Please explain clipping

If anyone would be so kind as to explain to me the term clipping. I would like to know why it takes place? and what are the sonic effects when clipping occurs? , etc.
I'll keep this short as I don't know the real technicals of it. Clipping occurs when an amp cannot drive the speaker to produce either really high or really low frequencies at the volume currently being asked of it. As far as my understanding goes, the amp responds by playing the signal at a lower frequency if it's clipping at high frequency, or vice versa. I'm still not sure if this damages the amp, so I tend to avoid it. The sonic difference would probably be a slight distortion of the sound at the upper and lower ranges. Most audiophiles appear to solve this problem by having an amp (or two) that will destroy the speaker long before it clips. It occurs to me I may be spouting crap here. Let's see if other, more knowledgeable folks describe it the same way...
Clipping occurs when one attempts to play music louder than the amplifier can deliver. It is characterized by extreme harshness and edginess in the high frequencies. The musical signal is literally 'clipped' off, resulting in a squared off signal. This square wave contains tremendous high frequency energy content which can only be dissapated as heat, usually in the tweeter, which can be damaged or destroyed. The amplifier can is also at risk as this excess energy, or current, can cause permanent damage by breaking down the P-N semiconductor barriers in transistors, diodes, and even capacitors. Speaker and amp fuses often do not provide adequate protection from clipping. If you must have loud music, get an amp with plenty of power and invest in efficient speakers.
Clipping occurs when an amplifier stage cannot supply either enough output voltage swing or current to the load. It is not really related to frequency, except that speakers often have a lower impedance in a certain frequency range (usually lower freq) and that is where the amp would tend to run out of gas when attempting to respond to a high amplitude (loud) passage in that range. This has been discussed in other threads at some length, you should search. The result of the rapid change in voltage / current when the limit is reached, is generally a high frequncy transient (spike) that typically would just be too much for the high frequency drivers, thus fried tweeters. Such damage is therefore usually the result of inadequate electronics rather than inadequate speakers. It has also been pointed out in another thread that clipping can occur in any preamp / amp stage and have sonic implications, but is usually thought about with respect to the final amp because of the physical consequeces. :)
Sf- Clipping is the inability of an amplifier to increase the amplitude of the input signal while maintaining the same waveform shape as the input signal. The peaks of a sinewave will no longer be round, but "clipped" flat. (This would be an example of severe clipping). Some amplifiers produce excessive amounts of distortion when this occurs and don't recover quickly or gracefully from clipping. Regrettably, without nearly limitless amounts of power, clipping is bound to occur sometimes. I seem to remember an MIT test where the reproduced sound of a pair of scissors still showed signs of slight clipping even with 2,000 watts available. Single-ended triode designs,and tubes in general, tend to handle this situation in a manner more pleasing to the ear, though they obviously have greater measureable distortion products than other topologies.
Hope this helps.
Sfrounds, Jcbtubes has given the best explanation. As an aside, there is evidence that between well designed amplifiers, the amp's recovery characteristis from clipping may be a key determinate in listener sound preference. Some amps go into short-term instability after clipping, while others seem unfazed.
Jcbtubes, I thought I knew what clipping was, but now I wonder. From your post, it appears that clipping is not amplitude related, as I thought it was -- did the 2kw amp produce clipping even at very low volumes? Was it more a function of slew rate or other factors? Are there two kinds of clipping -- one related to design and one related to overstepping the design? Sorry for all the questions, but I would appreciate any insight. Thanks for the post.
Clipping is simply the amplifier trying to output more power than it's power supply is capable of. It is not a function of slew rate but simply the amplifier is being asked to provide more power than it's power supply is able to handle. The term clipping came into use because of the oscillographic picture of a sine wave output of an amplifier in clipping looks as if it had the top cut off. Clipping causes distortion based in the odd harmonics. Slew induced distortion or Transient Intermodulation Distortion is a function of amplifier bandwidth or speed. If an amplifier has power in reserve it cannot be considered in clipping until that reserve is used up. That is why the spec of headroom is important. If an amplifier has a headroom spec of 3db it can pass 2 times the power it is rated for. Unfortunated speaker loads are not linear over frequency. While the nominal impedance may be 8 ohms the impedance may at some frequencies be as high as 20 ohms or as low as 2 ohms. It is possible that at the majority of frequencies our amplifier can supply enough power to drive the speakers but if the impedance requires more power than the amp can muster, the result is, you guessed it! Clipping.
So many points of view .... Here's mine : )

For an amplifier or "gain stage" ( any form of amplification, be it in a preamp, phono preamp, etc..) to remain "linear" or "clean", it would amplify all incoming signals equally. The input and output waveforms would look identical with the only difference being "amplitude" or "volume" i.e. the signal goes in "weak" and comes out "strong" but looks like a mirror image with the same overall shape and duration.

