Power Handling

[soundminded]

Now that I am awake and last night's wine has worn off, I remember what the voltage and current versus time of an RC circuit connected to DC looks like. Upon applying voltage, all of the voltage drop is across the resistor. This happens because the capacitor has no charge built up on it. The curent through the capacitor in the dielectric is called "displacement current" which is the charges magrating towards the plates. As charge builds up on the plates, the voltage across the resistor and the current in the circuit falls asymptotically to zero. The voltage across the capacitor rises asymptotically to the battery voltage. The rate depends on the RC time constant which in turn depends on the values of the capacitor and resistor. (BTW, many timing circuits depend on this principle where an adjustable resistor adjusts the time constant and a triggering device such as an SCR or triac switches another circuit.) Since the voltage across the resistor is the battery voltage instantaneously, the instantaneous power delivered to the resistor is V(BB) squared divided by the resistance of the resistor. For a 12 volt battery and an 8 ohm tweeter, the initial instantaneous power is therefore 12*12/8= 18 watts. For a 24 volt battery connected to a 4 ohm tweeter, the initial instantaneous power dissipation is 144 watts.

Will the tweeter fail. Like any fuse element, the tweeter voice coil has a maximum time current curve. If the power dissipation curve based on the RC time constant and the battery voltage crosses the maximum time current curve for the tweeter voice coil, it will melt. If it doesn't, it will survive. The tweeter could also fail if the power applied causes the initial cone excursion to exceed the tweeter's capacity.

That's how I remember it in the daylight.

[kkantor]

You remember it correctly. The peak voltage exists for an infinitesimal moment, and decays exponentially.

Generally, it is more illuminating to look at these things in the frequency domain, not the time domain. For example, how do you extend your analysis to the real world, where speakers are complex reactive loads, and crossovers contain more than single caps?

[kkantor]

There are some self-contradictions in your description of "compression" and what it means. I leave it to you to analyze the situation as you see fit. To me, the point of the paper was to encourage the use of (Rane) compressors to reduce the effective power of the amp to a level appropriate to the speakers.

You are exactly correct when you say that an amp that will provide 30 Watts to the tweeter will blow the tweeter. At least with a small amp, you have the warning sign of clipping, if you choose to heed it. THE SAME input and gain that will blow the tweeter with the small amp with do it with the big amp.

But I grow weary of saying this. It is pretty much established as a fact these days, but I had no true hope of changing anyone's opinion. All I wanted to do was present a case, and some data, which I hope to supplement. Then everyone can make what they want of it.

[kkantor]

I hooked up a high voltage supply (HP6515A), and set it for 1000 Volts DC. I connected it in series with 1 uF film cap, then shorted it directly to a cheap 1" metal dome tweeter; car audio type, Taiwanese, 3.3 Ohms DCR.

There was a slight click.

I discharged the cap, and repeated this 3 more times over the course of about a minute. No discernable damage to the tweeter.

Granted this supply will limit quickly at 10 mA, and so can only sustain 1000V for a few uS. So I repeated the process with a 500V, 0.5amp supply, with an extra 2 uF cap on across its output to provide a current burst. Same tweeter. More of a "pop" sound. Still no damage. In this last case, I estimate the peak current to the tweeter as being over 50 Amps, repeated several times.

As I said, V's and I's don't kill tweeters. Heat does.

[roundhome]

Hello Mr Kantor

Here is one for you i,m running four AR 9 tweeters in a set up just like the LST but using a DBX driverack pa electronic crossover so there is nothing between the amp and the speekers but a FNM2 fuse. On the other drivers there is a FNM4.

The amp is a pioneer spec four 180 watt into 4 ohm it is one of four in the system. i'm not virtical byamping one amp for the tweeters one for the mids one for the 8 inch woofer and a 250 watt for the woofers all matched amps. .005% intermodulation distortion at 1 watt. .001% harmonic distortion at 1 watt. these are bob carver amps.

The question is will all four tweeters handle 100 watts continues? AT this level the amps are no were nere clipping.

The crossover is a 24 db The dbx has a mic and a spectrom built in

Thank you

Jim

Ps i;m ready to start doing the test on the LST crossover if you could step me through the test that need to be done please.

