opamp phase compensation

[Kevin Weddle]

Does the phase comp. capacitor placed at the input reduce the capacitance because it's in series with the BC capacitance of the opamp transistors? The opamp in general kind of bewilders me because of the feedback resistors dictating exact gain and handling all the current. So I would not be surprised if I was wrong.

[audioguru]

Hi Kevin,
What is a phase compensation capacitor at an opamp's inputs?
I have seen a capacitor added between inputs to stop the input transistors from rectifying a strong nearby radio signal.

Most opamps have a built-in phase compensation capacitor.
Because an opamp has a very high open-loop voltage gain of about 200,000 and a lot of transistors, its phase-shift exceeds 180 degrees at very high frequencies that would cause it to oscillate when negative feedback is applied. So the internal phase compensation capacitor is added to reduce the open-loop gain to below 1 at the frequency when the circuit's phase-shift becomes excessive. Then it can't oscillate.

An old 741 opamp has its internal phase compensation capacitor selected to reduce its open-loop gain to 1 at 1MHz. It is just a single RC network (90 degrees max phase shift) so the max gain at 100kHz is 10, the max gain at 1kHz is 1000 and the max gain at 10Hz is 100,000. Negative feedback can be applied with two resistors to make any smaller amount of gain you want, below these frequency/gain limits.

The input resistance of an opamp is very high. So very little current is needed in the negative feedback resistors' voltage divider. The output transistors are complimentary emitter-followers that can drive about 20mA max into a load.

[Kevin Weddle]

I've read the cap can be put input to ground.

To start, I'm not attempting open loop gain at all.

Now, if I'm applying a high frequency that may cause excessive phase shift, why wouldn't I just select the correct gain and avoid adding the capacitor. I'm not trying to avoid a high frequency, I want to safely amplify it.

[audioguru]

I have no idea why you would want to short the high frequencies of the input signal to ground with a capacitor. Maybe the pcb is layed out with the input and output side-by-side causing capacitive coupling of high frequencies causing oscillation. Then a capacitor at the input shorts the high frequencies to ground at the input to prevent oscillation.

You forgot to attach the circuit you are talking about. Inverting or non-inverting?
You forgot to tell us which opamp you are using.
You forgot to tell us what frequency is the signal. Old opamps don't work at high frequencies.
You forgot to tell us the amount of closed loop gain you want.
You forgot to tell us what is the max p-p output amplitude required at the high frequency.

Maybe the opamp you selected doesn't have a good enough high frequency response for your high frequency.
Maybe the opamp you selected has a slew rate that is too low for your high frequency and the required output amplitude.

The output of an old 741 opmp can slew to an output of 28Vp-p up to a frequency of only 9kHz. Above 9kHz its max p-p output amplitude is less.
The max frequency that an old 741 opamp can produce with a gain of 1000 is only 1kHz.
The max amount of gain from an old 741 opamp at 1MHz is only 1.

[Kevin Weddle]

Maybe I'm off, but I think your saying the cap is there for prevention, should a high frequency occurance trigger an oscillation. Though my book suggests that it's also used to extend the 20db/decade open loop rolloff, which is more stable than the 40db/decade rolloff found as you increase the frequency. The 20db/decade bandwidth is widened by the cap.

[audioguru]

I don't know which circuit and i don't know which opamp you are talking about.
Fully compensated opamps (a few are not fully compensated) are made with built-in frequency compensation that has 20dB/decade rolloff so that their gain is below 1 when their internal phase-shift would cause oscillation when negative feedback is used if the gain was higher.

A fully compensated opamp still has overshoot and ringing on sharp transitions especially if it has a small capacitor load. Is that what you are talking about, having a capacitor added to the input to make it more stable?
A lead compensation capacitor can be added in parallel to the negative feedback resistor to also make an opamp more stable:

[indulis]

... if I'm applying a high frequency that may cause excessive phase shift...

Applying a high frequency signal to an opamp circuit doesn't change the circuits phase shift. Opamp "compensation circuits" change/alter the gain/bandwidth of the circuit and is fixed... the poles and zeros that the compensation circuit introduces don't move around just because you change the frequency of the input signal.

The transfer function of Audioguru's circuit is