Here is a decent description I found blatantly copied from another forum that describes my point I think:
Let's see if I can clarify.
Most modified sine inverters put out simple square waveforms, actually only two pulses, timed and width-modulated to be somewhat lined up in time as a sine wave would be. In other words, a modified sine wave inverter generates a blocky wave which has none of the smoothness of a sine wave, but fewer harmonics than a simple square wave. But that's about all.
There are some modified sine inverters that put out simple stepped waveforms that come closer to sine, but they are still "blocky". Either of the two then attempt to filter out all of the unwanted harmonics with big capacitors and or inductors. Neither ever attains the lack of harmonics that a sine wave offers, and it's the harmonics that are the nasty part.
A "true" sine wave inverter does exactly the same thing- it approximates an analog sine wave with digital steps, with the big exception that it creates many, many digital steps instead of just two or four, and since the steps are correspondingly tiny, it's super easy to filter out any remaining harmonics with a simple filtering network- the result is an actual sine wave that approximates an analog sine wave well enough that no-one (and no equipment) has any issues with the result.
It's hard to communicate in words but diagrams are easy- here- In the illustration a squarewave is shown in green, a modified square wave is shown in blue, and a true sinewave is in red.
The basic reason that modified sine inverters suck is that any energy that is not exactly on the red line is "harmonic" energy, and it manifests itself as RF interference, excess heat in induction motors, etc. In this diagram, all the energy your refrigerator
motor won't get to use is in the yellow areas, which is energy that will get converted to excess heat and RFI.
In this diagram is shown what a "better" modified sine wave inverter might create:
But the big deal is that no matter what the theoretical waveform "should" look like using a modified sine inverter, depending on the load the actual waveform can easily get as messy as this:
(and this is what mine pretty much looked like under load)
...and just for the sake of it, this is more or less what most true sine inverters create, just proir to the final filtering stages:
(although some do it slightly differently, the results are the same- a "true" sine wave! )
There's actually a lot more to all of this, both mathematically and electronically, and what I've presented here is definitely a simplification, but hopefully it gets the point across.