[Guest Bret]

>THE SAME input and gain that will blow the tweeter with the small amp with do it with the big amp. <

Ken, my friend (I hope you don't find being referred-to as "friend" offensive even though we've never met), please take a deep breath.

Never once did I miss the point of what you were saying and I, for one, deeply appreciate your illucidation of what *really* is happening.

I never had a "need to know" at this level and so remained ignorant until this thread enlightened me.

What I'm probably doing that makes you think I'm not convinced of the truth of what you say is overlaying what this means, as a practical matter, to the guy in the trenches, on the sales floor. I'm sure the perspective is quite different in the designers' world. The facts are the same, it is simply the application of those facts that we have differing perspectives on. Your facts are not in question; I am completely convinced.

I hereby acknowledge the absolute correctness of everything you have said, including "clipping doesn't destroy tweeters" and "it is easier to blow a tweeter using a large amp than a small one," and "neither a large nor small amplifier better 'protects' tweeters." I accept every bit of that as fact.

Hang with me for just a second pretending to be the sales-guy on the floor at. . . Billy-Bob's Audio Emporium; Because many people like to drive their systems very hard, many, many times a "too small" amplifier will not get their speakers as loud as they want them to go. I'm talking frat-guys, not audiophiles. A great big honkin' Ampzilla, on the other hand, would get plenty loud.

So in practice, selling Joe Fratguy an inadequate amplifier with his speakers meant a greater likelihood that you would see those speakers back in your shop with the tweeters blown. The cause of this is no longer in question, you have explained it perfectly. Joe Fratguy has no clue about clipping, he just knows cold beer when he smells it. Compression is something that happens somewhere in his car's engine.

I completely understand and accept that if you are trying to blow a tweeter it is more easily done with a large amp than a small amp.

Until now I assumed that it was because there was a DC component to the clipped waveform. I have been educated and I appreciate your efforts as well as that of the others on this forum, even if I don't seem to be agreeing well.

At this point it "feels" like you are "ready" to be outdone with me. But what is probably happening is what often happens with me; My meaning gets lost in all the words.

I'm turning-off my grovel-mode now. I've never known someone so difficult to agree with!! :-)

Bret

[kkantor]

Sorry if I came off too strong. This is my idea of fun. Otherwise, why would I be doing it? Anything that gives me an excuse to connect 1000V to a tweeter is worth it.

OK, so now you are convinced. Your mission is to convince two others before you shuffle off this mortal coil.

Or not.

[Guest Bret]

>Anything that gives me an excuse to connect 1000V to a tweeter is worth it. <

LOL; love it. Yeah, that was a cool experiment. And pretty bal. . . bold of you if you were in the same room with it. Gotta like a man with the strength of character to live his convictions.

This is real, true, story. Really. Once upon a time I had a college friend who had once gotten across a spark plug wire of a running automobile while he leaned across its fender. Not understanding the situation very well, he mistakenly translated this into "a car battery will knock you on your posterior." I corrected him. His disbelief ran deep. The hood was up on his non-running Mustang II and another friend was inside the car goofing off with the radio or something.

To properly demonstrate that I wouldn't draw enough current to hurt me, I reached over and grabbed both terminals on the battery. At that moment the friend inside the car decided to get out and while dismounting managed to punch the horn button with his elbow.

I was not electrocuted. I did, however, leave an impressive hair and tissue sample on the hood catch. Everyone in the general vicinity was, of course, delighted to the point of nearly wetting themselves. For several minutes. Several long minutes. I've had shorter weeks.

I have been careful about practical demonstrations since.

Now convince two others? Well, I've already started on one. I can complete this assignment unless I permanently run down the curtain tonight.

Bret

[kkantor]

1- With pink noise or music signals, 4 tweeters together will easily handle 100W "continuous" power. However, they would not be invulnerable under all conditions. For example, there are some frequencies where a constant sinewave might cause damage. I wouldn't worry about it, though.

2- You might want to put a high quality blocking capacitor between the amp output and each tweeter, to supplement the fuse in case of transient overload. Something in the range of 100uF, 100V should do the trick.

3- Honestly, it would be a total pain in the rear to try to completely explain the testing of the transformer, step by step! It would take pages and pages of writing and I would probably forget all kinds of little details. If you aren't already versed in doing this kind of measurement, do you have a friend who can help? Or, send me the transformer and I'll test it. By the time I try and write down the connections and settings for sources and loads and levels for equipment I can't even see, it will be hours.

