Let's Talk About Amplifier Clipping.

[ken kantor]

I have attached the RaneNote 128 right here. This note will explain in simple, semi-technical language why it is that clipping DOES NOT blow tweeters by introducing additional harmonics. If anyone wants a link to the paper, here's one:

www.adx.co.nz/techinfo/audio/note128.pdf‎

In practice, clipping-induced harmonics only add a tiny amount of additional power load to the tweeter, usually a fraction of a percent.

At first, I wanted to disbelieve the author, as I had been taught that clipping-induced harmonics were the culprit behind the majority of tweeter failures. However, the mathematics, when actually carried out, show otherwise. Further, I have conducted many real-world experiments using actual drivers and crossovers of a wide variety of types. It became clear that the author was 100% correct. Clipping, per se, is not what damages tweeters. The "clipping/harmonic" theory is tempting, but incorrect.

-k

 

[Pandovski]

Excellent. Thank you for sharing and setting things straight.

It's interesting how single marketing misinformation spread by few people became such widely know "truth" on a global scale for the past 40 years and still going strong.

Link to the parent thread for future reference: http://www.audiokarma.org/forums/showthread.php?t=547335

Maybe starting a cult isn't that hard to do... :scratch2:

 

[fhawk618]

Article is a good read... I have seen it before somewhere.

FYI--the document has apparently been removed and the link is no longer valid, but the pdf attachment still works.

I never really considered the harmonics theory in clipping as the cause of blown tweeters, but rather the ever increasing duration of the square wave plateau which is seen as ever increasing durations of DC bursts by the speaker. I am certainly nowhere nearly as knowledgeable as you on this topic, but that was always my impression/assumption... that the large square wave plateau of a heavily clipped sine wave essentially becomes bursts of fairly high voltage DC, and that was the cause of driver damage... please correct me if I am wrong.

 

[petehall347]

so it isn't clipping that's the cause but it is the clipping that is the cause .
that's how i read it anyway .

 

[fhawk618]

so it isn't clipping that's the cause but it is the clipping that is the cause .
that's how i read it anyway .

Basically... but it is not due to the harmonics of a clipped signal... that is the point, as I read it... another episode of "MythBusters"... but my theory may be incorrect... waiting for Ken to chime back in...

 

[JBL GUY]

Here is my post from the other thread.

For those that are interested

Can you answer the questions and what conclusion do you come to?

 

[GuyK]

Thanks for the link, now I understand.

 

[ofajen]

so it isn't clipping that's the cause but it is the clipping that is the cause .
that's how i read it anyway .

I'd say the key passage from the article is:

"Now it is easy to see how the high frequency portion exceeds the 5 or 10 watts tweeter rating. Sure, clipping is producing extra harmonics but they never approach the levels of the amplified high frequency source signals."

When you keep turning up the gain, the amplifier is driven into clipping on the LF content, so the low frequency content can't get louder. Some small additional HF content is produced by the clipping, but mainly as you continue to turn up the gain, you are increasing the HF levels of the program content until they also reach the clipping point, by which time you have probably already destroyed your tweeters, due to too much HF power.

So, clipping of LF content is probably occurring, but it isn't the cause of the tweeter damage. Most amps are powerful enough to destroy tweeters before they are clipping the HF content, though they may well be clipping the LF content severely.

Otto

 

[ken kantor]

Article is a good read... I have seen it before somewhere.

FYI--the document has apparently been removed and the link is no longer valid, but the pdf attachment still works.

I never really considered the harmonics theory in clipping as the cause of blown tweeters, but rather the ever increasing duration of the square wave plateau which is seen as ever increasing durations of DC bursts by the speaker. I am certainly nowhere nearly as knowledgeable as you on this topic, but that was always my impression/assumption... that the large square wave plateau of a heavily clipped sine wave essentially becomes bursts of fairly high voltage DC, and that was the cause of driver damage... please correct me if I am wrong.

Sorry, the link was working just a few hours ago, but I guess they didn't want the traffic.

-k

 

[ken kantor]

Basically... but it is not due to the harmonics of a clipped signal... that is the point, as I read it... another episode of "MythBusters"... but my theory may be incorrect... waiting for Ken to chime back in...

If you read earlier in the old thread, I explained why your theory is not correct. But, I am just about played out on this topic, so I don't really want to argue or explain yet again. I will just state, unequivocally that:

1- "Flat" sections of waveforms are NOT DC. They are not related to DC. They do not behave like DC. Flat waveform sections have rich harmonic content. There is no reason to call them DC, or DC-like. This is very basic Fourier and signal theory stuff.

2- Speaker cones do NOT follow the input waveform at their terminals. That is just not how speakers work. Their job is to make air pressure waves that follow the voltage, which means that the actual cone motion is related by a complex equation to the terminal voltage. The terminal voltage may look "flat," but I assure you that the cone is still moving during those moments.

3- Speaker voice coils are NOT cooled very much by motion, nor overheated by lack of motion. To a very great extent, coils are cooled by radiation to the pole piece and into space. They are not cooled by air or air flow, to a large extent. (I, and many others, have designed drivers with enhanced airflow or pumping in the gap, only to find that there was little or no improvement in power handling.)

I consider these statements as known facts in speaker engineering, and not open to casual debate. Hopefully, this will be my last post asserting them, at least for the time being. Believe them, or not, as you wish.

-k

 

[ken kantor]

Here is my post from the other thread.

For those that are interested

Can you answer the questions and what conclusion do you come to?

Sheesh, I am not going to take your quiz. But, here's one for you:


1- Assume a situation of 25%, heavy clipping at 150 Hz into a 2-way loudspeaker. (Most clipping happens between 100 Hz and 200 Hz.)

