Power Handling

[rrcrain]

I located what appears to be a few decent explanations of amplifier clipping. Any comments or feedback on these authors articles?

http://www.st-andrews.ac.uk/~www_pa/Scots_...ping/page1.html

http://www.rane.com/note128.html

What neither author addresses is the difference in how solid state and tube amps clip.

[kkantor]

Thanks! Good work!

The thing is, like so much in audio, people put all this work into testing and analyzing various urban legend, without ever doing a proper experiment to validate the basic assmuption. Wouldn't it be interesting to see if small amps REALLY DO blow out tweeters, or if people just ASSUME they do.

I already cited the Rane note above. The St. Andrews work was well done, and agree with it. The only problem was that while it totally disproved the idea that clipping damages tweeters (at least via currently proposed mechanisms), it was a bit political in the end about actually coming out and saying that in so many words.

In regard to my first paragraph, there are several million speakers of my own design currently in field service, and for these I have good knowledge of their service records and failure modes. They range from Best Buy and CompUSA multimedia products to $20K+ systems. I also have borne technical oversight for many times more. I've done thousands of tweeter autopsies, so I am not speaking from a contrarian theoretical perspective.

My experience tells me that clipping is almost never the cause of tweeter failure. So I was faced with math on one hand, and practical experience on the other, and it became impossible for me to keep believing the anecdotes and easy explainations.

[Guest Bret]

>My experience tells me that clipping is almost never the cause of tweeter failure.<

Ken, you hedged!

As I understand this: Power equals Volts times Amps. I understand that transistors "leak" current at high temperatures (thermal runaway). Perhaps as the transistors get hotter they "leak" high frequency current. Or maybe the additional high frequency current has another, more likely origin. . .

Perhaps when a tweeter gets warmer and warmer by being driven hard it draws double the current for every 10 degree C rise in temperature (should be true, right?). For every double in current there is a doubling of "power" even at a constant voltage. So perhaps it is *current* being "drawn" by the over-warm tweeter that is doubling (or more) the Power running through it. So even though the voltage (and therefore amplitude?) of the input signal has not changed the "wattage" of the signal has doubled?

So in an 80w speaker system with a 10:1 LF/HF distribution, a 10w tweeter, being driven at 7w, which has a rise in temp of 10 degrees C would pull twice the current effectively doubling the Power or wattage to 14w which is beyond its capacity and would melt the voice coil. The average 7w of power to the tweeter would only happen if ~70w average was being asked of the 80w amplifier meaning that the amplifier probably IS clipping very, very badly.

That would seem to be a mechanism by-which the tweeter could melt that would coincide with amplifier clipping but would not be caused by amplifier clipping. Clipping might not be the culprit, it would only be a likely bedfellow.

That would explain why ferrofluid in the gap, cooling the tweeter, would raise the effective power-handling. By preventing the tweeter's temperature from rising dramatically it would prevent additional current draw at any given voltage.

Hey, you guys have this education and know-how, I'm just guessing. If that's not plausible, tell me so.

Maybe I am looking for mayhem in all the wrong places.

Bret

[Guest Sean]

Ken: I don't believe that computer simulations for something like this are very accurate. That is because of all of the variables involved. You would literally have to test a specific speaker with a specific amp under specific conditions to know exactly what to expect. Anything else would be strictly a generalization.

First of all, clipping a test tone and clipping a musical passage are very different things. Test tones are typically singular or at least "simple" in nature and have a given peak to average ratio. Musical passages are complex sums of multiple frequencies with very different peak to average ratios at any given time or frequency. This is besides the fact that clipping in one frequency range can effect power delivery in another frequency range simultaneously.

Secondly, not all amps clip or recover from clipping in the same manner. Changing the amp and / or type of signal being processed and / or load can give you completely different results.

Thirdly, speakers are very complex loads. They do not maintain the same levels of resistance, inductance, capacitance, phase angles aka "complex reactances", etc... as frequencies are varied. As such, the load that an amplifier sees at clipping in one frequency range can be very different / contrast quite radically with what it sees in a different frequency range. If you don't think that one factor effects the other, think again. Combine this with the multitudes of frequencies, amplitudes and waveforms of the signals being processed with the variances that can be found amongst amplifiers and how they overload and you've got a completely unpredictable ball of wax.

