A typical reply from a bias which is convinced it is unconditionally always correct.jaddie said:
Yes, clipping does have to do with it. I quoted someone who works in an establishment supplying PA systems some of which cost tens of thousand Euros. He does understand what clipping can do.
If you are not convinced of their expertise, this is their website.
Olimpus Music | Home
Tens of thousands of euros is pretty small scale in the PA system world, and the fact that they're only selling them (rather than, say, designing them) means I wouldn't give them more than a cursory look.
Appeals to authority won't work here, I'm afraid. I don't run a music shop, but I do design PA systems. I've also had a few articles published in Live Sound International Magazine.
Anyway. The additional harmonics that are produced by clipping are a tiny tiny fraction of the power increase resulting from pushing an amplifier into clip.
The real increase in power comes from the increase in RMS signal levels.
Question: why would a speaker care whether a signal is clipped or not?
Chris
Appeals to authority won't work here, I'm afraid. I don't run a music shop, but I do design PA systems. I've also had a few articles published in Live Sound International Magazine.
Anyway. The additional harmonics that are produced by clipping are a tiny tiny fraction of the power increase resulting from pushing an amplifier into clip.
The real increase in power comes from the increase in RMS signal levels.
Question: why would a speaker care whether a signal is clipped or not?
Chris
No clipping itself doesn't over powering does but I think you sorta get that.
Clipping doesn't work like that either, there is an infinite range of clipping from just shaving the tops of the highest peaks to total compression of the signal so the amount of extra power generated will also range from insignificant to near instant speaker death. Another point to consider is that a voice coils impedance rises as it is heated so after some time in operation even at nominal power levels the amount of power dissipated drops, this effect is known as power compression. Combine that with the relatively heavy wire used in a woofer or subwoofers VC and these drivers are relatively immune to momentary amp clipping, but the small gauge wire used in high frequency drivers cannot tolerate even small overpowering peaks the fine VC wire behaves like a fuse and will burn open at the weakest point so those drivers are much more susceptible to the damaging effects of amplifier clipping.
I'm not convinced. I didn't see anything on that site that indicates a good technical understanding of clipping.
I've done extensive analysis of this, posted here somewhere.
There are several misconceptions rampant in the subject.
1. Clipping places DC on the voice coil. False. Clipped AC (audio) is not DC, not at all. It's alternating. For a clipped AC wave form would only be comparable to DC when it's peak value and RMS value match that of a similar DC voltage. As long as the signal is alternating, that doesn't actually ever happen. However, it comes close when the signal is a square wave. Audio clipping does not produce square waves! Distortion at that level would not even resemble audio, and would be unacceptable to anyone.
2. Clipping causes excessive high frequencies to burn tweeters. False. Clipping of music doesn't significantly raise the power in the HF spectrum at all. I have spectrum analysis that proves this. The HF content doesn't change at all for clipping amounts that are audibly tolerable, like a several dB or so above clip threshold. Above that, the audible distortion is horrific, but the HF content may rise a little. Total RMS is what goes up.
3. Clipping an amp is worse than the equivalent level of unclipped audio. False. I have a graph, already posted, that shows the rise in RMS power of a signal both clipped and not. Clipping causes an increase in RMS power, but the same signal unclipped shows RMS power that rises even faster and much higher.
4. And finally, speakers are damaged by clipping because it's clipping. False Speakers are damaged two ways: 1. by excessive RMS (heating) energy over heating the voice coil. 2. by over-excursion of the voice coil. 1 is caused by both clipping and no clipping, but the unclipped version actually applies more RMS per increase in level, as clipping is a form of limiting. 2 occurs mostly in woofers, and is a peak function. Clipping is a peak limiter.
The concept of clipping being a worse mechanism for damaging drivers is widely held. The reason goes back many years to the days when a very popular line of high power amplifiers would break into oscillation when forced to clip. The oscillation produced high frequency RMS energy far above the levels occurring in normal music spectral distribution. The fix, back then, was to avoid clipping by putting a higher power amp into that application. This resulted in higher power amps driving tweeters and lower power amps driving woofers, an odd condition that was actually a rather poor but expedient work-around. The root cause was an amplifier design flaw. That cause is long gone.
Now, if you'd like, I can post the analysis graph set again. But I'm sure it's in this thread somewhere already. The analysis was done with both tones and contemporary music with dense high-loudness processing. The measurements are real, not simulated.
This is all essentially a repost, provided for those that won't take the time to read or dig, but would rather parrot the popular, but errant, concept.
edit: Looks like I'm wrong. It's not all posted here. The topic comes up so often I loose track of where the discussion was. Evidently not here. I'll post the graphs....
Last edited:
I checked with mine. They said they don't care.
I checked with my ears, they prefer less clipping.
The problem with clipping is it usually happens on full power.
A square wave is 1.414 times the power of a sine wave.
So an amp matched to a speaker with sine waves is well out of spec for square waves.
No, that's not how it works.
Clipping is not square waves. Music is not sine waves.
The curves below are based on sine waves with successive data points representing levels as the clipping threshold is exceeded. The straight traces are unclipped. RMS levels are graphed.
The curves below represent the spectrum of real audio, music, with the bottom trace being made at a level where no clipping occurs, then each trace representing successively higher levels above clipping. The top curve is 10dB above clipping threshold, and totally unlistenable.
The curves below represent the spectrum of real audio, music, with the bottom trace being made at a level where no clipping occurs, then each trace representing successively higher levels above clipping. The top curve is 10dB above clipping threshold, and totally unlistenable.
While it's true that Fourier states all signals are a load of sine waves, it's pretty much irrelevant to this discussion.
Chris
Chris
Square waves are still just a load of sine waves - "short-term DC" is a misnomer. If there's a time component, it's AC by definition.
Chris
DC is a continuous unchanging voltage and applying that to a speaker produces no sound. But even a square wave switches from it's positive max to negative minimum every cycle which could be anywhere from 20 to 20000 times a second, and that is audible. So yeah clipping does not equal DC.
so thinking that dc is added to ac at clip is wrong , i thought that's what raised the RMS level?
<sigh>
Let's not have this pointless argument. Music can be de-constructed into sine waves, but the waveform itself is a composite, not a sine wave. A group of sine waves of various frequencies and phase relationships can be combined to make a complex wave form. So what?
It's irrelevant to the discussion because what's presented to the speaker and amp is not a sine wave in most cases, and even if it were, clipping doesn't produce square waves, and the relationship of RMS power to the degree of clipping is accurately represented in the graph I posted.
Your point is tangential at best. Can we drop it please?
Yes, true. A perfect square wave is an infinite load of sine waves. There's no DC with a time component. Bingo.
Google the definition of DC. "The unidirectional flow of an electrical charge". Pure DC must also not change within a time window of concern.
Look at any audible waveform at an amplifier output. Unless there's DC offset, everything there is bi-directional, and time-variant. DC offset is a defect, it should not be there. DC offsets in acoustic waves are not transduced by ears or microphones.