The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Channel Select Filter requirements
https://designers-guide.org/forum/YaBB.pl?num=1181718260

Message started by aaron_do on Jun 13^th, 2007, 12:04am

Title: Channel Select Filter requirements
Post by aaron_do on Jun 13^th, 2007, 12:04am

Hi all,

i have a question about channel select filter (CSF) design... My thinking is that the 2 main requirements are

1) attenuate interference so that IIP3 of following blocks does not need to be good

2) attenuate out-of-band noise including interference (considered random)

So the thing is the CSF center frequency (f0) can change by around 10% for example. This can lead to an attenuation of the desired signal. However, my CSF is only 3rd or 4th order, and most likely the signal will only be attenuated by 3 dB at the very most (due to the frequency offset). Requirements 1 and 2 will still be met with a 10% deviation in f0. Furthermore, this 3 dB loss doesn't seem to have much affect on the system NF since the CSF is near the end of the RX chain. So is it fair to say that i can live with the 10 %variation in f0?

thanks in advance,

Aaron

Title: Re: Channel Select Filter requirements
Post by ACWWong on Jun 13^th, 2007, 2:53am

Given a filter type and order, the choice and variation of bandpass filter f0&fbw (or just fc for lowpass) is a tradeoff between attenuation of out-of-band signals (so lower fbw the better) and too much signal filtering (causing ISI/closing of eye diagram, so higher fbw better).
System simulations with the expected modulation in something like matlab will tell you the impact of the variation of f0&fbw.
If your f0 is changing by 10%, then it is likely fbw is also changing. btw 10% is quite good control, are you using external component or calibration ?

Title: Re: Channel Select Filter requirements
Post by aaron_do on Jun 13^th, 2007, 7:42am

well...10% is a bit of a guess...I used resistive degeneration to get high IIP3 and designed it so that the passband gain and center frequency would be "immune" to resistor variation since R variation seems to be higher than C variation. Also, the overall biasing is referenced from a bandgap so i don't expect the current to change much. The manufacturers quote a 10% variation in C so that's what i based my estimation on.

Do you think I can get away with that or should I go for a tuning loop? I was told that it wouldn't be necessary but i always had doubts. Just looking for an outside opinion :)

Not so familiar with ISI...better start reading again...

thanks,
Aaron

Title: Re: Channel Select Filter requirements
Post by Berti on Jun 14^th, 2007, 8:15am

Hi Aaron,

I think you should use simulations to analyse how much degradation in EVM
you get due to your channel filter (signal attenuaten at the passband edge).
From this you can derive whether you need to tune your fiter or not.

Cheers, Thomas