The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> dynamic range of filter
https://designers-guide.org/forum/YaBB.pl?num=1243783961

Message started by smarty on May 31^st, 2009, 8:32am

Title: dynamic range of filter
Post by smarty on May 31^st, 2009, 8:32am

Hi All,
I may sound a bit naive, but I was interested to know is there any relation between dynamic range and SNR for analog filter. If so what exactly does they relate to.

Thanks,
SBR

Title: Re: dynamic range of filter
Post by raja.cedt on May 31^st, 2009, 11:05am

hi SBR,
for any stable circuit Power=k*SNR*BW,you can verify this for some passive filter (k is a constant will different for different circuits).

Thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by aaron_do on Jun 1^st, 2009, 5:45am

Hi Rajasekhar,

could you please show how you got that relationship? I know that Power>S*BW where S is the signal power at the output, but your equation also has an /N term in it...

For the dynamic range and SNR i don't remember the exact definition but I believe its the ratio between the signal power which causes a predefined THD to the signal power which gives the minimum allowable SNR. Probably want to check that up or hope that someone else answers this post.

cheers,
Aaron

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 1^st, 2009, 6:01am

hi Aaron,
i donno about THD but i assumed that SNR is same as dynamic range.
Take a simple rc low pass filter: SNR=v*v*/kt bw=1/r*c power=v*v/r....so you can verify the rule what i have given. And i learned this from one of tsividies student thesis ,there he has proven this simple example analytically and for some big ckts like 2nd order system through simulation. if you are interested i will send you the thesis.

Thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by smarty on Jun 1^st, 2009, 7:16am

Hi Rajasekar,
I had also read like Dynamic range is same as SNR but was not really clear in that definitive. If you can share the thesis which you are mentioning that would be of help to me.

Thanks,
SBR

Title: Re: dynamic range of filter
Post by aaron_do on Jun 1^st, 2009, 8:07am

Hi guys,

from my understanding, dynamic range is the range of values which the signal can take and provide a useful output. At the low end it is limited by noise and at the high end it is limited by distortion. That's where I got the relationship. But I'm not really sure if its defined exactly right. And also you should be careful when you use the word SNR since I believe some people use it to mean the ratio of signal power to noise power where noise is from random sources such as flicker noise and thermal noise, while some people use SNR to mean signal power to any unwanted power such as flicker noise and unwanted harmonics and interference.

I'm quite interested in the result Rajasekar. Could you give me the name of the student and i'll try and find his thesis. No need to send it to me. I think i've seen a relationship like what you wrote but i don't remember it too well...

thanks,
Aaron

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 1^st, 2009, 8:48am

hi Aaron
ya i think your argument is correct,but when you are taking about SNR people think about lower limit (i think distortion is difficult to find rather than noise ).Whats your opinion on dynamic range calculation?.Here is the thesis you asked.

http://www.ee.iitm.ac.in/~nagendra/dissertation/manuscript_hyperlinked.pdf

thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by smarty on Jun 1^st, 2009, 8:55am

Hi Rajasekar,
Thanks for the reference.

Best Regards,
SBR

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 1^st, 2009, 9:01am

hi SBR,
can you please define what do you mean by Dynamic range? is there any difference between SNR and Dynamic range.

Thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by smarty on Jun 1^st, 2009, 9:19am

Hi Rajasekar,
My understanding for dynamic range is the ratio of highest and lowest signal level in the circuit. and SNR is the ratio of signal level to the noise level.
But I heard most people say DR is same as SNR, which I was bit confused.

Thanks,
S.Baanurathan

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 1^st, 2009, 10:40am

hi S.Baanurathan,
Dynamic range is the usable signal range,in one way you can say it is equal to SNR,but as he reported some time distortion also will make scene

Thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by aaron_do on Jun 1^st, 2009, 7:23pm

Hi Rajasekhar and SBR,

thanks for sharing.

