Controling Cone Breakup In An Active System

[Ray Tremblay]

How do you control cone breakup in an active system assuming you were running a two way setup? Example: RS180 with an RS28.

[Todd G.]

Re: Controling Cone Breakup In An Active System

How do you control cone breakup in an active system assuming you were running a two way setup? Example: RS180 with an RS28.

You don't control it so much as you try to avoid it. The use of steep slopes and carefully placed xover points are your best bet. EQ can help some, but not much really. Ideally you want the breakup to be way down so as to be inaudible and thats a function of your crossover freq and slope.

[Ray Tremblay]

Re: Controling Cone Breakup In An Active System

Is it a good idea to use passive notch filters in addition to an active crossover, or would it be more advisable to run more drivers (3 way setup) in narrower bandwidths?

[Deward Hastings]

Re: Controling Cone Breakup In An Active System

How do you control cone breakup in an active system assuming you were running a two way setup? Example: RS180 with an RS28.

LR4 at a nominal 1600 Hz. Not only eliminates cone breakup but also has you 8-9 dB down at the 5th harmonic distortion peak at 1800 Hz. (which is related, of course to the cone breakup). There's still a bit of a problem that comes from the RS180's "hump" between 1 and 2 kHz. . . . you can drop that with a shallow notch or reduce it a bit by pushing the low-pass down to 1500 Hz. Another trick is to add a passive component or two to flatten the driver. And if you've got a crossover that can do it use a B3 at 1800 or thereabouts on the tweeter, or a LT if that's available. Combined with the drivers own characteristics that gives something more like LR4 overall . . . phase tracking, on axis lobe, all that good stuff. A digital crossover lets you do all sorts of things because the delay lets you correct for offset without worrying about phase/frequency response shifts.

Play with the options in PCD . . . it's easier to see what happens than to explain it.

[johnk...]

Re: Controling Cone Breakup In An Active System

How do you control cone breakup in an active system assuming you were running a two way setup? Example: RS180 with an RS28.

Cone breakup is nominally a form of linear distortion. The linear aspects of it can be controlled (or eliminated) by designing a filter transfer function which brings the driver response into agreement with the desired target acoustic transfer function. For example, if the want the woofer to have a B4HP response at 40 Hz (like a ported box) and a 1.5k LR4 LP response at the crossover response then if then, if we call this acoustic target Ta, the filter transfer function is given as Tf = Ta/Td, where Td is the driver SPL response measured in you box on the design axis. Once you have Tf it is just a matter of emulating it with a passive or active circuit, or digitally.

You can get an idea of this at my discussion of stored energy and driver response. These drivers were filtered digitally to match the desired acoustic transfer function. The deviation form the ideal CSD is due to possible inaccuracies in the digital filter TF and nonlinear distortion which can not be eliminated by the filtering.

[Deward Hastings]

Re: Controling Cone Breakup In An Active System

Cone breakup is nominally a form of linear distortion.

Linear? I recall seeing strobe photos of cone breakup in radial and tangential modes over 40 years ago, and they looked anything but “linear”, and Klippel describes cone breakup as nonlinear:

“Our approach is based on
physical modeling of the transducer.
It works only for electrodynamical
transducers coupled with an acoustical
system having a few number of
resonances like a closed or vented
box. We can reduce the distortion
generated by motor and suspension
nonlinearities but not the nonlinearities
in the cone and in the following
multi-dimensional transfer path."

Unfortunately all the AES articles with pictures seem now to be behind a price tag, and I’m not a member any more. While the simple breakup modes may be associated with “linear” resonances in theory (like the center of the cone and the periphery moving in opposite directions at some particular frequencies) those resonances are so far removed from the driving signal that “correcting” them with filters on the signal is all but impossible. It’s rather like trying to stop a rattling window by filtering the signal to your subwoofer.

http://www.klippel.net/scanner/c5-scn.pdf

http://www.klippel.de/download/News/AN31%20cone%20Vibration%20and%20Radiation%20diagnostics.pdf

While many of the effects we see are resonance related

http://www.linearx.com/products/accessories/LS310/LS310_01.htm

even those resonances are sufficiently disconnected from the drive signal that an anti-resonant filter is not going to help. A rocking cone resonance cannot be cured by a filter in the signal path.

Or scan down this page a bit for an image of a Kevlar cone “breaking up”.

http://www.soundinvestment.co.za/Infinity/PremiumSystems/Cascade/Pages/CascadeTech.htm

Avoidance is the only practical cure.

[Jeff B.]

Re: Controling Cone Breakup In An Active System

Linear? I recall seeing strobe photos of cone breakup in radial and tangential modes over 40 years ago, and they looked anything but “linear”, and Klippel describes cone breakup as nonlinear:

“Our approach is based on
physical modeling of the transducer.
It works only for electrodynamical
transducers coupled with an acoustical
system having a few number of
resonances like a closed or vented
box. We can reduce the distortion
generated by motor and suspension
nonlinearities but not the nonlinearities
in the cone and in the following
multi-dimensional transfer path."

