This thread split off from 1812 Overture speaker damage
_________________________________________________________________________________________________
If the woofers were destroyed, they received too much power. Clipping vs not has little to do with it.
I do have a 21" sealed subwoofer around here somewhere. Might give it a go.
Chris
_________________________________________________________________________________________________
If the woofers were destroyed, they received too much power. Clipping vs not has little to do with it.
I do have a 21" sealed subwoofer around here somewhere. Might give it a go.
Chris
Clipping, by definition, must be less power than if the peak was allowed through.
Those that say "clipping kills speakers" are mistaken. Guitar speakers are routinely fed signals in excess of 100% THD, and we don't see an avalanche of failures whenever the "overdrive" pedal is pressed. The existence of rock music proves that clipping alone does not kill speakers.
Speakers die because they have received too much power. When you take a musical signal where the peaks are at the onset of clipping an amplifier and raise the level by +3dB, the peaks are chopped off. Being short-duration, this is of little consequence. The harmonic structure barely changes, and the changes that do occur are also of a very short duration.
What is of consequence is that the long-term average power has doubled.
Since voice coils typically have a thermal time constant of the order of seconds (not milliseconds), it must be the average power level that eventually causes them to burn out.
Chris
Those that say "clipping kills speakers" are mistaken. Guitar speakers are routinely fed signals in excess of 100% THD, and we don't see an avalanche of failures whenever the "overdrive" pedal is pressed. The existence of rock music proves that clipping alone does not kill speakers.
Speakers die because they have received too much power. When you take a musical signal where the peaks are at the onset of clipping an amplifier and raise the level by +3dB, the peaks are chopped off. Being short-duration, this is of little consequence. The harmonic structure barely changes, and the changes that do occur are also of a very short duration.
What is of consequence is that the long-term average power has doubled.
Since voice coils typically have a thermal time constant of the order of seconds (not milliseconds), it must be the average power level that eventually causes them to burn out.
Chris
Yes but.... When the peak is not allowed thru the time that it is a maximum is much longer. Maybe 10X longer. That means a lot more current thru the voice coil. Amps also heat up fast when clipping.
I took at look and I have only 2 digital recordings of the 1812 Overture. One is by the London SO and has about 31 dB between average and peak. The cannons sound like fireworks. The other is Andrew Litton and the Dallas SO. It has a not so unusual 25dB range between the beginning of the piece and the the cannons. But the cannons are clipped! When declipped and the peaks are allowed thru there is at least 5dB more range, putting it near the London SO recording. The Dallas cannons sound better, clipped or not.
The "maximum" is at a much lower level than the unclipped signal, though.
Let's take a simpler example: a sine wave.
Take a sine wave, peak amplitude 1V.
Now, clip it to 0.5V. You'll get something resembling a square wave, with a peak amplitude that's 6dB lower than the sine wave.
Which one will deliver the most power to a connected load?
Chris
Let's take a simpler example: a sine wave.
Take a sine wave, peak amplitude 1V.
Now, clip it to 0.5V. You'll get something resembling a square wave, with a peak amplitude that's 6dB lower than the sine wave.
Which one will deliver the most power to a connected load?
Chris
Yes but when you clip it, you don't get 1/2 the voltage, you get 100% of the voltage for the duration of the clip.
If I measure the current flowing thru an amp it's easy to tell when the amp is clipping without even hearing it. Current flow (and thus power) will jump up drastically at the onset of clipping.
Here is part of the reason why. This is not the output of a power amp, but the digital recording of the 1812 Overture with the DSO. The extreme closeup is one of the cannon shots that is clipped. On the left channel (white) you see the original recording as ripped from CD. On the right (red) you see the same signal de-clipped. See how much longer the clipped signal spends at 100%? It's significant and will mean much more power going thru the circuit.
As to whether or not a few cannon blast peaks will kill a speaker or not, I can't say. I doubt that a few quick booms would burn up a voice coil. If the cannon peaks damage speakers, it's more likely an over excursion problem. Of course the amps might well be clipping even if the source isn't, but it shouldn't be clipping long enough to hurt on this recording.
