Yes, on sinewaves.
But what djk refers to is something else.
I've also read the JBL led study (have it stashed somewhere
) about listeningbto music at loud volumes.
Basically what they say is that we listen to music at an average level way below that of what would be sinewave clipping, when only some music peaks may clip sometimes, because that does not sound "ugly".
When we don't care much about that (typically a DJ environment) we start raising volume, louder portions of music start to clipo more, but we tend to "forgive" that (we are having fun, aren't we?).
That causes the *average* level of music to rise spectacularly (I couldn't believe it when I saw the graphs); bass (which is what usually clips first) does not rise much (it's clipping-limited) but mids and highs continue to raise *a lot*.
Even if not clipped themselves, the much higher levels burn tweeters.
Bass *is* clipping the way you describe, but it's too many octaves (even their harmonics) away from what would channeled into the tweeters.
Yes, in a "let's blast the neighbouthood with my boombox" distortion you'll reach squarewaves at every frequency (ugh!!) but what worried JBL was their Hi Fi stuff being blown at home parties and such, where the system sounds loud but still is perceived as "basically clean"
If I find the study I'll post it here.
If somebody finds it first, he's welcome to do so.
But what djk refers to is something else.
I've also read the JBL led study (have it stashed somewhere
) about listeningbto music at loud volumes.Basically what they say is that we listen to music at an average level way below that of what would be sinewave clipping, when only some music peaks may clip sometimes, because that does not sound "ugly".
When we don't care much about that (typically a DJ environment) we start raising volume, louder portions of music start to clipo more, but we tend to "forgive" that (we are having fun, aren't we?).
That causes the *average* level of music to rise spectacularly (I couldn't believe it when I saw the graphs); bass (which is what usually clips first) does not rise much (it's clipping-limited) but mids and highs continue to raise *a lot*.
Even if not clipped themselves, the much higher levels burn tweeters.
Bass *is* clipping the way you describe, but it's too many octaves (even their harmonics) away from what would channeled into the tweeters.
Yes, in a "let's blast the neighbouthood with my boombox" distortion you'll reach squarewaves at every frequency (ugh!!) but what worried JBL was their Hi Fi stuff being blown at home parties and such, where the system sounds loud but still is perceived as "basically clean"
If I find the study I'll post it here.
If somebody finds it first, he's welcome to do so.
"What do you think clipping is, it's the increase in average power when the top of the sine wave is cut off almost approximating a square wave. "
Read my post again, you totally missed the point.
The difference between a sine and a square wave is only a 3dB increase, the increase in average power reaching the tweeter of program material driven 10dB into clipping on the brief music peaks is trivial compared to the 10dB increase in long-term average-power during the majority of the time the amplifier is not clipping.
Read my post again, you totally missed the point.
The difference between a sine and a square wave is only a 3dB increase, the increase in average power reaching the tweeter of program material driven 10dB into clipping on the brief music peaks is trivial compared to the 10dB increase in long-term average-power during the majority of the time the amplifier is not clipping.
Waveforms are funny things.
It is quite easy to fry a tweeter with a 100Hz sine as long as the amplifier is clipping.
I assume a passive multiway speaker is being used.
A (perfect) sine is a pure tone with no harmonics at all but the moment the amp is clipping it produces a square wave which contains all the sines which fit inside it.
So if I drive a 200W amp into clipping it will produce, for example, 10 000Hz at the full 200W. I don't know of (m)any tweeters which will cope with that for very long.
This can easily be shown to be true with an analogue modular subtractive synth like the one I've got. Just take the output of a low frequency sine generator and feed it into a filter. The result is extremely boring as the tone either passes unmolested or is filtered out resulting in silence. But if you take the same sine, run it through an amplifier module and drive that into clipping before going into the filter things are very much different.
Suddenly there are plenty of harmonics for the filter to get its teeth into and the result is very, very similar to feeding the filter a square wave signal instead of a clipped sine.
Btw usually amplifier modules have an adjustable output level so that the signal which reaches the filter is of the same level as before the amp.
Bottom line a clipping amp produces vast amounts of harmonics at full power which are not present in the unclipped signal. No crossover will be able to save the tweeter.
It is quite easy to fry a tweeter with a 100Hz sine as long as the amplifier is clipping.
I assume a passive multiway speaker is being used.
A (perfect) sine is a pure tone with no harmonics at all but the moment the amp is clipping it produces a square wave which contains all the sines which fit inside it.
So if I drive a 200W amp into clipping it will produce, for example, 10 000Hz at the full 200W. I don't know of (m)any tweeters which will cope with that for very long.