Unfortunately all the AES articles with pictures seem now to be behind a price tag, and I’m not a member any more. While the simple breakup modes may be associated with “linear” resonances in theory (like the center of the cone and the periphery moving in opposite directions at some particular frequencies) those resonances are so far removed from the driving signal that “correcting” them with filters on the signal is all but impossible. It’s rather like trying to stop a rattling window by filtering the signal to you subwoofer.

http://www.klippel.net/scanner/c5-scn.pdf

http://www.klippel.de/download/News/AN31%20cone%20Vibration%20and%20Radiation%20diagnostics.pdf

While many of the effects we see are resonance related

http://www.linearx.com/products/accessories/LS310/LS310_01.htm

even those resonances are sufficiently disconnected from the drive signal that an anti-resonant filter is not going to help. A rocking cone resonance cannot be cured by a filter in the signal path.

Or scan down this page a bit for an image of a Kevlar cone “breaking up”.

http://www.soundinvestment.co.za/Infinity/PremiumSystems/Cascade/Pages/CascadeTech.htm

Avoidance is the only practical cure.

I don't think that anyone is saying that there is no non-linear distortion involved. A 15dB peak in response will also certainly raise harmonic distortion at this frequency, in addition to the other factors that contribute to it. However, I agree with John, it is primarily linear distortion that we are dealing with. This is why crossover networks can be useful in controlling it.

[johnk...]

Re: Controling Cone Breakup In An Active System

[QUOTE=Deward Hastings;1512547]Linear?

The question was how to control breakup, not how to eliminate it. Cone breakup is fundamentally a linear phenomenon, just as any decent driver is fundamentally a linear transducer. This is evidenced by the CSD response I show at my web site for the two drivers considered. If it were not basically linear the CSD plots would not come close to the ideal CSD response. The complicated motion of a cone in breakup should not be confused as being the result of nonlinearities. Breakup can be reasonable well predicted by a completely linear (though complex) physical model. Cones constructed of homogeneous, isotropic materials tend to have more organized looking break up. Kevlar cones, and other cone made of woven material looks much more unorganized because of the properties of the cone are not isotropic.

I agree that breakup, per say, can not be stopped by filtering but the effects of break up on the radiated sound as measured on axis can be controlled by application of the correct filters. For example, how does SL control the breakup of the W22 in the Orion? He uses a notch filter. This “controls” the linear aspects of breakup because it reduces the energy applied to the driver at the frequencies where breakup results in response peaks. But filtering, for example, it will not eliminate the possible effect of elevated distortion at frequencies for which breakup occurs at frequencies harmonically related to the applied fundamental since these are nonlinear effects.

Cone rocking is generally a nonlinear form of sub harmonic distortion and not breakup in the conventional sense. Being a form of nonlinear distortion it can not be controlled by linear correction.

[Deward Hastings]

Re: Controling Cone Breakup In An Active System

For example, how does SL control the breakup of the W22 in the Orion? He uses a notch filter.

A notch filter at 5kHz *AND* a 4th order crossover at 1440 Hz. That's some pretty serious avoidance, and represents a pretty serious effort not only to get signal off the cone in the breakup range but to get off it before the breakup can act as a distortion amplifier for the 3rd harmonic distortion at 1.6kHz. That's what explains why ORION doesn't have any "metal cone sound". It's not "controlling" the breakup, it's avoiding it altogether. I would not be interested in listening to the W22 "full range" with a notch filter that merely "controlled" the breakup peak to flat . . .

[AJ]

Re: Controling Cone Breakup In An Active System

DCX2496. No analog active crossovers are suitable without external passive filters any drivers really.

[philiparcario]

Re: Controling Cone Breakup In An Active System

Active analog crossovers without any passive filters will work fine with the right drivers. The key is using a highend full range like a jordan 92 as a mid. It lets you cross low enough with a lot woofers or sub woofers. It is very flat from 80hz to 4000hz. then pick a smooth tweeter like morel's cross it above 3000 and your are fine take a look at my album on the album page just click on my name. I use a rane 3 way active. crossed at 90hz and 3500hz. the problem is drivers are very costly. even though I bought the drivers used, they still cost me 775. list they run over 1450

[johnk...]

Re: Controling Cone Breakup In An Active System

That's some pretty serious avoidance, and represents a pretty serious effort not only to get signal off the cone in the breakup range but to get off it before the breakup can act as a distortion amplifier for the 3rd harmonic distortion at 1.6kHz.

