Clipping Amps and Blowing Tweeters

This is one of those topics that comes bouncing back about once a week here at AudioKarma, the notion of low-power amps destroying speakers. When I first heard of this theory decades ago, yeah, it made sense. Clip a signal, create sharp edges in time, thus create high frequency energy. Makes sense to me. I grasped that idea and ran with it for years.

More recently learned people have been implying this isn't really the issue. Their views make sense as well. In a recent thread on this very topic here at AK ("Amp power Vs. Speaker"), pustelniakr made this comment:

pustelniakr said:

Clipping a signal (limiting positive and/or negative excursions) limits the power at the clipped frequencies. As a signal approaches a square wave, what you get is a fundamental and all of the odd harmonics. Indeed, some additional energy IS being fed to the tweeters. Is it enough, by itself, to fry the tweeters and midranges? Lets see some hard math. I'm interested. But it is far from necessary.

Click to expand...

When someone says data is far from necessary, I take that as a challenge. Heck, I'm much more than interested myself. Thus, I decided to whip up a little program to simulate the effects of clipping and measure the results. First of all, I must say, I'm no programmer, and it has been years since I've written a VEE program, but VEE is a simple language created pretty much exactly for these types of "what if" scenarios. After many hours of sitting in front of a terminal feeling stupid for forgetting so much of the basics of VEE, I finally got something useful. I've decided to share the results here. I'm not claiming to be the ultimate guru on this topic, so if I've done something wrong in my logic, or made bad assumptions, please set me straight.

I was hoping to share the program with anyone interested, but I haven't gotten it to work in the run-time environment, only the development environment. Perhaps in another week or so... At any rate, I took screen shots of the clipping simulator with several different degrees of clipping going on, so we could see how much energy is in the higher regions. I'll go over the features of the screenshots in detail for the first picture, barely clipping, and then show the other pictures which have higher degrees of clipping, and greater high frequency energy.



The screenshot has two knobs and two displays. The knobs are used to set the degree of clipping and the displays show the effects of clipping. The upper knob, "Desired Amplitude," sets the amplitude of the desired wave, if the signal weren't clipped. The units of this knob are volts, so think of this as the maximum unclipped amplitude that is desired. To the right of this is a display that indicated the maximum peak power this would develop into an 8ohm load. In this example, the knob is set to 40, which implies the signal will travel between +/-40V unclipped. The display indicates that this is 200W at the peak voltage. The lower knob sets the actual clipping voltage that a particular amp may exhibit. In this example, I've got it set to 39.75V, or in other words, 250mV is being clipped off of the top and bottom of the waveform. The calculated power at clip is 197.5W. The "Clipping Amount (dB)" display shows how much of the signal has been clipped, and in this example is 0.05446dB - not much at all. The Waveform display shows both waves, with the unclipped wave in yellow and the clipped wave in blue. You can just barely see the yellow tops above the clipped blue ones. The Spectrum display again has both waves displayed, but we're really only interested in the blue clipped one. For this example (and the others as well), I put the markers on the 200Hz fundamental, and the third harmonic at 600Hz. The Spectrum display tells us that the third harmonic is at 22.49mV. I added a handy power calculator below the Spectrum display to translate this to power into 8ohms. In this case, it is a whopping 63uW. If you can't think in microwatts, it is 0.000063W.

By the way, the wasted space near the Clip Threshold knob is for other versions of this software with greater capabilities. One version allows me to set Vcc and Vee Clip Threshold independantly, to simulate asymmetrical clipping, and another version allows me to control the voltage of the clip area to simulate latch-up or power supply droop. Many more knobs, but not so useful for this thought exercise. One other thing, the ugliness of the traces is due to the resizing of the pictures in order for AK to accept them. I could have (should have) remade my panel to the appropriate size and regrabbed the screenshots, but I'm lazy.

The other three examples I did have greater amounts of clipping to really get a feel for what kind of energy is on tap in a clipping situation. I have the Clip Threshold set to 30V, 20V, and 10V. This gives rise to Clipping Amounts of 2.5dB, 6dB, and 12dB, respectively.