If you already know this stuff, and I am misjudging your experience, maybe you can ask more specific questions?

[ninohernes]

This has been a very interesting thread. For the record, I have never blown anything before. I play my AR-2's loudly in my recording studio with an Adcom 125watt per channel amp, which is very clean, and I have never had any problems. The only thing that has happened, is the woofer has reached its maximum excursion, making a very loud "crack" noise, but this has only happened once. The power handling of the AR-2 is great. The woofer is great with lots of power, and since it has two tweeters, the system is even more thermaly stable. (The woofer in my AR-2's have a large alnico magnet, looks just like the one in the AR-3, except a bit smaller.) I did Kens test, placing my hand over the drivers after playing loudly, and I couldent feel any heat. The big magnets do a nice job of drawing heat away. AR speakers have always preformed very well for me, under what I consider "rigorus" conditions. A lot of the material that I play through my AR-2's is umcompressed, like when I am mixing multitracks. I would like to get a new amp, a Bryston amp. I listened to these at the AES convention, and loved them. But being a senior in high school, I dont have the money (saving for college is enough!) and my Adcom seems to be just fine.

I have also never had any trouble with my AR-3's or my 4x's. I have used my 4x's with my Adcom without any trouble, but they normally reside in my protable system with an Onkyo 50 watt per channel receiver, which produces clean power.

[soundminded]

I=heat. IR heating is due to the resistive properties of the voice coil. Frequency Response analysis is much more useful for analyzing steady state response of periodic waveforms and for applying it to and understanding the response to non periodic waveforms than it is for transient analysis. In this case we have a single transient plus a steady state waveform at zero hz. Because any ideal capacitor will completely block dc current, only a transient analysis would be particularly useful here.

This case in many ways is analogous to other one time transient electrical events such as the startup current in a motor in series with a fuse or other overload device. The motor is analagous to the capacitor and the fuse is analogous to the tweeter voice coil. The current in the circuit is high initially until the motor builds up speed creating back emf to limit it. The motor on startup has a time/current curve. The fuse, circuit breaker, or thermal overload has a time current curve too. If the time current for the motor crosses the time current curve for the fuse to the right at any point, the fuse will blow before the motor reaches full speed. The selection of overloads in motor starters to allow motors to come up to speed without blowing but blow on an upset fault in the motor is therefore critical.

Here's another example. This is an important concept for power distribution engineers. One of their functions is to perform coordination studies which looks at the comparative time current trip curves for different overload protection devices in a system to be sure that a fault will take out the circuit breaker or fuse furthest down on the system limiting the disruption as greatly as possible. In a badly uncoordinated building in a company I once worked for, a defective coffee urn took out half a building. Properly selecting fuse types and adjusting the instantaneous trip settings on circuit breakers is critical for proper coordination.

If you increase the capacitance sufficiently, you will increase the RC time constant and eventually the tweeter should blow assuming that the battery or power supply can generate enough current not to be limited by its own internal impedence. A bench supply or a series of 9 volt batteries may not be able to but an automobile or marine battery will as even a modest car battery can deliver hundreds of amps.

[Guest dogmeninreno]

>Here's a good analysis of the situation:

>

>http://www.rane.com/pdf/note128.pdf

>

>

Ken, That link provides about the best indication to this thread ending. I will vouch for the comments about relay coordination in the power distribution field since I was in it for 35 years and am a E.E. "retired". The relatively small scope of a loudspeaker and it's reaction of component drivers or impedence is very complex as is a power grid protective relay design. The difference as I see it is the costs associated with a power system protection and a loudspeaker. I may be off base but I think the discussions are worthwile but without tests, hard to prove the individal points. Dale in Reno.....

[Guest Brian_D]

Don't get too sucked in...

Rane is trying to sell product as were the proponents of the theory that Rane is attempting to dispel.

Tom, in answer to your question about type of amplifier, the lower power amp was a Yamaha Class A/B rated at 100 watts per chanel in to 4 ohms. The replacement were Mackie 800's rated at 200 watts per chanel. We always used the Yamaha in A/B mode, it would trip 15 amp breakers within 10 minutes in class A. (But it sounded great!)

We bought 4 of those Mackie amps, they're extremely versatile. We use one per CV earthquake, (Bridged mono to 4 ohms = 800 watts) and one per pair of full-range cabinets.

