The actual control circuits can run a simple SSR or a mechanical relay to do the actual connecting and disconnecting as the input power goes up and down.
Its standard in every one of my systems.
I even included it in the 'Tiny GTI, maybe' thread. By unhooking the main power transformer every time the input power drops below the minimum feed back power point for more than a few seconds it automatically disconnects from the line.
So all that is powered from the line is the control circuit at that point and that has a power drain of less than 2 watts including the control transformer if one is used in the control circuit at all. For comparison my LED digital alarm clock has a input power rating of 10 watts.
I have posted several very basic voltage sensing circuits that are specifically designed just to handle that line connection function. Each one costs only a few dollars to make and can either run a mechanical relay or a SSR. I prefer SSR's because there can be a lot of actual cycling during a typical day.
As far as the actual operating efficiency the standard iron core 50- 60 cycle conversion is actually surprisingly efficient. I have done many different home made versions of the IGBT type H - bridges and I use the standard IR2110 IC's designed just for direct control of Mosfet or IGBT half bridge switching devices. They have the correct control functions built right in and automatically drive the gates at full saturation voltage with incredibly fast rise and fall times. Plus they use a standard logic level compatible input signal.
That IC gives a real life application the highest possible switching device efficiency they can run at so very little input energy is wasted by the actual switching itself. Its the same efficiency a HF type switcher will have but only running a much lower frequency.
That leaves the iron core transformer as the big efficiency loss device. But typical transformers operating efficiency is still 85 - 90 % over its full operating range. The smaller transformers do have some what less actual efficiency when compared to bigger transformers though. However again by using a small toroid core transformer that efficiency can be raised back up.
Anyone can check with any transformer manufacture and get information on the actual efficiencies of their transformers and then know exactly how efficient they are. The numbers are alway higher than everyone seems to think that a LF transformer is. I wish I knew why its such a common myth that LF transformers are inefficient. They are one of the most efficient voltage or current converting devices ever created.
On my actual designs I have measured and confirmed actual conversions efficiencies of around 85% at peak power levels.
The as calculated losses from the switching devices is less than 5% of the total input energy at its least efficient point. The other 10% loss is the transformer.
Not to bad for home made!
Anyone using a Mosfet or IGBT based H - Bridge that is driven off of IC's designed specifically for voltage controlled switching devices will likely be able to confirm this. It standard commercial inverter drive circuitry. I have even found some of the cheap 12 VDC to 120 VAC portable inverters use these IC's or other cousins of them for driving the output H - bridges that work on 170 volt DC to create the Ac output itself.
Most home made HF switching power supplies will not be any higher in efficiency either. Getting over that 85% hump requires more expensive parts and more complex control.
Its not that I am against HF based designs. I just have not seen any real life commercially made switching power supplies that had efficiencies much over 85% except for the HF fluorescent light ballasts used for the large 4 - 8 ft tube lights. They are commonly rated at 90 - 96 % peak efficiency.
I am glad a few people are now taking the home made GTI concept far more seriously! The more people doing it the better!