The only thing that thermally damages speakers is power... more specifically: average power over time.
I'll explain...
If you take a given amplifier, let's say 100 watts and operate it just below clipping with music material, the "Crest Factor" of the amplifier's output is equivalent to the "Crest Factor" of the program material.
"Crest Factor" is the difference between the average level of the signal and its peak level. For example, a pure sine wave has a "crest factor" of 3dB, meaning that it's peak level is 3dB higher than its average level. We all know that 3dB represents a power factor of 2, so another way to look at it is that the peak power of the signal is twice that of its average level. So, if we play a sine wave on our 100 watt amplifier, just below its clipping level, the average power (over time) the speaker is needing to dissipate is 50 watts.
A true square wave, by comparison, has a crest factor of 0db, so it has equal average and peak power. Our 100 watt amplifier, playing a square wave, unclipped, into our speaker requires that the speaker dissipates 100 watts of power (twice the heat as a sine wave).
Music has a significantly higher crest factor than sine waves or square waves. A highly dynamic recording (Sheffield Lab, Chesky, etc.) typically has a crest factor of 20dB or more, meaning that its average power is 100 times lower than its peak power. So, if we play our 100 watt amplifier just below clipping with the typical audiophile recording our speaker is only needing to dissipate 1 watt of average power over time.
Modern commercial recordings typically exhibit crest factors of around 10dB, meaning that the average power is 10 times lower than the peak power. So, our 100 watt amp just below clipping would deliver an average power over time of 10 watts that the speaker has to dissipate.
Okay, so what happens when we clip the amplifier (which we all do at times). When the amplifier enters into clipping, the peak power no longer increases, but here's the KEY... THE AVERAGE POWER CONTINUES TO INCREASE. We can often tolerate a fair amount of clipping... as much as 10 dB or more above clipping with a reasonably dynamic recording... a bit less with a compressed commercial recording.
So, if we turn the volume up 10dB higher than the clipping level with our Sheffield Lab recording, we have now reduced the crest factor of the signal reaching the speakers by 10dB... so instead of needing to dissipate 1 watt average, we are asking the speaker to dissipate 10 watts average, and we're probably ok.
If we turn up the volume 6dB past clipping on a compressed commercial recording (or bass music recording), we have taken the crest factor of the signal from a starting point of 10dB to only 4dB, asking the speaker to dissipate an average power of 40 watts instead of 10 watts... that's FOUR TIMES the average power, which generates four times the heat.
SO, in most cases, the reason clipping can damage a speaker really has nothing to do with anything other than an increase in average power over time. It's really not the shape of the wave or distortion... it's simply more power over time.
When someone plays Bass Mekanik clean (unclipped) on a 1000 watt amplifier the average power is 100 watts (10dB crest factor). You can also make 100 watts average with Bass Mekanik by heavily clipping a 200 watt amplifier.
If someone is blowing a woofer with 200 watts of power due to a lack of restraint with the volume control... they will blow it even faster with a 1000 watt amplifier because they will probably turn it up even more and now they have more power to play with... this is the recipe for aroma of voice coil.
When woofers are rated for power, an unclipped signal is assumed. We use test signal with a crest factor of 6dB for power testing and can run a speaker at its rated power for hours and hours on end without thermal or mechanical failure. For example, a W1v2 can dissipate 150 watts average power for eight hours or more with signal peaks of 600 watts. So, we rate the speaker for 150W continuous power. This way, when a customer needs to choose an amp for it, they will hopefully choose one that can make about 150 W clean power... Even if they clip the bejeezus out of that amplifier, it is unlikely that the speaker will fail thermally. This is a conservative method, but it needs to account for the high cabin temperatures in a car (think Arizona in the summer) which significantly impacts heat dissipation in the speaker. A top plate that starts at 150 degrees F is not as effective at removing heat as one that starts at 72 degrees F in the lab... and this affects the ramp up of heat in the coil.
A clipped signal carries more average power than a non-clipped signal... this is the only aspect of it that affects a speaker thermally.
It doesn't matter if the signal is clipped before the amp at the preamp level or after the amp.
A sine wave can damage a speaker in a matter of seconds given enough power and in a matter of minutes at fairly moderate output... it depends on the impedance of the speaker at that particular frequency... if the impedance happens to be low, it might go up in smoke in a couple of minutes.
The fact that a signal is clipped does not make it inherently damaging... if the average power of the clipped signal is low it won't ever damage a speaker. At higher power levels, the fact that a clipped signal carries more average power over time can result in damage.
The fact that tweeters have low inductance, do not employ low-pass filters and have small, delicate voice coils makes them more susceptible to damage from a clipped signal than a woofer or mid.
Does that make sense?
It just points out that the old statement of an amp that is too small will damage speakers more than a more powerful amp is entirely dependent on the use of each amp... if you clip both amps to the same extent, the more powerful one will blow speakers faster.... but it is possible to make a small amp operated into clipping produce as much average power as an unclipped larger amp (even though the peak power is greater on the big amp).
When you clip an amp you not only increase distortion, you also compress the dynamic range of the signal... the distortion isn't what kills the speakers (except tweeters in some cases)... it's the dynamic range compression that really does it.