With all of that in mind, i have to question ANY "simulation" and the accuracy of the results obtained running such a simulation. Sean

>

[kkantor]

Sean, hang on... I'm not talking about "simulations" of real speakers. I simply asked you to propose and describe mathematically some mechanism by which tweeters are be blown more readily by a clipped amp than by a higher powered amp. No need to read anything more into that simple and straightforward request.

You started to talk about harmonic structures, but you backed off before you gave any numbers or equations I could look at.

[kkantor]

Bret,

In all due respect, you seem to be trying to beat a dead horse, and your technical comments are, well, bizarre. For example, leakage current is not even a term typically applied to transistors. And hotter tweeters tend to draw less current.

Please, just drop it. Either you believe me that clipping amps were never the problem they were made out to be, or you should conduct a few experiments to prove (or disprove) it to yourself, and report back with data. The last thing the discussion needs to to have a bunch of spurious technical-sounding idea-lets thrown up against the wall to see what sticks.

Infinite numbers of techno-fairtales may be postulated without so much a single correct assumption underlying them, and it is possible to "win" by innundation. So I'm out of here. This whole thread has sunk into the muck of uninformed audio hobbyist debate, where belief structure is more important than facts.

[Guest Bret]

Look, Ken, you are confused about my intention. Of course my comments were bizarre, they were never intended as a hypothesis (that being an *educated* guess). There was no place to go but the Twilight Zone after I read the following:

“In general, therefore, it seems quite plausible to assume that damage may sometimes occur due to any one of the following mechanisms, or various combination of them:

HF creation by clipping of the LF.

Rise in the HF level of the original signal which – although limited by clipping by an amount similar to that which occurs at LF – rises to a high level.

Leakage of LF power.

Non-themal [sic] effects due to an increase in the peak displacement or accelerations.”

From the article linked-to by rrcrain (thank you):

http://www.st-andrews.ac.uk/~www_pa/Scots_...ping/page1.html

Now, with all due respect to you, you have asserted that amplifier clipping is *not* the cause, nay-saying cause #1, above. Fine.

This article asserts that cause #2 (compression) isn’t as pronounced in practice as assumed in the rane article:

“Indeed, one striking feature of the above is that the change in the relative levels at HF/LF is surprisingly small given that we are clipping around 35% of the time and that the overall compression is more than 3 dB – i.e. the mean power is being held to less than half the value it would have in the absence of clipping. In principle we could expect both harmonic generation due to clipping and any “compression” of alternating LF/HF sections to result in a rise in the relative HF power level. However in this example any such result seems quite modest. “ Fine.

#3 above. . . I haven’t a clue how that works. But what I do see is that this requires high power. Well, given enough power it’s entirely possible that we just “blew” the tweeter by giving it too much power even if that power is unaffected by clipping, compression, or cross-over leakage. Fine.

#4, Non-thermal effects don’t explain the melted wire or the smell. Fine.

Evidently I’m too stubborn to understand what killed the dead horse you say I’m whipping. I have yet to see a “generally accepted” cause-and-effect explanation.

My purpose in proposing that something “strange and unusual” is happening is merely my acceptance of your observation about clipping as gospel. In fact, it is precisely my lack of understanding of how a speaker (as a motor) might react that lead me to think, “Well, fine, maybe resistance doesn’t increase with temp.”

As far as this discussion goes, I have yet to see anything definitive. Just as I think “Ah ha!” regarding compression, along comes someone with some math that says, “Probably not.”

So, if we’ve reached a conclusion I missed it.

>The last thing the discussion needs to to have a bunch of spurious technical-sounding idea-lets thrown up against the wall to see what sticks. <

Well, it stills needs something. You may have proven to yourself and us that clipping is not responsible for blown tweeters, but we don’t seem to have said (with any degree of certainty) what does.

>This whole thread has sunk into the muck of uninformed audio hobbyist debate, where belief structure is more important than facts.<

I have no clue what you are talking about. I haven’t seen anyone assert a belief as in “Does-too,” “Oh yeah? Does not!”

But rather than be the non-technical fool you suffer not-gladly, I’ll shut up.

