# How to decouple a negative feedback loop in DC?

My question is open ended, but I provided an example to explain the question better. R19 and R8 in the middle of the pic form a negative feedback loop. They set the gain of the amplifier.

I need to decouple the NFL in DC, in order not to fry the circuit, but if I add capacitor, Q2 will not be powered on.

My first idea was to add a voltage divider to bias Q2 to it's value, but that doesn't work, because it screws up all the calculations for the NFL.

I actually need to build this and our professor said it's probably to hard to do, but I'm asking to see if anyone has any ideas.

EDIT: Added a pic without the currents and voltages.

• It would be a lot easier to decipher that drawing if you turned off all those voltage and current boxes. Commented Oct 25, 2017 at 19:19
• Added a pic without the boxes. Commented Oct 25, 2017 at 19:26

You need to provide a DC path for the differential pair inputs. It is usually done like this:

The DC path to the differential pair goes through R1. You need to make sure that the R1 impedance is large compared to the C1 impedance at your frequency range, and your feedback resistors in your case, in order to avoid loading effects.

In your case you will heave C1 in the feedback path. You need to make sure the feedback resistors are large compared to C1 at your frequency range as not to affect the gain.

The control-range for gain is far too large. And the rin of Q2 needs modeling.

Add 50Kohm from base Q2 to ground. Make R8 (in series with the FET) much much larger, perhaps 1MegOhm.

The Ic of Q1/Q2 is 0.1milliAmp. reac is 0.026v/0.0001Amp = 260 ohms; scaled up by beta, that likely is 260*100 = 26,000 ohms. So you need Darlington diffpair.

• Adding a resistor to the base of Q2 breaks my negative feedback loop and it doesn't produce oscillations any more. Commented Oct 27, 2017 at 13:35
• Also adding a Darlington differential pair makes it not oscillate and I don't really know why, because the currents coming out of the pair are exactly the same as before. Commented Oct 27, 2017 at 13:53
• Doing all the things you said here makes it not oscillate. Commented Oct 27, 2017 at 13:59