This can easily be shown to be true with an analogue modular subtractive synth like the one I've got. Just take the output of a low frequency sine generator and feed it into a filter. The result is extremely boring as the tone either passes unmolested or is filtered out resulting in silence. But if you take the same sine, run it through an amplifier module and drive that into clipping before going into the filter things are very much different.
Suddenly there are plenty of harmonics for the filter to get its teeth into and the result is very, very similar to feeding the filter a square wave signal instead of a clipped sine.
Btw usually amplifier modules have an adjustable output level so that the signal which reaches the filter is of the same level as before the amp.
Bottom line a clipping amp produces vast amounts of harmonics at full power which are not present in the unclipped signal. No crossover will be able to save the tweeter.
Would you please make up your mind and choose just one?
100Hz and 10000Hz not exactly the same.
To clear things: I very much doubt a 100Hz squarewave can burn a regular tweeter, even less a Driver+Horn. Talking about passive crossovers, of course.
And if you are talking 10000 Hz, well, *any* waveform at that frequency and 200W will burn any tweeter I can think of.
We are talking the kind which uses a coil and a magnet.
Piezos are something very different.
PS: and if you are saying that "the 10000 Hz harmonic of a squarewave clipped 100Hz frequency, when produced by a 200W amp, will burn a tweeter connected to its output through a passive crossover", the answer is: I very much doubt so.
Simple math: 100 Hz, clipping or square wave. Assume the tweeter crossover is about 1.6 KHz. Only harmonics 17th and above get to the tweeter. Assuming 200W of power from the amplifier, that's about 4.5 watts to the tweeter. So although Charles' math is incorrect, some(*) tweeters may overheat after a few minutes at that power level.
(Also assume perfect square wave and crossover, and zero inductance tweeter.)
(*) Tweeters intended for systems rated at about 50 watts or below, and assuming the tweeter sensitivity matches the midrange driver and doesn't require padding.)
(Also assume perfect square wave and crossover, and zero inductance tweeter.)
(*) Tweeters intended for systems rated at about 50 watts or below, and assuming the tweeter sensitivity matches the midrange driver and doesn't require padding.)
Last edited:
Would you please make up your mind and choose just one?
100Hz and 10000Hz not exactly the same.
To clear things: I very much doubt a 100Hz squarewave can burn a regular tweeter, even less a Driver+Horn. Talking about passive crossovers, of course.
And if you are talking 10000 Hz, well, *any* waveform at that frequency and 200W will burn any tweeter I can think of.
We are talking the kind which uses a coil and a magnet.
Piezos are something very different.
PS: and if you are saying that "the 10000 Hz harmonic of a squarewave clipped 100Hz frequency, when produced by a 200W amp, will burn a tweeter connected to its output through a passive crossover", the answer is: I very much doubt so.
The 10 000Hz was merely an example as it certainly is a multiple of 100 and a nice round number.
Let's try again:
We feed an amp a sine of 100Hz and drive it into hard clipping, the result is the amp puts out a 100hz square wave. Now lets just say the amp produces an output of 30V at that point. This square wave contains all possible multiples/harmonics of 100 at the full 30V.
If the tweeter xover is at 2000Hz the tweeter will now try to reproduce 2000Hz, 2100Hz, 2200Hz, 2300Hz etc all the way up to 20K and beyond at 30V.
Obviously in real life the first few frequencies above 2k will be attenuated due to the crossover but the tweeter does at least see all those above 4kHz at the full voltage minus any crossover insertion losses.
Ok, then let's disregard any mention of 10000 Hz since it's misleading.
What you say is that we have a 100Hz wave, we clip it into a squarewave, will have a ton of harmonica ans the higher ones will reach the tweeter.
Fine. So far so good.
I don't disagree in that, but on the actual threat to its integrity.
Another part you wrote in a confusing way is that:
To begin with the worse offender:
Written that way, without qualifications, it can be read as a statement that each and every harmonic will be 30V.
Fact is, if the 100Hz signal is 30V , harmonics will always be a lot lower than that, depending on "distance" from 100 Hz.
There's also another misstatement mentioning:
In fact we will have only odd harmonics, meaning 2100, 2300, etc , but not 2000, 2200, etc.
But this is only a minor statement compared to the first one.
Mind you, I agree that high harmonics will reach the tweeter, but not on their perceived power.
Let's calculate at least the most powerful ones.
One easy to grasp formula (so as to make it more intuitive) is:
It's clear that if we are talking about an Nth harmonic, its voltage will be 1/N that of the fundamental.
And remember that Power is proportional to V squared, so 1/3 the voltage is 1/9 power, and so on.