From the Orion manual, " UB7 in conjunction with R49 forms a 5k Hz notch filter to correct for a midrange driver out-of-band resonance. "

In other words, it is a linear correction to a linear response aberration resulting from breakup, and is intended to make the driver roll off smoothly in accordance with the LR4 LP filter.

A notch filter has no impact on the nonlinear aspects of the W22 or any driver. The choice of crossover point does as it limits the amplitude of the fundamental, thus ultimately reduce the amplitude of the HD by the reduction of the amplitude of the fundamental due to the crossover. With an LR 4 crossover at 1.44k Hz we are talking about an 8dB reduction at 1.6k thus an 8 dB reduction in distortion relative to the flat band level.

I don’t think we are in any disagreement on these points. The disagreement seems to be on whether or not breakup is nominally linear or not, for which I’ll stand by what I have said previously. The bottom line remains that the linear aspects of the driver SPL aberrations resulting from breakup can controlled by appropriate filters, nonlinearities can not. It then becomes a matter of assessing what is or is not linear.

[jkim]

Re: Controling Cone Breakup In An Active System

the linear aspects of the driver SPL aberrations resulting from breakup can controlled by appropriate filters, nonlinearities can not.

I agree. Technically, if we consider complex wave terms, linear distortions of a system change the relative amplitude of terms that are already present in the signal at their frequencies. But nonlinear distortions add new terms to it (i.e., waves at new frequencies).

Cone breakups themselves are linear distortions since the attenuation of input signals at the breakup frequencies can control them without affecting any other terms in the signals. But nonlinear distortions, which may occur at the breakup frequencies (excited by lower frequency signals; but the degree of excitation may vary), cannot be controlled this way. The only way is to attenuate the lower frequency signals that excite the breakups, but by doing so we can alter the signals at unwanted frequencies as well.

This fact suggests what should be a key element in a crossover design with a metal cone driver. It is a low crossover point given a filter type, not excessively suppressed cone breakups. It is unnecessary to attenuate the breakups far below the filter's target rolloff. A low crossover frequency with a reasonably steep filter (e.g., acoustic LR4) is the key.

-jAy

[Deward Hastings]

Re: Controling Cone Breakup In An Active System

The bottom line remains that the linear aspects of the driver SPL aberrations resulting from breakup can controlled by appropriate filters, nonlinearities can not. It then becomes a matter of assessing what is or is not linear.

I think that a good part of my dissatisfaction is with the implication of the prevously (and commonly) used term “correct”, which Mr. Webster tells me means “to make or set right”, or “to alter or adjust so as to bring to some standard or required condition”. With regard to cone breakup a notch filter does the first not at all and the second only in the loosest sense, since the “standard or condition” is “nominally linear”, which in fact is just what it is, “nominally” . . .

The name is not the Country.

First, response in the breakup region is not “smooth” like the response curve of a notch filter . . . even if the filter does produce “nominally flat” response there will still be a hodge podge of peaks and valleys in the response. When the cone is “breaking up” what that means is that pieces of it are moving in different directions (it’s not a simple resonance), and little if any of it is then under the complete control of the voice coil or any of the rest of the driver’s simple mechanical or electrical systems. While we can sometimes establish a pattern to the “breakup” with a steady sine wave (those strobe pictures) under drive from music cone motion in “breakup” is for all practical purpose chaotic . . . something no filter can “make or set right”.

And it’s not just that this messes up signal in the breakup range . . . when the cone is going every which way it is also modulating every other signal on it. None of the “transforms” that make the pretty computer simulations capture that. They’re busy doing all their calculations assuming that cone breakup is a simple linear phenomenon, like a resonance. You get an altogether different impression if you put a multi-tone test on a driver and then excite it in its breakup region . . . and it’s even worse if you excite the breakup with multiple uncorrelated sine waves, or “music”. What you then see around the tones outside the breakup region doesn’t look very “linear” to me. It doesn’t sound good, either. It looks like “spectral contamination”, and as soon as the cone starts “breaking up” (which is whenever there is any signal on it in the breakup region) you can hear it across pretty much the entire spectrum. I think it is something to be avoided . . . completely, absolutely and with extreme prejudice . . .

Paper cones of course do the same thing, often worse. They just don’t have that tattle-tale peak in their "linear" response to call attention to it, and give such clear warning what to avoid.

Other than that yes, we pretty much agree.

[Deward Hastings]

Re: Controling Cone Breakup In An Active System

A low crossover frequency with a reasonably steep filter (e.g., LR4) is the key.

That is generlly correct. Using a RS180 for example a LR4 filter at 1600 Hz. will put the top of the breakup peak around 40dB down, which is sufficient. What one commonly sees, however, is 2nd order filters called LR4 "acoustic", with the breakup peak barely 10dB below reference signal. That is clearly inadequate, and will produce clearly audible distortion in normal use. Such inferior crossovers should not be used with metal cone drivers.