# Problem with understanding the working of MFB bandpass filters.

#### Yuvraj Abhimanyu

Joined Jun 10, 2015
15
Hey guys!
I'm working on an bandpass amplifier project using OP Amp NE5532

Project Requirements:
Input Signal: Electric Guitar Output.
Output Signal: Amplified(variable) Signal form 82hz to 3.2khz

I'm trying to impliment a bandpass filter with variable gain such that by itself the input signal is lightly saturated at the output and as i boost the signal it becomes highly saturated!
MFB topology filters are used here because of the high Q, however I dont mind changing it to Sallen-Key or any other topology.

My questions are
1.Am i right in using MFB or should i change to Sallen-Key.
2.With reference to the circuit diagram below, the Signal cannot be seen after R1. I do not understand why!
3.After calculating R and C values using websites such as http://www.k7mem.com/Electronic_Notebook/filters/act_bpfil.html#Requirements ,
http://sim.okawa-denshi.jp/en/OPtazyuBakeisan.htm and using formule i arrive at some centre frequency (say x). But practically the centre frequency will
be miles away form the calculated frequency. I fail to understand why!

Plz help

#### Papabravo

Joined Feb 24, 2006
20,369
I need to understand how you are measuring your signals. DVM, oscilloscope, other?
What are the characteristics of an "electric guitar" output in terms of amplitude and drive capability.
What is the notation on the resistor from the + terminal to Vcc/2. It looks like "J10K".

For example if the guitar output was 700 mV P-P, then at the junction of R1&R2 it would be about 90 mV P-P

#### AnalogKid

Joined Aug 1, 2013
10,551
First thoughts: Active filters are critically dependent on the gain settings. If you really are trying to get both variable gain and bandpass functions in one opamp, don't. Your drawing shows a fixed gain filter. You should leave it that way and do the variable gain somewhere else. I haven't run your numbers, but some website calculators are not clear about multipliers. For example, do you enter 1 (kHz) or 1000 (Hz)? It varies from site to site. Do the results from the two calculators agree?

The R1-R2-C1-C2 node is a low impedance point, but not a virtual ground, so something should be visible there.

ak

#### joeyd999

Joined Jun 6, 2011
4,971
I will point you to this thread, assuming it will be helpful.

Secondly, keep in mind that the Q (bandwidth) and loop gain of a MFB are interrelated. In fact, loop gain determines the Q, and vice versa.

So, varying the gain will simultaneously vary the Q. Is this what you intend?

Alternatively, you could vary both R1 and R2 in such a way that the parallel combination of the two remains constant. This will adjust your gain without affecting Q or center frequency. But it would be rather difficult to do.

#### Papabravo

Joined Feb 24, 2006
20,369
You haven't indicated you expectations for stopband attenuation. I mention this because it will affect the number of sections a properly designed filter should have. Since your passband is comparatively narrow in logarithmic terms, 1.59 decades, you need to be aware that with a single stage you are unlikely to get an impressive rolloff in the transition band

#### crutschow

Joined Mar 14, 2008
32,843
Below is an LTspice simulation of your filter.
The center frequency is 700Hz with a gain of slightly over 12dB
The -3db points are 613Hz and 793Hz.
What center frequency and bandwidth did you want?

A good program for designing active filters is the free FilterPro software from Texas Instruments.

#### Attachments

• 1.8 KB Views: 0