Lab 4 - ECE 421L
Authored
by Nicholas Banas,
Banasn1@unlv.nevada.edu
2/23/15
This lab covers the gain-bandwidth product and slew rate of op-amps.
Op-amp LM324 Non-Inverting Bandwidth
The
bandwidth of the LM324 in the non-inverting configuration can be
calculated by using the GBP (Gain Bandwidth Product) of the chip with
the closed loop gain of the circuit design. From the data sheet
the GBP is:
While
the GPB will vary somewhat with different circuit topologies, this is a
good ballpark estimate for these calculations. The calculations
boil down to GBP/Acl=BW. The base bandwidths for our
non-inverting topologies are 1x=1.3MHz, 5x=260kHz and 10x=130kHz.
Using a spice model for the LM324 we can calculate the values of the 3dB BW for the different topologies:
You
can see from the measurements that BW1x is 882kHz, BW5x is 148kHz and
BW10x is 66kHz. These are significantly lower than the estimates
from the data sheet. We'll take some real measurements and see
which is closer. A quick note on the x1 measurement, the spice
sim for 1x seems to have an issue around 700kHz that makes for
inaccurate data. I took the measurement from 0dB instead of -3 as
it seems to be very close to where the -3dB should be.
The Bode Plot above illistrates how the output decreases at a steady rate after the 3dB BW has been reached.
1x Amp at 3dB | 5x Amp at 3dB | 10x Amp at 3dB |
| | |
The
real world measured results are fairly close to the sim results. The
measurements seem to suggest that the GBP is closer to ~700kHz.
Variations in the LM324
So
why does the data sheet have such a high GBP. Well the data sheet
also lists the test condidtions that the chip was under to measure the
GBP listed. The largest difference in our setup from the listed
conditions is the Vcc. Lets retake the measurements at 1x gain
with 30Vcc and see the differences.
You can see the difference is significant. It appears that the 3dB frequency should be much closer to 1.3MHz now.
5Vcc 3dB Frequency | 30Vcc 3dB Frequency |
| |
The
3dB frequency is now 1.1MHz, much closer to the 1.3MHz listed.
While the chip is able to be operated at small Vcc's, they appear
to effect the performance significantly.
The Inverting Amp
We created a basic inverting amplifier using the following schematic.
The
calculations for the inserting topologies are slightly different, with
the equation being GBP/(1+R2/R1)=BW. So our estimates are
-1x=650kHz, -5x=220kHz and -10x=120kHz.
These
calculations differ again. The -5x and -10x can be explained by
the reduced Vcc, but the value for the -1x seems to imply that the BW
should have been estimated with GBP/ACL=BW. We will see what the
real world measurements are.
Once again, the bode plot shows the decline in the BW over higher freqencies. The -1x seems to have worked correctly this time.
-1x Amp at 3dB | -5x Amp at 3dB | -10x Amp at 3dB |
| | |
The lab measurements show 3dB frequencies of: -1x=813kHz,
-5x=120kHz and -10x=55kHz. The values for -5x and -10x are very
close to the sim while the -1x is actually higher than the sim.
Once again, these values seem to suggest that the correct
calculation should be GBP/-Acl=BW.
Measuring the Slew Rate
We designed the following circuit to measure the slew rate of the LM324.
The
sim plot shows the response to a sine input. You can see how the
output can not keep up with the input. While there is some error
due to the turn around response, it is a very close estimate to just
use (voltage increase)/(rise time) to get the V/us value for the
slew rate.
Testing
the in the lab shows that the circuit seems to work as designed.
With a voltage rise of 1V and a rise time of 2.2us, the measured
slew rate is ~.45 V/us.
We are able to use the same circuit with a pulse input.
The graph shows a similar result to the square input as with the sine input.
The
lab measurement shows an even nicer output with a smaller effect from
the turn around. The measurments with the square wave are 1V over
2.15us. This shows a slightly higher slew rate of .465 V/us.
From the data sheet:
This
is very close to our measured value. This is because we used a
similar setup to the one used for the calculations. Not show in
our schematics is a 5k load resister that we inserted after we started
the test. The measurements without the resistor were very inaccurate,
with almost no slew restriction. This is because the slew limit
is strongly influenced by the current sourced/sunk by the op-amp.
A
note on the difference between the square and sine inputs: while the
same setup can be used for both inputs, the frequency or input voltage
for the sine must be much greater. Measurements in the lab the
lab seem to indicate that the minimum V*f of the sine input must be
about 4 times larger than the square input for usable results. We
were able to use the same setup because we used the min values for the
sine for both inputs.