active filter design topologies

[ananthesh bhat]

Hello all,

I'm starting an active filter design with in .18u technology. Its a low pass butterworth approximation with 0dB gain and with frequency tuning capability. I'm confused about selecting the topology for the implementation. There are topologies like Sallen–Key topology,Tow-Thomas biquad filter topology and Multiple feedback topology. For my filter the cut off frequencies are 1M, 5M and 10M. For these cutoff frequencies which topology will be more suitable? Is there any advantage going with a particular topology leaving the other?

In Tow-Thomas biquad topology, I have to use three opamps and thus power consumption will be more. But will I gain anything by selecting this topology?

By the way power specification for the filter is 20mW.

Please help me in this!!


Thanks.....

 

[D.A.(Tony)Stewart]

Is this the only requirement?

 

[ananthesh bhat]

I have mentioned the specifications for the design. trying to decide upon the topology.

 

[BradtheRad]

I once constructed a biquad filter (using 3 op amps). It requires a set of matched capacitors and resistors. If you want to tune its frequency, you'll need a ganged potentiometer.

Have you found any op amps that perform at the frequencies you're talking about?

A passive RC filter is easy to tune its rolloff frequency, simply with one potentiometer.

 

[ananthesh bhat]

I once constructed a biquad filter (using 3 op amps). It requires a set of matched capacitors and resistors. If you want to tune its frequency, you'll need a ganged potentiometer.

Have you found any op amps that perform at the frequencies you're talking about?

A passive RC filter is easy to tune its rolloff frequency, simply with one potentiometer.

Hello,

I'm planning to design the opamp to work in this frequency range. To make an opamp to work from 1M to 10M will not be issue i guess. My question is whether Sallen-key topology can work in this frequency range ? Or should I go with bi-quad?
In Sallen-key I can use only one opamp.

 

[FvM]

You posted in the analog IC design section, does this mean that you intend to implement the filter completely on-chip, not just design the OP?

Which tuning method?

Provided you plan on chip filter and electronical (e.g. voltage controlled) tuning, this would suggest gm-C topology.

 

[ananthesh bhat]

You posted in the analog IC design section, does this mean that you intend to implement the filter completely on-chip, not just design the OP?

Which tuning method?

Provided you plan on chip filter and electronical (e.g. voltage controlled) tuning, this would suggest gm-C topology.

Hello,

I will implement the on chip tuning later.For the initial tape-out, I will go for off-chip tuning.

First I have to stick to the topology. I have decided to stick to Sallen-key topology with single ended opamp and supply voltage as 1.8V in .18u. Is this feasible?

 

[Taher_Selim]

For Sallen-Key cut-off frequency is calculated by w=1/(sqrt(R1*C1*R2*C2)). So theortically it should work by tuning these values. I mean no much dependencies on the Opamp.
However you should be aware that Sallen-Key is sensitive to small variation of R's and C's values.

I don't know what other biquads you can work with (becuase there are tow-thomas, KHN, ..) but if you mean tow-thomas it hasn't this sensitivity.
One advantage for tow-thomas is you can control gain for your input signal and filter the noise.

So tow-thomas is less sensitive, immune to noise, configurable gain.
Sallen-key is simple, small area and power.

Regards,

 

[ananthesh bhat]

For Sallen-Key cut-off frequency is calculated by w=1/(sqrt(R1*C1*R2*C2)). So theortically it should work by tuning these values. I mean no much dependencies on the Opamp.
However you should be aware that Sallen-Key is sensitive to small variation of R's and C's values.

I don't know what other biquads you can work with (becuase there are tow-thomas, KHN, ..) but if you mean tow-thomas it hasn't this sensitivity.
One advantage for tow-thomas is you can control gain for your input signal and filter the noise.

So tow-thomas is less sensitive, immune to noise, configurable gain.
Sallen-key is simple, small area and power.

Regards,

Hey Ali,

Thanks. This is what I was looking for. Yes. I have power and area constraints. So as you suggested, going for Sallen-key will be suitable for my design. Any idea on how much should be the open loop gain for the OPAMPS for this topology? Will a single stage folded cascode with 60dB gain be sufficient?

 

[Taher_Selim]

What I can read from first post in this thread that you need 0dB filter.
So be aware that Transfer function of Sallen-Key depends on gain of the opamp (H= G/ (second order filter equation)). so you should work with gain between 1 and 10 (according to your specs).
Regards

 

[KlausST]

Hi,

Will a single stage folded cascode with 60dB gain be sufficient?

--> 60dB gain at your desired cutoff frequency should be sufficient.
--> 60dB of DC gain: most probably not.

GBW is more informative.

Is distortion of the output signal problematic?

Klaus

 

[Taher_Selim]

--> 60dB gain at your desired cutoff frequency should be sufficient.
--> 60dB of DC gain: most probably not.

May I add a note please that he wants to work as a Butterworth filter.

Regards,

 

[KlausST]

Hi,

May I add a note please that he wants to work as a Butterworth filter.

sorry, i don´t get it.
What has filter characteristic to do with open loop gain of an OPAMP?

Klaus

 

[Taher_Selim]

I thought you talk about peaking gain at cutoff frequency which is not needed in butterworth filters. So I thought gain at cutoff frequency is not needed.

Regards

 

[KlausST]

Hi,

sorry, i should have written that i refererred to this:

Any idea on how much should be the open loop gain for the OPAMPS for this topology? Will a single stage folded cascode with 60dB gain be sufficient?

But i can´t answer if a single stage is sufficient.

Klaus

 

[ananthesh bhat]

What I can read from first post in this thread that you need 0dB filter.
So be aware that Transfer function of Sallen-Key depends on gain of the opamp (H= G/ (second order filter equation)). so you should work with gain between 1 and 10 (according to your specs).
Regards

Hey Ali,

If I make the circuit to work in Unity gain feedback mode, then we dont have the gain dependency in the transfer function right? As my specification for the filter is 0dB, I can operate in unity feedback.

So all I have to worry about is, if I'm designing a low pass filter with cutoff frequency as 10MHz then I should design the opamp with a pole near 50MHz and a gain of about 60dB. Is this correct?