Talk:Clipping (audio)/Archive 1

Possible vandalism
check out that repair section, I'm pretty sure all those "$FIX_ME"s are ad-libbed and I don't know how to revert a page... Filter1987 03:29, 20 May 2007 (UTC)

More general
Clipping happens with signals that are not audio, too, but I don't see an article for it. Move to clipping (signal) or something instead? — Omegatron 18:17, 31 May 2006 (UTC)
 * Good point, sounds ok to me. Lgreen 05:41, 1 June 2006 (UTC)
 * Clipping (signal processing) sounds better to me. --Kjoonlee 03:50, 20 July 2006 (UTC)
 * I agree, it mirrors the naming of sampling (signal processing) at least.
 * There seem to be several clipping articles: Clipping (audio), Clipping (photography), Soft clipping, and Clipper (electronics). My first thought was to merge many of them into one, but that may end up to be too long of an article.  An alternative is to put the generic+summary information in one top-level article (eg. what clipping is, what general kinds of things might cause it), and have the more specific things in separate articles (eg. this article could include just audio-specific specific clipping causes/effects/concerns).  I took a shot at the second alternative, and created Clipping (signal processing) as a top-level summary article.  If that option is kept, this article could remove some of the general clipping information (the "Digital clipping" section was moved over completely...  many other parts of that page were borrowed from this one, but still much of this page is audio-specific, and the general one will become less audio-specific over time).  Or, if someone wants to take a shot at merging them all together to see how that looks, that would work too.  --Interiot 05:51, 22 May 2007 (UTC)


 * Today soft clipping was a redirect to clipping (signal processing) which doesn't really explain what soft clipping is. I changed soft clipping to a redirect to the gain compression article, which explains gain compression and then says "Gain compression is, rather, the same concept as "soft clipping"." Does that make sense? --76.209.28.72 15:08, 6 July 2007 (UTC)
 * The intro to clipping (signal processing) says "Clipping may be described as hard, in cases where the signal is strictly limited at the threshold, producing a flat cutoff; or it may be described as soft, in cases where the clipped signal continues to follow the original at a reduced gain." --Dawdler 23:14, 20 August 2007 (UTC)

Frequency Domain effect of clipping
"In the frequency domain, clipping produces strong harmonics in the high-frequency range. Extra high-frequency weighting of a signal is more likely to damage tweeters than a signal that was not clipped as higher high frequency power levels are delivered to the tweeter. However most loudspeakers are designed to handle signals with abundant high frequencies, like cymbal crashes, which have a greater high-pitch frequency weighting than amplifier clipping could produce.[citation needed] Therefore damage attributable to this characteristic is rare."

When you do calculate the fourier series for a square sine wave, you find the corners of the square correspond to extremely high frequencies. As far as I'm aware this causes significant damage to most loudspeakers at the amplitude required to produce clipping.


 * The power level at which clipping is involved is an important factor in figuring damage. Clipping can happen early in the signal chain after which subsequent pieces of equipment might easily be able to handle the clipped signal, including the final amplification stage and the loudspeaker. On the other hand, if the final amplifier stage is clipping and that amp has been attached to a loudspeaker whose destruction limits have been closely matched to the amp's unclipped capabilities, then yes, the loudspeaker will likely be damaged by high frequency output related to clipping. Large woofers are less likely to be damaged by HF from clipping. Binksternet (talk) 18:31, 28 August 2008 (UTC)


 * Have a look at table 1 in this white paper: http://www.audiovisualdevices.com.au/downloads/rane/note128.pdf And note that the power level of harmonics drop as the frequency increases. Although the square wave will have harmonics at essentially infinite frequency, the power level of those infinite frequency harmonics approaches zero. As the harmonic frequency increases, the power level of that harmonic decreases. So even with square wave clipping harmonics, most of the harmonic power is still at the low frequencies. The white paper is worth reading because it explains how clipping does increase high frequency power levels via overall power increase rather than harmonics. Sprexumn (talk) 08:01, 26 October 2008 (UTC)

Power in excess of its ??
I reverted a nonsense edit in the lead paragraph:


 * "Driving an amplifier into clipping may cause it attempt to put out power in excess of its maximum capacity, which are customarily done with a "clean" sine wave signal just at the onset of clipping."

