Power . How much do we need...

[Orb]

Forgot to mention one other critical aspect and that is the speaker loading/impedance.
Not going to rehash what most of us have read, but it is worth considering that a speaker's impedance usually fluctuates between low and higher frequencies, which can impact an amplifier if its performance-spec (including power reserve that I have raised earlier) are not up it.
It is interesting to see charts showing an amp's performance to clipping for different impedances.
And again this is compounded that real music-recordings do not behave like a single static sinewave at a set frequency.

Cheers
Orb

 

[j_n]

But that is of the audio track when the power supply is NOT clipping! This is why a cheap, low powered amp can destroy the tweeters of an expensive, high power speaker: the waveform of a clipping signal has a huge boost in the level of high frequency energies, courtesy of the sharp corners of the clipping, which get fed straight through to the tweeter. The kids have a loud party: bye, bye, little speaker! Frank

I have had two different scenario's for blown tweeters.

1. Total power: vc burnout as a result of beating the living daylights outta the system. Autopsy of the tweeter shows a dark brown mess, burned epoxy and such. I have one such example I use as part of my tours...a combination of "what happens when you don't use superconductors, how a charged particle moving in a magnetic field causes a force orthogonal to both, how lorentz forces build up in magnets very quickly, and of course, what happens when you loan your system to your son to use at a beach party where he points the speakers away from the turntables..

2. Ultrasonics. This is the most interesting of the two intellectually. Clipping of the musical content causes ultrasonic energy. There is one location within a standard voice coil tweeter which is susceptible to overband content, this is the location where the voice coil end wires exits the vc form and heads out of the flux field to the terminals of the driver. I've had several different manufacturers product fail the exact same way.

Specifically, the unsupported tweeter lead wire is still within the magnetic flux of the gap for a small length, and this unsupported region will flex in response to current. For stimulus below the resonance frequency of this unsupported wire length, this section is better cooled than the vc, so there is no problem. When US content causes resonance in the flying wire, this unsupported section will flex until slip plane dislocation compromises the physical integrity of the wire..failure will occur. The failure location will look necked down at the break, consistent with the look of a tensile stress failure. Microscopic examination of the ends will show the slip plane texture on the wire surfaces. repeated flexure failure.

I see this failure mode confused with overheating many many times. Both occur, and it is dependent on spectral content as well as tweeter VC design. The explanations of tweeter burnout and power/clipping is many times muddied...because nobody considers the resonance failure.

Cheers, jn

 

[Robh3606]

I look at if from a system standpoint. I look at the speaker sensitivity and listening distance and know up front what my SPL tarket is at the listening position. Then look at the power required and if the speakers can safely handle it. Then I add some headroom and there you have it.

When I did my HT I wanted to be able to reach the 105db reference level with 10db of headroom above that. I don't tyically listen at reference level but even if I do the set-up sounds effortless with the added headroom.

The point is there is no magic number. It is system, room and SPL dependant.

Rob

 

[Grainger49]

Of course, there is no way for me to wade through all 11 pages of responses. So I will give my standard answer.

You need enough to drive your speakers, in your room (the larger the more power needed to reach a physical volume level) to the volume you like to listen to.

No one else should tell you how to listen.

 

[amirm]

There are however clear cases where it is more our bias expectations that anything else at play .... and the power required to drive a given speaker is one of them ... Most speakers are grossly inefficient. The sensitivity numbers do not do justice to the vast inefficiency of our loudspeakers.

The wiki on this is pretty good and goes even further: http://en.wikipedia.org/wiki/Loudspeaker#Efficiency_vs._sensitivity

"Loudspeaker efficiency is defined as the sound power output divided by the electrical power input. Most loudspeakers are actually very inefficient transducers; only about 1% of the electrical energy sent by an amplifier to a typical home loudspeaker is converted to acoustic energy. The remainder is converted to heat, mostly in the voice coil and magnet assembly. The main reason for this is the difficulty of achieving proper impedance matching between the acoustic impedance of the drive unit and that of the air into which it is radiating (at low frequencies improving this match is the main purpose of speaker enclosure designs). The efficiency of loudspeaker drivers varies with frequency as well. For instance, the output of a woofer driver decreases as the input frequency decreases because of the increasingly poor match between air and the driver.

