Identifying Enclosure Resonances

[Mr. Widget]

Earl K and I have been having a discussion about finding, identifying, isolating, and eliminating cabinet resonances that was begun following an ordeal I have been going through while experimenting with a new type of cabinet construction.

After building a pair of prototype speakers, I determined that I was not pleased with the "quality" of the bass response. I ran the usual frequency response measurements and impedance curves (see below) and didn't see any apparent anomalies. I scratched my head and wandered around lost in the desert for a bit.

Once I eliminated outside sources as the problem, my first inclination was to blame standing waves within the cabinet. After making modifications that satisfied me that they were not the problem, I thought about cabinet resonances... these cabs are way over built and I found it difficult to believe that this would be the problem, but with nudging from others I took a much closer look at the LF impedance curves at a very high resolution.... bingo!

Essentially, the idea is that every resonance, including those within the driver itself will show up in the impedance curve. Even something as minor as changing the construction of the basket will affect the impedance curve.

[Mr. Widget]

Here is a plot showing the resonant peak that was in my speakers and the effects on that peak by making changes.

[Mr. Widget]

I added some sandbags at key spots, (identified by trial and error) and ended up with the resonances being reduced to the level show in these curves. I had hoped to do better and will rebuild the cabinets using a different technique, that said, I post these to show the level of objective improvement... this yielded a speaker that is quite listenable and the disturbing, soggy bass has tightened up considerably. If I didn't have the ability to measure this, it would have been a much greater task to "fix" the problem...

In another conversation, Techbot has said that the relatively low cost "Woofer Tester" can acquire these types of impedance plots so others should be able to hunt down and destroy resonances and standing waves that may be affecting the performance of their systems too.


Widget

[duaneage]

It is unfortunate but true that any mechanical movement will affect the impedance. I have always looked at the impedance curves for tuning before using a microphone. It is surprising what varying degrees of fill can do.

Could you feel resonance on the cabinet or horns at the offending frequency?

[Mr. Widget]

It is unfortunate but true that any mechanical movement will affect the impedance.

I see it as a bonus, since without an anechoic chamber it is really difficult to accurately measure lower frequencies. Obviously I agree that damping that mechanical movement is a pain, but at least we have a fairly easy method of looking for some anomalies.

Could you feel resonance on the cabinet or horns at the offending frequency?

It was the cabinet, and oh yeah baby! When I checked out the region with a frequency sweep, nailing the frequency was quite easy!


Widget

[Lancer]

the relatively low cost "Woofer Tester" can acquire these types of impedance plots

Just for fun - here's the 1500AL mounted in a 4.00 cubic foot sealed test cube. One should note the standing wave at ~ 350 Hz.

The standing wave can be seen in the impedance curve. It is the blip at ~ 350 Hz. With the oscillator set at that frequency the most horrid sound emanates from the system as would be expected. The internal dimensions of the enclosure are 484 mm cubed.

Note that 484 mm corresponds to a wavelength of ~ 712 Hz hence the standing wave will have a frequency of ~ 356 Hz.

The addition of 2" thick OC fiberglass on all panels is sufficient to reduce the standing wave ~ 350 Hz. Impedance drops to "normal" and the objectionable sound is greatly attenuated.

This doesn't necessarily have anything to do with the project. I just thought it might be interesting to some that the standing waves generated in a cube are definitely objectionable and are viewable in an impedance run.

Some closed box data:

driver mounted normal
4.0 cu ft gross
no fill
Qtc = 0.56
Fc = 49.8

driver mounted normal
4.0 cu ft gross
2" OC fiberglass on all panels except baffle
Qtc = 0.54
Fc = 47.8

[Earl K]

Nice Info guys .

- I might add that Widgets boxes were built using a method of layering up pre-kerfed mdf over a substantially beefy internal skeleton.

- This approach was adopted so that he could execute curved walls ( 2 of 6 ) .

( multiple fixes ) ,,,,,,, these yielded a speaker that is quite listenable and the disturbing, soggy bass has tightened up considerably. If I didn't have the ability to measure this, it would have been a much greater task to "fix" the problem...

- I find it quite intriguing that "disturbing, soggy bass" translates into a blip that is a fraction of "1 ohm" , on a high resolution impedance run .
- Man, 2 ohms must really suck !

Here's a porduct link to WT2 ( at Parts Express ) for those interested in using it to track down a possible cause for "disturbing, soggy bass" .

[Ian Mackenzie]

Just for fun - here's the 1500AL mounted in a 4.00 cubic foot sealed test cube. One should note the standing wave at ~ 350 Hz.