If I measure the current flowing thru an amp it's easy to tell when the amp is clipping without even hearing it. Current flow (and thus power) will jump up drastically at the onset of clipping.
Here is part of the reason why. This is not the output of a power amp, but the digital recording of the 1812 Overture with the DSO. The extreme closeup is one of the cannon shots that is clipped. On the left channel (white) you see the original recording as ripped from CD. On the right (red) you see the same signal de-clipped. See how much longer the clipped signal spends at 100%? It's significant and will mean much more power going thru the circuit.
As to whether or not a few cannon blast peaks will kill a speaker or not, I can't say. I doubt that a few quick booms would burn up a voice coil. If the cannon peaks damage speakers, it's more likely an over excursion problem. Of course the amps might well be clipping even if the source isn't, but it shouldn't be clipping long enough to hurt on this recording.
Attachments
As far as I can tell the 1V pure sinewave will deliver about 4 dB more power to the load than a sine that's been flat topped at 0.5V You can doulbe check me on that one.
However, we are saying basically the same thing from different angles.
You say get a bigger amp with more headroom that won't clip so your speakers will be safer.
I'm saying you already made the mistake of turning it up into clipping, and now there is a whole lot of time spent at 100%
... and if you look at the rest of the waveform, you'll note that the rest of the signals have been reduced by a factor of two, or 6dB.
ie, if we maintained the original average signal level, the unclipped waveform would peak +6dB over the clipped section.
... Which means more power delivered.
Chris
No, that's not what I'm saying.
I'm saying that exceeding the power handling capability of a speaker will cause it to overheat and stop working.
I'm also saying that whether the signal was clipped or not is irrelevant. Clipping does not destroy speakers. Too much power does.
The notion of "100% signal level" is irrelevant, because the speaker does not care if the waveform is clipped or not.
If we have two amplifiers with identical gain but one can provide twice the output voltage of the other, and then use a music signal that takes the larger amplifier to the onset of clipping, then:
- The low-level signals would be at identical levels
- The larger amp would deliver up to 6dB more power where the smaller amp would be clipping
ie, the larger amp will deliver more power overall.
Therefore, using a larger amp to "avoid clipping" and therefore "prevent speaker death" doesn't actually work, which is what I've been saying all along.
Chris
I disagree because music isn't a pure sine wave. The crest factor is much higher in music than a sine wave. A few peaks here and there that go past what the speaker can stand as a continues voltage aren't going to hurt it. But once you start clipping the peaks to flat tops you are burning up a lot more power. Much more than if the peaks were not clipping. I'll try to figure out how much.
I have set speakers on fire with actual smoke and flames by using too much power. No clipping, just way more power than the drivers could handle. But that was a gross mismatch.
This is now getting off topic from the tread and is worth discussing elsewhere.
I have set speakers on fire with actual smoke and flames by using too much power. No clipping, just way more power than the drivers could handle. But that was a gross mismatch.
This is now getting off topic from the tread and is worth discussing elsewhere.
And here we are in a new thread.
I was curious about how much power dissipation would result from a clipped musical signal vs one where the peaks are allowed to play thru to their full value. Last night I took a look.
Looking at several different tracks and different music genres, classical, rock, pop, jazz. The average to peak ratio of the tracks was anywhere from 13dB to 30dB. All tracks had peaks within 1.5dB of full scale, and where then normalized to peak at 0dBFS. After normalization, each track was amplified 10dB to cause hard clipping. Once clipped, the tack was lowered by 10dB to return it to its original level, but now with clipping.
At this point there were two versions of the music track. The original version normalized to 0dBFS and a version clipped by 10dB. The clipped tracks had no value over -10dBFS, just a lot of flat tops at that level. This gave an imitation of a signal that was either played to full voltage on an amp with plenty of headroom, or a signal that was badly clipped by an amp with limited voltage.
The tracks were then analyzed for RMS value to compare.
To my surprise there was very little difference. The clipped and unclipped tracks were very close in RMS value. The clipped tracks tended to calculate out as ~1dB louder, but mostyly it was a tie. 6dB clipping was also tried with the same results.
