Design of Second Order Bandpass

Active Filter For Super Tweeter Speaker

LEE CHING YEE HOR WEI JIE
A15KE0099 A15KE0077
DHARSHINI A/P SADACHARAM DARANISWARAN A/L THIAGARAJAN
A15KE0049 A15KE0047
Abstract—This paper presents the results of a reproducing frequencies which the tweeter may
detailed on designing an audio filter for a super tweeter produce with a distortion.
speaker. Components calculation and, simulation and
experiment result are shown. In detail, the comparison The problem given was to create a audio
in theoretical and experiment of voltage gains, cut-off filter for a super tweeter. Details of the super tweeter
frequencies, slopes of roll-off, phase angles at cut-off, speaker was analyzed so that a band pass filter with
the ranges of phase angles and percentage of error
cutoff frequency in between 20kHz and 100kHz
being conducted. Suggestions on possibilities of
speaker applications are list up. were to be implemented. Sallen-Key equal
component was used to design the filter so that it
Keywords—operational-amplifier (op-amp), low cut- would be affordable and with a high roll-off rate for
off frequency (fL), high cut-off frequency (fH), center the consumers. Bessel response type was chosen due
frequency (fC), quality factor (Q), bandwidth (BW), to its maximally flat phase delay which preserves the
active filter, roll-off-rate, phase angle, magnitude
wave shape of filtered signals. The objective of this
response.
experiment was to obtain the desired magnitude and
I. INTRODUCTION phase response of the active filter and to study the
A filter is type of circuit that is used to pass characteristics of active filters based on the number
certain range of frequency and filter out unwanted of poles and Q-factor.
frequency. It is an essential circuit in most of the
II. METHODOLOGY
electronic circuits. This property of filters is also
called frequency selectivity as it selects the An audio filter for super tweeters is designed
frequency that it wants to pass. There are two type with a minimum gain of 26db. Components such as
filters which is active filter and passive filter. This resistors of 1kΩ, 120Ω,4.7kΩ,10kΩ,1.2kΩ, 330Ω,
filter can be differentiated by the elements used in capacitor of 10nF and op-amp741 is used.
the construction of the circuit. Active filter is mainly Equipment such as the function generator, power
consisting of active components such as amplifiers supply, oscilloscope and multimeter are used
and transistors while passive filters contains passive throughout the experiment. The list of components
elements such as inductors. Active filter usually used is shown in the appendix.
needs to depend on external power supply
meanwhile passive filter does not need any sort of The experiment was started off by identifying
power supply for its circuit. The advantage of active the specifications of the audio filter. Based on the
filter over passive filter is selective frequency, cost
specification available on the cut off frequency of
effective (no inductor) and no loading effect. To get
20kHz to 100kHz, other specifications were
a better filter performance, employing a higher slope
of roll-off and higher gain-bandwidth can be chosen. decided. Sallen-key component was applied in the
The band pass filter passes only a specified band of filter to produce an affordable yet high roll-off rate
frequencies and rejects all others and quality factor audio product to the consumers. Bessel response
(Q’s factor) is an indication the selectivity of a band type was chosen because it has almost a flat
pass filter. response and no overshoot.

