Designer’s Guide Community

smarty Community Member Offline Posts: 46	choosign filter Jun 18^th, 2009, 11:21pm Hi All, �I am not sure if this has been already discussed in the forum, but I wanted to know from a given specs for filter how would we choose whether to implement using gm-C or opamp-RC. what decides which one to use. It might be basic but I was keen to know and if so how should I go about the task?. Thanks, SBR
Back to top	IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: choosign filter
Reply #1 - Jun 19^th, 2009, 1:26am

Ohhh smarty - another question would be much easier to answer.
There are several technologies and by far more than 20 different topologies to construct a lowpass filter. And the problem is - they influence each other !
The choice depends on a lot of subjects/requirements:
frequency, order of the filter, tolerance requirements, cost, technlogy (lumped or integrated), tunability, with/without zeros, power consumption, supply voltages, .....
Sure that I forgot something.

In general, I would say that gm-C can be recommended for higher frequencies, for integrated technology and/or when tuning is required.
But, as you see there are several other aspects - and a general answer is certainly NOT possible and not appropriate.
Regards

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

smarty

Community Member

Offline

Posts: 46

Re: choosign filter
Reply #2 - Jun 19^th, 2009, 2:25am

Hi Mr. Buddy,
�Thanks for the reply.
But my understanding was, gm-C would suffer from linearity perspective( which is one of the important criteria for filter) than the other counterpart, then in that case why is gm-C preferred, for sure people suggest that for high frequency ( I assume the other poles of op-amps) could affect the filter performance.

In general, what I really after was. Say we have a specs for Cut-off frequency, dynamic range, noise specs, group delay specs ( just to name a few) and this is for Integrated filter and I am not really caring for power, then how really should I go through the process.

Should I go from books for any approximation( butterworth, bassel, chebeshev...) or any other way.

Thanks.
SBR

Back to top

IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: choosign filter
Reply #3 - Jun 19^th, 2009, 2:49am

smarty wrote on Jun 19^th, 2009, 2:25am:

........But my understanding was, gm-C would suffer from linearity perspective( which is one of the important criteria for filter) than the other counterpart, then in that case why is gm-C preferred, for sure people suggest that for high frequency ( I assume the other poles of op-amps) could affect the filter performance.
...... Say we have a specs for Cut-off frequency, dynamic range, noise specs, group delay specs ( just to name a few) .......
Should I go from books for any approximation( butterworth, bassel, chebeshev...) or any other way.

At first, You are right in saying that dynamic properties/linearity are other aspects I forgot to mention.
At second, my understanding of your question now is that you are not sure about the several steps in designing a filter.
Therefore, at first the approximation has to be selected - and this depends on your damping requirements expressed by frequencies and corresponding damping.
Here you have several alternatives e.g. with or without zeros. On the other hand, this choice influences also the filter degree which is required - and also the delay characteristics (if they play a role). You see, it ends in a compromize between several conflicting aspects.
But that�s normal live for analog engineers and makes a single and final answer to your question difficult.
Regards
buddypoor

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

loose-electron

Senior Fellow

Offline

Best Design Tool =
Capable Designers

Posts: 1638
San Diego California

Re: choosign filter
Reply #4 - Jun 21^st, 2009, 6:33pm

Two questions are being asked here, and I see them being dumped into a single question.

Question 1 - What is my pole/zero placement necessary to get the frequency response desired?

Question 2 - What is the circuit level implementation needed to implement my pole/zero placement?

First question can get answered with some some simulation tools, matlab, or even (its a little awkward, but do-able) on an Excel spreadsheet. That or a good filter book (Zevrev is the classic) will get that question answered.

Second question is an issue of frequency and linearity. - High frequency filters (GHz land) tend to get done with passive methods (all LC sections or microwave with resonance cavities and similar.) As you come down in frequency and put it inside a chip, gmC filters get used in the 50MHz to 1GHz region. (YMMV) and linearity issues are a concern, but it can be controlled in a similar manner to how one deals with linearity in an RF design. As you come lower in frequency (200MHz to DC) you tend to use op-amp and RC methods. Linearity is much less of an issue because the high open loop gain of the op-amps will make for good linearity when in a closed feedback system.

As an aside, tuning of a gmC filter needs to be updated frequently due to thermal variance (even with compensation) and op-amp RC systems tend to be a bit more set once and then retune on the next power up.

Tuning systems tend to be a big part of the design effort so don't forget to budget time for that.

Jerry

Back to top

Jerry Twomey
www.effectiveelectrons.com
Read My Electronic Design Column Here
Contract IC-PCB-System Design - Analog, Mixed Signal, RF & Medical

IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: choosign filter
Reply #5 - Jun 21^st, 2009, 11:46pm

Quote loose-electron: First question can get answered with some some simulation tools, matlab, or even (its a little awkward, but do-able) on an Excel spreadsheet. That or a good filter book (Zevrev is the classic) will get that question answered.

Instead I recommend one of the powerful filter design programs available for download from the web. They are very easy to use and - more than that - give some hints for realization (mostly RC).
Although ZVEREV is the classical one, I would not recommend it.
It is too detailed. When you don�t intent to become a specialist in filter design, use instead filter software or some other introductory papers.
But, be aware that no single and final answer to your design problem is possible. It�s always a compromize between conflicting requirements.
Regards

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

smarty Community Member Offline Posts: 46	Re: choosign filter Reply #6 - Jun 22^nd, 2009, 5:04am Hi, �Thanks for your inputs, will do some work and will share my findings with you all.. Best Regards, SBR
Back to top	IP Logged