I will explain the system:

i am trying to transfer data from tx node to rx at UHF band on 400 MHz. I am trying first only wired transfer then i will try wireless.i am assuming only gaussian noise. My bandwidth is 1 Mhz. I am using gardner timing recovery with 2 samples/symbol. I am using 2nd order loop filter. please let me know any other information is required?.

What kind of interpolation filter should be used in timing recovery block of the receiver? I have seen some papers where linear, cubic, quadratic farrow filter is used.

Is there any particular criteria for this? Do i need a farrow filter or any other filter? Does this has any affect of the convergence of the timing recovery? If my input and output samples/symbol are same can i use any simpler filter?

  • 1
    $\begingroup$ There are so many options to this and I fear my comment will only lead to more questions (which I won't have time to go into a lot of detail unfortunately). That said let me suggest that you look into polyphase filter structures for timing recovery. It would be ideal for your case in that it can completely run at 2 samples per symbol yet provide the interpolation needed limited only by how many filter coefficients you want to use. I may have another post further detailing this approach and if so I will update the link to that. $\endgroup$ – Dan Boschen Mar 4 '20 at 1:46
  • 1
    $\begingroup$ Here is one related post: dsp.stackexchange.com/questions/30998/… $\endgroup$ – Dan Boschen Mar 4 '20 at 1:47
  • 1
    $\begingroup$ And here is the other detailing the polyphase interpolator approach specifically. I hope this is helpful! dsp.stackexchange.com/questions/51810/… $\endgroup$ – Dan Boschen Mar 4 '20 at 1:48
  • $\begingroup$ i have gone through your post its mainly linear filter. but still questing remains should the interpolation filter be linear, cubic, quadratic farrow filter? and any effect on the timing convergence? $\endgroup$ – nancy Mar 4 '20 at 6:04
  • $\begingroup$ @nancy he recommend a linear filter (and so do I, using non-linear filters is a bad idea in almost all cases of communications engineering...), so I think that is answered. Timing convergence is something that's impossible to say with the information you're giving, and defined by the specific implementation you decide to do. $\endgroup$ – Marcus Müller Mar 5 '20 at 10:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.