Help with Multiple Feedback Bandpass filter for ECG signal

Thread Starter

vaishya9

Joined Jul 9, 2023
6
I am trying to create a 2 stage Mutiple feedback bandpass filter for a band of Frequenz of 8 to 18hz for ECG Rpeak detection. The 1st stage has cut off frequency band of 4.5 to 13hz and 2nd stage have a band of 10 to 26hz.

I am using OPA2338 and working voltage of +v=3.3v and -v=0v. Circuit is referenced at around vcc/2=1.65v

The problem I am facing is that the input signal and output signal are output of phase by 180degree. I am not getting the reason why this is happening. I have analysed the circuit on spice and pcb both and getting the same respone. The circuit phase response is linearly decreasing from-60degree to 60degree for the band of8 8 to 18hz respectively and also the group delay of the filter is not constant and has a maximum groupt delay of 54ms.
The only thing that I can suspect is that the ecg signal is composed of mutiple frequency and as phase, gain and group delay frequency depedent and as the phase and group delay is not constant they are responding differently for different frequency. Can anyone help me know the exact reason?circuit of MFBP filter.PNGfilter out_ip.PNGphase resposne.PNGgroup delay.PNG
 

Thread Starter

vaishya9

Joined Jul 9, 2023
6
MFB filters invert the signal. They are based around an op-amp configured as an inverting amplifier.
Yes correct, as the input is connected to inverting terminal it should invert but I am using 2 such MFB filter in cascaded form, the overall phase should be same as input right?
 

Ian0

Joined Aug 7, 2020
10,046
Yes correct, as the input is connected to inverting terminal it should invert but I am using 2 such MFB filter in cascaded form, the overall phase should be same as input right?
Agreed.
What does the signal look like at the output of the first stage?
 

Ian0

Joined Aug 7, 2020
10,046
What is the Q of the filter?
Do you need a band pass filter? Obviously, you need a steep high frequency roll off to remove mains hum, but is there anything below the frequency of interest that needs to be eliminated? Of would a simple 1st order roll off to remove the DC level be enough?
 
Top