Re-thinking of my D-amp...full bridge or not.

[ChocoHolic]

Originally posted by SSassen
.... now where did I put my coffee? Pour me another!

[ChocoHolic]

... another simple principle was derived in this thread.

Distorsions in class D

Our class D amp guru's lead me step by step to an interesting design principle. (Thank's again!)
Last posting of page 11 shows how a full bridge could be driven with an unbelievable simple self resonating principle.

...did not try it, but it is still in my mind for my subwoofer...

[classd4sure]

Wow Choco, long time no see in class d land.

Pat, have a look at the Mueta schematic pdf at www.mueta.com you'll see an example of a differential amp in full bridge and I think you'll be able to see how the gain is set.

[pat allen]

thanks, good schematic.
I can recon the integrator, the oscillator, the difference amp for the full bridge, wich is what i am trying to add to my amp, and wait, there is a transfo...
im lost.

[classd4sure]

pay no mind just look at the differential amp

[pat allen]

this is exactly what i did, except that i take the signal before the lc filter... (did you see my test sch?)
the mueta difference amp got like 1,12 gain...so they must compensate for the remaining wanted gain into the rest of the feedback loop...looks simple.
am i close or not?

[ChocoHolic]

...I think the intension of classd4sure is:
instead of R13, R14 & C8 use the attached circuit.
This differential amp has a gain of 1/8.2, but two times
of the input compared to the former feedback network.
So the new feedback resistor has to be 4.1 times less than
the original (46k+1k), means 11.46k. I simplified it to 12k.
The low pass is performed also in my attached dif amp by the two capacitors.

Please be aware that all inconvinieces of feedback systems including insatbility can also happen in class D amps.
The active dif amp is adding another pole at high frequencies (depending on the chosen OP amp) and could theoretical make the frequency compensation of the entire amp more difficult.


cd4s:
...yes, I have been quite silent in any DIY topic for a long time. My bosses are keeping me busy...

[ChocoHolic]

But ugh..!
It is quite ugly to run an OP amp with such massive rectangular signals. Behaviour of normal OP amps may differ very much from the ideal OP in this application.... and very fast OP amps might directly start to oscilate, because the dif stage has very low gain and heavy feedback.
Last but not least, the combination of 1k and 8k2 allows rail voltages up to approx. +/-40V, above you may drive the OP amp into clipping (also depending on OP voltage supply and OP amp type)....
If you want to use such a diff amp it is probably better to pick the feedback signal from behind the output filter. But be careful the output filter has quite a crazy phase shift behaviour and you would probaly have to redesign the entire feedback oscillation system....

I am sorry, but I feel things will remain difficult in realty, even if the basic approach seems simple and straight forward.

[classd4sure]

Many excellent points raised.

That's why the ideal would be true differential feedback, along with true differential inputs.

[pat allen]

Hi, thanks for the comments.

I have put that aside a bit, i received my new pcb !!!

i assembled one, it fired up right away !!
my new protection feature works very well, the microprocessor monitor DC, pump up and pump down.

square wave at the output is super clean, no ring.