Tweeter failure was initialy blamed on a faulty crossover component, but after that component was replaced we still blew two tweeters. We havn't blown one since we switched to the Mackies.

-Brian

[Guest Bret]

>Don't get too sucked in...<

I dunno. . . I'm pretty sucked-in, not by the idea of limiting amplifiers to end tweeter destruction, but by the reasons that were given for why tweeters blow after an amplifier clips.

(Okay, Ken, or before given the same input.)

Seems to me this is dangerously close to what Soundminded said and explains exactly what anecdotal evidence suggests. It also explains why an amplifier sounds as it does when it goes into compression and why that's so audible; There's a big change in the audible spectrum being produced.

The "purposefull slowing" of the clipping indicators or protection circuits also seems to mirror my personal experience of driving amplifiers too hard. I swear I can hear when an amplifier goes into compression long, long before peak meters or clipping indicators tell me the amp is clipping. (or it may be that I'm hearing the power supply or maybe even the Romex running-out of juice)

It also seems to confirm the practical experience many of us have had, even if we can't quite prove why.

Occam's Razor looks to be striking again.

[kkantor]

OK, this is all pretty simple and straightforward. Are you making a point about power handling that I am missing?

(Other than the very obvious point that enlarging the cap increases the danger to the tweeter.)

[kkantor]

I am not an objectivist or a logical positivist, so forgive me:

People experience and believe many things that they cannot prove, or even defend. Often this happens when they are primed by expectation and explaination.

Hey, we all want to be different, above all that. But who is?

[rrcrain]

>If you increase the capacitance sufficiently, you will

>increase the RC time constant and eventually the tweeter

>should blow assuming that the battery or power supply can

>generate enough current not to be limited by its own internal

>impedence. A bench supply or a series of 9 volt batteries may

>not be able to but an automobile or marine battery will as

>even a modest car battery can deliver hundreds of amps.

If you increase the capacitence value in the crossover network, it's no longer tuned for a tweeter. Also, a nine volt battery connected directly to the terminals of a tweeter voice coil is very capable of overheating the coil.

[soundminded]

My only point is that the risk of damaging the tweeter with DC depends entirely on the circumstances. But I consider it a real possibility. Probably purely academic though. On the other hand, think up a scenario preposterously stupid and sooner or later someone will try just that, or something much worse you never even dreamt of.

[kkantor]

OK, Agreed.

However, my position is both theoretical and empirical. (I'm a loudspeaker designer by trade.) When repeated applications of 1000V DC steps through a 1 uF cap into a cheap tweeter do no harm, I think it is fair to say that one must postulate a pretty extreme circumstance wherein "DC" harms a tweeter.

[Guest Sean]

Momentary transients of high amplitude and high levels of harmonic energy that are of an extended duration are not the same thing. The tests that you conducted are therefore legit for some purposes, but not for the demonstration purposes that you are using them for here.

The primary notes may be "high" in amplitude, but their duration and harmonic structure are limited by the actual notes played. When clipping takes place, the amplifier is not only saturated, but both the intensity and duration of the primary note and phenomenally high level of harmonics generated are increased in duration. The bottom line is that the RMS or "heating power" of a clipped signal is MUCH higher than a signal of equal peak power applied.

As a side note, a decent amp doesn't "pass DC" when clipping. "Flat topping" and "passing DC" are two different things. Flat topping aka "square waving" simply raises the RMS value of the signal drastically. As Ken alluded to in a previous post, but for the wrong reasons, it is the heating or rms power that blows tweeters. Clipping increases the average power applied to the tweeters, the spectrum that the energy is applied to and the duration that it is applied for. When it comes to tweeters with a small voice coil and limited heat dissipation, more power over a wider bandwidth for a longer duration of time = smoke Sean

>

PS... Much of my livelihood comes from MAKING circuitry "clip". I get to view all kinds of distorted waveforms for 8 - 12 hours a day, 5 - 6 days a week.

[kkantor]

Sean,

As a man who hates vague generalizations, perhaps you can enlighten me with some specifics about the following quotation of yours:

"Clipping increases the average power applied to the tweeters, the spectrum that the energy is applied to and the duration that it is applied for."

