Fine.
I chose that value on purpose to show *one* case in which the VC would burn with no mechanical stress involved.
And being a pure sinewave, would make crossover slope irrelevant.
Now, if you use a 100Hz sinewave, which *is* what we are discussing here, you won't kill the tweeter with its harmonics.
It was already shown in earlier posts, with accompanying Math.
I chose that value on purpose to show *one* case in which the VC would burn with no mechanical stress involved.
And being a pure sinewave, would make crossover slope irrelevant.
Now, if you use a 100Hz sinewave, which *is* what we are discussing here, you won't kill the tweeter with its harmonics.
It was already shown in earlier posts, with accompanying Math.
No crossover has the power to infinitely attenuate. Even a 4 pole filter will let something of the original sine wave through.
Then there are two possibilities: Either there is something wrong with your maths or with your interpretation of the results because I can do what I described in reality without any problems at all.
You can also over-stress tweeters with a bass drum whose fundamental is usually around 80-120Hz while the loudest part, the transients, are more in the region of 8kHz.
As for the filter issue Kesh mentioned if we stick to a 100Hz sine wave and a crossover point of 2000Hz the original signal would be attenuated by over 96dB using 24dB filters. Which is close enough to nothing for me.
There is a third possibility: that your understanding of the mechanism involved is incorrect. Your rejection of the well-understood laws of harmonic progression lends weight to this possibilty. If you're burning out tweeters by feeding them a 100 Hz square wave you need to look for another reason - for example, an incorrectly implemented crossover. For example, a "crossover" with a single series capacitor might let through enough of the lower harmonics to overheat or over-excurse the tweeter. This would tend to be the case in lower-spec speakers with poor-quality tweeters.
If we have 4 pole filters then we've pretty much made our tweeters safe from this issue, so we're missing the point, which is how tweeters fail.
On the other hand if our square wave has a rise time of 0.000015 sec from zero to 40V we actually have a 20kHz 40V sine right there.
(rise time of a sine being roughly 0.3 x period)
Which could very well lead to tweeter failure, be it thermal or mechanical.
Shorter rise times lead to higher frequency components but 20k seems appropriate in these days of digital audio as it is about the fastest rise time digital with a 44.1k sample rate can produce since it is low pass filtered at around 20k.
(rise time of a sine being roughly 0.3 x period)
Which could very well lead to tweeter failure, be it thermal or mechanical.
Shorter rise times lead to higher frequency components but 20k seems appropriate in these days of digital audio as it is about the fastest rise time digital with a 44.1k sample rate can produce since it is low pass filtered at around 20k.
You didn't read earlier posts, did you?
In post #63 I explain:
Repeating this calculation with your 20KHz front edge , created by a rise time of 0.000015 sec (your numbers) we now have a duty cycle of=
0.000015 sec/0.01Sec (meaning 100Hz frequency)=0.0015=0.15% (PLEASE check the Math) so the average power dissipated by that voice coil will be= 200Wx0.0015=0.3W .
Another misconception you write:
No, the amplitude of a harmonic is:
so, the "20KHz sinewave", being the 20000/100 th harmonic=200th harmonic, and its *voltage* will be 40V/200=200mV which amounts to a power of: 0.2V x 0.2V / 8 ohms= 5 milli Watts.
By the way, djk says about the same, only using dB instead of % , in post #59 .
Now that I reread older posts, I see that all this was already explained to you.
What's the point of wasting our time repeating the same 1000 times just because you cover your ears and shout "I can't hear you .... I can't hear you .... "
Why Do Tweeters Blow When Amplifiers Distort?
Scroll down to 'moderate clipping' and you'll see what I mean.
The amps full range output is 24.7V rms in this case, the woofer receives 20.5V rms and the tweeter 8.9V rms.
But the voltage peaks the tweeter receives exceed 35V on this scale while the woofer only sees peaks of 30V.
The spikes in the tweeter voltage are the leading edge of each squared off wave.
Even that is visible in that measurement.
Scroll down to 'moderate clipping' and you'll see what I mean.
The amps full range output is 24.7V rms in this case, the woofer receives 20.5V rms and the tweeter 8.9V rms.
But the voltage peaks the tweeter receives exceed 35V on this scale while the woofer only sees peaks of 30V.
The spikes in the tweeter voltage are the leading edge of each squared off wave.
Even that is visible in that measurement.
You don't read what you cut and paste, do you?
We have been discussing from the beginning *your* statement that harmonics of a 100Hz squarewave , even with the fastest rise time possible, and with a 2KHz crossover can burn a tweeter coil ..... and demonstrated by many different methods that it does NOT happen.
We calculated the power of all harmonics and in no case they added up to a dangerous level .... by a LONG way.
Now you refer to a test which you clearly do not understand and *think* it proves your point ... which it does not.
Instead of the original clipped 100Hz waveform we are discussing, you refer to a Rod Elliott's study which is absolutely unrelated.
He does NOT analyze a clipped 100Hz frequency but a MIX of frequencies, arbitrarily chosen, from 160Hz to 13 KHz
And the intensities are also arbitrarily chosen at very high and unrealistic values.