Driver ratings based on the SPL for a given input are called sensitivity ratings and are notionally similar to efficiency. Sensitivity is usually defined as so many decibels at 1 W electrical input, measured at 1 meter (except for headphones), often at a single frequency."

Notice the bolded section from me. Efficiency is a marketing number for speakers so one has to believe that some stretching of the truth is going on in some or many cases.

My experience certainly mirrors that of Frantz in that the math of efficiency does not agree with the amount of power I see necessary to drive the speakers.

 

[Orb]

I have had two different scenario's for blown tweeters.

1. Total power: vc burnout as a result of beating the living daylights outta the system. Autopsy of the tweeter shows a dark brown mess, burned epoxy and such. I have one such example I use as part of my tours...a combination of "what happens when you don't use superconductors, how a charged particle moving in a magnetic field causes a force orthogonal to both, how lorentz forces build up in magnets very quickly, and of course, what happens when you loan your system to your son to use at a beach party where he points the speakers away from the turntables..

2. Ultrasonics. This is the most interesting of the two intellectually. Clipping of the musical content causes ultrasonic energy. There is one location within a standard voice coil tweeter which is susceptible to overband content, this is the location where the voice coil end wires exits the vc form and heads out of the flux field to the terminals of the driver. I've had several different manufacturers product fail the exact same way.

Specifically, the unsupported tweeter lead wire is still within the magnetic flux of the gap for a small length, and this unsupported region will flex in response to current. For stimulus below the resonance frequency of this unsupported wire length, this section is better cooled than the vc, so there is no problem. When US content causes resonance in the flying wire, this unsupported section will flex until slip plane dislocation compromises the physical integrity of the wire..failure will occur. The failure location will look necked down at the break, consistent with the look of a tensile stress failure. Microscopic examination of the ends will show the slip plane texture on the wire surfaces. repeated flexure failure.

I see this failure mode confused with overheating many many times. Both occur, and it is dependent on spectral content as well as tweeter VC design. The explanations of tweeter burnout and power/clipping is many times muddied...because nobody considers the resonance failure.

Cheers, jn

Ah yeah great point on the 2nd one, forgot about that and is a real interesting one especially when considering-expanding other factors such as a wide bandwidth amp and interraction with ultrasonics generated elsewhere (crude example being a poorly designed nos dac ).

Cheers
Orb

 

[Phelonious Ponk]

I'm no engineer and there's a lot I not only don't know and wouldn't even understand if you explained it to me. But among the things I've heard, over and over again, until they've finally sunk in is that loudspeakers are inefficient and very inconsistently inefficient at that. And so manufacturer's ratings are not reliable. I've also learned, as has been discussed here, that an amplifier can have lower watts than another, yet have higher headroom. It's pretty difficult to know what to do with such inconsistency if you're not an engineer. But then there's what my ears taught me: More powerful amps sound better. They sound better at low volume. They sound better at high volume. They sound better into efficient and inefficient speakers. Many years ago, in my passive midfi days, I had a pair of Altec Lansings that were very efficient. I was driving them with a Kenwood integrated amp rated at 60 watts. I sold the Kenwood and bought a Harman Kardon that was basically monoblocks in a single case - separate paths, separate transformers, etc. It weighed a ton. The dealer I bought it from, a guy I knew, told me that Harman's watts were louder. I didn't know what he meant by that, I still don't really understand it, but I know that Harman, at 45 watts per channel, just seemed to make those Altecs breathe easier.

Probably expectation bias . But whatever it was, I have, ever since then, been a believer in excess headroom. That belief was only reinforced when I got into actives. The really powerful active monitors with bigger amps for the woofer than for the tweeters, and more wattage (whatever it may mean...) than even seems to make sense for the size of the drivers, always seem to sound the best.

Or at least that's what I hear. Oh, and by the way, really good, heavily amped actives deliver, to my ears, the purest, cleanest, most transparent midrange in the world. Low power SET, runining at anything close to realistic volume, sounds like a special effect by comparison. MHO. YMMV. That doesn't mean it's not attractive, by the way.

Tim

 

[Ron Party]

Efficiency is a marketing number for speakers so one has to believe that some stretching of the truth is going on in some or many cases.

