequencies" was discussed earlier in this thread (45-47).

You have nearly the opposite build philosophy to Dr Geddes, with almost no stuffing / damping, so I'm interested to know your response to the question of post 47.
 
From those posts i’ll take this:

A stiffer structure takes more energy to excite it so it may not have as great a response amplitude.

Which reinforces my thot.

Add in that at HF there is less energy to excite anything and perhaps you can see my approach of getting potentisl resonances up high and just not excite them. That has been a successful approach so far.

I use a mechanics stehoscope to find problem areas.

dave
 
Fraud alert... for all you folks who think Klippel is above reproach and think it is always OK to link commercial websites as if they were objective.

DDF gave me a bunch of relevant links and resources to dig through. This is grand; exactly what I asked for in post one.

I've already shown that I realise the Klippel info is an advertorial, and needs to be read selectively:

Case study 2 shows how good selective bracing can be "The 350Hz resonance has been reduced 25dB", but it's a picked example of an expensive measurement tool improving a cheap speaker, the opposite of the usual DIY scenario.

...so yea, it is OK. Everyone is playing together nicely.
 
If it is in the bandwidth of the potential resonance and the energy is insufficient to excite it, it is damped by the box material (turned to heat) — i use plywood, the interface between plies is a good energy absorber especially at HF. If the energy is not in the bandwidth of the potential resonance it is reflected.

dave
 
I may use something like that.

That particular product = glue + fillers, and is quite bulky. Since I can mix my own with locally available materials, and not pay for shipping, that's probably how I'll roll.

You have a source for hollow ceramic microspheres? I've used this before and when you rub it between your fingers you can feel the spheres like it's really gritty. Whether it works better than anything else close at hand, I have no idea. One of the things I want to test as I mentioned earlier. It did radically change the 'ol knuckle rap test!
 
If the energy is not in the bandwidth of the potential resonance it is reflected.

Not the whole story since most energy to excite resonances comes from the speaker/baffle interface. The energy not at a frequency to cause a resonance, bounces around in the walls until it disipates as heat. In my boxes i usually couple the driver to at least 3 panels other then the baffle so as to share any energy amongst them to provide more material to damp it.

dave
 
i use plywood, the interface between plies is a good energy absorber especially at HF.

Yea, that BBC report stated:

"This (stiffness/resonance) is a measure of the minimum attenuation of a sound wave through the walls of the loudspeaker cabinet and as such is a figure of merit of the material for this purpose [...] It will be seen that, on this basis, of the wood based materials, birch ply gives the highest value".

Chipboard etc <0.3
Plywood ~0.4
Fibreglass ~0.8

They didn't list MDF, but I assume it would be at the low end due to its softness.

Birch was the only ply with a high ply count tested. All else equal, I guess other plies would rank roughly in proportion to the base material's hardness ...hence your excitement about getting ply made entirely from strand woven bamboo.

A few Australian natives are very hard, hence I'm leaning towards a core of locally made construction ply that's about twice the stiffness of birch (F27 vs F14), and that costs about a quater as much.

If the energy is not in the bandwidth of the potential resonance it is reflected.

Thanks.
 
A note on good damping materials:

Damping is always done by friction. Friction is created in a material as the particles move against each other. It is always beneficial to add in some particles in the form of a filler. I use 3M micro-spheres which are very small hollow glass beads. Solid ones are also available. These spheres rub against one another and dissipate the energy.

The green glue is OK, but no where near as good as it would be if filled. As a damping compound the binder should not set-up to be solid. The best that I have found is a soft 2-part polyurethane from a place like Innovative Polymers. Then fill it with the spheres into a paste. That stuff works exceptionally well.
Newbie Query so pardon.
How are these damping panels made with micro spheres ? If held between two panels with some force they will be as good as solid one body. I mean the spheres will not be able to move as much to create friction and dissipate heat. Are these filled loosely in hollow space ? I am interested because this may be helpful in turntable plinth Building.
Thanks and regards
 
You have a source for hollow ceramic microspheres?

Yea, this:

Microspheres | Domcrete

Or, for smaller amounts, art supply places: they are used as a casting filler material.

...but I'll try the basic stuff before I go nuts with CLD and space-age composites.

Whether it works better than anything else close at hand, I have no idea.

Me either. This is my first try at using good published data to choose between materials, relative thicknesses, and so on.

Previously I went with common wisdom (some of which will have been based on measurements, some not) and over-engineering, but that means I've made stuff a lot heavier than it needs to be.
 
A high Q resonance is over a very narrow bandwidth while a low Q resonance has a much wider bandwidth. The latter often means that it is more often perceived.
When applied to the signal! In this case resonances make the signal louder. A low Q resonance creates a wide bump in the frequency response which is relatively easy to perceive. A high Q resonance is only a problem if it lines up with a note in the music. If this doesn't happen then it does nothing.

Sound radiation from the cabinet is not the signal you want to listen to it is a low level distortion you don't want to listen to. The Harman work did not address this case and is therefore largely irrelevant. The level of sound radiation from typical cabinets is 40dB or more below the signal at frequencies away from resonances. This is quiet enough to be inaudible. The problem is that high Q cabinet resonances can raise this unwanted sound by 20dB or more and in particularly severe cases up to the level of the signal. This is audible as was shown by the experiments reported in the BBC papers.

So the wise DIYer first seeks to have no resonances in the passband of the driver/s. If that cannot be achieved then they seek to damp, damp, damp. And the wisest will ensure it is the resonances in the midrange that are the best damped because these are the more audibly intrusive.

More importantly a big Q resoance needs to have large amounts of energy pumped into it over a very narrow bandwidth, something that does not easily happen with music as the energy driver, particularily if the resoance is at HFs. The resonance potential is there, but rarely excited.
Music consists of a sequence notes. These notes are a set of harmonically related tones that are sustained for a large number of cycles. They are almost ideal for driving high Q resonances with only a sustained pure tone being better. Which is good news because it is primarily resonances that add the richness and colour to sound.

When a high Q resonance is driven effectively all the energy available from the source is transferred to energy in the resonant motion. A person getting the timing spot on when pushing and pulling a child on swing is doing this. It may take a significant number of cycles to get the cabinet radiation to the absolute maximum of say 30dB above the equivalent non-resonant floor of 40dB or so below the signal but this is missing the point. The cabinet radiation will get louder at the fastest rate possible given the work available from the driver (or person in my example). So it may have risen "only" 25 dB louder when the driving note is over but it will then take many cycles to decay because it is a high Q resonance.
 
Digikey sells accelerometers, but their search engine is useless on my cell phone. There is a piezo electric one that is about $10 that works well. You will most likely need to build a preamp for another $10 or so.

Then there is the simple method to test for box wall integrity. Tap each panel with a rubber mallet. A dull thud is good. A bass drum is not.

Low frequency leakage is also level dependent. At some energy level you get buckling behavior. Not usually a problem in home use.

Another technique is to enclose sand in the cabinet walls. If too heavy try vermiculite.
 
Status
Not open for further replies.