Caveats: This was done in the digital domain, no analog signals were harmed. The clipping was perfect and not likely to match that of any real amplifier. That's the next avenue of investigation.
I was curious about how much power dissipation would result from a clipped musical signal vs one where the peaks are allowed to play thru to their full value. Last night I took a look.
Looking at several different tracks and different music genres, classical, rock, pop, jazz. The average to peak ratio of the tracks was anywhere from 13dB to 30dB. All tracks had peaks within 1.5dB of full scale, and where then normalized to peak at 0dBFS. After normalization, each track was amplified 10dB to cause hard clipping. Once clipped, the tack was lowered by 10dB to return it to its original level, but now with clipping.
At this point there were two versions of the music track. The original version normalized to 0dBFS and a version clipped by 10dB. The clipped tracks had no value over -10dBFS, just a lot of flat tops at that level. This gave an imitation of a signal that was either played to full voltage on an amp with plenty of headroom, or a signal that was badly clipped by an amp with limited voltage.
The tracks were then analyzed for RMS value to compare.
To my surprise there was very little difference. The clipped and unclipped tracks were very close in RMS value. The clipped tracks tended to calculate out as ~1dB louder, but mostyly it was a tie. 6dB clipping was also tried with the same results.
Caveats: This was done in the digital domain, no analog signals were harmed. The clipping was perfect and not likely to match that of any real amplifier. That's the next avenue of investigation.
re: clipping kills
The reason we say clipping kills is because when we start to clip we don't lower the threshold, but actually raise the volume. This means that the clipped signal has the same peak as the peak without clipping (without going over the ceiling); but a higher volume.
We're not lowering the ceiling, we're raising the floor.
The reason we say clipping kills is because when we start to clip we don't lower the threshold, but actually raise the volume. This means that the clipped signal has the same peak as the peak without clipping (without going over the ceiling); but a higher volume.
We're not lowering the ceiling, we're raising the floor.
Thanks for doing the testing, Pano. Interesting results.
Would it be fair to say, then, that driving a small amplifier up to clipping is no more or less harmful than a larger amplifier delivering the same average signal level?
The caveat being, of course, that a larger amplifier increases the likelihood of mechanical damage to the speaker.
The amplifiers I've tested (a limited number, admittedly) have always shown very sensible clipping behaviour, but they tended to be PA amps where any weakness is frowned upon.
Chris
Would it be fair to say, then, that driving a small amplifier up to clipping is no more or less harmful than a larger amplifier delivering the same average signal level?
The caveat being, of course, that a larger amplifier increases the likelihood of mechanical damage to the speaker.
The amplifiers I've tested (a limited number, admittedly) have always shown very sensible clipping behaviour, but they tended to be PA amps where any weakness is frowned upon.
Chris
Seems at first glance that it's kif-kif as they say in North Africa, or six of one - half dozen of the other. The higher voltage amp will push the driver farther toward its x-max, for sure. And that seems to be what was causing woofer damage with the Telarc CD. Of course it was clipped, too.
Clipping behavior in amps is a major topic, but yes, many good amps are well behaved at clipping. One thing that clipping is going to do is push out a whole slew of harmonics. what all that harmonic energy will do to your tweeter would be good to know.
Clipping behavior in amps is a major topic, but yes, many good amps are well behaved at clipping. One thing that clipping is going to do is push out a whole slew of harmonics. what all that harmonic energy will do to your tweeter would be good to know.
This statement is not accurate. Clipping does kill speakers, especially tweeters, as clipping generates square wave with rich upper harmonics. This kills the tweeters. (trust me, I have done this many times
)Your logic with guitar speakers is not valid, since typical guitar speakers are big fullrange drivers, no tweeters.
Adason, I once thought as you did.
However, a member "Jaddie" talked me round: https://www.diyaudio.com/forums/pa-...r-required-bi-amped-2-system.html#post5358495
The additional power in the harmonics is basically negligible.
Chris
However, a member "Jaddie" talked me round: https://www.diyaudio.com/forums/pa-...r-required-bi-amped-2-system.html#post5358495
The additional power in the harmonics is basically negligible.
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.