INTRODUCTION

MFB filters are simple looking design filters, but it is difficult to calculate the values to meet

specifications Such filters are used for equalization purposes, analysis and other tasks like

Sound to Light converter or fully functional Vocoder. Vocoder is a device that takes a music

source as one of its input and vocals the other input, thus allows a guitar, keyboard etc or

complete ensemble to be made to speak or sing. The "speech" signal from a good and reliable

vocoder is very intelligible, and is "ear candy", thus the very best kind for experimental

musicians.

The source impedance should be kept low as compared to the input resistance, and usually

such filters are driven using OPamp buffer. If a source having high impedance is used, it

increases the net input resistance, thus resulting in unpredictable centre frequency and so the

response.

OBSERVATIONS & CALCULATIONS

The center frequencies were derived based on the frequency equation of a band pass filter

which is given by

1 𝑅1 + 𝑅2
𝑓𝑜 = √
2𝜋 𝐶1 𝐶2 (𝑅1 𝑅2 𝑅3 )

Based on the resistors and capacitors values given in the table, following frequencies were obtained.
R1(Ω) R2(Ω) R3(Ω) C1(nF) C2(nF) f0 (Calculated) f0 (Theoretical) % Error
82000 2700 160000 220 220 35 31 12.90322581
82000 2700 160000 180 180 43 40 7.5
82000 2700 160000 150 150 51 50 2
82000 2700 160000 120 120 64 63 1.587301587
82000 2700 160000 100 100 77 80 -3.75
82000 2700 160000 82 82 94 100 -6
82000 2700 160000 56 5.6 138 125 10.4
82000 2700 160000 47 47 165 160 3.125
82000 2700 160000 39 39 199 200 -0.5
82000 2700 160000 27 4.7 218 250 -12.8
82000 2700 160000 22 2.7 319 315 1.26984127
82000 2700 160000 18 1.5 473 400 18.25
27000 820 56000 47 47 507 500 1.4
27000 820 56000 39 39 611 600 1.833333333
27000 820 56000 27 2.7 882 800 10.25
8200 510 18000 47 4.7 1151 1000 15.1
8200 510 18000 39 39 1388 1400 -0.857142857
8200 510 18000 27 27 2000 2000 0
8200 510 18000 18 1.5 3294 2800 17.64285714
8200 510 18000 12 1.8 3683 4000 -7.925
8200 750 18000 8.2 8.2 5518 5600 -1.464285714
8200 1200 18000 4.7 4.7 7800 8000 -2.5
8200 1200 18000 2.2 2.2 16665 16000 4.15625

In any band pass filter design (or indeed any filter) procedure, one of the important parameters

to be taken care of is 'Quality Factor, Q'. Q of a filter decides its bandwidth, and this is especially

compelling in the case of band pass filters. Majority of audio applications systems will need a

maximum Q of approximately 4, which is better suited to a 1/3 octave filter set. A very few

applications would require closer filters than that, and in order to have a 1/3 octave band filters

covering the audio band (that is 20-20kHz) would require 30 separate filters at a time.

Generally, for even greater resolution this might increase very dramatically.

We know that for an octave EQ (10 band), the frequencies are generally as follows
Similarly, for ½ octave (20 band) the frequencies are typically in the range shown in the below

figure

And lastly for 1/3 octave (30 band), the frequencies are lying in the range shown below

Since the frequencies give in the table are lying in the third range of frequencies, therefore, it is

obvious that the proportion of octave used is 1/3 and quality factor Q up to 4. Q beyond 4

decreases the bandwidth of the filter and so the required frequencies might not be achievable.

Center frequencies can be calculated using equation

1 𝑅1 + 𝑅2
𝑓𝑜 = √
2𝜋 𝐶1 𝐶2 (𝑅1 𝑅2 𝑅3 )

For Band 15, 882 Hz was obtained based on the resistors and capacitors values given in the

table against the band.

𝑓
𝐵𝑊 =
𝑄

But we must have Q to find BW.

So to find Q, we have

1 𝑅3
𝑄= √ = 4 𝑓𝑜𝑟 𝐵𝑎𝑛𝑑 15 (𝑏𝑦 𝑗𝑢𝑠𝑡 𝑝𝑢𝑡𝑡𝑖𝑛𝑔 𝑡ℎ𝑒 𝑣𝑎𝑙𝑢𝑒𝑠)
2 𝑅1 ||𝑅2

Thus

𝐵𝑊 = 220 𝐻𝑧 𝐵𝑎𝑛𝑑 15

In order to find closed loop voltage gain, there is an equation

𝑅3
𝐺=
2𝑅1

Corresponding to band 15 values of the two resistors are 27K and 56K for 15. So based on those

values, gain of almost 1.091 was obtained for Band 15 (just put the values and get the answer).