Understanding negative feedback OpAmp

(Original post by uberteknik)
I've used a servo mechanism to illustrate the operation. Feedback for an amplifier on its own uses exactly the same principle but makes it more difficult to visualise and describe.

Don't think of feedback purely in terms of amplification.

Think of it in terms of a summation junction producing an 'error' signal prior to the amplifier. i.e. the actual input signal summed with a sample of the required output and the resultant is an -ve feedback error signal passed to the amplifier.

The error signals function is to always drive the amplifier output to reduce the error between the sampled output and the input reference.

.

The output is required to track the input faithfully. Notice the summation box prior to the amplifier. Also you should be aware that servo amplifiers have a very high open loop gain typically >100x10³. because of that, a very small signal is all that's required at the input of the op-amp terminals to cause the output to produce a very large voltage swing. i.e. in this case 100uV would be enough to drive the output to 10V.

For the purpose of this explanation, assume the output needed is a 1:1 ratio with the input to control the position of say one axis of a robot arm

A transducer on the controlled device monitors the actual performance parameter and the transducer signal is passed back (fed back) to the summation junction which then subtracts the transducer output from the reference control signal.

i.e. the subtraction provides the error signal which is passed to the high gain servo amplifier.

Case 1 Output equal to input

If the controlled device is in the correct position, then the transducer output will match the reference signal exactly. The summation junction subtracts the two signals and zero error signal is produced. Nothing is amplified and no signal is produced to drive the motor which stays put. So far so good.

Case 2 Output signal exceeds input signal

If the transducer monitored signal exceeds the input reference signal (output overshoots requirement) it represents the robot arm position exceeding the required position.

The summation junction subtracts the more +ve output from the reference input and thus produces a -ve error signal which is then amplified. The -ve error signal is greatly magnified which then reverses the servo motor and the control device is forced to reverse direction to reduce the overshoot error.

Case 3 Output signal less than input signal

If the transducer monitored signal is less than the input reference signal (output undershoots requirement) it represents the robot arm undershooting the required position.

In this instance, the summation junction subtracts the less +ve output from the reference input and thus produces a +ve error signal which is then amplified. The now +ve error signal is greatly magnified which continues to drive the servo motor in the same direction only faster to the required position.


An error signal is only produced if there is a difference between the output and input. That difference only needs to be uV to cause a large change in the output to reduce the output error back to zero.

In this way, the output signal is 'slaved' to the input reference and the error signal produced by the summing junction dithers back and forth about the input reference signal producing a continuous a.c. error input signal to the amplifier. The amplitude and polarity of the error signal is dependent on the amount of overshoot or undershoot of the controlled device (robot arm) position.

You should note that there is an inherent delay between the timing of the input signal commands and the transducer output responding to the command signal, caused by the signal path propagation delay and electromechanical component inertia.

This delay changes the phase relationship between the output and input and if that phase difference exceeds 180⁰ then the error signal will produce the opposite effect and +ve feedback will result to create instability and locking up the output at one of the end stops.

Exactly the same principle applies to -ve feedback in an amplifier on it's own where the feedback signal is a fraction of the output provided by a potential divider.