For the equation derived in the appendix it has two problems. In passive filters we don't really talk about power dissipation, although the equation may be correct. For active filters, orders of magnitude doesn't begin to decribe how far off the equation could be. For example A.19,

Pi = 8kTf₀S/N

Let S/N = 10 and f0 = 1 MHz. Then Pi = 3.2016*10^-13 W. I haven't heard of an active filter operating with less than the order of microwatts. The problem comes from the authors use of an output swing of Vdd which almost never happens in active filters.

BTW the author mentions

Quote:

The dynamic range specified is the range of input signals over which THD ≥ 40 dB and S/N > 0 dB are maintained.

So DR is basically a ratio of two signal strengths while SNR is the ratio of signal to noise strength. In my opinion it makes sense that if you want an overall SNR of 20 dB, then the maximum signal strength gives THD of 20 dB and minimum gives SNR (where N is random noise) of 20 dB.

Anyway overall the thesis looks quite interesting so thanks for sharing. I will have a look.

cheers,
Aaron

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 1^st, 2009, 8:35pm

hi Aaron,
why in passive filters power dissipation is not important concern?You just read that thesis and tell me because some where i read in that thesis regarding this.

Thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by aaron_do on Jun 1^st, 2009, 11:24pm

Hi Rajasekhar,

I usually take power consumption to mean the amount of battery (or whatever supply) power used. Power consumption is especially important in design for mobile applications which rely on batteries.

When I think passive filters, I usually think LC type (or microwave) filters which are commonly used as band-select filters and even image-reject filters. Such filters don't consume any power at all from the battery. There is some insertion loss but it is usually kept at a minimum...

I'm not really sure what other type of passive filters people use. Sorry if i'm being confusing here. I'll take a look at the thesis but it will take me some time as i'm quite busy.

cheers,
Aaron

EDIT: I just realised this thesis is about companding filters. Maybe the fomulae are correct....will keep reading.

Title: Re: dynamic range of filter
Post by vivkr on Jun 2^nd, 2009, 12:56am

In addition to the work done in the thesis mentioned here, more information may be found by looking at publications from Yorgos Palaskas (on Google or IEEE Xplore), also a former student of Tsividis, and a collaborator of Nagendra Krishnapura (who is the author of the thesis mentioned in the previous post).

I remember having attended a presentation on dynamic range in filters by Palaskas a few years ago. As Aaron correctly pointed out, this is primarily relevant to active filters since passive filters consume no battery power.

Regards,

Vivek

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 2^nd, 2009, 1:09am

hi vivkr,
that relation i gave will work for active filters and whatever, because in that thesis he proved for class A and class AB power amplifiers.Please correct me if am wrong.

Thanks,
Rajasekhar.

Title: Re: dynamic range of filter
Post by aaron_do on Jun 2^nd, 2009, 1:51am

Hi Rajasekhar,

he didn't actually prove the formula, he simply showed scenarios where it worked. A theory can only be disproven by (counter) example.

That said, I think the formula is much more relavant for companding filters which is the subject of the thesis. The reason is because companding filters have high output signal swings (output of the inner filter) for most of the dynamic range while conventional active filters such as linear active-RC filters only have high output swings when the input signal is large. You have to remember that a conventional active filter will not have high output SNR when the signal is small while a companding filter will have very high gain when the signal is small and thus achieve significant SNR gains.

When I gave you the example using equation A.19 I assumed an output SNR of 10 dB, however, this thesis is talking about an output SNR of 54 dB which is nearly 20000 times higher. A conventional filter cannot achieve 54 dB SNR over a very wide dynamic range (58 dB in this case).

BTW this is really an excellent thesis so thanks so much for showing it to me,

cheers,
Aaron

Title: Re: dynamic range of filter
Post by raja.cedt on Jun 2^nd, 2009, 2:30am

hi aaron_do,
thanks, i will work out and tell you.
Thanks,
rajasekhar.

Title: Re: dynamic range of filter
Post by Asmodeus on Jun 19^th, 2009, 1:49am

i feel dynamic range n SNR are same, provided their is no noise source other than the ckt itself.
SNR takes into account other noises too, apart from the intrinsic noise of MOSFET.

Based on it, i guess DR is always greater than or equal to SNR :)