Let's jump to the worse case one, 10V/12dB, to see what sort of energy there is. In this case, again, we have the levels set for a clean signal if our amp could supply 200W into 8ohms. But, the amp we're using has power supply rails so low that it can only pass a 12.5W signal without clipping. Right, we've got it so loud that we should have a 200W amp, but we're using a 12.5W amp. 12dB of signal is clipped right off. The third harmonic is 3.84V, or 1.843W. Frankly, IMO, if someone is clipping this severly, they deserve to blow a tweeter or two.

All in all, less power is generated on a clip than what I was assuming going into this experiment. Several interesting things become evident when playing around with this software, but nothing earth shattering. Going further, I would like to add a summer to sum the total power of the harmonics. Or, better yet, be able to set what the crossover frequency is to the tweeter to sum just those frequencies. And, have a variable frequency clipped signal, to boot. This would give a much better idea of the total heating power delivered to a tweeter voice coil. Alas, it was hard enough to bang out this little bit of code in a week for me.

I hope this was useful to other AKers. I know it actually has been an eye opener to me.

Drunken morons touching the volume knob will cause blown tweeters.

coonmanx said:

Drunken morons touching the volume knob will cause blown tweeters.

Click to expand...

This has been clinically proven numerous times, and I concur.

House de Kris said:

This has been clinically proven numerous times, and I concur.

Click to expand...

The only blown tweeter here was caused by a drunken guest turning up the volume. So I guess this must be the correct diagnosis for blown tweeters after all.

Ken Kantor (of NHT) has posted here on AK on this topic, and IIRC he is in agreement with your findings. i.e. A clipping low-powered amp is unlikely to damage a tweeter.

To the OP -

Very impressive work, and quite interesting.

Here is a pretty good thread that might be of interest. It occasionally gets off track, but worth reading...especially page 7 onward. You might consider getting in touch with one of the very knowledgeable posters - JBL Guy...I would think that he might be able to give you useful input as regards the logic you build into your simulator.

http://www.audiokarma.org/forums/showthread.php?t=555539&page=7

For the record, I did not say that "data" is unnecessary.

What I said was that clipped signals from the bass end, rerouted to tweeters, is not necessary to blow tweeters. You have enough ability to blow tweeters with un-clipped high end amplification. That is because, once bass clipping is reached, as you continue raise the volume control, "actual" high frequencies "are" further amplified, and "do" go to the tweeters (by design). The bottom end is clip-limited, so the high end amplification is less evident.

The energy distribution between low and high end differs by orders of maginitude. The bass may be clipped, but the highs are not, and can be further amplified, to the point of blowing tweeters.

Power distribution to speakers can be conservatively anticipated to have high frequencies at levels at least 15 db down from power levels sent to woofers. That is quite a bit, but it is overly conservative to illustrate the sufficiency of the principle I am explaining. Amps are designed to provide flat response across the entire audio spectrum. That means that signals to tweeters "can" be amplified to the output level of the amp (power normally sent to woofers). Real music and natural sound is not composed flat, hence tweeter voice coils can be finer than a human hair, while woofer voice coils are of significant gauges. Underpowered amps have no problem amplifying high frequencies to levels that will fry tweeter voice coils, without any regard whatsoever to what is happening to low frequency output levels (that means "clipping notwithstanding"). A 20 watt amp is able to produce 20 watt high frequency signals. I know of no tweeter designed to take 20 watts. Look up the energy distribution, across the audio spectrum, for natural music and sound. If you don't know what I am talking about, educate yourself. Then come back and spit at me.

I have been repairing tweeters blown by under-powered amps for over 30 years. I can also expain why they blow. I can blow them right in front of you, and let you watch it happen.

If you are going to dispute a statement, make sure you are disputing the correct statement. Underpowered amps can blow tweeters, and they do blow tweeters, but via a different mechanism than previously believed, by myself and many others.

Rich P

I agree with Rich P post #7 (and the countless others) about an Overdriven, Underpowered amp damaging speakers. No it is not a myth. And no, I don't use is it as an excuse to promote the use of high power amps. I do promote keeping whatever amp you chose, whether it 20 wpc or 200 wpc out of clipping, and that you chose one within the recomended input power range recomended by speaker manufacturer.

