Complex Filter Design

[MichaelHelal]

Question: Why is there a peaking in the complex bandpass filter response?

Hello everyone
I am designing 5th order butterworth Complex Band Pass filter, I finished the design of the 2 stage miller opamp.
After running the simulation to check the filter response and by comparison with the ideal one, I find a peaking in the response of the actual designed one,, does anyone know why this peak happen?

Note: Opamp GBW is around 300 MHz and the cut-off frequency of the filter is 1.5 MHz
Note2: in the red curve, I use actual resist

ors and capacitors while in the green one I use ideal res&caps

 

[Dominik Przyborowski]

The simplest (conceptually) way to check it is to perform PZ analysis. The real passive elements has ca 10% parasitic capacitance to ground which modifies transfer function. There is also a question about your opamp PZ location. By closing the loop, any opamp zero is forming directly a pole.
Start with model of opamp and PZ analysis. Maybe split filter to low pass and high pass circuits can help in debugging.

//edit
Please notice your peak has only 0.5dB, it is not too much

 

[MichaelHelal]

The simplest (conceptually) way to check it is to perform PZ analysis. The real passive elements has ca 10% parasitic capacitance to ground which modifies transfer function. There is also a question about your opamp PZ location. By closing the loop, any opamp zero is forming directly a pole.
Start with model of opamp and PZ analysis. Maybe split filter to low pass and high pass circuits can help in debugging.

//edit
Please notice your peak has only 0.5dB, it is not too much

Dear Dominik, thank you for your sharing
PZ analysis may be too complicated to locate the pole or the zero as it's a huge circuit.
I tried to replace the designed opamp with a macro-model and started tuning gain and BW and the response improved by increasing the BW.

 

[MichaelHelal]

The simplest (conceptually) way to check it is to perform PZ analysis. The real passive elements has ca 10% parasitic capacitance to ground which modifies transfer function. There is also a question about your opamp PZ location. By closing the loop, any opamp zero is forming directly a pole.
Start with model of opamp and PZ analysis. Maybe split filter to low pass and high pass circuits can help in debugging.

//edit
Please notice your peak has only 0.5dB, it is not too much

I didn't get what you said about any zero is forming a pole in the closed-loop, can you explain this?

 

[danadakk]

If G(s) was a zero you can see that it would form a pole in the denominator
after you do the analysis -

Regards, Dana.

 

[timof]

The SPICE does not tell you where the problem is coming from?