Active Filter Layout Questions

[Henry-cool]

Hi~~

I designed a forth active RC lowpass filter
I use two second active RC filter acheive it , -3dB cutoff frequency is 600KHz
When I designed filter in schematic, it works well.
After I finished Layout, I simulated RCX(RC Extraction).
But, The results is not good.

Filter has 2~3dB peaking near 600kHz Frequency.
I don't know how I should fix my layout.

Plz let me know how to approach the solution.

 

[pfd001]

It depends on your filter architecture. Parasitic cap can alter the pole and zero position

 

[LvW]

The peaking you have mentioned is indicating some kind of Q enhancement.
Very often, this is the case when the transit frequency of the used opamp is not high enough (at least 20 times higher than the pole frequency).
By the way: Which filter topology did you choose ?

 

[memed09]

Analog design is much more diffucult then digital. In schematic design did you arrange the AS,AD,..etc.?Because these parameters are all set in schematic design and tr.s are excepted to be ideal. But in layout it's completely different. All parameters change in layout. And also in schematic you can ignore parasitic effects but in layout all efects must be considered....

 

[Henry-cool]

Thank you for reply.

Filter topology is 4th Butterworth Filter(Active Filter)

Butterworth 4th Filter doesn't have ripple and peaking in theory.

Still, I try to find any solution, but it is not easy.

First, I guess any pole or zero are added because of parastic caps and resistors.

But, Parastic cap values are relatively small(several tens femto F) because Filter

cap array's value is 10pF above.

The pole or zero's location may be separated from original pole or zero's location.

Do the pole or zero affect the circuit?

or Is there another problem?

 

[LvW]

..............................
Do the pole or zero affect the circuit?
or Is there another problem?

Of course, affect parasitic poles (zeros occur not very often) the response, however, as mentioned already, I suppose it´s the parasitic phase shift of the opamp which influences the filter response.
Therefore again: What is the GBW of the opamp and what topology are you using ?
(topology is the kind of circuit: Sallen-Key, multifeedback, or something else; Butterworth describes only the chosen approximation).

 

[memed09]

Are you using 2 cascaded 2th order filter or one 4th order filter topology?
if you are usign 2 cascaded filters then the output of the first stage can affect the second stage.
and i think that the problem is related to poles and zeros.

 

[Henry-cool]

Thank you for reply

Filter type is Active RC Filter(Thomas 1)

It is made up of two cascade biquad filter(4th Butterworth Filter)

OP amp's GBW is about 70MHz in 1p Load.

I'm not sure the solution. I guess ..OP amp's Gain, BW, and Parastic cap, resistors make peaking..

 

[LvW]

Filter type is Active RC Filter(Thomas 1)
It is made up of two cascade biquad filter(4th Butterworth Filter)

I don´t recommend the Thomas-1 structure to you. The 2nd order block consists of three opamps, thus, your 4th order filter contains six opams - all connected in series ! As each opamp contributes some parasitic phase shift it is not surprising that the filter response shows some sort of peaking (Q enhancement).
The chosen structure (Thomas-1) is useful only if you are interested in the specific features of this structure - namely owpass, highpass and bandpass at the same time at three different outputs.
I can recommend (a) multifeedback topologies or (b) Sallen-Key structures with a total of only two opamps. Choice (a) is less sensitive to parameter tolerances and choice (b) is less sensitive to opamp parasitics.