Quote Originally Posted by jack70
A couple of points that aren't being made clear here... (perhaps missed by some?):

Electrical Watts (amp power) and Audio Watts (acoustic energy that hits yer ears), are 2 different things.

Acoustic (perceived by our ears) energy is an "absolute thing". If you went into the quietest room in your house, you'd find the background sound level was around 30-40 db. That's an Absolute figure.

The Electrical energy of the amp is totally dependent on the efficiency of the speakers. You could have an amp that puts out 100 electrical watts through an 8-ohm speaker... this would be equal to 28 volts across it (sq rt of 100x8). If the speaker is very efficient it'll "sound" loud (acoustic energy). If it's very ineficient, it'll sound much quieter for the same input going into the amp. The amp-speaker interaction is the key here.

You could have a 10 watt amp conected to a speaker with 115db "efficiency" that puts out 125 db of acoustic energy (loudness to your ears). You could have a 500 watt amp connected to a low effiecient speaker (70db) that sounds much quieter in comparison.


Its' true that perceived "loudness" is a logrithmic (power) function... it requires "more & more" power to hear "a small amount" of loudness increase.

The other issues discussed (frequency aberations) and (peak power) are both more complex issues, because they depend on more variables. Generally, audiophiles use high power amps because they have the reserve electrical power to reproduce the kind of (very-small) peaks in reproduction that make music sound more real... these are instantaneous (fast) pulses of power... the kind you'd hear from the sound of a hammer or a gun. It's not a constant acoustic output, but of a short duration. It's impossible to reproduce that "reality", through any speaker with a smaller amp (all things considered). But having that extra "power reserve" of a big amp isn't valuable unless the speaker system is matched well to the electrical circuitry of that particular amp.

Speakers are VERY inefficient devices. They waste 90% (more or less) of the amps energy in mechanical ways. The acoustic power they put out is but a fraction of the electrical power coming into them. They also have dozens of times the distortion of the electrical signal coming into them. That's also why speakers are such a neat thing to play around with for most audiophiles -- they all have inherent defects and a "ways to go" to reach perfection.
Hi Jack,

You are correct in concluding that the loudspeaker and amp are key factors, but the size of the room and listening distances are equally important factors. You are also confusing "sensitivity" and "efficiency." Efficiency is always rated as a percentage. Sensitivity is a measure of pressure-level at a given distance using a specified input . The two are related, but VERY different.

Using acoustic watts as a measure of power needs isn't going to work very well. A 36-inch bass drum can produce a peak power of around 25 acoustic watts. Speaker efficiency is typically between 1% and 10% (can be lower or higher, but this is typical) so we'd need from 10 to 100 times that 25-watts just to produce the power of the 36-inch bass drum. That would be from 250 to 2500 watts! However, if you put that drum in your living room, along with the rest of the orchestra, and then have them play at peak levels, you'll be deaf in short order. In a huge performance hall the power is needed, but not in your home. If you take a sound system that can reach 100dB at Row M in Radio City Music Hall and put it in your bathroom it will produce a high enough pressure level (SPL) to exceed the pain threshold.

The industry settled on sensitivity measurements years ago because they are far more useful than efficiency numbers. As a system designer they allow us to *estimate* the output we can expect, and by applying a few "rules of thumb" we can easily insure that we'll be able to produce the target SPL levels.

Working with watts and efficiency really falls apart when we look at something like an ESL, as they are driven by voltage, and could be said to be "wattless speakers." Working with sensitivity allows consistency even with odd driver types. Your numbers (70dB and 115dB) for sensitivity aren't at all typical, but your point is valid. More commonly we might end up comparing a loudspeaker that has a sensitivity of 82dB/w/m with one that has a sensitivity of 97dB/w/m (the latter being the sensitivity of my ESL's). The former loudspeaker will only reach 102dB *peak* at one meter when driven by a 100 watt amplifier, but the latter will reach 107dB with only a 10 watt amplifier. If the goal is to produce 110dB peaks then the 82dB/w/m speaker will need well over that 100 watts, but the second will reach that level with only 20 watts.

Of course the above example also shows the danger in any over-simplification, because if you try to drive my stats with any commercial 20-watt amplifier it will most likely shut down instantly. ESL's present a nasty load to the amp, and the load is yet another factor to consider when determining amplifier requirements.

Rod Elliott, and others (most notably Bob Carver) have written extensively on amplifier requirements and works like the "Handbook for Sound Engineers" are full of info on determining the power requirements for almost any situation, and it is a topic that is far to complex to discuss on an audio forum. It takes chapters, and anybody with the level of interest to tackle all the reading and the math simply needs to get a few good books. Nobody want's to read an entire course on the topic in a string of forum posts.

Crest factor hasn't even been mentioned.

Without going into too many details, some of the important factors to consider are:

* Room size.
* Room acoustics (dead, live, etc.)
* Distance from source to listener.
* Loudspeaker sensitivity.
* Loudspeaker load (as seen by the amplifier).
* Types of music to be reproduced.
* Listener preferences (some like it louder than others).

