If output going to clip , will it burn the mosfet ??

[phase_accurate]

I once had a circuit idea for reducing charge current (specially at startup) with enlarged high-side storage caps. I have never tried it though.
You have the ubiquitous 100nF cap as usual. Additionally you have a larger one (the one for prolonged on periods of the high-side switch) that is connected to the main storage cap via a resistor in parallel with a diode. Dischargeing of the additional cap is done via the diode and chargeing via the resistor.

Regards

Charles

[titanchen68]

1.The 100nF cap you said , Is it mean the bootstrape cap??

2.So , I have to change my bootstrape cap to 100uF , maybe 220uF???
I still get some confuse , I will try it later.

[phase_accurate]

Here is what I mean:

Regards

Charles

[classd4sure]

Originally posted by titanchen68
oh?? should be using 100uF ??? do you mean C25 on my schematic??
my schematic used 1uF // 104 be the bootstrape cap yet.
But the member "SSassen" told me should use 470nF be the bootstrape cap.
He said that "the bootstrap capacitor should be between 0.1 ~ 0.33uF, yours is too large.
I use 1uF//104 yet.... but he said too large.
And when i use 1uF//104 , i can not turn on the amp normally.
The output of the error_amp going to saturate.
When i change c25 to 470nF, it work find.

Are you seriously using three mosfets in //?

Originally posted by Pierre
Sorry but experience makes me disagree with ClassD4sure also.

1uF is optimum. I have played with a lot of versions of my class-D amp, based on a IR2113 chip with bootstrap cap. I started with 47uF, but it produced a very audible "click" at turn on.
I reduced it and found that 1uF produces almost no audible "click" while still maintaining full charge even clipping with a 10Hz signal.

You can have a look at the IR app.notes and find that for typical applications (200-400KHz, usual mosfets, 12-15V Vcc, etc), you theoretically need less than 100nF, although it is good to have at least 10 times that.

That's all fine, but doesn't begin to tell the complete story, or give a better explanation as to why his output stage falls apart at clipping.

IR app notes calculate the minimal value required and don't tend to take into account 20% clipped at 20Hz.

A click at turn on can be considered a secondary effect and could easily be delt with via other means than simply reducing the bootstrap cap value.

Charles, that's a real neat idea and I think would be the easiest bandaide.

[titanchen68]

Originally posted by classd4sure


Are you seriously using three mosfets in //?

Yes , i am. beacause I want to load to 1ohm.

[titanchen68]

I try to change my bootstrape cap to 100uF.
But It can't turn on normally.
I test the output of the error_amp is going to saturate.
so , i change back to 1uF , It became working.
does my C38 too small??

[zkaiser]

titanchen68, hi

you again have mistake in yours sch. , like
you cannot use same cap for startup delay on ir2110 and for power supply decoupling
diode D5 is connected with its kathode to gate of one of upper mosfets insted to +rail (+55V)

BTW my oppinion is to you need to use 10uF bootstrap cap (tantalum) and to keep 100ohm gate resistor and add some dead time at inputs of ir2110, just add appropriate RC part.

Best regards
Zeljko Kaiser

[titanchen68]

Originally posted by zkaiser

you cannot use same cap for startup delay on ir2110 and for power supply decoupling
diode D5 is connected with its kathode to gate of one of upper mosfets insted to +rail (+55V)

1.I am not sure about what you mean ??

2.have i need to add RC dead time control??
what is advantage for ??