Okay, after a good bit of googling I may have found a resource that will appeal to you:
A Review of Inverter Design and Topologies -- Trace Engineering
Also a bit of a fuller explanation (still probably less detailed than you would like)
here:
Thanks that answered my questions. No surprises there.
F.Y.I. There are essentially 3 kinds of inverter outputs.
Square wave. This is the cheapest and is the noisiest electrically . Switching happens at line frequency (60 Hz). There is some kind of low pass filtering on the output, but this is only really good for running electrical motors and they are going to run hotter than normal. Running electronics from one of these would be a bad idea. Square edges contain a huge quantity of higher frequencies at odd octaves (3rd, 5th, 7th... octaves). This is what is called harmonic noise. Square wave inverters originated in older mechanical inverters that used mechanical switches to create the square wave output.
Modified square wave . This is essentially a very poor mans version of a sine wave. Switching frequency is at three times line frequency (180 Hz) and is again full of harmonic noise. Not as bad as a square wave output, but still nothing to run sensitive electronics from. Makes motors and stuff happier.
"Pure" sine wave. This is actually generated using square waves switching at much higher frequency than line frequency. The duty cycle of the square wave is adjusted on the fly to create an reasonable approximation of a sine wave. The higher the switching frequency, the smoother the approximation of a sine wave is and the easier it is to filter out harmonic noise (which is the expensive part). This is what modern inverter designs use. A Class D audio amp works in a similar fashion if that comparison helps.
The low frequency inverters discussed in previous threads use a ferro-resonant transformer to low pass filter a square wave output switching at 60 Hz. I am familiar with ferroresonant transformers having used them in industrial applications since the 80's. In particular I spec'ed Sola MCR power conditioners as part of any process control system I worked on back in the 90's when that was my day job.
The Sola MCR Hardwired Series provides excellent noise filtering and surge suppression to protect connected equipment from damage, degradation or misoperation. Combined with the excellent voltage regulation inherent to Sola/Hevi-Duty's patented ferroresonant design, the MCR can increase the...
solahevidutysales.com
An MCR/CVS does a decent job of cleaning up line noise and provides some AC line regulation as well, but they are large, heavy, expensive and have a limited load dynamic range. These are normally only used in applications where the load doesn't change since they will overheat if lightly loaded (below 20% of capacity). Typically you size them so the load is between 40 to 70% of the rated capacity of the MCR/CVS. These are noisy too (acoustically).
I have a hard time understanding why a ferro-resonant transformer would be considered a great choice in an inverter application when the possibility exists that there might not be a load connected to the inverter. I have seen CVS regulators burn up when run with the load disconnected. I guess a smart inverter could detect the fact there is no load and shut down to protect the transformer, but then you wouldn't have a power output. The efficiency of one of these would not be great either.
Incidentally large variable speed motor controls (called VFD's) are used to run extremely large motors rated in the hundreds and thousands of horsepower. These VFD drives use a version of the "Pure" sine wave drive concept. These change the output frequency to control the rotary speed of synchronous AC motors. VFD circuitry has been refined greatly over the last 40 years and is both extremely reliable and inexpensive. This is the same electronics that make electric cars possible.
I can't imagine that modern inverters wouldn't use the same power switching circuitry as VFD's. I can also think of no reason to use 80's technology instead. I can believe that a $200 inverter from China could be crap, but that is because they have shaved every penny out of the design, not because it uses a modern switching topography.
P.S. Modern switching systems can be designed to handle huge surge currents. You should see what kind of demands an electric motor makes on the VFD drive when operating a dynamic load. 10x surge currents are not unusual.