Final Project - EE 420L
Authored
by: Roman Gabriele Ocampo
Email: ocampor5@unlv.nevada.edu
Date: May 5, 2014
Design Of A General Purpose Op-Amp
Prelab:
Lab
Description and Goals:
- The goal of this lab is to build a general purpose op-amp using
only two CD4007 chips and as many resistors and capacitors as needed.
- Using the models created in Lab 8, design and simulate the op-amp in LTspice with different gain loads.
- Build the op-amp on PCB and obtain the experimental results.
- Compare the simulation and experimental results.
The Op-Amp:
In the
process of creating this op-amp, the first major design constraint was
the limitation of only two CD4007 chips. If one looks at the CD4007 datasheet,
each chip has 3 pairs of NMOS and PMOS transistors. Each N/P pair is
gate connected, which means that while each chip has 6 transistors (12
total), only 3 per chip (6 total) can be used if independent gates are
desired.
As
such, sacrifices must be made. In class, a
simple op-amp is created using a differential pair (with a current
mirror load and tail) as the first stage with a push-pull amplifier as
the second stage. Such a configuration would require at least 8
independet NMOS and PMOS gates (though the push-pull stage and the
biasing current mirror can take advantage of the NMOS/PMOS tied gates
and/or the tied drain/sources). Since only 6 independent gates are
available, resistors must be used in place of the transistors in both
the diff pair and the output stage. The final design is shown below.
The pin associated with each transistor is included. The letter "a"
denotes the upper CD4007 chip, while the letter "b" denotes the lower
CD4007 chip.
The
gain is significantly decreased in this design (as shown later). In
this design, M6, M8, and M10 come from one chip (to ensure good
matching for the M8/M10 current mirror) and M7 and M11 come from the other chip.
Open-Loop Gain:
The
open loop gain of this design is only 25dB, which is roughly 17V/V.
This is terrible. Ideally, the open loop gain of an op-amp would be
infinity. The simple op-amp discussed in the book has a gain of
1,000V/V. The op-amp's low gain will result in significant errors when
put into different closed loop gain configurations (shown below).
Noninverting Topologies:
The closed loop gain in the noninverting topology is ideally Av=1+R2/R1.
For Av=1:
From the simulation, the gain is 0.86V/V. From the experiment, the gain is 158m/194m = 0.81V/V.
For Av=5:
From the simulation, the gain is 3.55V/V. From the experiment, the gain is 624m/192m = 3.25V/V
For Av=10:
From the simulation, the gain is 5.84V/V. From the experiment, the gain is 968m/184m = 5.26V/V.
Inverting Topologies:
The closed loop gain of the noninverting topology is Av=-R2/R1
For Av=-1:
From the simulation, the gain is -0.66V/V. From the experiment, the gain is -110m/182m = 0.60V/V.
For Av=-5:
From the simulation, the gain is -3.15V/V. From the experiment, the gain is -544m/188m = -2.89V/V.
For Av=-10:
From the simulation, the gain is -5.30V/V. From the experiment, the gain is -876m/188m = -4.66V/V.
Conclusion:
From
the simulations and experimental results above, we can conclude that
the op-amp I designed does indeed exhibit operational amplifier
behavior, albeit having an extremely low open loop gain. In the
different closed loop gain configurations, the output does achieve
different gains for different configurations, but it is nowhere near
the ideal values due to its low open loop gain.
To improve
this design, more chips should be made available. Rather than having
resistors on the diff pair, improvements can be made by substituting a
PMOS current mirror load. For the output stage, a proper push-pull
amplifier can be included. Including these changes would significantly
increase the open-loop gain of the op-amp, reducing the closed loop
gain errors.
Return to the listing of my labs
Return
to the whole class reports
Return
to the EE420L site
Return to CMOSedu.com