Problem with switched capacitor biquad

[shyboy]

Hi, all,
I recently built a switched-capacitor biquadratic building block for a high order lowpass filter...It was a lowpass notch biquad. I tested it out with Spectre. The result showed good dc gain and high frequecy gain as expected. However, the pole and zero locations deviated a lot from the expected values. And the transition was not as sharp as I thought. I tried to change some capacitor values to change the pole-zero location. It turned out that the dc gain changed a lot while the pole and zero lacations almost remained the same..Could anyone help me out of this problem? Thanks a lot!

shyboy

 

[Vamsi Mocherla]

Well,cnsidering your problem, your circuit should respond to changes in the capacitor values. But do not change the capcitors which are switched. Change the capacitors which are used for cap purposes only.

Secondly, it would be good, if you can give the specifications of your design. May be there is some problem in the amplifier which is hitting the DC gain. Or else, you must be tweaking the switches and the relevant caps. DO NOT DO THAT.

 

[shyboy]

Hi, Vamsi Mocherla,

Thanks a lot for your reply...

The transfer function I am going to realize is :
0.3333(s^2+1.76948)/(s^2+0.448s+0.78 )
The above function has been normalized to 3.6MHz. After bilinear transformation, I have the z-domain transfer function which is supposed to have a zero at around 4MHz. The biquad I use is a general purpose one. I find all the capacitor value by comparing coefficients of the z-domain transfer function. With these capacitances calculated, the actual zero freq from simulation is just slightly bigger than 1Mhz and the pole is almost a decade lower than the zero, 100+ KHz.

And I don't quite understand what you mean by 'tweaking the switches and cap'...Are you saying I am usign the wrong switching phase?

Thanks again.

 

[Vamsi Mocherla]

Like I was thinking, there could be a problem with the OPamp. Just out of curiousity, what type of filter are you implementing?I mean Cheybshev/Butterworth?

 

[flushrat]

try to use ideal opamp to see if it is caused by opamp. If so, then use a opamp model whose poles and gain can be controled to see the influence of the gain and poles of opamp.

 

[shyboy]

Like I was thinking, there could be a problem with the OPamp. Just out of curiousity, what type of filter are you implementing?I mean Cheybshev/Butterworth?

It's a 6th order elliptic LP filter by cascading three LPN biquads..

Added after 4 minutes:

Thanks guys!

I think there is real problem with my opamp..When I change the bias voltage, dc gain of the biquad would change...But if i test the opamp alone, the output of the opamp is independent of bias voltage...Think the capacitive load is seriously affecting the performance of the opamp

Added after 3 hours 42 minutes:

Hi,
I am using Spectre to simulate my circuit...Just want to know is there an existing ideal opamp model in Spectre I can use directly? I checked with some guys in the same lab, seems none of us can find one...Thanks~~~

 

[Vamsi Mocherla]

try and see if you have ahdlLib ( I am not sure about the case sensitivity). You will see an ideal opamp there. Best of luck. Or else why dont you do a Macromodel. You will find good information in Allen Holberg book.