Lab 6: Filter Design Project
ENG214: Circuit Analysis Laboratory
Tim Laux, Eric Brokaw, Thomas Approvato, Alin Bojkovic
The College of New Jersey
December 11th 2014
Table of Contents
1.
2.
3.
4.
5.
6.
7.
8.
9.
Application
Requirements
Research
Calculations
Solution
Equipment
Procedure
Results
References
Application
A singular loudspeaker is generally
incapable of reproducing the entire
audio spectrum with a linear
frequency response and without
distortion.
Most professional and high-end
systems use two or more drivers,
each catering to a specific range of
frequencies.
Each loudspeaker needs to be driven
by a signal with frequencies in its
linear range of operation.
Figure 1. Three-way speaker system
Requirements
We picked out the HiVi M4N, a
commercially available driver.
After examining its frequency
response plot, we determined that
it responded linearly between
100Hz and 5kHz. This makes it a
low-midrange driver.
Figure 2. HiVi M4N
Therefore, we require a band-pass
filter which has -3dB cutoff
frequencies of 100Hz and 5kHz.
Figure 3. HiVi M4N Frequency Response
Research
There are two ways to filter audio
signals: before or after amplification
(active or passive crossovers)
Before amplification (active):
Better overall sound quality
Highly tunable
Less expensive
Smaller/lighter
Requires multiple amplifiers
Figure 4. Active Crossover
After amplification (passive):
Requires only one amplifier
Lower complexity
Potentially expensive
Bulky/heavy
Power losses and non-linearities
Figure 5. Passive Crossover
Research
We chose the active filter route.
The two popular active filter
topologies are Sallen-Key and
multiple feedback (MFB).
Figure 6. Sallen-Key
We chose the Sallen-Key topology
because of its simplicity and its
suitability for our application.
In order to pass a wide band of
frequencies, we need to cascade
two filters, one high-pass and one
low-pass.
Figure 7. Multiple Feedback
Research
There are three major responses
possible from an active filter.
o Bessel
o Butterworth
o Tschebyscheff
We chose a Butterworth response
because of its passband flatness
and its relatively sharp transition
into the stopband.
Bessel was not steep enough, while
Tschebyscheff introduces some
ringing in the passband.
Figure 8. Comparison of different
filter responses
Calculations
We used Op-Amps for Everyone by Texas Instruments to design our filter
according to our needs.
Solution
First, we used LTSpice to
confirm the design
worked.
Then, we swapped in the
closest E12 capacitor
values and the closest
E24 resistor values. We
resimulated with these
values.
We were able to achieve
acceptable performance
even with the adjusted
values.
Figure 9. Schematic diagram of the filter
Solution
Bill of materials
Op-amp
1. LM324 Quad Op-amp
Carbon film resistors
1. 12K
2. 15K
3. 22K
4. 30K
Ceramic capacitors
1. 0.1F (2x)
2. 1nF (3x)
Total cost (single quantity) : $0.98
Figure 10. Circuit on a breadboard
Equipment
HP 54645D Oscilloscope
Agilent 33220A Function
Generator
Figure 11. HP 64645D Oscilloscope
Elenco XP-581 Quad Power
Supply
Breadboard
Figure 12. Agilent 33220A
Function Generator
Procedure
1. Using sources from online about op-amps and filter design techniques,
we drew the schematic for our filter
2. We built the circuit on a breadboard
3. We tested this filter using frequencies ranging from 10Hz to 60kHz
The op amp was powered by a 12V supply
The function generator was used to create the test frequencies
The output was probed with the oscilloscope and the peak-to-peak
voltage was recorded at each frequency
4. We created the circuit using LTSpice
5. We compared our experimental data with our calculated data using
LTSpice
Results
Band-pass Filter Gain vs. Frequency
Gain (dB)
3
0
-3
-6
-9
-12
-15
-18
-21
-24
-27
-30
-33
-36
10
LTSpice
100
1000
10000
Frequency (Hz)
Figure 13. Gain vs. Frequency Plot (Simulated and Measured)
Results
The results from
LTSpice were very close
to the ideal figures.
The measured results
were close, and had
errors less than 10%.
Taking component
variations into
consideration, our
results were
satisfactory.
Ideal
LTSpice
103 Hz
Measured
Cutoff frequency 1
(-3dB)
100 Hz
107 Hz
Cutoff frequency 2
(-3dB)
5000 Hz 5010 Hz
5400 Hz
-3dB bandwidth
4900Hz
5293 Hz
4907 Hz
Figure 13. Results comparison
LTSpice
Measured
Error
(cutoff frequency 1)
2.96 %
6.76 %
Error
(cutoff frequency 2)
0.20 %
7.69 %
Figure 14. Percent error
References
Carter, B. (2001). Active Filter Design Techniques. In Op-Amps for
Everyone.
HiVi Speaker. (2006). M4N Full Frequency. Retrieved from Swan Speaker:
[Link]
Maxim Integrated Products. (2003, February 4). A Beginner's Guide to
Filter Topologies. Retrieved from Maxim Integrated:
[Link]
bibin3210. (2012, May 8). Active vs. Passive Crossovers. Retrieved from
HiFi Vision: [Link]