OK, but, just exactly how much does a certain amount of clipping increase the average power to the tweeters? No need to educate me about the principles involved; I know them. No need to cover every case. I just want to see some hard numbers.

No baby stuff about "odd harmonics" or "DC" or "saturation." Show me the facts. I want to see exactly how much extra power is sent to the tweeter in a hypothetical case of music being clipped. Pretend I am playing typical music at an average power of 100W, and the peaks of the music are just at the 50V rails. The tweeter is getting 20W of average power.

Now turn up the input voltage by a factor of 1.5. Terrible clipping results. How much power is now going to the tweeter??? We know that if the amp was big enough for there to be no clipping, the tweeter would be getting about 50W of average power. But how much, exactly, are those nasty clipping harmonics adding to this? Is the tweeter getting 75W? Or is it, in fact, now getting about 55W? Assume that we can ignore energy over 50 KHz, where the tweeter is too inductive to absorb much power at all.

In the first case, it is plausible that clipping might accelerate the demise of the tweeter. In the second case, is doesn't really change anything compared to using a larger, unclipped amp.

I feel like I am wasting a lot of time on this thread as similar (wrong) assertions are being made by a parade of people. You, on the other hand, seem to have the technical background and mindset to back up what you say. I'm calling you out: prove it. Define the analytic signal of your choice to roughly represent music. Show me the power above 3,000 Hz. Clip it. Show me the power again. I'm no math wiz, but I think I can handle this first-semester stuff.

[Guest Sean]

I'm going to try and present this in a way that the "average joe" can follow along with. As such, i'm not "talking down" to you Ken, just trying to help others "catch up".

I think that the biggest and most common mistake that is made is that one assumes that the unclipped higher power amp would deliver the same signal that the clipped amp does, just at a higher amplitude. It doesn't.

Let's say that we have a 50 wpc amp and a 100 wpc amp. For sake of clarity, we are using a sinusoidal waveform that has a peak envelope power of 100 watts centered at 2.5 KHz.

When we apply this signal to the 100 wpc amp, the amp reproduces the signal in normal fashion. Given that a sine wave demonstrating a 100 watt peak envelope powerhas an average power of 25 watts, the momentary burst of power is easy to deal with. The amplifier follows the rise in amplitude to the peak, reaches the peak and descends down in smooth fashion. There is a linear escalation and decline of amplitude performed in a timely manner.

When we apply this same signal to the 50 wpc amp, the amp is driven into "clipping" aka "saturation" or "overload". Due to the lack of headroom involved and non-linearity of the waveform reproduced, harmonic distortion is generated and the peak to average power ratio is altered. On a square wave, which is what happens with severe clipping, the average power is the same as the peak figure. As such, even though we only have a 50 watt peak compared to a 100 watt peak with the other amp, we also have 50 watts of average power compared to the bigger amps 25 watts of average power. That means that we've got twice the heat dissipation to deal with even though we've got half the peak power. What's even worse, that heat is generated for the entire duration of the note rather than just a portion of it. Not good.

On top of this, we have to take into account several other factors. Remember the harmonic distortion that i mentioned above? Well, harmonics are signals that are created when a signal is "clipped". These are multiples of the original signal. As such, we not only have power being fed into the tweeter trying to reproduce the primary 2.5 KHz signal, but we also have additional signals that were created at 5 KHz (2nd harmonic ), 7.5 KHz ( 3rd harmonic ), 10 KHz ( 4th harmonic ), etc... These distortions of the original signal will be generated ad infinitum or at least out to the bandwidth limitations of the amp.

The stresses of the increased power being dissipated and the additional harmonics being reproduced cause even more problems now. Not only is the amp trying to reproduce the original signal, it also has to deal with the added power demands presented by the harmonics. Whereas we started off with a clean 2.5 KHz signal that had a peak of 100 watts and an average power of 25 watts, the tweeter now has to deal with 50 watts peak, 50 watts average at 2.5 KHz AND all of the extra energy shoved into it from the harmonics that appeared. How much extra energy will depend on how hard the amp is driven into clipping and the individual design of the amp itself. Obviously, this is "more of a bad thing". The added harmonics also produce a more complex waveform, causing the amp to work harder trying to keep track of what is going on / perform error correction on itself.