Example: in our example of 100Hz base frequency, the 2KHz component amplitude would be 1/20th the amplitude of the fundamental , so for 40V peak we would have 2V peak=-26dB or 0.5W *peak*
BUT Rodd Elliott arbitrarily sets the 2KHz signal 3 dB below the lowest frequency one.
He assumes a 20 dB higher level, 100 times the power !!!
And that talking of *undistorted*, below clipping signals!!!
Of course, *all* high frequency signals will rise even more after clipping.
So what? It's not a 100Hz signal, the original one discussed (and accepted by you) but a completely different (and artificial) thing.
Talk about cheating !!!
No doubt you now want to include this absolutely unrelated signal!!!
To show what you are NOW referring to (and which is not what we are analyzing here):
So, next time you bring a witness in your behalf, at lest choose one who speaks about what we are discussing here.
Thanks.
We have been discussing from the beginning *your* statement that harmonics of a 100Hz squarewave , even with the fastest rise time possible, and with a 2KHz crossover can burn a tweeter coil ..... and demonstrated by many different methods that it does NOT happen.
We calculated the power of all harmonics and in no case they added up to a dangerous level .... by a LONG way.
Now you refer to a test which you clearly do not understand and *think* it proves your point ... which it does not.
Instead of the original clipped 100Hz waveform we are discussing, you refer to a Rod Elliott's study which is absolutely unrelated.
He does NOT analyze a clipped 100Hz frequency but a MIX of frequencies, arbitrarily chosen, from 160Hz to 13 KHz
And the intensities are also arbitrarily chosen at very high and unrealistic values.
Example: in our example of 100Hz base frequency, the 2KHz component amplitude would be 1/20th the amplitude of the fundamental , so for 40V peak we would have 2V peak=-26dB or 0.5W *peak*
BUT Rodd Elliott arbitrarily sets the 2KHz signal 3 dB below the lowest frequency one.
He assumes a 20 dB higher level, 100 times the power !!!
And that talking of *undistorted*, below clipping signals!!!
Of course, *all* high frequency signals will rise even more after clipping.
So what? It's not a 100Hz signal, the original one discussed (and accepted by you) but a completely different (and artificial) thing.
Talk about cheating !!!
No doubt you now want to include this absolutely unrelated signal!!!
To show what you are NOW referring to (and which is not what we are analyzing here):
So, next time you bring a witness in your behalf, at lest choose one who speaks about what we are discussing here.
Thanks.
JMF,
so how do you explain the 35V peaks the tweeter sees?
The original wave form frequency or shape doesn't really matter that much, all I'm saying is that the leading edge of any square wave is of high frequency at full output voltage regardless of the square waves fundamental and Rodd Elliotts example shows that quite nicely unless you are wilfully blind to it.
The fundamental frequency could be 10Hz, 100Hz, 1000Hz or anything else, the only thing that really matters for the extreme peaks is the rise time. The faster the rise time the higher the frequency needed to achieve it.
And this may well lead to failure of the tweeter. Either through overheating or, as djk pointed out, mechanical failure.
so how do you explain the 35V peaks the tweeter sees?
The original wave form frequency or shape doesn't really matter that much, all I'm saying is that the leading edge of any square wave is of high frequency at full output voltage regardless of the square waves fundamental and Rodd Elliotts example shows that quite nicely unless you are wilfully blind to it.
The fundamental frequency could be 10Hz, 100Hz, 1000Hz or anything else, the only thing that really matters for the extreme peaks is the rise time. The faster the rise time the higher the frequency needed to achieve it.
And this may well lead to failure of the tweeter. Either through overheating or, as djk pointed out, mechanical failure.
And who says the tweeter sees 35V peaks?
A simulation with poorly chosen and contradictory values?
So poorly chosen, that it "shows" the HF content has *higher voltage* than the LF signal, which is impossible in any clipped sinewave.
You must understand something about simulation: you can start with physically impossible values, and of course get physically impossible results.
What's news about that?
There's even an acronym for that: "GIGO" , meaning "Garbage In, Garbage Out".
Specially you are ignoring a certain "Fourier" guy, who seems to know everything about this.
Joseph Fourier - Wikipedia, the free encyclopedia
Fourier transform - Wikipedia, the free encyclopedia
Fourier Series -- from Wolfram MathWorld
In particular, he is ***THE*** Authority to study stuff such as this:
Fourier says otherwise, and the Math explaining it was posted many times above.
Don't want to overload the post with Fourier's formulas time and again, read them in full in post#50.
Mmmmhhhhh , big doubt.
Who do I believe? Fourier or "Charles Darwin"?
From Rodd Elliott's own mouth:
And mechanical failure was discarded by yourself in post#77
A simulation with poorly chosen and contradictory values?
So poorly chosen, that it "shows" the HF content has *higher voltage* than the LF signal, which is impossible in any clipped sinewave.
You must understand something about simulation: you can start with physically impossible values, and of course get physically impossible results.
What's news about that?
There's even an acronym for that: "GIGO" , meaning "Garbage In, Garbage Out".
Really? You are inventing new Physics.
Specially you are ignoring a certain "Fourier" guy, who seems to know everything about this.