My experience certainly mirrors that of Frantz in that the math of efficiency does not agree with the amount of power I see necessary to drive the speakers.

As does mine. Amir, have you played around with the Salon2s? My local high end dealer who carries these speakers told me they really begin to sing when feeding them 1000 W.

 

[DonH50]

The issue of efficiency vs. frequency has long been understated... Not documenting this means the fairly severe impedance excursions most speakers exhibit are "hidden" factors driving how much power is really needed. Another factor is how we hear, i.e. the old loudness contours. That means a speaker with lower frequency response almost inevitably requires more power than one with more limited FR because more power is required to make the LF's louder to match the (perceived) loudness of the highs.

I consider loudspeaker efficiency numbers very useful, but only as general guidance on where to start.

As an aside, I would love to know what peak power a given speaker can take, based upon its drivers' maximum excursions. Of course, that also probably varies over frequency...

To throw out some numbers, in the midrange 1 dB is almost imperceptible to most people if you change the volume level (I am NOT talking about volume matching for AB tests) and takes about 30% more power. Around 3 dB is a noticeable change and requires twice (2x) the power. Doubling the volume is 10 dB and takes 10x the power. If we start looking at headroom over the average level, then 20 dB means 100x the average power, and 30 dB is 1000x the average power. So, even with just 1 W average, if we never want to clip (a debatable though worthwhile goal), it's easy to see why people often prefer amps capable of delivering 100 W or more. And, why a few dB increase in speaker efficiency means so much.

 

[Gregadd]

Interesting comment on SS vs. tube power... I would have guessed the opposite, with about 2x the power in a tube amp to match the bass tightness of a SS amp. In the midrange and up, I expect power to be a wash, with the tube amp sounding a little "sweeter". I have always felt bass to be one of the weaknesses of tube amps, primarily due to their relatively high output impedance (and low damping factor). BTW, assuming a good amp with stiff rails, the SS amp rated at 600 W should put out 800 W into 6 ohms; the tube will put out 300 W no matter the load as long as the tap is there and it is matched to the speaker's impedance.

88 dB/1 W/1 m, say we want it to reach 120 dB peaks at our listening spot, maybe 3 m away. That means we need an extra (120 - 88) = 32 dB, and another 6 - 12 dB for distance depending upon the room gain (size etc.) Let's say an even 40 dB, or 10,000 W (!) Maybe 110 dB is more reasonable, yielding 1000 W -- closer to your target goal. In fact, 600 W is about 28 dB above 1 W, so the speaker should output 116 dB at 1 m and maybe 110 dB at the 3 m listening spot driven by 600 W with a little help from the room; close enough! Good "guess", Gregadd! Leaves the tube amp lagging a bit, though, as it will be 3 dB into clipping at that power. We probably won't notice as it will only happen on the loudest peaks.

I thought this was about amps; it's a "guess the speaker" game? Alas, I cannot play, as I am unfamiliar with most of the newer speaker brands. 6 ohms rules out Magnepan, however.

Onwards - Don

p.s. I am an "objectivist" who has owned tubes until very recently, and sold out to SS due to cost and other considerations. I always did like falling outside the box...

Of course the rooms must be considered

 

[Phelonious Ponk]

The issue of efficiency vs. frequency has long been understated... Not documenting this means the fairly severe impedance excursions most speakers exhibit are "hidden" factors driving how much power is really needed. Another factor is how we hear, i.e. the old loudness contours. That means a speaker with lower frequency response almost inevitably requires more power than one with more limited FR because more power is required to make the LF's louder to match the (perceived) loudness of the highs.

I consider loudspeaker efficiency numbers very useful, but only as general guidance on where to start.

As an aside, I would love to know what peak power a given speaker can take, based upon its drivers' maximum excursions. Of course, that also probably varies over frequency...

To throw out some numbers, in the midrange 1 dB is almost imperceptible to most people if you change the volume level (I am NOT talking about volume matching for AB tests) and takes about 30% more power. Around 3 dB is a noticeable change and requires twice (2x) the power. Doubling the volume is 10 dB and takes 10x the power. If we start looking at headroom over the average level, then 20 dB means 100x the average power, and 30 dB is 1000x the average power. So, even with just 1 W average, if we never want to clip (a debatable though worthwhile goal), it's easy to see why people often prefer amps capable of delivering 100 W or more. And, why a few dB increase in speaker efficiency means so much.