2- Assume a 2nd-order crossover at 2,000 Hz.

3- Assume 100Wrms into the woofer.

4- Estimate roughly how much actual power the clipping harmonics send to the tweeter.

5- Compare this to the statistically expected amount of unclipped tweeter power resulting from 100Wrms into the woofer playing rock music.

(This is a very simple problem, and should not even require a calculator.)

-k

 

[ofajen]

I have attached the RaneNote 128 right here. This note will explain in simple, semi-technical language why it is that clipping DOES NOT blow tweeters by introducing additional harmonics.

BTW, this pretty much takes care of my question about amp ratings, Ken. Thanks.

Otto

 

[nailer]

If you read earlier in the old thread, I explained why your theory is not correct. But, I am just about played out on this topic, so I don't really want to argue or explain yet again. I will just state, unequivocally that:

1- "Flat" sections of waveforms are NOT DC. They are not related to DC. They do not behave like DC. Flat waveform sections have rich harmonic content. There is no reason to call them DC, or DC-like. This is very basic Fourier and signal theory stuff.

2- Speaker cones do NOT follow the input waveform at their terminals. That is just not how speakers work. Their job is to make air pressure waves that follow the voltage, which means that the actual cone motion is related by a complex equation to the terminal voltage. The terminal voltage may look "flat," but I assure you that the cone is still moving during those moments.

3- Speaker voice coils are NOT cooled very much by motion, nor overheated by lack of motion. To a very great extent, coils are cooled by radiation to the pole piece and into space. They are not cooled by air or air flow, to a large extent. (I, and many others, have designed drivers with enhanced airflow or pumping in the gap, only to find that there was little or no improvement in power handling.)

I consider these statements as known facts in speaker engineering, and not open to casual debate. Hopefully, this will be my last post asserting them, at least for the time being. Believe them, or not, as you wish.

-k

Teaching is a hard profession. How do they do it with a new batch of us appearing every semester/year?

 

[fhawk618]

Thank you Ken for correcting me... as I posted, I wanted your input... as I may have been misguided for many years...

I just kind of go with the flow and what I know... whatever that may be, and I am certainly wrong on many things... thanks for your insight...

 

[ken kantor]

BTW, this pretty much takes care of my question about amp ratings, Ken. Thanks.

Otto

I'm sorry, I keep logging on to answer your question, only to be distracted playing Internet Whack-A-Mole with the clipping enthusiasts.

Setting power ratings for a speaker is never a science and is based largely on the experience of the designers, along with commercial pressures. If one tries to establish an absolutely safe rating, where the speaker would survive any frequency for any amount of time, at any plausible ambient temperature, few speakers, even large ones, would reach 10 Watts. Rating a speaker that way would be commercial suicide, and would keep people from using their speakers anywhere near their actual potential.

At NHT, we rated speakers over a range of amp power. At the low end of the scale, we considered the sensitivity of the speaker and its likely application, and picked a Wattage that we felt would give people at least a decent and expected sound level. At the high end of the scale, we picked an amp the would push a speaker to its limits in normal use. One that, if it was exceeded, would probably not wring more performance from the speaker, and would only increase the risk of damage.

Of course, there are plenty of situations where folks were very happy running speakers with smaller amps in, say, desktop applications. And, plenty of reports of SuperZeros, etc, being successfully run off of 200W/ch.

-k

 

[ken kantor]

Teaching is a hard profession. How do they do it with a new batch of us appearing every semester/year?

Indeed. If people here would help, and the sysops are willing, I would be up for creating some FAQ-style stickies on key topics.

-k

 

[ken kantor]

Thank you Ken for correcting me... as I posted, I wanted your input... as I may have been misguided for many years...

I just kind of go with the flow and what I know... whatever that may be, and I am certainly wrong on many things... thanks for your insight...

Thanks, I appreciate this.

There is no shame in believing what you were taught. We all believed it. In the larger scheme of things, it is good to be cynical and demand a high standard of proof when it comes to overturning generally established facts. It may be frustrating to those trying to introduce new learning, but the alternative would be chaos. So, I do respect the cynics here, particularly the ones who engaged in dialog.

-k

 

[chicks]

Thanks for all your informative posts, Ken. I've often wished for a forum here where only factual, scientifically valid posts were allowed, moderated similarly to the Harbeth and Hydrogenaudio forums. Wishful thinking, I know, but your participation here is greatly appreciated.

 

[89grand]

Ken, I remember a while back I said I quit and it was pointless. Nice to see you are more dedicated than myself. It's hard to keep repeating yourself over and over and over again, but you deserve credit for continuing the fight against incorrect information, and you are certainly more knowledgeable than me, so hopefully coming from you it sinks in.

The sad part is, half the forum tomorrow will again say clipping ruins tweeters. I hope I'm wrong.

 

[frankxbe]

Ken Kantor wrote in post # 1 :

I have attached the RaneNote 128 right here. This note will explain in simple, semi-technical language why it is that clipping DOES NOT blow tweeters by introducing additional harmonics. If anyone wants a link to the paper, here's one:www.adx.co.nz/techinfo/audio/note128.pdf‎

This link seems to work at the moment


Ken:

Thanks taking the time to educate some of us ' phools' young and old (including me ) and dispel some long held misinformation many of us believed to be true at least in respect to how I presently understand what I read here today and priorly similar things elsewhere .
Thanks to others here that have contributed valuable comment.

Obviously not being an EE or not otherwise having a professional background in electronics or audio it was good to see some information
well presented in laymans terms.
While it may or may not merit additional discussion I for one am not sufficiently qualified for that.

This thread should perhaps be a sticky on All audio forums it has my vote for a sticky at AK here!
Lets submit it to the politburo for consideration

Best regards