asymmetry within band-pass filter pass-band

[jschonbe]

Hello,
I am working on a high frequency q-enhanced two-pole band-pass filter. I am having some difficult obtaining the desired flat pass-band that I am striving for. Basically, there is a "humpiness" occurring on left side of the pass-band. I have quantitatively measured the amount of magnetic coupling between the inductors in the LC tank and applied this to simulation. Ideally, I would be solely depending on capacitive coupling opposed to magnetic coupling. I have included the ADC schematic & simulation. If anyone has any suggestions on what might be causing this problem, it would be greatly appreciated.

Thanks

 

[zappedagain]

While not on this type of filter, I've seen this effect when there is a mismatch of the center frequency for each pole; I'm not sure why you are seeing it in simulation though.

I do see one difference; the tank resistance (Q-factor effect) on the left side is R1//R1, while the load on the right side is R1//R2.

You should check and make sure there aren't any other 'parasitic' parameters set for the components that are not visible.

Where is dB(vir_gnd_1) on the right side plot? Are the two traces identical (buried under db(vir_gnd-2), or is it off the scale?

Do you get the expected response if you remove C5-C9? This may isolate if it is a loading problem on the current source or a coupling between the two stages problem.

Is the 'GND' coupling at LX and LY real, or are these two circuits isolated (AC coupled)? With your isolated current source LX and LY really are in series...

Looks like fun...

John D

 

[jschonbe]

Thank you for your response zappedagain. I investigated the effects of loading as you discussed and don't believe it to be the culprit. The vir_gnd were stacked on one another and act as a sanity check to verify that that node was indeed acting as an AC virtual gnd. After further investigation, I believe the "hump" to be a product of the capacitive coupling. Specifically, capacitive coupling is a form of quadrature coupling (phase relationship between voltage and current). To test this theory I needed to add some sort of in-phase coupling, such as resistive. I added variable Rc and Rcross resistors in parallel with the Cc and Ccross capacitors and found that I was able to independently neutralize/create humps on both the left and right sides of the passband.