# Stability and Positive/Negative feedback

I often read that positive feedback is used to get oscillators (since the output of the circuit is summed up to its input signal and so leads to instability), while negative feedback to get amplifiers (since its output is subtracted to its input and so leads to stability).

But it seems to me that this intuitive reasoning is in contrast with the analysis of stability: in fact a negative feedback system is not always stable and a positive feedback system is not always unstable (we may check their stability by using the Nyquist criterion, for instance).

So why is it often used the first sentence?

• Positive acceleration feedback is standard method of compensating for varying inertia in servomechanisms. (Varying inertia is commonplace in robotics) – Chu Jun 21 '19 at 15:51

You are not reading the words carefully. Tomato sauce is used to make a pizza, but not everything made with tomato sauce is a pizza.

Yes, positive feedback is used to cause oscillation. But that statement does not say that a circuit with positive feedback will always oscillate.

Similarly, negative feedback is used to achieve linear stable operation. But, the statement does not say that a circuit with negative feedback will always be stable and linear.

• Ok, but why should we prefer using positive feedback for oscillation, and negative feedback for amplification? – Kinka-Byo Jun 21 '19 at 15:05
• Because positive feedback tends to make things unstable and negative feedback tends to make them linear. We use tomato sauce to make a pizza because that is how you make a (proper) pizza. – Elliot Alderson Jun 21 '19 at 15:09
• Perfect, thank you very much. – Kinka-Byo Jun 21 '19 at 15:14
• The last sentence of the answer deserves a comment: Negative feedback is ALWAYS stable!! However, this is a pure theoretical/mathematical consideration because in reality each system with negative feedback (at DC and for lower frequencies) will turn into positive feedback for rising frequencies (due to unavoidable phase shifts). Therefore - depending on the loop gain - some circuits with negative feedback (at lower frequencies) can be unstable. Using the term "negative feedback" we apply it to the desired behaviour only - however, for larger frequencies it does not apply anymore. – LvW Jun 21 '19 at 17:35
• @LvW Thanks for clarifying, that was indeed what I meant to say. – Elliot Alderson Jun 21 '19 at 17:36

In general, circuits need the operating point controlled (the quiescent voltage and current of the amplifying devices), and that requires negative feedback of some sort.

With quiescent values being at DC (zero Hertz) thus the DC behavior being Negative, because DC_positive feedback would be an unstable operating point, we often classify the polarity of feedback at steady-state DC conditions.

• Yes - a very helpful answer. For example: Even a linear/harmonic oscillator needs negative feedback at DC and (of course) positive feedback at one single frequency only! – LvW Jun 21 '19 at 17:46