| Author |
Message |
neo
Joined: 19 Aug 2004
Posts: 57
Location: Beijing
|
 13 Dec 2004 7:45 tell me about the groupdelay~~ |
|
|
|
could u tell me something about the groupdelay of the filter, which is used in the baseband of the WLAN around 10M. some material maybe..
u see, if the dc signal delays 40ns and the signal at 10MHz delays 60ns, how could we cancel the error caused by the groupdelay?
|
|
| Back to top |
|
 |
okguy
Joined: 01 Mar 2002
Posts: 560
Helped: 6
|
| 13 Dec 2004 13:31 Re: tell me about the groupdelay~~ |
|
|
|
Your previous filter was not linear! With linear filter such Bessel, or equiripple, the group delay remains constant over a large bandwidth.
Search articles about equiripple filters.
|
|
| Back to top |
|
|
brmadhukar
Joined: 21 Jun 2002
Posts: 844
Helped: 29
|
| 13 Dec 2004 14:29 Re: tell me about the groupdelay~~ |
|
|
|
Hi,
It is not really such a critcal thing. I assume you are using a standard part for the filter.
If you find very apprecaible difference between this filter and an other one with linear delay, then only you need to consider this. However you may measure the delay and then use a compensating structure along with you RRC. Hope this helps.
B R
M
|
|
| Back to top |
|
|
neo
Joined: 19 Aug 2004
Posts: 57
Location: Beijing
|
| 13 Dec 2004 15:31 tell me about the groupdelay~~ |
|
|
|
| what confused me mostly is i cannot find any material talking about the group delay, seems it doesnnot matter much, all i found only talk about the simplest 1st order ones, why?
|
|
| Back to top |
|
|
Borber
Joined: 01 Jan 1970
Posts: 1492
Helped: 112
|
| 13 Dec 2004 17:32 Re: tell me about the groupdelay~~ |
|
|
|
| Ideal transfer function has linear phase with frequency. This means that all components of the spectrum of the signal has equal delay. Group delay is nothing but derivative of phase characteristics. So group delay is constant. Any deviations from constant causes change of pulse shape on the output. Question here is how much shape distortion you can allow. To minimize distortion, flatten group delay, you can implement phase equalization to transfer function (filter). Filters other than Gaussian or Bessel has steeper edge but introduce nonlinear phase responce so they usually have phase equalizing cells at the end.
|
|
| Back to top |
|
|
zmliu
Joined: 21 Mar 2004
Posts: 92
Helped: 4
|
| 14 Dec 2004 3:49 Re: tell me about the groupdelay~~ |
|
|
|
Group delay variation across passband is important parameter of baseband filter.
For constant group delay,use linear phase filters such as Bessel.
If Both high phase linearity and sharp roll-off magnitude response are required,delay equalization technique is available.
|
|
| Back to top |
|
|
neo
Joined: 19 Aug 2004
Posts: 57
Location: Beijing
|
| 14 Dec 2004 5:16 Re: tell me about the groupdelay~~ |
|
|
|
| Borber wrote: |
| Ideal transfer function has linear phase with frequency. This means that all components of the spectrum of the signal has equal delay. Group delay is nothing but derivative of phase characteristics. So group delay is constant. Any deviations from constant causes change of pulse shape on the output. Question here is how much shape distortion you can allow. To minimize distortion, flatten group delay, you can implement phase equalization to transfer function (filter). Filters other than Gaussian or Bessel has steeper edge but introduce nonlinear phase responce so they usually have phase equalizing cells at the end. |
u said "Question here is how much shape distortion you can allow", what did u mean, is it decided by the sample rate of the baseband AD, or by the bit rate of the digital baseband module?
by the way, i solve a 2nd order system for my group delay, but the formula seems very complicated, makes me a little uncomfortable!~~~hehe
|
|
| Back to top |
|
|
Borber
Joined: 01 Jan 1970
Posts: 1492
Helped: 112
|
| 14 Dec 2004 9:05 Re: tell me about the groupdelay~~ |
|
|
|
| Shape distorsion is meant as consequence of filter transfer function, it's phase response (group delay) and not as a result of digital processing of the signal. Usually, the shape distortion is visible as ringing on the edges of pulse.
|
|
| Back to top |
|
|