The standing wave can be seen in the impedance curve. It is the blip at ~ 350 Hz. With the oscillator set at that frequency the most horrid sound emanates from the system as would be expected. The internal dimensions of the enclosure are 484 mm cubed.

Note that 484 mm corresponds to a wavelength of ~ 712 Hz hence the standing wave will have a frequency of ~ 356 Hz.

The addition of 2" thick OC fiberglass on all panels is sufficient to reduce the standing wave ~ 350 Hz. Impedance drops to "normal" and the objectionable sound is greatly attenuated.

This doesn't necessarily have anything to do with the project. I just thought it might be interesting to some that the standing waves generated in a cube are definitely objectionable and are viewable in an impedance run.

Some closed box data:

driver mounted normal
4.0 cu ft gross
no fill
Qtc = 0.56
Fc = 49.8

driver mounted normal
4.0 cu ft gross
2" OC fiberglass on all panels except baffle
Qtc = 0.54
Fc = 47.8

Like pointing to a distant Moon gazing at you on a still night there are also the more ponderious issues a lower frequencies:

That of Pressure waves as detailed by the quote below are best attenuated by a thick loosy layer of open cell foam with a laminated thin film plastic layer such open cell carpet under felt to attenuate such pressure resonances.

"Unless an enclosure is spherical or ellipsoidal, all or some of its sides

will consist of plane surfaces clamped at their edges. Such a clamped

panel will have its own acoustic output when forming part of an

energised loudspeaker, derived from sound energy within the enclosure.

The output consists of standing-wave modes at higher frequencies and

pressure modes at those frequencies where the wavelengths exceed the

internal enclosure dimensions. Adjacent panels may be similarly excited

by vibrational energy from the drive unit chassis.

Theoretically a clamped panel has a well defined vibrational series in

both longitudinal (volume stiffness) and bending modes. A further mode

is due to the panel mass resonating with its own and the enclosure's air

volume stiffness. Stevens found that in a typical reflex cabinet this

latter resonance appearedat almost twice the fundamental enclosure

resonance, a condition verified over a range of tuned system frequencies.

[edgewound]

JBL' Enclosure Design Reference Manual states that an enclosure should never be constructed with indentical dimensions, such as a cube, nor dimenesions that can be divided equally into one another...or multiples of one another, due to exciting unacceptable panel resonances. When building a rectangular box, the sides should be odd multiples of one another to reduce the panel resonance and also be well internally damped to thermodynamically eliminate internal turbulence.

[Lancer]

Yeah, we often point people to the JBL Enclosure Design Reference Manual. It's good stuff.

http://www.lansingheritage.org/html/.../jbl-plans.htm

http://www.jblpro.com/pub/manuals/enclgde.pdf

For those with BassBox 6 Pro you can look up standing waves in the online help.
Testing Loudspeakers by D'Appolito has some stuff too.
Many other sources out there as well.

The point of the cube was to exacerbate any standing wave and take an impedance measurement to identify it. The cube made it pretty easy to hear and measure as would be expected.

Awhile back someone argued that a cube was perfectly fine. I hope it worked out for them.

[duaneage]

Wow Ian, I got goosebumps reading that very sexy description of a vibrating speaker enclosure.


I am gonna need 20 minutes of quality time after that .....

[duaneage]

I see it as a bonus, since without an anechoic chamber it is really difficult to accurately measure lower frequencies. Obviously I agree that damping that mechanical movement is a pain, but at least we have a fairly easy method of looking for some anomalies.




It was the cabinet, and oh yeah baby! When I checked out the region with a frequency sweep, nailing the frequency was quite easy!


Widget

When I first built speakers 15 years ago I had nothing more than a signal generator and meter. I would place sawdust on a cabinet wall and look for the dust to move as I swept the frequencies. Touching the cabinets would sometimes affect the test. The dust would fall off the sides when I hit a magic note. After jamming a brace here or a gusset there I would end up with a reasonably tight cabinet.

One could use piezo elements with wood blocks fastened to measure vibration but the dust trick still works for me.

[Ian Mackenzie]

Wow Ian, I got goosebumps reading that very sexy description of a vibrating speaker enclosure.


I am gonna need 20 minutes of quality time after that .....

Yep,

Mrs Lansing loves it when I whisper groovy stuff in her ear..

Its an old trick but a goodie for absorbing the stuff 250hz and below, a very different kind of effect to fibreglass.....that said its basic physics.

[yggdrasil]

Hi Widget.

I just stumbled over this page: http://users.tpg.com.au/users/gradds...udspeakers.htm

Maybe there are a few usable ideas.

[Mr. Widget]

Stacked laminations are certainly one way to do curved cabs... but, boy what a PITA for a 4' tall cab!

That is a good link. Someone posted it before on one of the Project May threads.


Widget