The high pass filter is cascaded with a low pass

filter. The values of components of the filter are
then calculated. Extra amplifier was added to the
circuit to achieve the desired gain. Calculations and
circuit design of the filter is shown in the appendix.
Figure 1. The bigger size is woofer, while the smaller size is
tweeter
Simulation was then carried out via Multisim to
examine the result of calculations. Values of
A wide range of speakers are available in components were being adjusted in simulation to
the market with each playing it’s specific purpose. obtain the available components in the market.
The super tweeters are used to produce ultra high Simulation result of frequency response of the filter
frequencies. It’s purpose is to create more realistic was discussed later. The circuit is the implemented
sound field in the speaker system. They are usually in hardware. Connection of components was done
used to supplement the sound of tweeters by by referring to circuit in Figure 1. Op-amp was
connected to DC power supply of ±15V. Function 𝑅𝐹3
𝐴3 = 1 +
generator was connected to positive terminal of op- 𝑅3
amp to produce AC input waveform with tunable 10𝑘
frequencies for the band pass filter. Connection of 12.44 = 1 +
𝑅3
oscilloscope at input and output of band pass filter 𝑅3 = 874 ≈ 1𝑘Ω
was to observe the functionality of the system.
Before turning on the power supply, circuit
connection was checked using multimeter. III. RESULTS FROM STIMULATION
To prevent stimulation error in our experiment,
Measurement and collection of data were we had use Multisim software to stimulate our circuit
proceeded. Frequency of input voltage, input design before we start to connect it in real world. The
voltage, output voltage and phase angle of both value of the components that we use in stimulation is
wave forms were collected for analysis of data. exactly what we will use in experiment. This can
Experimental gain of the filter system was estimate what results we will get in this experiment.
calculated based on the voltages obtained. Below had showed the schematic diagram and
Experimental result was being compared with the results in graphs from Multisim software.
simulation result. Discussion and conclusion were
made after that. A. Schematic Diagram
By referring figure 2, we used Bode plotter
Lastly, the project was demonstrated before instead of oscilloscope in Multisim. It shows bode
being introduced to the market. plot graph and phase angle graph directly. However,
we use oscilloscope in real world and get bode plot
Stage 1 (High Pass Filter) graph and phase angle graph by calculation which
α1 = 1.732
A1= 3-α1 = 3-1.732 = 1.268
1
kLP1=0.785 𝑘𝐻𝑃1 = = 1.274
𝑘𝐿𝑃1
C1=10nf
𝑘𝐻𝑃1 1.274
𝑅𝐴 = =
2𝜋𝑓𝐿 𝐶 2𝜋 × 20𝑘 × 10𝑛
= 1013.8 ≈ 1𝑘Ω
𝑅𝐹1 = 𝑅𝐴 𝐴1 = 1𝑘 × 1.268
= 1268 ≈ 1330Ω will be shown after this.
𝑅𝐹1 1268
𝑅1 = = Figure 2. Band Pass Filter Circuit Diagram
𝐴1 − 1 1.268 − 1
= 4731.34 ≈ 4.7𝑘Ω B. Graph

Stage 2 (Low Pass Filter)

α2 = 1.732
A2= 3-α2 = 3-1.732 = 1.268
kLP2=0.785
C2=10nf
𝑘𝐿𝑃2 0.785
𝑅𝐵 = =
2𝜋𝑓𝐻 𝐶 2𝜋 × 100𝑘 × 10𝑛
= 124.9 ≈ 120Ω
𝑅𝐹2 = 2𝑅𝐵 𝐴2 = 2 × 124.9 × 1.268
= 316.75 ≈ 330Ω Figure 3. Bode plot graph for Magnitude Response
𝑅𝐹2 316.75 Based on the magnitude response of the filter,
𝑅2 = =
𝐴2 − 1 1.268 − 1 the roll-off-rate can be calculated as follow:
21.10 + 13.54
Roll-off rate=
log 17.50𝑘−log 1.73𝑘
= 1181.9 ≈ 1.2𝑘Ω = 34.48𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒

Stage 3 (Amplifier)
𝐿𝑒𝑡 𝐴 𝑇 = 26𝑑𝐵 = 20
20
𝐴3 = = 12.44
1.268 × 1.268
Assume RF3=10kΩ
 20 0.44 5.16 11.72727 21.38394 38.15

21 0.44 5.32 12.09091 21.64918 32

22 0.44 5.4 12.27273 21.77882 26

23 0.44 5.44 12.36364 21.84292 22

24 0.44 5.56 12.63636 22.03244 16.08

25 0.44 5.6 12.72727 22.09471 12.25

Figure 4. Maximum gain of 24.04dB
30 0.44 5.64 12.81818 22.15653 7.3

35 0.44 5.48 12.45455 21.90656 -6

40 0.44 5.24 11.90909 21.51757 -20

45 0.44 5 11.36364 21.11035 -32

50 0.44 4.68 10.63636 20.53586 -45

55 0.44 4.4 10 20 -52

Figure 5. 𝑓𝐿= 17.50kHz, fH = 99.03kHz 60 0.44 3.84 8.727273 18.81757 -63

C. Percentage Error 65 0.44 3.56 8.090909 18.15995 -72

(i) Roll-off rate 70 0.44 3.4 7.727273 17.76052 -79
𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑟𝑟𝑜𝑟
75 0.44 3.12 7.090909 17.01404 -90

40 − 34.48 80 0.44 2.96 6.727273 16.55678 -89.15

= × 100%
40 90 0.44 2.76 6.272727 15.94913 -95
= 13.80%
95 0.44 2.5 5.681818 15.08975 -103

(ii) Low cut-off frequency, 𝑓𝐿 100 0.44 2.32 5.272727 14.44071 -109.2
𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑟𝑟𝑜𝑟 105 0.44 2.22 5.045455 14.05801 -113.3

20 − 17.5 110 0.44 2.08 4.727273 13.49221 -117

= × 100%
20 115 0.44 2 4.545455 13.15155 -123.4
= 12.50% 120 0.44 1.88 4.272727 12.6141

(iii) High cut-off frequency, 𝑓H 125 0.44 1.74 3.954545 11.94193

𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑟𝑟𝑜𝑟 130 0.44 1.7 3.863636 11.73992