If an input signal requires more gain than the amplifier can deliver, the peaks of the signal are distorted or clipped off (depending on the severity of power limitations ). This type of distortion can occur due to saturation of either the active devices such as the transistors, tubes, mosfets, etc... found in the driver or output stages OR the power supply being taxed beyond its' capabilities. Think of it as running a cars' motor well beyond redline ( saturation or "overload" ) or a cars' motor that is drinking fuel faster than the fuel pump could supply it ( starvation ).

Once "clipping" occurs, distortion of all types start to take off. Harmonics ( odd and even multiples of the main frequency ) are the primary "culprit", but other forms are also generated. This means that a 100 Hz signal that was "clipped" would produce notes at 200 Hz, 300 Hz, 400 Hz, etc... all the way up into the high treble region and beyond. Some amps, like tubes, will reproduce more even harmonics ( 100 hz x 2, 100 hz x 4, etc..) whereas most transistor amps will reproduce more odd harmonics ( 100 hz x 3, 100 hz x 5, etc...). Many integrated circuits ( IC's) tend to reproduce ALL of the harmonics. These are generalizations at best, but i hope that you can follow along.

Most people feel that even order harmonics ( x 2, x 4, etc..) like a tube clips tend to sound more natural and less harsh than odd order ( solid state ) harmonics. Since many IC's reproduce ALL the harmonics, many people say that everything is "COMPLETELY smeared" when they are used in circuitry. Quite honestly, I will agree that all of these things sound "different". Even amps using very similar parts can sound different due to various circuit designs, bias levels, etc... Each type of device has different strengths and weaknesses and that's why we have SO many different designs out there.

Since harmonics ( even, odd or all of them ) are reproduced ON TOP of and along with the musical signal that is already present, the upper frequency ranges are now seeing WAY more power than what they would normally see. On top of this, severe clipping results in increased duration of the notes, i.e. "hard / extended smearing", etc... Mids and tweeters now have to handle a greater percentage of power AND do it for a longer duration of time. As such, their smaller motors ( voice coil assemblies and magnet structure ) might not be able to dissipate near as much energy ( primarily heat ) as a woofer could. The result is a burned winding ( resulting in either a "short" or "open" in the voice coil), a deformed voice coil former that drags or rubs, a broken "flex lead", etc.. Woofers are rarely damaged when amps clip due to their sturdier motors and increased capacity for heat distribution. However, it is possible to damage a woofer if the amplifier is "passing D.C." (direct current) or it is not "built like a tank". If there is a lot of DC voltage present for an extended period of time, the woofer can even "flame out".

Besides the harmonics that are generated from pushing an amplifier stage too hard, I.M.D. (InterModulation Distortion) usually skyrockets also. What IMD does is to take and mix the signals ( both music and harmonics ) and creates even more chaos. Whereas you might have had a signal at 125 hz and 200 hz, you will now have signals that are the sum (125 + 200 = 325 hz) and the difference (200 - 125 = 75 hz). Add all of these multiples and divisions ON TOP of the energy already being reproduced by the harmonics and "regular music" being passed through the system and you have a real "energy mess" that the speakers are trying to deal with and dissipate.

Needless to say, music sounds "better" and "cleaner" when the gain stages are "coasting" and not being pushed. Speakers and amplifiers also typicall last longer if not "pile driven into clipping". Just like anything else in life, the more that you abuse it, the more likely it is to fall short of your expectations in reliability.

Something that is similar to overdriving amplifiers and clipping taking place occurs in speakers. This is called "compression", but is a WHOLE 'nother ball o wax. Somebody else can fill you guys in on that. I'm too tired tonight... : )

In the meantime, i hope this helps... Sean
Great responses (wow, Sean), if I may say. Just to add on ss vs. tubes, mentioned above: tubes have high voltage-low current operation and thereby sound more "forgiving" in extreme driving situations than ss (lower voltage-higher current) i.e., the *voltage* requirement is rarely surpassed by the sudden dynamics, say, in an orchestral piece (but the current is!!!).

Speaking of orchestra: I've seen dB change from 25-70 on orchestral music (same volume setting, ofcourse). Rock can be more forgiving -- unless you have electronically generated sounds (at 20Hz, 35kHz, etc) that can be very difficult to amplify...

Ozfly- Your questions, I believe, were answered in Sean's terrific response. ( I'm glad that there are those here with much faster typing skills than I !! ) Thanks, Sean!