But the relationship between the average power an amp can produce (RMS?) and the peaks it can deliver is not fixed, is it? Isn't it possible that the right design in a 50 watt amp could deliver higher peaks than some 100 watt amps? And isn't the possible length of the peaks also variable? And couldn't this be critical if your transient peaks are orchestral crescendos instead of snare drum rim shots?

Forgive me if this has been covered; I haven't read the full 11 pages.

Tim

 

[mep]

My experience certainly mirrors that of Frantz in that the math of efficiency does not agree with the amount of power I see necessary to drive the speakers.

I agree as well. My Def Tech BP7000SC speakers are rated at 92dB sensitivity which is fairly high. The specs further say the speakers are compatible with amps from 20 watts to 1000 watts. You would think that if you had a 100 watt tube amp or SS amp driving a 92dB speaker that you are golden. Sorry for the Jadis Defy 7 MKII lovers, but I never heard my speakers until I ran them with Phase Linear 400 Series 2 amps which put out over 200 watts per channel. I was dumbfounded at how much different they sounded with over twice the power (which is still only slightly more than a 3dB difference in loudness). Doubling the power woke the speakers up. And these speakers already have built in 14” subs with an 1800 watt amp driving them.

And all of this tells me a few things. First of all, I don’t think these speakers are as efficient as they claim to be or that efficiency rating was taken at one particular frequency where the speaker is more efficient than at any other frequency. I can’t imagine driving these speakers with 20 watts. Another thing that is entirely possible is they fudged the impedance value as well. The specification sheet only says they are “compatible with 8 ohms outputs.” I’m not sure what the hell that means. Does that mean the speakers are really nominally 7 ohms or 9 ohms? Or, does it really mean they are closer to 4 ohms and my SS power amps are putting out twice the power into the lower impedance load which of course a tube amp can’t do? That would make sense to me in light of what I heard if I was listening to them with a 100 watt tube amp and then heard them with a with a SS amp that could put out over 400 watts into a 4 ohm load.

And now I’m driving these speakers with the Krell KSA-250 which was measured by Stereophile to put out over 300 watts into 8 ohms and over 600 watts into 4 ohms. And it just keeps doubling down as the frequency is halved. Do the speakers sound more powerful now? Yes. Do I feel like the amp is loafing along? Yes.

As long as you like the sound of the first watt, my answer is that I will take as many more of those watts as I can get. Maybe there comes a point where it gets ridiculous, maybe not. As a note on my setup, my speakers are approximately 8’ apart from center to center, 3.5’ from each side wall, 6’ from the rear wall, and my listening position is about 13.5’ from the speakers.

 

[RogerD]

Came across this article.....seems like the divide between low power and high will never really be answered.

http://www.soundsgoodtome.us/2011/12/06/musical-ecstasy-size-doesnt-matter/

 

[fas42]

But the relationship between the average power an amp can produce (RMS?) and the peaks it can deliver is not fixed, is it? Isn't it possible that the right design in a 50 watt amp could deliver higher peaks than some 100 watt amps? And isn't the possible length of the peaks also variable? And couldn't this be critical if your transient peaks are orchestral crescendos instead of snare drum rim shots?

Another way of looking at things, and particularly with the idea of some amps having "louder watts", is to consider current delivery capability. It's fairly easy to have an amplifier generate large voltage swings, nice , clean, low distortion, when no actual drive has to be delivered to a speaker. Once real, sizeable current has to be delivered, this is what separates the men from the boys. As people have pointed out, it gets even worse when you have widely swinging impedance curves for the speaker system: they can get down to 1 ohm, which is getting mighty, mighty close to a short circuit. Big amps have to be delivered at that moment, even if it is doing nothing more than keeping the passive crossover happy. And where does that amperage originate from? From the smoothing capacitor bank -- the transformer and rectifier has no chance of getting the energy through in time -- and if that capacitor size is insufficient only one thing can happen: the voltage rails plummet like a stone, at least initially, you're getting exponential decay of voltage.