This discription of clipping and it's effects on a speaker from Wikipedia explains it better than I could. It's reffering to an oscilloscop tracing.

"Difference between clipped and maximum unclipped waveforms. Because the clipped waveform has more area underneath it than the smaller maximum unclipped waveform, the amplifier produces more output power. (See the waveform to the right for an example.) This extra power can cause damage to loudspeaker components, including the woofer, tweeter, or crossover, via overheating.

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".

Here: http://sound.westhost.com/tweeters.htm

Read it, and then go argue with the author.

Rich P

pustelniakr said:

For the record, I did not say that "data" is unnecessary.

What I said was that clipped signals from the bass end, rerouted to tweeters, is not necessary to blow tweeters. You have enough ability to blow tweeters with un-clipped high end amplification. That is because, once bass clipping is reached, as you continue raise the volume control, "actual" high frequencies "are" further amplified, and "do" go to the tweeters (by design). The bottom end is clip-limited, so the high end amplification is less evident.

Click to expand...

Oops, my bad, I thought the word 'data' was the 'it' referred to in the last sentance. Either way, it was a good incentive to do this thought experiment.

Just curious, did you see your name in quoted text in my post, see red, stop reading, and bang out a reply? Had you read what I wrote, you'd see that I am in complete agreement with you, and have supplied the math to back it up. In a short summary statement, there is very little tweeter energy created by clipping a non-tweeter signal.

pustelniakr said:

The energy distribution between low and high end differs by orders of maginitude. The bass may be clipped, but the highs are not, and can be further amplified, to the point of blowing tweeters.

Power distribution to speakers can be conservatively anticipated to have high frequencies at levels at least 15 db down from power levels sent to woofers. .... Look up the energy distribution, across the audio spectrum, for natural music and sound. If you don't know what I am talking about, educate yourself. Then come back and spit at me.

Click to expand...

Luckily, I've been an active observer of the spectrum of music since the late 70s, so I've got an idea of what you speak. Yet, I have no intention of spitting at you, since your understanding seems to be in line with my understanding of the topic you're currently bringing up. Still, this is all unrelated to the point of the original post.

pustelniakr said:

I have been repairing tweeters blown by under-powered amps for over 30 years. I can also expain why they blow. I can blow them right in front of you, and let you watch it happen.

Click to expand...

Wow, that's really cool. But, it has nothing to do with my post. The point was not the all-too-typical AK thread underpowered amps blowing tweeters due to clipping. You may have noticed I mentioned at the end that clipping power is small.

pustelniakr said:

If you are going to dispute a statement, make sure you are disputing the correct statement. Underpowered amps can blow tweeters, and they do blow tweeters, but via a different mechanism than previously believed, by myself and many others.

Rich P

Click to expand...

And which statement do you feel I'm disputing? To be clear for you, I'm suggesting the power delivered to a tweeter from a woofer signal being driving into clipping is not very much power. Just as you say, low power amps blow tweeters via a different mechanism than previously believed. The point of this post is to refute the previously believed mechanism (clipping as the culprit), and to provide data to support such a claim. Do you really think the only reason people post on Internet forums is to argue and fight? There may actually be people out there in the world who are supportive as opposed to confrontational. I really thought I was providing you the hard math to support that clipping, by itself, it not enough to fry midranges and tweeters. Just as you had asked for.

charles 1973 said:

I agree with Rich P post #7 (and the countless others) about an Overdriven, Underpowered amp damaging speakers. No it is not a myth. And no, I don't use is it as an excuse to promote the use of high power amps. I do promote keeping whatever amp you chose, whether it 20 wpc or 200 wpc out of clipping, and that you chose one within the recomended input power range recomended by speaker manufacturer.

Click to expand...

Again, the point here is not to claim it is impossible for low power amps to destroy any speaker component. Rather, that the extra distortion energy produced in a clipped waveform doesn't amount to a lot of power. The tweeter will most likely seem more power from the music than it will from a clipped woofer signal.

charles 1973 said:

This discription of clipping and it's effects on a speaker from Wikipedia explains it better than I could. It's reffering to an oscilloscop tracing.

