Nope. The sub should play exactly what it's fed. Switching freqs shouldn't cause any more distortion than playing only one. In fact if you were to actually look at the output of an amp on an o-scope with music playing, you would see nothing but a huge mess. All the freqs do get blended together. Not just in the mixing board, but in the instruments themselves as well. Strumming an open string on a guitar produces the fundamenetal freq as well as the harmonics. It's the harmonics that make the instrument sound real.
Now translate this to speakers. If a speaker can only play one note at a time without distorting, all you ever heard from a set of headphones would be a big mash of distortion. Same with your midranges and tweets. A good driver of any type should be able to play the entire range of its assigned frequencies without distorting or rolling off too much on either end. The only reason you divide freqs up in the first place is because the design that makes a good tweeter (light and stiff transducer that can be moved by a motor with very low inductance) do not make for a good midrange or midwoofer. As frequency decreases the amount of air that must be displaced to keep the same relative SPL increases exponentially. THe limitations that keep large drivers capable of playing low freqs from doing well at higher ones are 1) the inductance of the voicecoil required to move a large cone assembly typically has a fairly high inductance. This causes the impedance of the driver to rise as the frequency increases. The result is less output at higher freqs. 2) The larger a cone is, the heavier is has to be to maintain rigidity. A heavier cone leads us back to problem 1.
Since the ideal of the perfect full range driver is impossible (at least until we can make an infinitely rigid diaphragm driven by a coil with an infinitely small inductance) you have to divide the frequency range up. The problems inherent with that are several. First every division of frequencies done by analog means shifts the phasing of the signal. This can be somewhat overcome by using certain filter designs (as long as you are willing to accept the negatives of that design) or eliminated by going with a digital frequency division network, of which there are a few out there. next you have to consider the phase behavior of the drivers themselves right around the crossover freq. A small difference in output phase can cause cancellation. Again this can be addressed with processing, but the tuning is a pain in the ass and the proper test equipment is neither cheap nor simple to use, not to mention the processors themselves. From there you have to account for pathlength differences which can be fixed with time alignment circuitry (phase response issues cannot be fixed with time alignment since it isn't a time domain issue as it is frequency dependant) but again the time and equipment ot do it right are a pain to use, but at least this time the processor that does the work is realtively widely available and reasonably affordable.
Now that you see the problems caused every time you divide freqs between drivers, you should understand that division where no division is necessary, i.e. dividing the sub-bass between several drivers when any one of them could do the job just as well as the group, adds a level of complication where there is no real gain. The same can be said for the results of adding additional drivers playing the same upper range freqs unless the drivers are either exactly evenly spaced (and I do mean exact) from the listener or each has its own channel of amplification and is time aligned exactly.
Again, nope. The speaker simply plays them both. There is no switching. One wave form is laid over the top of the other and they combine. The driver plays the result, which as I said above, can look like a real mess on the scope.
:
