LabGruppen, what to do?

[TOINO]

I have already tried that but in vain...

 

[Eva]

Try again...

 

[TOINO]

You can close the door again

 

[DjLeco]

This waveform that you call junk is carefully shaped to exhibit exactly the same average value as the own audio signal, so it does not produce any distortion, it only adds carrier residuals to the output, 630Khz (or whatever) and its harmonics. This is the trick. You are telling us that you don't understand switching mode electronics at all, and as a result, you just hate it, you call it junk and you discard it (typical human behaviour).

There is nothing wrong with shaping audio signals through high frequency carriers and integrating the HF stuff back to get the audio stuff. Rail modulation is actually a nice thing because output devices are operating at constant Vce at audio frequencies and this makes them more linear.

Very good said here, Eva!

Also at 7,5 Vdc between Emmiter and Colector at AB class bipolar stage, SOA, is ENORMOUS, instead greater Vce, for bipolar transistors!

Practically under 10Vce is maximum current disponibility from transistor as shown, best SOA possible!

http://postimage.org/image/4c20scdfz/

 

[Raul Freire]

Epoxy or just a thick lacqer coat to make the assembled PCB moisture-proof can make wornders. Sensitive parts are the big coils and capacitors as usual.

Here you can see an early prototype:

This included PFC or not? And it is of 4.8Kw?