My initial exposure to the issue of real power applied to speakers happened at Mike Bengfort's home in Minneapolis.  Mike had an old Pilot 240 tube amp with El84's in a push pull configuration.  The amp pushed about 10-12 watts. Mike also has a CARY SET 300B that pushes about 8 watts.  Both of these ratings are manufacturers ratings.  Both amps were used to drive a pair of Peerless/SEAS 2-way speakers with 8 ohm nominal sensitivity at about 85db/watt with 2nd order crossovers.  I believe the speakers used a Peerless 850122 and a SEAS 27TFFC.  I will digress briefly into tube land, then address the raw power issues applied to speakers.

What I heard at very mild spl was quite interesting.  The 8 watt Cary 300B sounded relatively terrible.  It had no dynamics, no drive, and didn't do much of anything well.  The only thing the Cary 300B might have done well is some holographic imaging with a saxophone.  I have subsequently learned that SET amps have incredibly poor dampening, and distortion that rises significantly when approaching maximum output.  The poor dampening is also affected by the impedance swings and phase angle presented by most loudspeakers.  Jack at Electra Print told me, "a guy really needs to have 100db efficient speakers when using a SET 300B setup.  This way the amp is only using that first watt of power.  It is that first watt of power from a SET 300B that makes it so special."  Jack has a plethora of experience with tubes, and is known to wind the best transformers in the United States.  Others have verified the need for very sensitive speakers and very flat loudspeaker impedance when using a single 300B SET amp. 

The old Pilot 240 El84 push pull sound was a totally different that the Cary 300B experience.   This Pilot 240 really knocked my socks off!  It was smooth, clean, and had very respectable dynamics.  It also pushed those 85db speakers louder than I would care to have them.  The 10-12 watt push pull amp played incredibly loud!  It was also very clean, and nicely smooth too.  I was confounded by the difference between the Cary 300b and Pilot 240 El84 push-pull, but couldn't disagree with the sound pressure and dynamics evident in that room.  Then I looked into the issue further, came to some conclusions.

Power handling among loudspeakers is a nebulous beast. There are several ways to market/advertise power handling. The most common methods are excursion and thermal.  When considering the thermal limits the issue is voice coil HEAT.  As the heat builds, the signal will encounter increased resistance (and compression). The common reference for the thermal power handling limitation is when the voice coil has a melt-down. This is how most manufacturers rate their speakers.  After this point is reached, the speaker will cease to function.  I have heard that some manufacturers put relatively wimpy inductors or capacitors in their speakers so the uses will toast them before the voice coil.  This is pretty smart.

A side-note analogy. Several years ago I purchased a cheap sleeping bag that had a 10 degree temperature rating. I used it on a short overnight hunting trip above Lincoln, MT when elk hunting in September. I learned that the 10 degree rating is the point where a human being would DIE in that sleeping bag! The temperature got down to about 20 degrees and I was very cold.  Such is the case with the power handling rating of a speaker.  They are rated at the point where further wattage would likely cause them to die.  A loudspeaker rated at 150watts when receiving 150watts will see the voice coils toasted.

Then there is the issue of X-Max. Linear X-Max is the fore/aft linear throw of the woofer.  The voice coil remains in the magnetic gap when linear.  However, the voice coil can go beyond the top plate.  There will be decoupling & distortion when this occurs. This rating is very significant, and easily derivable, but NONE (not that I have seen) of the commercial manufacturers use this for a wattage reference. It would force them to place a 5-10 watt nominal power handling sticker on their speaker. It is because a typical midwoofer/woofer setup will handle about 5-10 watts before its voice coil goes beyond the top plate. Some of the monster throw 15+mm one-way X-max subwoofer drivers will actually push 60 watts in the linear region when used in a sealed enclosure, but these are VERY uncommon.

There is some solace though. When your amp is pushing 10 watts nominal (not including the peaks), there is a huge amount of sound emanating from the speakers! Most folks don't realize that they normally listen to a good 3-4 watts of average power. This is precisely the reason that tube guys can live on a push pull that has 10 watts. Sure, there are dynamic peaks, but a tube amp deals with these very gracefully.

A hifi geek conveyed the information about normally listening to 5 watts through speakers, but I was in disbelief.  I then decided to test it myself. My speakers are about 85db/watt. My amp is a 120WPC Bryston 3B-ST. When listening VERY LOUD on the first track of Steven Curtis Chapman "Speechless" my volume knob is on about 60% "throttle". With that same throttle setting, using the same piece of music, I removed the speaker leads and put them across an 8 ohm resistor. I measured a slightly shifty @ 8 volts of music induced drop across that resistor. This mathematically equates to 1 amp x 8volts = 8 watts of power across the speaker terminals.  I was very surprised.

Hence, it might seem impressive to have 150wpc speakers, but this means almost nothing.  It is an almost meaningless exaggeration of what is happening.  Given 85db/watt speakers, a good El34 push pull (about 35wpc) or a good 60wpc stereo amplifier will get the job done with more than enough power.  It is also why a good 10 watts of push pull can sound incredibly good at normal listening levels.  Most often we listen to 2-3 watts of nominal power through our speakers.  The peaks are obviously higher (5x - 10x nominal), and some headroom is needed. 

Jim Dickenson became curious about my comments above and did some research for himself.  He commented thus:

The 1801's in my room at about 1 meter using about 2.83 volts (2.85) with a 400 Hz tone gave about 85 dB worth of spl on my rat shack meter. This is pretty much right on with what you have posted.  I then put on a CD without moving the volume knob from the measurement above. I listened and used the Fluke's ability to record voltages and give me a min max. The max voltage was about 7.5 V, which turns into somewhere around 8 watts. This level of volume was louder than I would normally want to listen to and I wasn't even at 10 watts! Hmmm, I wonder what the instanteous levels were? Needless to say, I am now more interested in lower power level amps....

The instantaneous levels are 5-10x higher than the average levels measured on Jim's multimeter.