Problems with this include several grammar ones, the ridiculous idea that an amplifier is attempting something, and the idea that an amplifier's maximum capacity is somehow connected to a clean sine wave signal. No and no: an amplifier's maximum output is greater than its maximum 'clean' output. The amplifier does not 'attempt' anything, it does something or it does not do something. Here is the correct sentence:


 * "Driving an amplifier into clipping may cause it to put out power in excess of its published ratings, which are customarily done with a "clean" sine wave signal just at the onset of clipping." Binksternet (talk) 21:42, 18 March 2010 (UTC)


 * Good work. Clearly improved. It is OK to supply a brief edit description and let your edits speak for themselves. --Kvng (talk) 14:16, 19 March 2010 (UTC)

More power, than the unclipped? :o
"Because the clipped waveform has more area underneath it than the smaller unclipped waveform" -- How is this possible? Looking at the clipped PCM signal it's obvious that the unclipped signal has a greater area underneath it, because substracting the clipped from the orignal gives a residual area (that tiny triangle like peak). Anyone care to explain this to me? Thanks. PAStheLoD (talk) 11:28, 15 January 2008 (UTC)


 * I think you are right. I had the same question while reading the article.  RodneyMyers  —Preceding comment was added at 03:46, 26 February 2008 (UTC)


 * A squarewave obviously has more area underneath it than a sine wave of the same amplitude. The fact that you can draw the Sine wave inside the squarewave proves it. 80.229.70.84 (talk) 11:53, 21 July 2008 (UTC)

However, I now have a different misunderstanding. What does clipping mean exactly? The overdriving of the amp or the actual distorsion (the amp can't produce the requested/recorded waveform, and "clips off" the peak)? Because in the latter case, there's actually less power than it would've, had the amp been able to produce the requested waveform. —Preceding unsigned comment added by PAStheLoD (talk • contribs) 00:46, 24 July 2008 (UTC)
 * Here's a graphic representation of one sine wave cycle at 0 dBFS compared to the same wave amplified by 1 dB such that it is now clipped. The red area shows how much more power the clipped wave has. The calculated rms power of the lower, unclipped cycle is 0 dBFS, of course, and the clipped one is 0.62 dBFS; about 40 percent of the 1 dB amplication process was cut off at the top. Binksternet (talk) 18:53, 21 July 2008 (UTC)
 * Thanks for clarification! I was looking at the picture on the article page. [[Image:Clipping.svg|thumb| On this picture the fainter wave has more area than the stronger black one.]] And on this image I was thinking about the wave that would have been on the graph without the hard limit and the actual clipped wave. I wasn't considering the actual original wave, that was amplified during "mastering".
 * Clipping is the overdriving of one segment of an audio chain beyond its current and voltage ability. Yes, an amplifier capable of reproducing the requested waveform is more powerful than one that isn't. However, when considering a finite system of audio gear, there's an absolute top limit of ability beyond which the signal gets clipped and there ain't no more. In this case, that is, when considering a single product, the clipped waveform has more power than an unclipped waveform from the same piece of gear, simply because you have to turn the damn thing down until it's not getting clipped. :-) Binksternet (talk) 01:05, 24 July 2008 (UTC)
 * Thanks for this fast reply. I got it, hopefully :) PAStheLoD (talk) 01:12, 24 July 2008 (UTC)
 * It sounds like everyone is on the same page now. I updated the article to say: "Because the clipped waveform has more area underneath it than the smaller maximum unclipped waveform, the amplifier produces more output power."  In other words, when you turn the amplifier dial up to 11 (into clipping), the amplifier produces more power (area under curve) than the largest signal it could produce without clipping. Sprexumn (talk) 04:37, 29 July 2008 (UTC)

