DC stabilizer for integrator (class D amplifier)

Thread Starter

moretti1993

Joined Sep 11, 2016
6
Hello guys.
I have this diagram of a class D amplifier that im trying to understand and i stumbled across this "DC stabilizer".
Obviously the first op-amp creates a triangular curve. But what's the purpose of "DC stabilizer" and how does it work? To me it looks like another integrator with feedback to the first one.

Thanks
upload_2018-11-17_20-46-56.png
 

OBW0549

Joined Mar 2, 2015
2,985
But what's the purpose of "DC stabilizer" and how does it work? To me it looks like another integrator with feedback to the first one.
That's exactly what it is. U3, R9 and C10 form an integrator which is used as an error amplifier, feeding back to U2 via R6 and R7, and which acts to keep the triangle wave at U2's output centered about zero, regardless of any DC offset in the incoming signal at SQ_AC. The RC time constant of the U3 integrator is a lot longer than that of the U2 integrator, to avoid distorting U2's triangle wave output.

This is a commonly-used method for getting rid of DC offsets in signals at the outputs of high-gain integrating amplifiers; see the figure at the bottom of p. 1 of this data sheet for another example, and also the diagram at the bottom of the last page.
 

Thread Starter

moretti1993

Joined Sep 11, 2016
6
That's exactly what it is. U3, R9 and C10 form an integrator which is used as an error amplifier, feeding back to U2 via R6 and R7, and which acts to keep the triangle wave at U2's output centered about zero, regardless of any DC offset in the incoming signal at SQ_AC. The RC time constant of the U3 integrator is a lot longer than that of the U2 integrator, to avoid distorting U2's triangle wave output.

This is a commonly-used method for getting rid of DC offsets in signals at the outputs of high-gain integrating amplifiers; see the figure at the bottom of p. 1 of this data sheet for another example, and also the diagram at the bottom of the last page.
Perfect! Apparently the setup is called "DC servo" Took that from the datasheet.
Thanks a lot for the quick answer and explanation. I'll proceed my study and ask if necessary :)
 

Thread Starter

moretti1993

Joined Sep 11, 2016
6
Okay im starting to grasp this circuit, but i have one more question.
What is the purpose of the resistors in the output stage? (680 ohm)
upload_2018-11-17_23-57-55.png
 

ebp

Joined Feb 8, 2018
2,332
The capacitors and resistors form a damped low-pass filter. The capacitors create a local path back to the supply rails for some of the high frequency component that exists because of class D operation. Since I'm not sure whether the inductor is actually in the circuit or not, I'm not entirely certain why the resistor would be used.

With the inductor in the circuit it is likely that there would be a very large peak in the filter response at resonance. In that case, the resistors would presumably be intended to dampen the filter. The net resistance does seem very high to me if the intent is just damping, however.
If the inductor were replaced with a short circuit, the capacitors would represent a very low impedance at the switching edges of the waveform, resulting in high amplitude current spikes and possibly making noise radiation worse. In that case, the resistors would limit the current amplitude through the capacitors. In this configuration, the RC network across each FET is much like a typical snubber to control voltage "spikes" and dampen ringing that might occur due to resonance of the FETs' capacitances with every blasted stray inductance in the circuit.

It is odd that MKP (metalized polypropylene film) capacitors would be used, but audio people have strange ideas. MKP is a very good choice for lots of analog circuitry because of low dielectric absorption and good temperature stability. Film capacitors are mediocre at high frequency, where monolithic ceramic types are superior. At 470 pF, C0G ceramics are small and inexpensive and at least equal to MKP in low frequency performance.

Multiple resistors in parallel are presumably to increase power handling and maybe to keep parasitic inductance low, though depending on the type of resistors, you'd have to be way up in frequency for the inductive reactance to start to approach the resistance - so maybe just for power handling. There isn't a lot to gain by fussing too much about resistor inductance when you use a film capacitor - they aren't terrible if the construction is "right" but they are generally much worse than ceramics for parasitic inductance.

Don't forget that an ideal voltage source (the power supply) should have zero impedance at all frequencies, so the +40 line and whatever is at the bottom should be close to a short circuit at high frequency. How well they are short circuited at high frequency depends on the amount of (HF-effective) capacitance between them.
 
Top