I want to learn about cone (woofer) break-up.

[MinusTheBear]

I tried searching for some Audioholics articles on cone break-up but couldn't find any. I understand different cone materials such as paper and metal "break-up" at different frequencies and have to be dealt with by the engineer/designer in the crossover region specifically. But what I am most interested in is to learn how to interpret cone break-up and if it can be found in various objective measurements such as frequency response, cumulative spectral decay, impedance etc. Also, if cone break-up is not dealt with properly its impact on overall perceived sound quality when doing listening tests.

I know this isn't talked about much so I thought I would throw this out there and hope to learn a few things.

Thanks.

 

[GranteedEV]

Waterfalls:

Lets start with a...poor..speaker:

take a look. see the 'fingers' in the upper midrange and treble, where the frequency response has alsp lost smoothness?

now compare that to a good waterfall

 

[MinusTheBear]

So would the area around 3kHz be most problematic? Since the waterfall plot shows resonances in the upper midrange and treble would this be perceived as "brightness" or perhaps "sibilance" with vocals in listening tests that could otherwise not show up in a frequency response graph?

How do we interpret the difference between woofer resonances produced by cone break-up and that of cabinet resonance? A take it cabinet resonance would not be happening in the upper midrange/treble?

 

[Swerd]

So would the area around 3kHz be most problematic? Since the waterfall plot shows resonances in the upper midrange and treble would this be perceived as "brightness" or perhaps "sibilance" with vocals in listening tests that could otherwise not show up in a frequency response graph?

Yes, the upper midrange around 3 kHz. It would sound bright or sibilant, depending on what music you are playing. Some music might not excite (or provoke) those resonances.

Many people blame this type of noise on the tweeter. But it is really woofer breakup that is not filtered out enough by the crossover. A good example of this, that I've heard, is the B&W 602.

How do we interpret the difference between woofer resonances produced by cone break-up and that of cabinet resonance? A take it cabinet resonance would not be happening in the upper midrange/treble?

When woofer cone resonances are insufficiently filtered by the crossover, in my experience, they sound much more prominent than cabinet resonance. Speaker measurements, like FR curves, made with a microphone will show the cone break up prominently. Typically, to measure cabinet resonances, you need different type of pick up, an accelerometer, that attaches directly to the cabinet surface.

Here are links to two sites that do a good job of explaining woofer breakup and what can be done while designing a speaker to avoid hearing it.

Crossovers 101
A well written (slightly oversimplified) outline of how to go from a raw unfiltered woofer without any crossover, to a proper design. It includes frequency response curves to illustrate each step.

Infinity White Paper on CMMD
This is longer, but discusses the various advantages and disadvantages to various woofer cone materials. It is well written, but keep in mind that its main purpose is to sell Infinity products.

 

[Swerd]

But what I am most interested in is to learn how to interpret cone break-up and if it can be found in various objective measurements such as frequency response, cumulative spectral decay, impedance etc.

You've now seen examples of woofer break up in frequency response and cumulative spectral decay (waterfall) plots.

You can also see evidence of woofer break up in impedance plots. In this example, you see a 7" woofer with a kevlar/paper cone. Scroll down until you can see the FR and Impedance plots. This is the woofer's raw unfiltered response, not in any specified cabinet, probably mounted on a large baffle.

Look first at the FR curve (top). There are some small irregular peaks just above 1 kHz, and several larger peaks above 2.5 kHz. At first glance, I would think the ~1 kHz peak isn't too bad, but I'd definitely crossover this woofer low enough to avoid the 2.5 kHz and higher peaks.

Now look at the impedance curve (lower figure). (There are two curves both in black. One is impedance magnitude vs. frequency. It has the large peak below 50 Hz indicates the Fs frequency of the woofer - ignore that peak, its not a problem. The other curve is impedance phase vs. frequency.) Above the large Fs peak, the impedance magnitude curve should gradually rise smoothly. Notice that just above 1 kHz there is some small unevenness in the gradually rising impedance. Most important is that the FR and the impedance curves both show something suspicious at the same frequency, just above 1 kHz. This is a good indication of the beginning of break up or a funky resonance developing between the kevlar/paper cone and the rubber surround. Notice also that this impedance bump is larger than the ones around 3 kHz where the FR curve shows worse looking break up peaks. This kind of impedance curve combined with the FR curve is enough for me to say, drop this woofer and look for something else.

Also, if cone break-up is not dealt with properly its impact on overall perceived sound quality when doing listening tests.

In my opinion, if you can hear woofer break up, it is the worst sounding noise an otherwise undamaged speaker can make. It means the crossover is poorly designed. Start over.

Most people blame woofer break up noise on the tweeter only because the noise is high frequency.

 

[MinusTheBear]

Thanks swerd. That was a nice correlation between the response and impedance issues. Could a response and impedance graph that his highly smoothed mask these resonances when interpreting the possibility of cone break-up?