Bret

[kkantor]

?????????????????????????

We all know what causes tweeters to burn out: being over-driven.

The ways tweeters overheat, the ways crossovers are suboptimal, the way large excursions come into play, all this is extrememly well researched, widely published, etc. This is what speaker designers learn in Jr. High.

None of this has to do with clipping. It has to do with being played too loud, pathological musical spectra, drivers and crossovers used improperly, etc.

Really, please absorb this, or there is little further basis for you and I to talk about this particular thread. Just go and search the AES or ALMA sites, or go to the www.uspto.gov and do some seaching.

[Guest Sean]

Ken, as i mentioned, there is NO way to "do the math" on something like this other than to do actual testing of the devices ( amps & speakers ) in use. The formula's that we have for "doing the math" are all based on symmetrical sine waves or square waves. Dynamic music waveforms that aren't distorted respond like neither of these. Nor do amps driving reactive loads that vary with frequency respond like a tone generator driving a dummy load.

Due to all of the design variables involved with the actual components ( amp & speakers ) and the intensity / duration of clipping, it would be possible to come up with anything from mild to wild responses. I think that this is why AR specifically mentions "no more than 10% clipping" in their power ratings. That's because a higher percentage of clipping will not only expose the speaker to a higher average power level, but that higher average power level would also be maintained for a longer duration of time and a wider bandwidth.

If you want to see what i'm talking about, take a look at peak and average power levels involved with a "symmetrical" sine wave and "linear" square wave. You can do this at any given frequency. Both should be set to crest at the same voltage. While comparing the average "heating" power involved with both waveforms in the time domain, look at these signals in the frequency domain too. Not only is the "clipped" waveform measurably more "potent" in terms of the amount of average or "heating" power generated for the same amount of time, that power is much wider in bandwidth due to all of the harmonics generated. This boils down to a single simple waveform versus multiple complex waveforms.

For those that don't know formulas, the average heating power of a sine wave is 25% of the peak envelope power generated. Due to the tapered shape of the waveform, the intensity of the power transfered is neither consistent or efficient. To top it off, you also only have one frequency that the power is being delivered at i.e. it is a simple waveform with low THD.

On the other hand, the square wave's average power is the same as it's peak value. That's because the peak is the average i.e. they both occur at the "flat plateau". This plateau lasts the entire duration of the note whereas on a sine, the value tapers as the voltage rises. On top of that, whereas the sine wave slowly ramps up to its' peak potential, the square wave is instantly on at full potential i.e. perfect rise time. This gives the square wave a much longer duration to transfer that energy i.e. no taper at the front or rear and consistent amplitude for the entire duration.

Besides all of that, the harmonics generated out of a severely clipped signal aka "flat topped" signal produces output signals that are 10 - 15 times the bandwidth of the primary frequency. It is the complex sum of these harmonics that give the "flat topped" square wave its' actual shape. If these multiple harmonics didn't exist, the leading edge of the "square wave" would begin to round, making it less distorted and narrower bandwidth.

In many cases, heavy clipping can increase the average power generated by anywhere from 3 to 10 dB's. This is why AR plainly stated their "fear" of severely overdriven amplifiers "cooking" their speakers in their manuals. This is further discussed regarding the amount of "compression" ( which is nothing more than "controlled clipping" ) used when recording "rock" music vs "classical" music. In their own literature, AR makes mention of a typical "rock" recording requiring 10 times the average power that a typical classical recording does.

As such, it's not the peak's that get ya. Whether derived from sheer brute force ( MEGA power capacity with phenomenally large peaks on a repeated basis ) or heavy sustained clipping ( very low peak to average ratio ), the generation of heat is what will do any driver in. The driver simply can't dissipate the energy fast enough, so they end up "melting down". What typically melts is the glue that bonds the voice coil to the coil former, the enamel that insulates one turn on the former from the next or the glue that bonds the voice coil former to the cone and / or spider. In some cases, you can have actual ignition aka "fire" spitting out of the driver if there is enough internal heat built up fast enough. If you're not driving the tar out of the amplifier and this takes place, time to check the amp for DC voltage leakage. Even very small amounts of DC applied to a speaker can do a LOT of damage in a very short period of time. Sean

>