L4: Putting OpAmps to Use: Followers, Gain, Inversion, Subtraction Feedback Dynamics

1
L4 Putting Op-Amps to UseFollowers, Gain,
Inversion, SubtractionFeedback Dynamics

• ICB EngineeringModeling and ControlEngineering
  of Compartment Systems
• Fall, 2006

2
Review Our final Model of an Op-Amp
3
The Simplest FeedbackThe Follower
4
What does this teach us?

• Output of follower converges to input
• Output of follower resists disturbances
• Op-Amps Behavior
• Trys to move output to get inputs to agree
• Eventually, inputs agree (almost perfectly)

5
The Op-Amp Follower
Why is This a Useful Circuit?
The Difference Between Forklifts Eyes and Arms
Power Steering in Cars and Ships
6
How about this one?
7
Using Op-amps for Gain
What are these?
8
Using Op-amps for Gain
Rtotal 2 R
V Out / 2
I Out / (2 R)
9
Using Op-amps for Gain
Out Converges to 2 X In
V Out / 2
10
What does this do?
Out Converges to -In
Converges to 0
11
Whats going on here?
12
Take Away Lessons

• Negative Feedback around a integrator loop causes
  the integrator output to converge to a value
  where the integrator input is 0.
• i.e. negative feedback drives error to 0.
• Many useful functions can be built this way

Error
13
Feedback Dynamics
14
What Determines How Fast We Converge?
The Gain! Low Gain Slow convergence High Gain
Fast convergence
15
Can we figure out exactly how fast?
What functions looks like themselves after
integration or differentiation?
A Exponentials!
What are the units?
What is ? The Time Constant
A Seconds
16
How far do we converge in one time constant?
This is
of the distance from out(0) to 0
or 63 of the way
17
How far do we converge in two time constants?
This is
of the distance from out(0) to 0
or 82 of the way
18
How far do we converge in three time constants?
This is
of the distance from out(0) to 0
or 88 of the way
19
Big Idea
of the way there
Each additional time constant gets us another
Self-Similarity All of the state at any time is
captured in the value of the single integrator,
so we can arbitrarily pick t0 at any point along
the way, and see the same (scaled) exponential
convergence with the SAME time constant.
20
What it the Gain is very High (e.g. 107)?
Then the time constant is very small! e.g. 10-7
seconds (100 ns).
Result System converges very quickly. We can now
shift out attention to the
quasi-static equilibrium, and assume the
convergence happens really really
fast (i.e. on a time scale finer than we care
about).