An Op Amp with 35 Degrees Phase Margin

This discussion relates to The Signal blog on settling time. A question was raised as to how I created two otherwise identical op amps, one with 90° phase margin and the other with approximately 35° phase margin.

Note that this “op amp” is not a real circuit design. It is a contrived circuit intended to convey some basic principles of slew rate and bandwidth and how they affect settling time. So I took some liberties.

The initial circuit (below) discussed in this blog on slew rate is intentionally “over compensated” to produce an idealized response with 90’ phase margin. It has an output stage formed with an infinite bandwidth ideal op amp provided in the TINA-TI circuit simulator.

  

For the blog on settling time, I wanted to easily change the phase margin of the above op amp without significantly changing the slewing behavior.

I first tried to replace the perfect, infinite bandwidth output stage with the generic op amp model in TINA. By changing the gain-bandwidth of this op amp, I could add a pole at an easily adjustable frequency. But when I put all the additional phase shift in the output stage it created an unnatural R/C response at the beginning of the slewing. Real op amps don’t have so much of this effect.

So I added this second op amp in the feedback path. This created the desired effect, more closely replicating the response of a real op amp. Note that the output stage op amp is not really required. I was in a hurry to complete the blog. :-)

The graph showing step response (below) was made by using the parameter stepping feature in TINA. R4 was stepped through values of 0Ω, 1kΩ and 3kΩ to create the gain steps of 1, 2 and 4. The gain-bandwidth of A2 is stepped from 10MHz (infinite, for purposes of this slow op amp) to 85kHz, to create the second pole in the loop response.

So, in short, I cheated. I didn’t create an amplifier with 35° phase margin. I created a feedback path that degraded the phase margin of the original circuit. The end result replicates the response of such an op amp in a convenient way to produce the needed graphs.

Note that the value of R1 is adjusted slightly to trim the offset voltage of the amplifier so that the detailed settling time graph (below) settles to 1V. R1 and R2 were included in this contrived op amp circuit to degenerate the input stage, creating a wider voltage range of small-signal settling. It also creates a more abrupt transition from slewing to small-signal to illustrate the points in the blog.

Hope that is all clear,

Bruce