Filter Project
Jen Lee, Rupa Pandya,
Sagar “the man” Rathie
April 18th, 2007 – PHYS
352 – Digital Electronics
Design:
The scope of our design included the
combination of a high Q split supply Bandpass-Biquad
Filter with the addition of an active
Characterization:
The combination of the chebyshev high pass filter and the biquad low pass filter created a response like that of a bandstop, or notch, filter. The output was attenuated to match the low pass and high pass combination transfer functions as shown below.
Transfer and other
Relevant Functions:
We were able to find the transfer functions of each respective design (biquad, highpass, and combined) and created charts showing their phase and magnitude bode plots. They are attached.
What we learned:
This lab was very beneficial to us in understanding how applying various filter designs in different configurations can drastically change the output of the device. The active elements of our filters were also a new attribute that we were able to work with, understand and see how the quality factor of these devices compare to simple analog circuits. Additionally, we were able to see hands-on experience as to how mathematical approaches to electronics can be very helpful in the lab.
|
Combined
Filters |
Biquad Filter |
|
||||||
|
Hz |
Gain |
Phase |
Hz |
Phase |
Gain |
Hz |
Gain |
Phase |
|
1.0E+00 |
5.4E-14 |
-98 |
1.0E+00 |
174 |
2.2E-10 |
1.0E+00 |
4.0E-13 |
90 |
|
7.9E+00 |
4.5E-13 |
-90 |
7.9E+00 |
180 |
2.3E-10 |
1.6E+01 |
6.4E-12 |
90 |
|
2.5E+01 |
3.2E-12 |
-92 |
3.2E+01 |
171 |
2.8E-09 |
7.9E+01 |
6.9E-11 |
94 |
|
3.2E+01 |
3.1E-12 |
-91 |
2.0E+02 |
0 |
6.1E-12 |
7.9E+02 |
3.3E-10 |
84 |
|
5.0E+01 |
2.0E-12 |
92 |
7.9E+02 |
0 |
3.8E-13 |
3.2E+03 |
1.2E-09 |
30 |
|
2.0E+02 |
3.0E-13 |
94 |
2.0E+04 |
-57 |
5.7E-17 |
1.3E+04 |
1.5E-09 |
10 |
|
5.0E+02 |
1.3E-13 |
100 |
6.3E+04 |
177 |
3.1E-14 |
3.2E+04 |
2.5E-09 |
-1 |
|
2.5E+03 |
5.4E-14 |
40 |
2.0E+05 |
170 |
3.4E-13 |
1.3E+05 |
1.4E-09 |
-20 |
|
1.6E+04 |
1.7E-15 |
1 |
5.0E+05 |
156 |
2.0E-12 |
3.2E+05 |
1.0E-09 |
-43 |
|
3.2E+04 |
5.5E-15 |
-60 |
2.0E+06 |
104 |
2.1E-11 |
4.0E+05 |
9.0E-10 |
-49 |
|
4.0E+04 |
1.1E-14 |
60 |
7.9E+06 |
16 |
4.5E-11 |
1.3E+06 |
3.7E-10 |
-66 |
|
7.9E+04 |
5.1E-14 |
177 |
|
|
|
1.6E+06 |
3.1E-10 |
-66 |
|
1.3E+06 |
1.0E-11 |
115 |
|
|
|
4.0E+06 |
2.2E-10 |
-71 |
|
7.9E+06 |
3.5E-11 |
17 |
|
|
|
5.0E+06 |
2.2E-10 |
-83 |
When choosing
a complementary filter to use in conjunction with the biquad the original plan
was to build a Sallen Key Chebyshev
high pass filter. However, the break
frequency was discovered to be far from the calculated value, thus we built a
Twin T Chebyshev high pass filter. A schematic is provided below.
The
relative resistor and capacitor ratios were assigned as indicated to the left
of the schematic. We did not have the
exact values of resistors needed, however we used the
closest values. R1 and R2 had values of
820 ohms, R3 390 ohms, C1 .1μF and C2 2.2μF. With an overall R value of 1000 ohms and a C
value of .1μF, the theoretical break frequency was calculated to be about 1.59kHz. The high
pass response was tested using a function generator and an oscilloscope. It was found that the break frequency
occurred at about 1.39 kHz. Considering
that our resistors were not of the exact ratio as expressed in the schematic,
this error of 12.6% was expected. 

The graphs above demonstrate the gain and phase behavior of the filter when a sine wave was applied to the filter. The high pass filter did show the expected behavior of attenuating low frequencies while allowing higher frequencies to pass. The phase also showed the expected behavior, with the phase change decreasing as the frequency increased.

Easily the most important aspect of our filter was the
use of the split supply Biquad bessel active filter. This device is a
combination of two-integrator feed back loops and an invertor which we
characterized to act as a low-pass filter.
This was done by taking the output after the inverter. For our design,
we used various resistor combinations to achieve a break frequency at about 10
KHz, however our actual value was around 11.2KHz. An error of 10.71% is
understandable considering the error discrepancies and magnitude difference
between resistors. We selected our capacitors to be exactly 22uF (C1 and C2) and
our resistor values were 227Ω ( R2, R4) and 10KΩ (R1, R3, R5, R6)
depending on the feedback desired.
Below, we have 2 graphs indicating the gain and phase when a sinusoidal
signal was applied to the filter. Our filter worked well as a low-pass,
allowing a majority of signals to pass before cutting off at 11.2Khz. However, ater the cutoff frequency, the
filter started passing signals again. This could possibly be due to the op-amps
being used.


Combined Filter Characterization
As per project
specifications, we were asked to combine a basic filter with a biquad of some
sort. Our group chose the Cheybyshev high pass filter in conjunction with the
bandpass biquad low pass filter. With the combination of the low pass bandpass
biquad first and then the high pass filter we expected to have a resulting
bandstop, also known as a notch filter. This
would collectively limit only a certain frequencies that would pass through the
filter. This frequency range was dependent on the characteristics of the filter
combinations itself, including the Q factor, active elements, RC combinations,
and overall break-frequencies for component circuits. We installed our
high-pass filter to input the output from the Biquad at the low-pass port. So
essentially we combined the high-passing elements of the Twin-T and the
low-passing elemnts of the biquad to create a region of conflicting activity.
This collective area is the actual band-stop range. Our break frequency for the
HighPass was 1.39KHz and for the Biquad was 11.2KHz. We hypothesized that the
Band-Stop range would occur between 1.39Khz and 11.2KHz, however, it was
actually larger, occuring between 99Hz and 10.2Khz. We believe the reason for
the offset was in response to the respective Q factors of each circuit. The
biquad’s Q factor actually decreases with frequency, while the active high
pass’ Q factor remains constant. Thus, at lower frequencies, the high-pass
filter could “overpower” the effects of the biquad to create a passing element
which attenuates at smaller frequencies.
Our circuit diagram is shown above and our graphs of Gain and Phase are
attached below. Additonally, we have attached our calculations for determining
Poles and Zeros for the filters. Yeay for Filters.

