opamp load capacitance for active filter design

[jimito13]

Hello all,

I am designing an active continuous time filter and i have chosen a Tow-Thomas biquad to realize my circuit.My question is the following :

During the design of the opamp and the simulation procedure via cadence spectre,which is the proper load capacitance that i should concider for my opamp circuit?? 1pF or 2pF should be enough? Is there a way to evaluate the proper load capacitance value for my opamp via some kind of simulation of the ideal circuit,that is to measure the capacitance seen at the output of the first opamp for example in the TT biquad?

Apparently i need this info in order to make an initial estimation for the gm value of my differential pair at the input stage to obtain the required transition frequency since 2*pi*ft≈gm1/(Cout+Cload).

So,i need an estimation for the sum Cout+Cload.

Any helpful answer would be appreciated.Thanks in advance.

 

[upvl]

My suggestion is just estimating a capacitor load at first, then you can use the "stb" in Cadence analysis to simulate your closed loop opamp. This analysis can give you accurate closed-loop characteristics, like loop gain, loop bandwidth. Next, you can tune your opamp according to these simulation results.
This procedure makes sense, because it's really hard for us to accurately caculated the exact equivalent load information, considering the effects of the feedback loop.

 

[LvW]

Hello all,

I am designing an active continuous time filter and i have chosen a Tow-Thomas biquad to realize my circuit.My question is the following :

During the design of the opamp and the simulation procedure via cadence spectre,which is the proper load capacitance that i should concider for my opamp circuit?? 1pF or 2pF should be enough? Is there a way to evaluate the proper load capacitance value for my opamp via some kind of simulation of the ideal circuit,that is to measure the capacitance seen at the output of the first opamp for example in the TT biquad?

Apparently i need this info in order to make an initial estimation for the gm value of my differential pair at the input stage to obtain the required transition frequency since 2*pi*ft≈gm1/(Cout+Cload).

So,i need an estimation for the sum Cout+Cload.

Any helpful answer would be appreciated.Thanks in advance.

Sorry, but I don`t understand the problem.
You speak about the "capacitance seen at the output of the first opamp" and in the next sentence from "the gm value of my differential pair at the input stage".
So, what capacitance you are speaking of? At the output of the first opamp stage (diff. amp) or at the opamp output?

Another comment to upvl's contribution: Loop gain as well as loop bandwidth are no "closed loop characteristics". Instead, both parameters are defined for an open loop!

 

[jimito13]

upvl : thanks for your suggestion.i will try and come back if i have further questions.

LvW : The Tow-Thomas biquad single ended topology consists of three operational amplifiers with different load capacitance seen at the output of each one due to the different feedback circuit elements.So,i am looking for an average estimation of the load capacitance for the op amp that will fit my design (if the estimation is feasible).if my opamp consists of a folded cascode OTA with gain boosting the Cload value i need will help me determine the gm value of my diff pair.That's what i meant.Thanks for your reply.

Any further suggestions will be appreciated by anybody.

 

[LvW]

OK, now the situation is a bit more clear to me.
As far as I've understood, you intend to design the opamp by yourself rather than to use a commercial IC, right?

Nevertheless, I do not understand why speak about "the capacitance seen at the output of the first opamp". (This is determined by the filter structure).
Or do you mean: The output of the first opamp stage?

 

[jimito13]

That's right,i intend to design my own opamp with some specifications i have derived from the filter structure.I have chosen a Tow-Thomas biquad for my circuit.

Exactly this capacitance concerns me : "the capacitance seen at the output of each of the three opamps".How can i determnine that??Do you have any idea?Maybe i will design an one stage opamp with gain boosting,so the output of the first stage will be the output of the overall opamp.I hope i am more clear to you now.

 

[LvW]

OK, two questions:

1.) By using the word "opamp" I think of a voltage controlled voltage source (VCVS) with a pretty low output resistance and a rather high input resistance - as well as a dc gain in the order of at least 80 dB).
Is this your intention? (I doubt, because you speak about a one stage amplifier).
In this case you can use this opamp in the classical TT biquad - and you can forget any load capacitance at the output of this opamp

2.) If you are going to use any other kind of amplifier (finite input/output resistor; gain resp. transconductance finite), you cannot use the classical TT topology.
Or do you speak about OTA`s? Then an OTA-suited topology (with grounded caüacitors) is to be applied.

 

[upvl]

That's right,i intend to design my own opamp with some specifications i have derived from the filter structure.I have chosen a Tow-Thomas biquad for my circuit.

Exactly this capacitance concerns me : "the capacitance seen at the output of each of the three opamps".How can i determnine that??Do you have any idea?Maybe i will design an one stage opamp with gain boosting,so the output of the first stage will be the output of the overall opamp.I hope i am more clear to you now.

Actually, you load is not a pure cap but a complex impedance, because you would have some feedback resistors for the opamp. If you don't take the resistor into account, the total load may be roughly C=(Cout//Cf)+Cin, where Cout is the capacitance of the output node, including the parasitic cap of this opamp and next opamp, Cf is the feedback cap, and Cin is the capacitance of the input node. But, as I mentioned, if you want your design to be accurate and avoid over design, Cadence provides "stb", which is very nice for such applications, since loop gain and loop bandwidth are more meaningful for feedback systems.

 

[jimito13]

thanks a lot upvl.i will try stb analysis and come back if i come up with any question.

 

[Megh]

@upvl,
Using "stb" is fine for Single ended structure. What if I have a fully differential opamp.
How would you suggest I should model the load seen by the opamp.
Thanks
Meghraj

 

[sutapanaki]

Why do you want to chose your capacitors in the filter based on stability? According to me those should be based on noise. I would chose caps and resistors based on noise, then design the opamps accordingly.

 

[upvl]

@upvl,
Using "stb" is fine for Single ended structure. What if I have a fully differential opamp.
How would you suggest I should model the load seen by the opamp.
Thanks
Meghraj

I remember when you use stb, you'll use a special macro model in analogLib. Now I remember it's just for differential case, but, if not, you can use vcvs to build a simple balun circuit to convert the single-ended signal to differential signal.

 

[sutapanaki]

iprobe in cadence is doing stability analysis for single-ended case and is performing a procedure similar to Middlebrooks method. There is another component from analogLib (I forgot the name) which can be used for differential case.
I doubt that building a balun with vcvs will give you all the benefits. Better build it with ideal transformers.

 

[dedalus]

There is another component from analogLib (I forgot the name) which can be used for differential case.

cmdmprobe

CMDM parameter: 1 - common-mode stability, -1 - differential-mode stability

 

[sutapanaki]

yes, exactly, that was it.

 

[Megh]

this cmdmprobe can be used with stb analysis right?
I tried using it it gives me some weird results.
Anyone has tried this?