135 0.44 1.62 3.681818 11.32125

100 − 99.03
= × 100% 140 0.44 1.56 3.545455 10.99344
100
= 0.97% 200 0.44 0.94 2.136364 6.593504

Table 1. Experimental Data

(iv) Maximum gain
𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑒𝑟𝑟𝑜𝑟
B. Magnitude Response
26 − 24.04
= × 100%
26
= 7.54%

VI. RESULT FROM EXPERIMENT

A. Data
Phase
Frequency Vin
Vout (V) Gain Gain (db) shift
(kHz) (V)
(°)
1 0.44 0.058 0.131818 -17.6005
Figure 6. The magnitude response of the filter
5 0.44 0.92 2.090909 6.406703 139

10 0.44 2.9 6.590909 16.37891 99

15 0.44 4.52 10.27273 20.23372 64

16 0.44 4.8 10.90909 20.75577 52

18 0.44 4.96 11.27273 21.04058 47

19 0.44 5.04 11.45455 21.17956 43.12

C. Phase Response V. DISCUSSION
A. Compare and Analysis
The circuit that we constructed composes
of 3 main parts which are low pass filter, high pass
filter and amplifier. The first part of the op-amp is a
low pass filter with a voltage gain of 1.268 and the
second part of the op-amp is a high pass filter with a
voltage gain of 1.268 also. An amplifier with a
voltage gain of 12.44 was used at the third part of
the filter in order to get an overall gain of 20. By
Figure 7. The phase response of the filter using Multisim, a maximum voltage gain of 24.04
dB was obtained which is 1.96 dB less than the
D. Roll-off Rate desired voltage gain of 26db. Meanwhile, the final
voltage gain obtained is 22.16 dB which was 3.84dB
less than the desired voltage. This might be due to
the variation in resistor and capacitor values
compared to the calculated values.

The intended cut off frequency for super

tweeters speaker filter design was in between 20kHz
and 100kHz. The cut off frequency obtained from
the simulation was in between 17.5kHz and
Figure 8. The roll-off rate of the filter 116.5kHz. This is because the values of capacitor
and resistors that was used were rounded off to the
Roll-of-rate applicable values in the experiment. The cut off
20.23−(−13) frequency obtained from the experiment was in
=𝑙𝑜𝑔15𝑘−𝑙𝑜𝑔1.5𝑘
between 15kHz and 60kHz. This might be due to the
=33.23 dB/decade resistance between the jumper wires used in the
circuit.
E. Percentage Error A second-order band pass active filter
Percentage error of maximum gain should have a phase shift of −90° at the cut off
|24.04−22.16|
= × 100% = 7.82% frequency. This is because there are 2 poles in the
|24.04|
filter where each pole contributes to −45° of phase
Percentage error of fL shift. However, the phase shift obtained from the
|17.5−15| simulation is 73.73 ° and -69.64° for the low and
= × 100% = 14.29% high cut-off frequency respectively. Meanwhile,
|17.5|
from the experiment, the phase shift obtained was
Percentage error of fH 63° and-64° respectively for the low and high cut-
|116.5−60| off frequency.
= × 100% = 48.50%
|116.5|

The desired roll-off rate was 40dB/decade

Percentage error of fo since it is second order band pass filter with 2 poles
|45.53−30|
= × 100% = 34.11% and each pole contributing to 20dB/decade. The
|45.53|
simulated roll off rate was 34.47dB/decade and the
Percentage error of Bandwidth experimental roll off rate obtained was
|93.5−45| 33.23dB/decade respectively. The experimental
= × 100% = 51.87%
|93.5| roll-off rate is almost as equal as the simulated roll-
off rate with just a difference of 1.24dB/decade. The
Percentage error of phase angle of fL Q-factor obtained for this super tweeter is 0.487
|73.73−63|
= × 100% = 14.55% through simulation and 0.667 through experiment.
|73.73|