Joseph Fourier - Wikipedia, the free encyclopedia
Fourier transform - Wikipedia, the free encyclopedia
Fourier Series -- from Wolfram MathWorld
In particular, he is ***THE*** Authority to study stuff such as this:
What you are saying over and over is that it burns tweeters, and it has been proven it's not so.
Fourier says otherwise, and the Math explaining it was posted many times above.
Don't want to overload the post with Fourier's formulas time and again, read them in full in post#50.
Mmmmhhhhh , big doubt.
Who do I believe? Fourier or "Charles Darwin"?
Does it?
From Rodd Elliott's own mouth:
Yeah, really?
Well, overheating was proven not to be enough.
And mechanical failure was discarded by yourself in post#77
Last edited:
Dear Andrew T , "Charles Darwin" himself chose it in post #44 , where he made a wrong statement
which is haunting him ever since, because *many* Forum Members proved him wrong, from different analysis tools .... which agreed, of course.
That's the beauty of Science: that it's consistent.
I don't understand why he *still* insists on it.
Maybe it's to "save face" or something, but this is not the way.
Read the post #44 in full, where he bases his assumptions on his being able to burn tweeters using his synthesizer, a very uncontrolled "experiment" (if you can call it.)
Then he confesses on not using a pure sinewave but a "kick drum simulation" which of course can have all kinds of harmonics, but is not what we are discussing here.
In a nutshell: he is using poorly understood anecdotal "evidence" to challenge well known Physics answers.
Just to make sure we read his error correctly, and it was not a typo, he repeats it again in post #49:
Incredible !!!
Even worse: not to be outdone, he doubles the bet in post #80
Double incredible !!!!
By the way, you spoke of a 1000Hz sinewave, driving the amp to clipping, and with a 2KHz crossover.
In that example yes, of course, starting at 10X "Charles Darwin frequency", it's reasonable to deduce that the tweeter will receive 10X the former power, and this will be very bad to its health.
But all this proves is that AndrewT is right and "Charles Darwin" is wrong.
Personally I have been corrected many times, and, if approppriate, accepted it graciously.
I'm here to learn and exchange experiences, like most others.
Thanks.
JMF,
so because it supports my point Rodd Elliotts work is faulty even though the behaviour shown is pretty much textbook?
Interesting…
Either way you appear to be under the impression that the harmonics occur spread out over time as neatly as they are spread out over frequency ie they happen consecutively.
This is of course wrong, they all start simultaneously.
The effect of this is that the leading edge of a square wave takes on the frequency of the its highest harmonic while the amplitude is closer to F1+F3+F5+F7+F9... and so on until the system runs out of bandwidth and minus a certain amount since the sines of the harmonics do not sit exactly on top of each other. The reproduction of a perfect square wave is of course impossible since it would require infinite bandwidth.
Which is also why it does not matter if the square wave is 10Hz, 100Hz or !000Hz as the steepness and therefore the frequency of the leading edge is limited by the clipping amps bandwidth which might be 50 or 100kHz.
Here is a little gif which should make this very clear, also notice that the leading and trailing edges at all times have a higher voltage than the original.
http://upload.wikimedia.org/wikipedia/commons/f/f8/SquareWave.gif
PS: I have not dismissed breakage due to mechanical failure as you make out.
The point was that the exact failure mode matters less than the fact that it failed.
so because it supports my point Rodd Elliotts work is faulty even though the behaviour shown is pretty much textbook?
Interesting…
Either way you appear to be under the impression that the harmonics occur spread out over time as neatly as they are spread out over frequency ie they happen consecutively.
This is of course wrong, they all start simultaneously.
The effect of this is that the leading edge of a square wave takes on the frequency of the its highest harmonic while the amplitude is closer to F1+F3+F5+F7+F9... and so on until the system runs out of bandwidth and minus a certain amount since the sines of the harmonics do not sit exactly on top of each other. The reproduction of a perfect square wave is of course impossible since it would require infinite bandwidth.
Which is also why it does not matter if the square wave is 10Hz, 100Hz or !000Hz as the steepness and therefore the frequency of the leading edge is limited by the clipping amps bandwidth which might be 50 or 100kHz.
Here is a little gif which should make this very clear, also notice that the leading and trailing edges at all times have a higher voltage than the original.
http://upload.wikimedia.org/wikipedia/commons/f/f8/SquareWave.gif
PS: I have not dismissed breakage due to mechanical failure as you make out.
The point was that the exact failure mode matters less than the fact that it failed.
"The point was that the exact failure mode matters less than the fact that it failed. "
Yeah, right.
Very few tweeter actually fail due to the coil burning up, the vast majority are from mechanical failure.
Yet we continue to spread the myth that this is somehow connected to amplifier clipping when it is the increase in stress during the non-clipped portion of the program material that is the culprit.
Get over it.
Yeah, right.
Very few tweeter actually fail due to the coil burning up, the vast majority are from mechanical failure.
Yet we continue to spread the myth that this is somehow connected to amplifier clipping when it is the increase in stress during the non-clipped portion of the program material that is the culprit.
Get over it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.