Hello everyone, I am trying to manually calculate FIR filter coefficients to compensate for Chromatic Dispersion (As an input I provide dispersion,distance,wavelength and baud rate,taps: Output should be weights of the filter). I have a formula from the Savory paper, but it does not seem to work. Any ideas? Regards, Danish
FIR Filter Coeeficeints?
Started by ●July 28, 2008
Reply by ●July 28, 20082008-07-28
On Jul 28, 4:56 am, "Danish Rafique" <spurtofbl...@yahoo.com> wrote:> Hello everyone, > > I am trying to manually calculate FIR filter coefficients to > compensate for ChromaticDispersion(As an input I providedispersion,distance,wavelength and baud rate,taps: Output should be > weights > of the filter). > I have a formula from the Savory paper, but it does not seem to work. > > Any ideas? > > Regards, > DanishDeja vu? Didn't you already ask the same question and get an answer from Jerry Avins? http://groups.google.com/group/comp.dsp/browse_thread/thread/32dc178e73593370/48c263bbfbef2f44?lnk=gst&q=dispersion#48c263bbfbef2f44
Reply by ●July 28, 20082008-07-28
Hey, Well, I did not quite get him actually. I mean I already know it's not simple; that's why I am putting up a question, no? Can you elaborate? What I did earlier was to use an adaptive algorithm object and let the filter converge. Then I extracted the weights from the object. I used these weights in another simulation using the Matlab filter(Weights,1,Corrupted_Sequence) and it worked fine. But, now when I am trying to calculate weights *manually* and then use it with filter(Weights,1,Corrupted_Sequence), it simply does not work. Can you suggest a formula for calculating weights in this scenario? Regards, D
Reply by ●July 28, 20082008-07-28
Danish Rafique wrote:> Hey, > > Well, I did not quite get him actually. I mean I already know it's not > simple; that's why I am putting up a question, no? > Can you elaborate? > > What I did earlier was to use an adaptive algorithm object and let the > filter converge. Then I extracted the weights from the object. > I used these weights in another simulation using the Matlab > filter(Weights,1,Corrupted_Sequence) and it worked fine. > > But, now when I am trying to calculate weights *manually* and then use it > with filter(Weights,1,Corrupted_Sequence), it simply does not work. > > Can you suggest a formula for calculating weights in this scenario? > > > Regards, > DSo instead of asking for clarification you just ask the same question over again and hope? Well, it's a method I suppose. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" gives you just what it says. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 28, 20082008-07-28
So there is no answer anyways? and by the way, no offense Sir; but coming hard on people is no solution to problems. There always are better ways to point out mistakes rather then adding to the junk. Regards, D
Reply by ●July 28, 20082008-07-28
Danish Rafique wrote:> So there is no answer anyways? > > and by the way, no offense Sir; but coming hard on people is no solution > to problems. There always are better ways to point out mistakes rather then > adding to the junk. > > Regards, > DWell, since you've decided that it's polite to tell people when and how they've been rude, please allow me to adopt your rules of etiquette for a moment: 1. Your question was vague. I'll respond to it directly as to why, but in the mean time you can re-read it and ask yourself "could I answer this question if I knew the subject matter". Then ask "Is asking impossible questions a nice thing to do?" 2. This is USENET. Posting the same question twice, particularly if you've already been answered in your original thread, is considered Bad Style for a number of reasons: - You become the rude kid insistently tugging on someone's sleeve, because if your question is important to you It Must be Important to Everyone Else. - Your posts get perilously close to spam. - It gives everyone another thread to track. - It significantly raises the possibility of duplicate answers. Since giving answers is work, no one wants to spend 20 free minutes telling you something someone else already did a fine job doing. (OTOH, If you honestly think that your post hasn't gotten through that's another matter -- post, but you say in the text or the subject line that it's a duplicate post, explain why, and apologize in advance for any inconvenience it causes to someone who's seen your original.) 3. This is USENET, not a web forum. Please leave context _in_; most newsreaders don't automatically display more than one post at a time, and often times servers will lose parts of threads -- each post, even replies, should stand by itself. 4. This is USENET. Folks get brusque, or even rude. I try not to, but don't always remember to be nice. If you can't stand the heat, get out of the kitchen. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" gives you just what it says. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 28, 20082008-07-28
On Jul 28, 5:56�am, "Danish Rafique" <spurtofbl...@yahoo.com> wrote:> Hello everyone, > > I am trying to manually calculate FIR filter coefficients to > compensate for Chromatic Dispersion (As an input I provide > dispersion,distance,wavelength and baud rate,taps: Output should be > weights > of the filter). > I have a formula from the Savory paper, but it does not seem to work. > > Any ideas? > > Regards, > DanishTry and ask your question in traditional filter terms: phase response, magnitude response, group delay, etc. It's not clear what you are trying to accomplish K.
Reply by ●July 28, 20082008-07-28
Danish Rafique wrote:> Hello everyone, > > I am trying to manually calculate FIR filter coefficients to > compensate for Chromatic Dispersion (As an input I provide > dispersion,distance,wavelength and baud rate,taps: Output should be > weights > of the filter). > I have a formula from the Savory paper, but it does not seem to work. > > Any ideas?What Savory paper? Do you have a URL that points to a copy? You're extracting a formula from a paper. Do you understand the formula or are you just blindly applying it? What tests have you done to see if it'll work? -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" gives you just what it says. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 28, 20082008-07-28
Danish Rafique wrote:> So there is no answer anyways? > > and by the way, no offense Sir; but coming hard on people is no solution > to problems. There always are better ways to point out mistakes rather then > adding to the junk. > > Regards, > DSo, you've read my post and complained about my manner. Are you going to respond to the content, and go back to Jerry's answer detailing what you didn't understand and why? Clearly your original question was defective, because it only elicited one answer and that one wasn't useful for you. But instead of trying to fix your original question (by responding to it with an expansion, or by explaining to Jerry why you didn't understand his answer), you just post the same exact question over again. How, pray tell, is that going to do anything but, in your own words, 'add to the junk'? -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" gives you just what it says. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 28, 20082008-07-28
Danish Rafique wrote:> So there is no answer anyways?Here is a tip: if there is no answer, try increasing the reward for the solution.> and by the way, no offense Sir; but coming hard on people is no solution > to problems. There always are better ways to point out mistakes rather then > adding to the junk.If you decided to teach folks about life, you'd better find some other place.> Regards, > DVladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
