opamp phase shift, how to think about it

[Asa Cannell]

I am having trouble understanding opamp phase shift. For instance, its
common to have 90 degrees open loop phase shift for a wide frequency
range, but if you were to connect the opamp in a non-inverting
configuration, there would not be any phase shift in the output vs.
the input in that 90degree range. Where is the phase shift? Or if you
were trying to make a phase shift oscillator, you would only need to
add the extra 90 degrees phase shift at one frequency to get it to
oscillate, so clearly the opamp is supplying the other 90 (plus 180
for the inverting input), but in the non-oscillating band, the phase
shift is zero, how does the phase shift suddenly appear?

Asa

 

[John Popelish]

I am having trouble understanding opamp phase shift. For instance, its
common to have 90 degrees open loop phase shift for a wide frequency
range, but if you were to connect the opamp in a non-inverting
configuration, there would not be any phase shift in the output vs.
the input in that 90degree range. Where is the phase shift?

It is still there exactly as it was before you closed the feedback.
That gain and phase shift data you see on the data sheet is the output
relative to the differential voltage on the two inputs. As long as
the opamp is not in saturation or slew limited output swing, that data
applies. That is a completely different thing than the gain and phase
between some other two points, like the + input and the connected
output and - input (which is really the voltage between the inputs).
When you connect up the feedback path, the output becomes two voltages
in series, with respect to the input. A very tiny one that is the
difference between the + and - input and a much larger one that is the
output. there is still a 90 degree phase shift between these two
parts of the total, but the differential input voltage is so small
compared to the output voltage (by the open loop gain of the
amplifier at that frequency) that you just don't notice it and only
see the output that is almost perfectly in phase with the + input (and
very nearly as large in amplitude). If the output did anything else,
there would not still be a tiny difference between the inputs so the
output would be driven toward the above described situation.

Or if you
were trying to make a phase shift oscillator, you would only need to
add the extra 90 degrees phase shift at one frequency to get it to
oscillate, so clearly the opamp is supplying the other 90 (plus 180
for the inverting input), but in the non-oscillating band, the phase
shift is zero, how does the phase shift suddenly appear?

Since the phase shift oscillator does not usually use the opamp as a
strictly differential input open loop amplifier, but has some
additional feedback across the opamp, the opamp is not providing a 90
degree phase shift to the oscillator loop. It can be connected to
have very near zero phase shift or very nearly 180 degrees phase shift
with a fixed gain. The feedback network that connects the output of
this gain block back to the input must provide 180 degrees of
additional phase shift to form a stable oscillator.

If you have a specific schematic in mind that would make this easier
to talk about.