Dampen metal cone resonance: How much is enough?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi

When working with metal cone woofer, some people say it is
necessary to filter the cone resonance with a paralel LC trap, or use a steep filter (18dB/oct or more) or both.

But, how much is enough?

If the peak of the main resonance is filtered and is 12 dB below the average level, is it OK? Is 15dB better?

Should I use the full arsenal to eliminate the resonance like a 4th order electrical filter plus a LC trap?
 
Hi gary f

I'm afraid what you heard is quite true. It is a pretty sharp peak at break-up. And not a small one, quite a few dB. The further away you are from the peak, the better. It can be used for a 2-way, but I find little benefit considering the hassle and the extra cost in crossover. Should you decide to go 2-way, you may well need the "full arsenal". I would rather use them for subs, especially bandpass.

Cheers
 
It depends on how far away it is from crossover and the crossover slopes. If it's a a point where the driver is 50 dB down anyway, I might not bother. If it's anywhere near the crossover point, I'd go at it with everthing I've got- including some experiments with viscoelastic damping.
 
Audiophysic is presently using metal-cone drivers which employ what it calls "active damping." The basic idea is to encircle the periphery of the metal cone with a tensioned band. The tensioning alters the resonant behaviour of the driver, and dramatically lowers the amplitude of the high-frequency break-up peak.

The active damping technique will likely allow the design of the crossover network to be simpler and more elegant than it would be otherwise.

hth, jonathan carr
 
diyAudio Moderator Emeritus
Joined 2001
JCarr:

You mean basically glue a rubber band around the back side of the edge of the cone, just below the point where the surround is glued to the cone?

By any chance, would you have any tension recommendations for various size drivers-say 4", 6", etc.?

What I would consider doing is: cut a rubber band, glue it back together at the length I wanted, then stretch it to where it is the proper length to go around the outside edge of the cone, and use a kitchen scale to judge the tension.

Just a thought.
 
diyAudio Moderator Emeritus
Joined 2001
Gary:

I haven't worked yet with metal cone woofers.

However, I can add that it is generally understood any output that is down 12 dB is considered "out of the running", sound-wise. That is why the crossover region of a 6 dB crossover is considered to be four octaves. It goes from -12 dB on the low frequency side to -12 dB on the high frequency side. Same thing for the crossover region for a 12 dB crossover-the region lasts two octaves-from -12 dB on the low frequency side to -12 dB on the high frequency side.

Having never worked with metal cones, I cannot say for sure. But that would be my assumption going in. Unless proven otherwise, minus 12 or 15 dB would make me feel safe.
 
For those with access to AudioXpress, the Sept issue has part one of a three part article detailing the design and construction of a 2-way using the Hi-Vi Research M8a metal cone woofer.

The author goes into great detail regarding the use of a high order Cauer/Elliptical XO's to deal with the issue that's being discussed here.

Jon Mark Hancock (the author) is a twice published member of the AES and seems to have a reasonal grip on loudspeaker design.
 
diyAudio Moderator Emeritus
Joined 2001
catapult said:
I'd want it at least 40dB down which usually means deploying the whole arsenal to kill it. If you loosely define the breakup peak as a form of distortion,

12dB = 25% distortion
20dB = 10%
40dB = 1%
60dB = 0.1%

Very true.

I have an idea on how Gary can test this, though. Run either a steady tone or "white noise" through his setup. He could use either a tone generator for a single tone, or use the "white noise" from tuning his radio to a frequency between stations. Using a voltmeter, he should adjust the "balance" control until one side gets 2 volts, and the other side gets 0.6 volts. Then he should hook the speaker which gets the 0.6 volts to the B speaker if his amp has one.

The speaker that gets 2 volts will now play 11 dB louder than the speaker that gets 0.6 volts. Starting off with A+B on the speaker selector switch, he will get both channels playing. Then he can switch the speaker selector switch to A, thereby cutting off the speaker that is 11 dB down. See if he can hear the difference. 11 dB is awful close to 12 dB.

If he can hear a difference, then he should consider going to a filter that suppresses more than 12 dB. I have a hunch he won't have to, but he should try it.

Just a suggestion. :)
 
However, I can add that it is generally understood any output that is down 12 dB is considered "out of the running", sound-wise.
Sorry, that's nonsense. 12dB down (25%) is clearly audible by almost anyone. No golden ears required. :nod: I think it was our local guru Planet10 that said the difference between a good speaker/crossover and a great speaker/crossover is how it behaves at 40dB down.
 
diyAudio Moderator Emeritus
Joined 2001
Catapult:

The issue is: If I can suppress a peak to the point where it is 12 dB under the midpoint, have I effectively eliminated it?

In answering, you correctly pointed out that a sound 12 db under the midpoint has 25% of the midpoint's power.

Okay, let us suppose that the peak in cone resonance is at 3000 Hz. The cone resonance is 12 dB down at 3000 Hz. If the cone had flat response, the speaker would be 18 db down at 3000 Hz, but the resonance brought it up to 12 dB down.

How would this translate into an increase in SPL?

Well, the formula for turning power into SPL is:
10 Log (V1/V2). Hence, when we a feed a speaker twice the power, it is 10 Log (2/1) = 3 dB. The speaker goes up 3 dB.

Here, we are increasing the power of the speaker to 1.25 of it's former power. So:
10 Log (1.25/1) = 0.967 dB.

The peak, at it's worst, causes the output to rise less than a single dB.

And this is narrow band-less hearable.

