Twl's response of "it depends" is about as accurate as you can get. Given the massive amount of variables involved ( speakers being used, their radiation and compression characteristics, the type of music being listened to, the individual recording being used, the size and acoustics of the room, your listening distance, etc...), it would be way too hard to be specific.
Having said all of that, i will make one generalization: The more efficient your speakers are, the smaller your room is and the closer you sit to the speakers, the less power that you will need. Kind of makes sense, huh ??? : )
If we look at this strictly from an amplifier to speaker power transfer point of view and use Classical music as a source, we find that the standard for peak to average listening ratios as recorded is appr 15 dB's. Using "old school" sealed speakers of the low efficiency variety, average power levels of 5 watts or less would be required for all but the largest rooms on an average basis. The power required to generate full peaks without compressing the amplifier would be ( theoretically speaking under ideal conditions ) about 160 watts to achieve 15 dB's of dynamic range. If we reduced the peak to average ratio to appr 10 dB's, which is probably more common in most recordings, 50 to 55 watts would be all that you would need. That is, so long as the amp was fast enough to respond to the demands placed upon it. Otherwise, you would run into problems with slewing induced distortion ( SID, which is a form of "clipping" ) from an amp with a slow slew rate.
Bare in mind that these figures are based on a speaker that is both steady in impedance and presents minimal amounts of reactance. In plain English, we are counting on having an 8 ohm load ( for example ) that stays very close to 8 ohms and does not produce a lot of reflected EMF or sharp phase angles. This would be considered "heaven" as far as a speaker load goes and very few speakers of this nature tend to exist.
In other words, the above "theory" only works on paper as the dynamic conditions of a speaker / amplifier are continually changing as power and spl are varied. In most cases, the changes are NOT very linear and can change quite a bit from low spl's to high spl's. In effect, i would count on needing quite a bit more "muscle" than the above figured 160 watts to deal with the "less than perfect" situations that one encounters in real world operation with less than "ultra high" efficiency speakers. How much more one needs would depend on ALL of the above criteria that i mentioned in the 1st paragraph.
Obviously, Twl's situation is a little different due to using highly efficient speakers. While his figures are all correct for a "theoretical" installation under ideal conditions, i've yet to see a speaker that actually transfers input power to output SPL's in a completely linear fashion. This is due to thermal losses within the driver itself and compression that takes place on longer excursions. Once again, that brings us back to needing ( or should i say "wanting" ) more power on hand so that one could more easily overcome deficiencies / lack of linearity in the system.
To throw a BIG wrench into the above works, let's try looking at the same system with the same peak SPL level using hard rock as a music source. The accepted norm of dynamic range for a rock recording is appr 5 dB's. Obviously, this is due to the use of large amounts of compression and the much more consistent "drive" or playing of notes associated with "harder & faster" music.
On the surface, it would appear that we would need a lot LESS power to play rock due to the greatly reduced amount of dynamic range. That is simply not the case when you look at the big picture. Remember, we are comparing apples to apples here i.e. peak SPL's with Classical to the same peak SPL with "rock". We are not comparing the average listening levels between the two, otherwise rock would require a lot less power. After all, how many people do you know that like to "jam" to rock music at the same level that one listens to Classical music at ??? Most would consider listening to "rock" music at that level as being "elevator music".
Since we used 15 dB's as a reference for peak to average ratios on Classical music and "rock" music typically only displays appr 5 dB's of dynamic range, the average listening level would have to be appr 10 dB's higher for rock than that of Classical music in order to obtain the same appr peak readings. With that in mind, the 5 watts that sufficed to power the speakers for Classical music is now required to sustain 50 watts on an average basis for rock music. While the peaks would still be hovering around 160 watts, the continual heat dissipated by the amp and speakers on an average basis would be 10 times as high. Needless to say, this is MUCH harder on the speakers and the amplifier and explains why "rock" music is much more demanding ( in terms of ruggedness and reliability ) when building a system.
To further confuse the issue as to how much power one needs, different speakers "spray" sound or pressurize the room in different manners. Danner touched on this subject in his post. While the sound coming out of all speakers "falls off" as it gets further away from the speakers, this effects some designs more than others. If listening at 10' from the speakers, a 90 dB 8 ohm "line array" ( like the Pipe Dreams ) will need less power to produce the same volume that a 90 dB 8 ohm "conventional" woofer / mid / tweeter type system would require at the same distance. This has to do with the dispersion pattern i.e. far and near-field radiation. As such, a 90 dB speaker can actually play louder into the distance than a speaker with slightly higher sensitivity due to differences in radiation patterns. SPL's from line arrays do not fall off as fast as a "conventional" design as listening distance is increased.
While there are formulas to figure out the exact distances that spl's will start to diminish at a given frequency, the typical results are that line arrays and other "acoustically coupled" designs will produce more sound per watt in a large room than a conventionally designed speaker would. Needless to say, comparing speaker sensitivities at face value is only useful if listening at 1 meter from the speakers. Otherwise, you have to take a lot of other factors into consideration.
My personal thoughts are that the quality of wattage used is more important than the quantity ( under most conditions ). As such, one would want to strive for a high level of linearity ( Class A ) for as high of a wattage as possible. This amp should also be fast enough to respond to large changes in amplitude ( fast rise time and slew rate ) to accomodate the pulsed and highly dynamic changes found in some types of music. That is, if you want to retain all of the speed and attack that instruments have when listening to them live. On top of this, if one had a huge reserve of power available, the system would tend to sound much cleaner when driven hard due to the system never being "pushed" AND easily control the speaker even if it were considered a "highly reactive" design. Both Bigtee and Gs5556 touched on this in their posts from slightly different perspectives but with the same goal in mind. As far as DrDiamond goes, i'm with him. So long as quality does not suffer, i'd rather have a LOT more power than not enough.
Obviously, not everyone has the same listening tastes, speakers and rooms. As such, what one "needs" and what one "wants" can be very different things. Each situation is independent of another and needs to be assessed individually. That is the reason that most of us stress the importance of in-home auditions when possible. After all, what works "good" for you might seem "dynamically challenged" for someone else in their system and vice-versa due to all of the variables mentioned.
I hope that this helps explain a few different aspects of what we hear and why in an understandable manner. I'd also like to say thanks to Pbb for having faith in me to be able to possibly explain such things. I'll try to extend the same amount of courtesy back to him in the future : )
As to JCB's comments, i basically agree with what he had to say, especially about the wealth of info that we have available to us here and on a few select other forums. I learn from others every day. As proven by this thread, everyone can contribute a little bit with it all ending up to make for a well rounded yet diverse perspective on any given subject. Sean
>