To discuss about the percentage of error,

Percentage error of phase angle of fH
|−69.64−(−64)| comparing to the simulation and calculation results,
= × 100% = 8.1% the error of maximum gain in term of percentage is
|−69.64|
7.82%. Both the low and high cut off frequency, 𝑓𝐿
Percentage error of roll-off rate and 𝑓𝐻 has a percentage error of 14.29% and 48.5%
|34.47 − 33.23|
= × 100% = 3.6% respectively. The central frequency, 𝑓𝑜 stands at
|34.47|
34.11% and the bandwidth has 51.87% of error.
Besides, the phase shift of both 𝑓𝐿 and 𝑓𝐻 have
14.55% and 8.1% percentage error respectively. (iii) Centre Speakers – A small speaker with
Finally, the roll-off rate percentage error stands at an ability to operate simultaneously with
3.6%. other speaker to get a surrounding sound
system.
There were a lot of factors that affected the
experimental error compared to the simulated (iv) Bookshelf Speaker - A compact
results. The op-amp (UA741) is thermal dependent. loudspeaker that is design for affordable
The temperature of the op-amp would have consumers. This speaker is great for one
fluctuated during the experiment as the experiment who seeking for a great sound with a low
was conducted for a long period of time. Other than price.
that, the resistor values that were used which is
manipulated to match the availability of the resistor
in the lab and the degradation of the resistors could (v) In-Wall Speaker - Is used as the same as
also have affected the experimental values. The bookshelf speakers. This speaker is design
resistance in the jumper wires also could have for a narrow place which it can hang in the
played a role in the difference of values. The wall.
oscilloscope and function generator used was of old
tools, which could have metrology error within. D. Problem Faced
We faced many problems during the
B. Precaution experiment. Firstly, we had to make some changes
To not confuse the circuit and make it in the calculation value due there is no desired
more organized, we used different colour of jumper resistor value in the lab. Then, we experienced a
wires in the circuit and labelled each of the wires for fluctuation in the results for the output signal thus
easier reference due to the clustered and many wires making it hard for us to obtain a constant data. This
in the circuit. For example, the VCC and VEE wire is due to the sensitivity of the probe of the
was labelled to be connected easier. oscilloscope. Finally, we couldn’t get output result
at our first attempt because our function generator
Furthermore, we also made sure every and oscilloscope were not working properly. The
connection in the circuit was well connected in the measuring tools in the lab is very old, which have
circuit using a multi meter before turning on the accumulated metrology error over the years.
power supply to prevent and short circuits in the
circuit. We also checked the op-amp IC using IC E. Solution
tester to ensure that it is of the right specifications To get the right resistor values, we used
and it is working before connecting to the circuit. series and parallel connections of several resistors to
obtain the desired resistance needed. Next, our
To ensure the accuracy of our results, we group used a different set of probes and hold the
made sure the amplitude scale of both channels is the probes in certain angles to remove the fluctuation at
same before taking any readings. The VCC and VEE the oscilloscope. For the measurement tool, we
value of the op-amp was obtained from the data couldn’t figure out at first and attempted reassembly
sheet of the IC. The value in this case is 15V and - of our circuit. Then, when we tried on a different set
15V respectively. It was made sure the DC voltage of oscilloscope and function generator which finally
did not exceed that value to prevent the clipping of showed the desired output. Besides, the input
output signal. voltage from function generator to the circuit should
be very low so that the output would not undergo
C. Application clipping.
(i) Home theatre Speaker – An exclusive
surround sound system which design to VI. CONCLUSION
mimic commercial theatre performance
and feeling that consumer can experience it From this experiment, magnitude
at home. The speakers come with a full responses of the bandpass filter were obtained by
package of every speaker that consumer observing the changes in output voltages with
need to providing sound from 360’ radius. different frequencies. Not being limited to voltages
only, phase responses of the filter could also be
(ii) High Fidelity Speaker – Hi-fi speaker is a obtained from the oscilloscope by comparing the
class of consumer home audio equipment input and output waveforms phase angles. In short,
that comes with high price. However, the the skill of identifying the characteristics of active
price is coming with an excellent quality of filter by knowing the number of poles and Q factor
sound reproduction. was gained. The higher the number of poles, the
higher the roll-off rate. The lower the Q factor, the
wider the pass band bandwidth, the further the low
and high cutoff frequencies to each other. The low
and high cutoff frequencies will determine the
natural frequency of the filter system.

REFERENCES
[1] K. Lacannette, "Application Note", National
Semiconductor, no. 779, 1991.
[2] "Home Theater Systems - Yamaha -
Malaysia", My.yamaha.com, 2016. [Online].
Available:
http://my.yamaha.com/en/products/audio-
visual/hometheater-systems/. [Accessed: 06- Mar-
2016].
[3] "Dxm Series | Smart Modular Wireless Speaker
| Modularity | How",Creative.com, 2016. [Online].
Available: http://www.creative.com/dxm/how/.
[Accessed: 06- Mar- 2016].
[4] P. W. Tuinega, A Guide to Circuit Simulation.
Englewood Cliffs, New Jersey: Prentice Hall, 1990.
[5] R. E. Thomas, The Analysis and Design of Linear
Circuits. 3rd ed. New York: John Wiley & Sons,
2000.
[6] S. Karni, Applied Circuit Analysis. New York:
John Wiley & Sons, 1988.