No wonder the crossover region is defined as between -12 dB points. The -12 dB point is the place where the suppressed driver's putput, if cut out completely, would result in a change of total output of less than a single dB.

We don't argue much if a speaker goes up or down 1 dB throughout it's range. In fact, we call that smooth response. I think a peak suppressed to the point where it's top is 12 dB under the midpoint is essentially fully suppressed, if the only effect is that it raises the total output less than a single decibel.

Am I wrong on this?

If the peak were suppressed to the point where it is 40 dB down from the midpoint, the rise in SPL would be:

10 Log (1.01/1) = .043 dB.

Is the difference between 0.97 db and 0.043 dB really so hearable that extra filters, complete with their sacrifices in phase and temporal response, must be used to knock that SPL increase down to 0.43 db from 0.97 dB? You're lucky if you can find a driver that smooth with no filters at all.

PS: If I were building this, I would indeed get that issue of AudioExpress to see how a professional worked out his metal cone crossover.
 
Just another Moderator
Joined 2003
Paid Member
OK guys I'm confused. I at some stage have got it in my head that -3db sounds like 1/2 the volume, but I have just done a bit of a check and -3db is actually half the power (roughly) right???

Now my understanding is that you need 10 times the power to get twice the volume (ie a 100W amp is only twice as loud as a 10W amp)

So firstly how is 12db down 25% of the power? I would have though it would be 1/16th the power. And secondly I think I have also seen somewhere that -6db is half the volume, but going on the above 10X rule then that wouldn't hold true either.

Back to the books for me I think!

Regards,

Tony.
 
diyAudio Moderator Emeritus
Joined 2001
Wintermute:

My understanding of the way it works is:

A) Decibel: The smallest volume difference that people can detect consistently. Tests run under laboratory conditions. Tests like: a single tone is raised up and down in volume, and the subject presses a button when he/she can detect a difference. How much this means with real music playing different frequencies at different volume levels is open to question. But a decibel is the smallest volume change that people can consistently detect.

B) 3 dB: The difference where people universally agree that one sound is somewhat louder than another. This is perceptible during music listening as well. Requires 2 times the power to achieve.

C) 10 dB: The difference where people perceive one sound as being approximately twice as loud as another. Requires 10 times the power to achieve.

Based on my reading, that seems to be the general consensus.
 
KW: I'll look around for the article tonight, but I'm pretty certain that the "10 dB sounds twice as loud" chestnut was recently debunked.

Also, 0.1-0.2 dB level changes can be reliably heard by listeners with good ears on music material (I've done that experiment). Unfortunately, level changes that small aren't perceived as level changes, but as changes in "quality." That's why you have to do VERY careful level-matching when setting up A-B tests. It's an old hifi salesman's trick- present the mark with what appears to be two amps playing at the same volume, but in reality, the amp being pushed is set a teensy bit louder. The mark will be amazed at how much more open and clear the louder amp sounds...
 
diyAudio Moderator Emeritus
Joined 2001
Note: On post # 12, I gave the formula for SPL rise with power rise as:
10 Log (V1/V2)

That should read 10 Log (P1/P2).

Not being an engineer, I thought I would put "V" as standing for "volume". This was a mistake. "V" universally means "voltage". The equation for voltage and SPL rise is very different.

Sorry for any confusion on that. :bigeyes:
 
diyAudio Moderator Emeritus
Joined 2001
SY said:
KW: I'll look around for the article tonight, but I'm pretty certain that the "10 dB sounds twice as loud" chestnut was recently debunked.

Also, 0.1-0.2 dB level changes can be reliably heard by listeners with good ears on music material (I've done that experiment). Unfortunately, level changes that small aren't perceived as level changes, but as changes in "quality." ...

Those are interesting things to know. It looks like a lot of articles are going to need asterisks in reprints! :att'n:

On this particular application, though, I would like to reiterate a few points. Gary has the peak of a metal cone speaker-in the "stop band"-suppressed to the point where it is 12 dB down from the midpoint. He wants to know if that is enough.

If my math is correct, (I wouldn't mind confirmation on that, because I am not certain), then that peak, at it's height, will cause less than 1 dB rise in the midpoint SPL level.

A real world speaker is considered smooth if the response fits into a "window" of ± 2 dB throughout it's midrange. The up-and-down variations of his speaker away from the suppressed resonance area will likely be greater than this one one area where there is a 1 dB peak.

Of course, this is all math and theory. I freely yield to anyone who has actually dealt with metal cone resonances in their own system and has found that further peak suppression is hearable and necessary. :)
 
Just another Moderator
Joined 2003
Paid Member
kelticwizard said:
Wintermute:

My understanding of the way it works is:

B) 3 dB: The difference where people universally agree that one sound is somewhat louder than another. This is perceptible during music listening as well. Requires 2 times the power to achieve.

C) 10 dB: The difference where people perceive one sound as being approximately twice as loud as another. Requires 10 times the power to achieve.

Based on my reading, that seems to be the general consensus.


Thanks Kelticwizard,

I think I remembered the two things but somehow got them muddled up, and have been thinking for ages that -3db was roughly half the volume level. Glad that's cleared up!!! could have given me all sorts of confusion when I start testing things!!!!! (That is if I ever stop reading this forum and order some speakers and make some cabinets!!! :) )

Now that you have said that, it somewhat rings a bell.

SY it will be interesting to here the views on the +10db not being twice as loud.

I have long been under the imression that the 10 times power figure was because of the logarithmic nature of human hearing. (though It was just something I read, and math is unfortunately not one of my strong points, so I haven't ever questioned it)

Regards,

Tony.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.