bessel OTA-C filter: help required

[EmbdASIC]

I need some help in designing a bessel OTA-C 2nd order low pass filter. I have designed my OTA, and now need some guidance to start designing my OTA-C filter.

Previously I have some matlab experience in designing digital filters only ! from where should i start?

 

[Braski]

first u have to choose your transfer function. Then u have to realize it with transconductor and capacitance. There are several architecture to achieve this goal. Look at Delayannis book. I post u a link where you can find it.

http://ifile.it/higq6to

The book gives several ideas to implement OTA-C biquad filters. I think it should help.

 

[LvW]

I need some help in designing a bessel OTA-C 2nd order low pass filter. I have designed my OTA, and now need some guidance to start designing my OTA-C filter.

My recommendation is as follows:
* OTA1: non-inverting integrator (input at pos. port) with C1 as load.
* OTA2: low pass 1st order (100% neg. feedback) and C2 as load; pos. input connected to output of OTA1.
* Overall feedback: Connection between OTA2 output and pos. input of OTA1.

This gives a good working lowpass of 2nd order with two grounded capacitors.

* Design formulas: pole freqiency wp=1/sqrt(T1T2); pole Q Qp=Sqrt(T2/T1)
with T1=C1/gm1 and T2=C2/gm2.
(values for wp and Qp derived from design tables).

 

[EmbdASIC]

Thank you braski and LvW.
Both your replies were of help towards my design.

@LvW
My design is fully differential so the grounded caps is now a differential cap. instead.
I have two queries here:
1. If i implement the filter, as chosen by the attached paper, then there is a slight change that i cannot correlate: the first feedback loop, as it does'nt go far enough to become 'overall feedback' (see separate fig.)
Could you address this difference?

2. Could you suggest some theoretical background for filter realization (or some text with design example like you have given here, for realizing a filter like that in the paper ! )

Thanks a lot !

 

[EmbdASIC]

Out of the so many techniques for filter realizatoin in the book "Continous time active filter design" by T. Deliyannis, "Ladder simulation method" is rather more straight forward.
Which method would you suggest as the simplest ?

 

[Braski]

My design is fully differential so the grounded caps is now a differential cap. instead.

if i could give you an advice, better if u use grounded capacitances of double value on both the terminals. It will occupy 4 times more area, but the grounded capacitances are always preferred.

About other theoretical background, i'm not expert in filter architecture, so i've to search something useful before telling you something else. However, there are tons of paper with filter example on IEEE. Try to search it.
In addiction, you have to be well aware of CMFB problematics (fully-diff systems), noise performance of your transconductors, linearity range...the transconductor design is very important, i think more important than the filter architecture.

I think ladder simulation is a good method, but i think TOW-THOMAS structure (9.6.1) is simplier (vi2 and vi3 grounded). However, i don't know if it applies to your specifications :|
the solution proposed by LvW is surely good and simple, and uses only 2 transconductors. Try it out!


Maybe the following paper could help. It's a bit long but gives u useful hints and addresses the problematics. Besides, if u could find the Tsividis book on continuous time filters (Integrated Continuous-Time Filters), i think it explains the architecture u posted (i thought it controlling the bibliography of the paper i posted).

And if someone knows where to find this book...it would be nice to let us know

Excuse me if i've been confusing. Hope i helped!

 

[LvW]

Out of the so many techniques for filter realizatoin in the book "Continous time active filter design" by T. Deliyannis, "Ladder simulation method" is rather more straight forward.
Which method would you suggest as the simplest ?

Hi embdASIC,

I would propose to follow the recommendation from braski and study the paper from Tsividis. It gives you a good overview over the various strategies to realize OTA-C filters.
By the way: The structure as formerly suggested by me is based on the very well known Tow-Thomas-topology (for opamps) which has only second order. And remember, the choice depends strongly on the required order. As mentioned by braski, for orders higher than 4 an active simulation of ladder structures is surely the best way.
Regards
LvW