To ease calculations, and ajusting values in your favor, let's consider 40V instead of the 30V suggested.
Why?
Because it represents 200W into 8 ohms, a common speaker impedance.
Also let's consider a perfect crossover filter, which applies no attenuation above the crossover point.
All in your favor, as I said, because I consider more power and less attenuation.
Now let's see what the poor tweeter will have to stand:
the first harmonic of 100 Hz reaching that tweeter will be 2100Hz, the 21th harmonic.
Its voltage will be 40V/21=1.9V , so its power will be=(1.9^2)/8=0.45W
the next will be 2300Hz , the 23rd, so V=40V/23=1.73V and P=0.38W and so on.
We see that we start with a very weak harmonic and each higher one is rapidly losing power, so even adding all them up will result in power that will hardly damage a Teeter of the type that would be used in a PA type cabinet.
Which is what we are talking about.
In the Hi Fi world , they might use a very weak silk soft dome tweeter with hair thin VC wire, which of course will be easier to damage, but there clipping is a definite no no and although some brief peaks will inevitably clip if used "somewhat loud" their duty cycle will be low or music becomes unbearable.
Last edited:
Don't remember exactly but I think the transistor equation was like
output= Gain* input + b/input +c/input*input + d/input*input*input ......goes on....
B,C,D are harmonics and they reduce in amplitude as you go higher in harmonics
Worst case, it will have to withstand a bit under 3.75 watts. Even a poor tweeter should be able to manage that.
(100 Hz square wave, 40 volts, 8 ohms (200 watts), crossover 2 KHz)
Last edited:
I think the idea of sounding twice as loud implies a subjective impression of loudness that will probably prove impossible to generalise and quantify.
Though not logarithmic, it reminds me the idea of a day's weather feeling twice as hot; impossible to know what this means subjectively, just that it bears no relationship to physics.
Though not logarithmic, it reminds me the idea of a day's weather feeling twice as hot; impossible to know what this means subjectively, just that it bears no relationship to physics.
I agree, Its too tough to tell what is half when it comes to sound.
used a tone generator
I thought -4.8db is half as loud at 200hz
my friend thought -5.0db at 200hz and then after hearing 1khz for 30 secs he said -6.6 db is half as loud at 200hz
10db is a biggg difference. Just as hot weather example loudness is too subjective.
Ok JMFahey but lets just look at the square waves rise time for a moment:
If the rise time is 50 microsec to go from zero V to 40V we must have a 20kHz component that reaches 40V. If it would be lower than 40V we would get a longer rise time because only a 20kHz sine is fast enough and all waveforms in the end are made up of sines.
I may have miscalculated the actual numbers but the idea is the same: Look at the actual rise time and find a sine of frequency x that satisfies that. The faster the rise time the higher the frequency needed to achieve it.
If the rise time is 50 microsec to go from zero V to 40V we must have a 20kHz component that reaches 40V. If it would be lower than 40V we would get a longer rise time because only a 20kHz sine is fast enough and all waveforms in the end are made up of sines.
I may have miscalculated the actual numbers but the idea is the same: Look at the actual rise time and find a sine of frequency x that satisfies that. The faster the rise time the higher the frequency needed to achieve it.
Just in case it isn't clear you need to be careful (or specific) when using "dB" and "perceived volume".
Electrical dB values (eg watts) do not correlate directly to acoustic dB values (eg SPL).
It's a generalization, but useful none the less:
To gain a perceived doubling of loudness (+3dB acoustically) you need 10x the power electrically.
Now, you can take into consideration a bunch of factors and come up with somewhat different results. If you were designing a system using electronic crossovers, for example, that might be useful and even recommended. But in the grand scheme of things, with a single amplifier and individual full-range speakers, you're better off using the quick guesstimate and using your time dealing with other things that might be part of your design considerations. It won't be far off results-wise and you won't be wasting money on unused power or risk not spending enough under-powering.
Electrical dB values (eg watts) do not correlate directly to acoustic dB values (eg SPL).
It's a generalization, but useful none the less:
To gain a perceived doubling of loudness (+3dB acoustically) you need 10x the power electrically.
Now, you can take into consideration a bunch of factors and come up with somewhat different results. If you were designing a system using electronic crossovers, for example, that might be useful and even recommended. But in the grand scheme of things, with a single amplifier and individual full-range speakers, you're better off using the quick guesstimate and using your time dealing with other things that might be part of your design considerations. It won't be far off results-wise and you won't be wasting money on unused power or risk not spending enough under-powering.
Last edited:
"all waveforms in the end are made up of sines."
A square is all the odd harmonics out to infinity. In practice it looks square after the 29th harmonic, and each higher higher harmonic has a lower value.
