Two common reasons op amps oscillate

When the feedback signal arrives at the inverting input, a perfectly damped response without delay, as shown in Figure 1, occurs. The op amp responds by ramping to the final threshold and slowly dropping off when the feedback signal detects closure at the proper output voltage.

The problem is exacerbated when the feedback signal is delayed. Because of the delay in the loop, the amplifier cannot immediately detect its progress to the final threshold, which in turn manifests as an overresponse in the form of a too fast move to the normal output voltage. Note that the more delayed feedback the faster the initial slope. The inverting input cannot receive timely feedback that it has reached and delivered the normal output voltage. It will overshoot the target and require many successive polarity corrections before the final settling time.

If it’s a small amount of delay, you might just see some overshoot and ringing. If it’s a lot of delay, then these polarity corrections can go on forever – and thus form an oscillator.

The source of the delay is usually a simple low-pass RC network. This is not a constant delay at all frequencies, but the gradual phase shift of the network from 0° to 90° produces a delay of first order approximation, td=RC.

There are two most common situations where an RC network can inadvertently form in our Circuit. The first case is a capacitive load (see Figure 2a). The resistor is the open-loop output resistance of the op amp, and of course the capacitor is the load capacitance.

In the second case (see Figure 2b) the feedback resistor and the op amp’s input capacitance form an RC network. Board connections at this sensitive circuit node are also an important factor in capacitance. Note that both Circuits have the same feedback loop, the only difference is that the output nodes are different. From a loop stability point of view, they create the same problem. Both of these factors of delayed feedback usually come into play — and it’s more troublesome if both are at the same time.

A little explanation is needed for the second case: a feedback resistor is usually not needed for a simple G=1 buffer, so a more common case is in a gain structure that uses a feedback resistor and resistor to ground ( See Figure 3). These parallel resistors in the R/C circuit form an efficient R.

We still have a lot to learn about Bode analysis of feedback amplifiers. So this simple visual representation of how delay or phase shift in the feedback path affects stability can help you diagnose and resolve some of the most common stability problems.