Binksternet has removed the discussion of this matter from the article on a technicality because there was no citation. The text removed on March 19 2010 at 14:37 came after: In the frequency domain, clipping produces harmonics at higher frequencies than the unclipped signal. This additional high frequency energy has the potential to damage a loudspeaker's tweeter via overheating The removed text was: However, these higher frequency harmonics have less energy than the lower frequency unclipped signal. Well designed loudspeakers usually do not overheat playing white noise (which has the same energy at all frequencies) near their maximum specified power level. Therefore damage attributable to clipping-induced harmonics is rare. I wrote that text and I think it's important to include because it addresses a common controversy that causes many people to look for information about clipping. It seems to me to be common sense info; nobody would argue that loudspeakers normally overheat playing white noise near their maximum specified power level, and I've backed up the first sentence with the Rane white paper posted in the talk section. But I don't have compliant citations that meet wiki's standards. I came up with that example as the most concise way to clearly express the idea that it's the power not the frequency that causes damage with clipping. Can anyone else figure out a way to convey that idea in a way that conforms to wikipedia's guidelines? Sprexumn (talk) 23:52, 27 March 2010 (UTC)
 * I question the wording; I think the choice of white noise as the example signal is a poor one. Virtually nobody plays white noise loudly on their loudspeakers.


 * In addition to the Rane paper ("Power Amplifier Clipping and its Effects on Loudspeaker Reliability"), here are some further sources:
 * "Loudspeaker Power Ratings", Community Loudspeakers, Chuck McGregor
 * "How Many Watts : Amps vs. Loudspeakers", ProSoundWeb, Chuck McGregor
 * "Frequently Asked Questions", JBL Professional
 * "Speaker Power Requirements: Answers To Some Often Asked Questions", JBL Professional
 * "An Investigation into How Amplifier Clipping is Said to Burn-Out Loudspeakers, and How Limiters Can Save Them", AES paper, Montgomery F. Ross of Rane Corp.
 * These sources differ on minor points but apparently agree that tweeters blow more often than woofers because tweeters are more efficient, and when the signal's power is increased, the woofer shrugs it off but the tweeter (or compression driver) accepts relatively more of it. Each paper comes at the problem from a different viewpoint, but, in agreement with the motive behind your wording, they do not put a high value on the danger of clipped harmonic frequencies. I have not been able to find it today, but I once saw a photo essay mounted online by a loudspeaker reconing guy. This photo essay showed the effects of mechanical vs. heat damage on a voice coil, and the effects of heat with a DC component. As I remember it, there were several photos of voice coil failure modes in which there was no discoloration from heat, and no physical damage from the voice coil slamming against its mechanical limits, but the tinsel lead was broken. The idea presented to the reader was that HF components of clipping may have caused the tinsel lead to serve as a fusible link. Binksternet (talk) 01:12, 28 March 2010 (UTC)
 * One more failure mode can be observed in bi-, tri- and quad- amped systems in which big paper cones are driven by their own dedicated amplifiers, following some kind of crossover. If a high-level high-frequency noise gets into the amplifier, the woofer cone cannot reproduce the signal, and just sits there absorbing the energy, turning it into heat. A weird and unusual failure mode, to be sure! Binksternet (talk) 01:16, 28 March 2010 (UTC)


 * I skimmed those papers but didn't see where they asserted or even agreed that tweeters blow more often than woofers. If I'm missing something please point out which passages support your assertions.  That last paper sounds interesting and relevant but unfortunately is behind a $20 paywall.  Although there are many people in the world who express the belief that too little power kills tweeters via clipping-induced distortion, they never mention the much bigger problem of overheating due to too much power killing tweeters, or make any attempt to distinguish between the two very different causes of failure.  The reality is that too little power does not kill speakers.  Perhaps it would make sense to create a separate web page for the "Too little power kills tweeters" controversy.  At any rate, here's a thorough exposition of my viewpoint that too little power does not kill speakers:  http://www.bcae1.com/2ltlpwr.htm Sprexumn (talk) 01:45, 5 June 2010 (UTC)