"Difference between clipped and maximum unclipped waveforms. Because the clipped waveform has more area underneath it than the smaller maximum unclipped waveform, the amplifier produces more output power. (See the waveform to the right for an example.) This extra power can cause damage to loudspeaker components, including the woofer, tweeter, or crossover, via overheating.

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".

Click to expand...

Now that I've actually looked at the amount of real power produced by a clipped signal, I'm believing that this Wiki quote is just perpetuating the myth. Like I said in the OP, I'm not the ultimate guru on this topic, just stating my beliefs.

This is an interesting topic and an interesting discussion. I for one would really like to know the exact mechanism of how clipping blows tweeters etc, apparently more often than straightforward over driving a particular speaker?... is this even correct?

I am still confused about whether the (now apparently discredited) reason for tweeter blowing really is attributable especially to misuse of low powered amps, or another mechanism? I seems to me that clipping at one frequency may be a lot more benign than the clipping artifacts produced by music, into an over driven amplifier of any power rating.

I hope I am not alone in this desire for the truth.

I for one would really like to know the exact mechanism of how clipping blows tweeters etc,

Click to expand...

Depends on the specific amplifier, and degree of clipping. Fundamental theory is nice. But, the real world doesn't always work that way.

Just one of the as of yet un-mentioned mechanisms across multiple threads would be the poorly/mimimally designed amplifier sections(there are too many to list) that respond to hard clipping by going into high frequency oscillation.

Somewhere there is a discussion at Rane. The general upshot was that it isn't the clipping that usually destroys tweeters, but that when an amplifier is turned up too loud, the high frequency content is increasingly amplified while the volume doesn't increas. For example, with a fifty watt amplifier, the treble content is maybe a few watts. If the amplifer is very over driven, the treble content would increas to 20 or so watts (when the amplifier doesn't have a bass note to clip) and this increase in energy is too much for most tweeters - i.e. its when the amplifier isn't clipping the bass notes that the tweeters get destroyed.

To put it another way, Unclipped musical peaks of 100 wpc might produce 10 wpc continuous RMS. In an attempt to get the same SPL's a 50 wpc amp driven into clipping might produce the same 10 wpc RMS as the 100 wpc amp produces. Add to that the increased energy of the High frequiency harmonics going to the tweeter, and this might blow a tweeter that was already near it's maximum power dissipation level, then agian it might not. Most speaker designers build them to be "Idiot Proof" to some degree. But some, like my Energy Pro-22's aren't very tolerant of abuse.

The problem lies with trying to get more out of the 50 wpc amp than most of us are likely to try. Often by the point of dangerous clipping we have either reached a satisfactory SPL, or we will turn the amp back before we get too far into clipping. However IF we turn the amp well into clipping, that 50 wpc amp might send 25 w RMS into the speaker, again with more of that energy going the tweeter. This is what fries tweeters. And the 100 wpc amp of course, driven that far into to clipping will fry the tweeter almost instantly. The theory is, that extra 50 wpc and 3 db of SPL discourages us from having to drive it into clipping.

I read an article over 30 years ago which warned about destroying tweeters by clipping underpowered amplifiers.

The gist of the article(as I recall it) was that when an amp clips, it produces triangular waveforms which have an extensive high frequency content. It is that high frequency content which takes out the tweeter.

https://en.wikipedia.org/wiki/Triangle_wave

That's what I recall.

skriefal said:

Ken Kantor (of NHT) has posted here on AK on this topic, and IIRC he is in agreement with your findings. i.e. A clipping low-powered amp is unlikely to damage a tweeter.

Click to expand...

I've blown literally a dozen tweeters since being involved in this hobby. Every time, I was using relative low power amps from 30-100 W/ch. Since the mid eighties, I've only used relatively high power amps from 200-500 w/ch and have NEVER blown a tweeter even at ridiculous listening levels. The worse tweeter killer I owned was the Pioneer SA-1040 integrated amp. It alone ate 6 tweeters on my Polk SDA-2s.

IMO ANY hard clipping amp is potentially dangerous to tweeters and less capable amps are more easily driven into clipping under normal circumstances.

IMO it's gamma rays reflecting from the longitudinal section of the second harmonic that blow tweeters.

My brain hurts!

Don't spend time trying to do any critical thinking or research...just have an opinion and stick to it! Much easier, and no brain pain.