Assume a 100W square wave at 100hz. The 3rd harmonic is -9.54dB, the 5th is -13.98, the 7th is -16.9dB, the 9th is -19.1dB.
Assume a 1Khz crossover. Now the 11th will be -20.8dB (0.83W), the 13th is -22.3dB (0.589W), etc. these go to the tweeter. As you can see, the total amount of power left to make it through an ideal 1kHz crossover (and on to the tweeter) is less than two watts (0.83 + 0.589 = 1.419W). Hardly a problem.
Yet the tweeters still blow up.
Driving the amplifier harder cannot increase the power, so why did the tweeter burn out?
Could it possibly have been to the increased power during the un-clipped portion of the program material?
The brief duration of clipping peaks on program material does not cause the failure of the tweeter, it is the increase in long-term average-power during the un-clipped portion of the program material the causes the failure.
In reality, most of the failures in the field were of a mechanical nature, the single-strand lead-out wires broke.
PS
Calculations beyond the 13th harmonic are left for the reader to add up. Hint: the sum is still trivial.
A square is all the odd harmonics out to infinity. In practice it looks square after the 29th harmonic, and each higher higher harmonic has a lower value.
Assume a 100W square wave at 100hz. The 3rd harmonic is -9.54dB, the 5th is -13.98, the 7th is -16.9dB, the 9th is -19.1dB.
Assume a 1Khz crossover. Now the 11th will be -20.8dB (0.83W), the 13th is -22.3dB (0.589W), etc. these go to the tweeter. As you can see, the total amount of power left to make it through an ideal 1kHz crossover (and on to the tweeter) is less than two watts (0.83 + 0.589 = 1.419W). Hardly a problem.
Yet the tweeters still blow up.
Driving the amplifier harder cannot increase the power, so why did the tweeter burn out?
Could it possibly have been to the increased power during the un-clipped portion of the program material?
The brief duration of clipping peaks on program material does not cause the failure of the tweeter, it is the increase in long-term average-power during the un-clipped portion of the program material the causes the failure.
In reality, most of the failures in the field were of a mechanical nature, the single-strand lead-out wires broke.
PS
Calculations beyond the 13th harmonic are left for the reader to add up. Hint: the sum is still trivial.
Last edited:
Those high freq corners of the square wave might be really high freqs at full output voltage, but their duty cycle is very low. You might look at that 20kHz spike, but it occurs only at the worners of thge 100Hz signal in question. SO it isn;t the same thing as feeding a full power 20kHz signal into the tweeter.
There used to be a Rane Note on clipping and speaker failure, I'll post it if I can find it. It made the case for compression blowing the speakers. The typical speaker faces music with huge low freq spikes and the treble content considerably lower in amplitude. As you turn up the material into clipping, those lows clip, but the smaller treble range continues to grow - relative to the lows. Yes, the clipped lows add some high freq stuff, but the whole high freq portion of the music is free to get louder and louder. If you had a 200 watt woofer and a 20 watt tweeter in an enclosure, and a 200 watt amp0lifier, at clipping you might have those power levels into each driver. Now raise the power to the cab to 400 watts. Or what would have been 400 watts. The woofer drive is clipped at the 200 watts the amp can create. The tweeter would now see 40 watts, well within the amp's ability, but a stress for the tweeter. Now push the signal to the 800 watt level. The woofer drive still clips at the amp's 200 watts, but now we have 80 watts into the tweeter, and the amp not even sweating. The idea is that it is the 80 watts of compressed material that blew that 20 watt tweeter, not the high freq components of the clipped low end.
There used to be a Rane Note on clipping and speaker failure, I'll post it if I can find it. It made the case for compression blowing the speakers. The typical speaker faces music with huge low freq spikes and the treble content considerably lower in amplitude. As you turn up the material into clipping, those lows clip, but the smaller treble range continues to grow - relative to the lows. Yes, the clipped lows add some high freq stuff, but the whole high freq portion of the music is free to get louder and louder. If you had a 200 watt woofer and a 20 watt tweeter in an enclosure, and a 200 watt amp0lifier, at clipping you might have those power levels into each driver. Now raise the power to the cab to 400 watts. Or what would have been 400 watts. The woofer drive is clipped at the 200 watts the amp can create. The tweeter would now see 40 watts, well within the amp's ability, but a stress for the tweeter. Now push the signal to the 800 watt level. The woofer drive still clips at the amp's 200 watts, but now we have 80 watts into the tweeter, and the amp not even sweating. The idea is that it is the 80 watts of compressed material that blew that 20 watt tweeter, not the high freq components of the clipped low end.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.