Hence big mutha power supply capacitance does its job, even on nominally low powered amps, of keeping the circuit in a correctly functioning state, hence definitely they have "louder watts" ...

Frank

 

[fas42]

With regard to sensitivity ratings, my understanding is that it is taken at 1kHz frequency typically, so if the FR has a huge spike at that point then, voila, high sensitivity speaker! But most systems have a response at 1kHz which aligns nicely with the majority of the spectrum, so it should not really be an issue ...

Frank

 

[LL21]

Another way of looking at things, and particularly with the idea of some amps having "louder watts", is to consider current delivery capability. It's fairly easy to have an amplifier generate large voltage swings, nice , clean, low distortion, when no actual drive has to be delivered to a speaker. Once real, sizeable current has to be delivered, this is what separates the men from the boys. As people have pointed out, it gets even worse when you have widely swinging impedance curves for the speaker system: they can get down to 1 ohm, which is getting mighty, mighty close to a short circuit. Big amps have to be delivered at that moment, even if it is doing nothing more than keeping the passive crossover happy. And where does that amperage originate from? From the smoothing capacitor bank -- the transformer and rectifier has no chance of getting the energy through in time -- and if that capacitor size is insufficient only one thing can happen: the voltage rails plummet like a stone, at least initially, you're getting exponential decay of voltage.

Hence big mutha power supply capacitance does its job, even on nominally low powered amps, of keeping the circuit in a correctly functioning state, hence definitely they have "louder watts" ...

Frank

i know nothing about amps...but i understand the capacitance (ability to store energy) is extremely important to generate consistently high current/high power without clipping, failing, weakening at strong signals or strong demands from the drivers...and thus, delivering the wide dynamic capability that helps create genuine loudness.

 

[Robh3606]

You would think that if you had a 100 watt tube amp or SS amp driving a 92dB speaker that you are golden.

Hello mep

Well did you figure that you loose 6db because of your listening distance?? That drops you down to 86dB where you sit. Add 100 watts or 20dB and you can hit 106dB peaks figure 109dB for the pair. Barely enough for clean peaks.

Rob

 

[LL21]

Hi Rob,

What do you consider adequate for clean peaks? i understand that quick transient peaks could easily hit 109db...but i would also say that is extraordinarily loud. I believe 109db would be noticeably louder than having a jackhammer where the speakers are, no? i am not trying to be controversial here, btw...i did my own calc and i think i am at about 116-117db...but i have no intention of going there!

 

[fas42]

Hello mep

Well did you figure that you loose 6db because of your listening distance?? That drops you down to 86dB where you sit. Add 100 watts or 20dB and you can hit 106dB peaks figure 109dB for the pair. Barely enough for clean peaks.

Rob

But do you want to do that on an ongoing basis? Came across this: FLUTES OR PICCOLOS COULD HARM YOUR HEARING, http://www.larrykrantz.com/flutesor.htm, something to think about, if you're after having those peaks happening continually ...

Frank

 

[Robh3606]

What do you consider adequate for clean peaks? i understand that quick transient peaks could easily hit 109db...but i would also say that is extraordinarily loud. I believe 109db would be noticeably louder than having a jackhammer where the speakers are, no? i am not trying to be controversial here, btw...i did my own calc and i think i am at about 116-117db...but i have no intention of going there!

Hello lloydelee21

I designed my set-up to be able to do 115dB peaks clean with no clipping. If you want a system to sound effortless you need to have enough power and SPL capability to cleanly reproduce those short duration peaks. I suggest you get your hands on a Digital Peak reading SPL meter. When you hit these 'extreme" peaks they do not sound loud. If anything it opens things up and it sounds unrestrained and natural. That said my average SPL levels are in the mid 80's and I can easily see peaks above 105dB depending on the source material. You may be hitting higher instantaneous levels than you think you are.

But do you want to do that on an ongoing basis?

Hello Frank

Looking at you own reference and orchestra can hit in excess of 120dB. I am not advocating listening at these extreme levels but the important point is if you can barely hit 110db it's going to fall short. Many people don't have 92dB speakers many are well under that. I think it's rather obvious that most systems simply cannot do clean transients at anywhere near their actual levels. Most systems are on the wrong side of the power curve because the speakers are simply not capable of hitting these peaks.

Rob