 

[Swerd]

Thanks swerd. That was a nice correlation between the response and impedance issues. Could a response and impedance graph that his highly smoothed mask these resonances when interpreting the possibility of cone break-up?

Yes, some manufacturers are honest and publish good data, and others smooth it to the point of hiding all the flaws.

 

[MinusTheBear]

BTW are there any other measurements that show this issue?

 

[Swerd]

BTW are there any other measurements that show this issue?

Not that I know of.

With raw woofers, SPL or impedance vs. frequency curves can easily show where the problems are.

The trouble with most commercially available speakers is that all you may have is a frequency response curve of the finished product. They probably won't show FR curves of the drivers without a crossover. If they did a really poor job with the crossover, you might be able to see break up noise in a FR curve - if excessive smoothing doesn't hide it. But if they did a half decent job, you might not see it.

Depending on the music you play, the break up noise may be prominent and ugly sounding or not at all noticeable. I've learned from experience to notice if a speakers adds an "extra edginess" to the midrange sound. At first it may sound "more detailed" than other speakers, but with music that hits the right notes, the edginess becomes irritating noise, eventually causing listener's fatigue.

 

[Swerd]

In the 1980s, Celestion first used the technique of laser interferometry to develop new drivers. It allowed optical measurements of the movement of a driver's surface without touching it.

It took me a while to find this, even with google . These images show, better than anything else, what break up mode is.

A dome tweeter moving as a piston

A dome tweeter breaking up

 

[GranteedEV]

BTW are there any other measurements that show this issue?

Metal cone breakups are a lot easier to notice on a frequency response graph as they`re often 5-15db peaks above the intended pass band. I've also noticed breakups pop up on harmonic distortion measurements:

Harmonic Distortion of a metal cone:

Notice the skyrocket-effect?

 

[yepimonfire]

In the 1980s, Celestion first used the technique of laser interferometry to develop new drivers. It allowed optical measurements of the movement of a driver's surface without touching it.

It took me a while to find this, even with google . These images show, better than anything else, what break up mode is.

A dome tweeter moving as a piston

A dome tweeter breaking up

polk audio uses similar technologies to design their "dynamic balance" drivers, reducing modal resonance and "break up" in their drivers. their article on it explains (somewhat simply) what it is and how it works http://www.polkaudio.com/education/tech_article.php?id=25 the article has animations of break up showing just how bad it can be.

 

[qingcong]

Am I right from interpreting the waterfall plot that the cone breakup is not dependent on amplitude?

 

[jliedeka]

I think that's right. Cone breakup is a function of frequency.

Notice that the breakup propagates down especially as 3rd and 5th harmonics which sound pretty discordant.

I would point out that breakup nodes don't always show up in the impedance graphs. If you do see hiccups in the impedance, that is indicative of something which could be cabinet resonance or a driver problem. Cabinet resonance issues generally fall in the mid-bass to midrange frequencies.

Jim

 

[GranteedEV]

Am I right from interpreting the waterfall plot that the cone breakup is not dependent on amplitude?

The cone breakup itself is not dependant on amplitude. But presumably, it's audibility to our ears may very well be.

 

[qingcong]

Metal cone breakups are a lot easier to notice on a frequency response graph as they`re often 5-15db peaks above the intended pass band. I've also noticed breakups pop up on harmonic distortion measurements:

Harmonic Distortion of a metal cone:

Notice the skyrocket-effect?

The typical frequency response plot, is it the system's response to the fundamental or the fundamental + harmonics?

 

[GranteedEV]

The typical frequency response plot, is it the system's response to the fundamental or the fundamental + harmonics?

It is a gated "Fundamental only" plot

 

[qingcong]

What do you mean by "gated"?

 

[qingcong]

So anyways, would this mean that two speakers with the same frequency response plots might sound totally different due to different THD characteristics?

 

[Swerd]

What do you mean by "gated"?

Gated refers to time, where the test microphone is opened and closed by computer test software over a brief timespan of a few milliseconds.

The audio test software (something like Praxis) sends a very brief transient pulse of pink noise to a speaker being tested. The speaker changes the pulse to sound, and the sound pulse propagates through the air to the measurement microphone, and back to the computer for analysis.

Of course, this sound pulse also travels everywhere else in the room too, reflecting from the walls, floor and ceiling and other surfaces, losing energy all the while. These multiple reflections will eventually reach the microphone, contributing to the overall impulse response. With test software that has an adjustable "time gate", you can select a short time window that picks up mainly sound from the speaker, but little or no sound from room reflections.

Many people have downloaded Room EQ Wizard (REW), free software designed to help identify bass problems due to room reflections. It does not have an adjustable time gate - the microphone stays open and picks up all the room reflections. As a result, REW does not work as way to test speakers. Many people try anyway and get very confused by the results they get.

So anyways, would this mean that two speakers with the same frequency response plots might sound totally different due to different THD characteristics?

That is much too broad a generalization .