Other things that most seem to overlook:

* Distortion levels - Sound can seem "ear splitting" at modest levels if there are objectionable distortion products. When someone says "my 30 watt amp is already WAY TOO LOUD when the volume control is at 3:00" it is almost always distortion giving them the impression of "TOO LOUD."

* Crest factor - This is a complex-signal issue. With even a slightly complex signal we have to consider crest factor. Two tones are enough. Assume we record 100 Hz. and 2000 Hz., and that each is originally produced by a 10 watt source. We have two 10 watt sources, but to reproduce the two tones with a single speaker and amplifier we need more than 20 watts. In fact, we need 40 watts, due to the crest factor of the two signals when they are mixed (the 2k signal rides on top of the 100 Hz. signal producing twice the voltage swing, and twice the voltage swing means four times the power. This is one of the reasons so many people go to bi-amping with their flagship loudspeaker designs, and why most pro loudspeakers are powered and bi-amped, but it is probably too complex to cover in forum posts. Check the literature (Rod has some good info on his site).

So, Jack, you are right even if a little confused over terms; A LOT has been glossed over here, but the basics still hold. If you take the sensitivity of a loudspeaker, adjust for the listening distance, ignore room support, and then use the desired peak SPL desired, you can calculate the amplifier power required. Good engineering practice requires that you add in a margin for error (at least 3dB) and I'd add a little more for other factors (like the crest factor). THEN I'd add a little more with an SS amp to make sure that protective circuits were either never partially triggered, or simply weren't needed.

I think the original post discussed 89dB/w/m. Here's how I'd *estimate* the amplifier requirements for my situation. I sit four meters from my speakers. With one watt I can expect a little over 83dB at the listening position (a little over due to room support) with one watt applied. There are times (RARE), when playing music with lots of bass, and showing off the system, that I've seen 112dB in my room, and I want at least 3dB margin, so I want to shoot for a peak output of 115dB. Assuming single amp and passive xovers we can now calculate the power we need to reach (about) 115dB.

1w = 83dB
10w = 93dB
100w = 103dB
1000w = 113dB
2000w = 116dB

(note that we have two speakers and two channels, so it would take almost 1000 wpc to reach that 115dB mark. AND, we still haven't considered the crest factor issue. How many speakers will even handle 1000 watts? If we add in 3dB more for to handle worse case crest factor issues, and another 3dB to alow a margin of error or to avoid partial protective circuit triggering, we're talking 4000 watts per channel!!!

Increasing the speaker sensitivity 10dB would reduce that by a factor of 10, to only 400 watts, so the speaker sensitivity is obviously a key factor.

Note that when the bass peaks are reaching 112dB, the average level is right around 90dB, and that the bulk of the high-energy is at the lower frequencies. If that weren't the case this level would be ear-splitting (and if the sound is distorted it will be ear-splitting at much lower levels). I use ESL panels that have a sensitivity of 97dB/w/m, and drive them with an amp that can deliver over 1000 watts to the load without working up a sweat. That's JUST for the mid-tweeter. The mid-woofer has it's own 600 wpc amp and the sub-woofer has it's own 1000+ watt amplifier. The system reproduces highly dynamic music at any desired level as effortlessly and with as much impact as any high-efficiency horn system (but with a much flatter response). Does ANYONE actually NEED that much power and output capability?
IMHO, ABSOLUTELY NOT.

However, 200wpc was NOT enough to drive my mid-woofers, and 400 watts was NOT enough to drive my sub, even at more normal (and common) "audiophile" listening levels (averages approaching 90dB). With the smaller amps the sound was LOUD but distorted due to all the momentary clipping on peaks. Not obvious to anyone who has never heard a system that is free of such problems, but very obvious when compared to the same drivers with larger amps.

Velodyne upped their sub drives from 400 watts to over 1000 watts, because 400 just wasn't enough. Ask them if you don't believe me.

InnerSound upped the power of the amp driving their mid-woofers from 200 watts to 600 watts, because 200 just wasn't enough. Ask them if you don't believe me.

Kilpsch Corner Horns can be driven to INSANE levels with a 10 watt amplifier, so the speaker certainly can make all the difference.

I use to sacrifice efficiency (as well as sensitivity and high output capability) to get lower distortion, wider bandwidth, and excellent transient characteristics, but that is no longer necessary. Today we can have our cake and eat it too (as long as money and floor space isn't an issue).

Yea, Jack, we glossed over a BUNCH of stuff, but I don't think any thing that was missed was actually relevant to the original question.

If I understood it properly, the original question could be rephrased this way: "If my speakers have a sensitivity of 89dB/watt/meter and 90dB will damage my hearing, then why would I want an amplifier larger than 2 watts per channel?" I think I answered that question in my response. If he's putting the system in a bathroom he'll need less power, in Radio City Music Hall he'll need a lot more, but in general, if he takes the sensitivity and adjusts for distance and desired SPL he'll see why he needs more than 1-watt. That's all he asked. Why more than a watt or two? Didn't I answer that fully enough?

(I could have just said "crest factor, distance, and a bunch of other stuff" but the explanation of sensitivity and distance seemed most appropriate. Do you think I was mistaken?

See ya,

Chuck