To top it off, amplifiers always "follow" the signal fed into it. In other words, if the signal rises in amplitude, the amplifier tries to follow that demand. The amp can't "think ahead" or contemplate what the signal is going to do, so it is always lagging slightly behind the signal. How much it lags will depend on how fast the amplifier is. As such, once a signal is fed into the amp that is beyond its' capacity, it goes into clipping or "saturation". While in the state of saturation, it is producing the aforementioned harmonic distortion.

Given that the signal rose above the power capacity of the amp, the amp can no longer track the amplitude of the signal i.e. it is "off the scale". In order to correct this, the amp has to wait for the signal to drop back below saturation to start tracking it again. As such, the amp will stay in saturation at max output for a longer period of time than the actual signal takes to occur due to the reaction time involved. This increases the amount of heat & energy that must be dissipated, both in the amp and in the tweeter. Depending on the speed of the circuit and how well it recovers from clipping, non-linearity / excess power output will occur. In comparison, the amp that remained linear never lost track of the signal and recreated the signal just as it was fed, both in proper amplitude and duration. On the clipped amp, not only is the primary note "dragged out" until the amp can respond, the harmonics are also generated for that extended duration.

Looking at the big picture, the tweeter now has to deal with:

1) A higher average power level due to the much closer peak to average ratio

2) More power that is spread over a wider bandwidth due to the addition of harmonics / more complex waveform

3) Power that has to be dissipated for a longer duration of time due to smearing / recovering characteristics of the individual amplifier circuit.

One can't express all of these variables mathematically because of the complexities involved i.e. the individual amplifier circuit in question, speaker impedances that vary with amplitude and frequency, etc... I hope that this explains some of my previous statements in a manner that Ken and others can follow along with. Sean

>

[Guest Sean]

OOPS!!! Open mouth insert foot. Then again, that's what i get for staying up all night on the computer. After going to bed and thinking about this further, i saw the light after waking up : )

There is one major mistake in this thread and it is mine. That is, i was using the classic explanation of square wave vs sine wave power distribution without applying common sense. While i could have let it stand as it was to see if Ken or someone else caught my mistake, i would rather correct it myself than to pass on "misinformation".

The mistake that i made is that we are dealing with a heavily clipped signal and not a square wave. There is a difference and i forgot to take that into account. My example is correct under extreme overdrive conditions, BUT is not suitable for use with the condition that we are applying the same exact signal to both amps. That's because the signal that is being "clipped" isn't totally "square", it is simply "severely distorted" or "trunctuated" in amplitude. The peak to average ratio does change, but the average power won't climb nearly as high as described.

As mentioned, those figures are for a "perfect" square wave, which we weren't using in this explanation. It is possible to generate a "perfect" square wave while listening to an audio system, but it would be SO severely distorted that i would hope that the listener would know better. As such, spurious distortions are created, but there nature is more closely related to "in band splatter" aka "Inter-Modulation Distortions" ( IMD ) than "Total Harmonic Distortion" ( THD ). This also skews the amount of power that the tweeter sees and further reduces the validity of my above argument. Harmonics are generated, but not to the extent that i initially commented on. I kept thinking square wave rather than "clipped sine wave". The levels of distortion are both quite high, but in order to achieve a square wave, the distortion characteristics would be through the roof. Bare in mind that the clipping level taking place as discussed here is already VERY high.

Boy, i'm doing a good job here of deconstructing my own post

With that all of that in mind, there are various manners in which circuitry clips. Some are far more adept at dealing with the situation and recover in a far more linear fashion than others. I have to re-think a lot of this. Ken may have more going for his argument than i initially gave him credit for, BUT, there are quite a few variables here that would be hard to simulate. After last night and the fight i got into with my girlfriend about "staying up all night on the computer", it might be a while before i respond with further thoughts though. All i can say about that one is "OUCH" Sean

>

[kkantor]

Sean,

Please give my message #3798 a re-read. Generalizations about how clipping alledgedly blows out tweeters are a dime a dozen. However, facts and math do not support this position. Are you prepared to show otherwise?

[kkantor]

The wise-ass in me is tempted to say that if you switched from a Mac to a PC, this would all go much faster, and you could get back to bed...

>hard to simulate. After last night and the fight i got into

>with my girlfriend about "staying up all night on the

>computer", it might be a while before i respond with further

>thoughts though. All i can say about that one is "OUCH"

>Sean