Re: [oc] DPLL

[Lior Shtram <liors@flextronicsdesign.co.il>]

Hi,

This is much too simplistic. When the channel acts as a sort of a 
filter, you get ISI and you need an equalizer. This will not only effect 
the bit clock sync but also phase sync and performance of the reciever.

This is however not a must in all systems, and does not relay on delay. 
Fading only makes your SNR worse which effects everything.
There are many methods to recover bit clock, to name a few:
1. Pass the signal through a non-linear filter, and filter it around the 
bit clock frequency.
2. Phase lock loop with feedback being the (S(clock-delta) - 
S(clock+delta))*(sign(S(clock))
    * not completely sure I got it right here, but this is about right 
anyway

you can get a lot of information about bit clock recovery from text 
books in digital communication.

So equalizer or no equalizer depends on your channel and is not a must 
for many satellite communication systems (doesn't hurt though usualy).

hope that helps, please send any questions.

regards,
Lior

haoguang.guo@philips.com wrote:

>Hi,
>           I have a problem about  bit synchronization when use QPSK in satellite communication. When design the demodulator , how can i get the
>bit clock?  Some one said because of the fading and unknown delay , you can not synchroniza to the transmitter . so must  use the equalizer to estimate the channel. Is it right?
>
>
>
>
>
>
>
>
>"Sam Gladstone" <samg@t-and-t.com>@opencores.org on 11/15/2001 06:29:28 AM
>
>Please respond to cores@opencores.org
>
>Sent by:  owner-cores@opencores.org
>
>
>To:     <cores@opencores.org>
>cc:      (bcc: Haoguang Guo/SHA/SC/PHILIPS)
>Subject:  Re: [oc] DPLL
>Classification:
>
>
>
>I take it you are demodulating back to a certain softbit size?
>There is several books out there that are pretty good. I will find the
>one I used when we were writing an 802.11a phy for a company.
>
>The main idea is to demodulate each dimension of the QPSKconstellation data
>seperately by
>breaking the real and imaginary up into two seperate demodulations.  Linear
>extrapolations can be used to generate softbits for a first order method.
>There are more cost effective, but
>harder to understand methods available as well.
>(Assuming that you have already done the proper power and phase
>corrections.)
>I think the number of softbits generated looks like this table.
>BPSK - 1 softbit per constellation
>QPSK - 2 softbits per constellation
>QAM-16 - 4 softbits per constellation
>QAM-64 - 6 softbits per constellation
>(QAMs get nasty because they are like combinations of different codings that
>have
>to have multiple softbits generated per demension because of grey coding
>with the
>modulution module. Yuck! )
>
>I will try to find the book name and send it out.
>
>Regards,
>  Sam
>
>----- Original Message -----
>From: <aolmo@grupoeyp.com>
>To: <cores@opencores.org>
>Sent: Wednesday, November 14, 2001 9:24 AM
>Subject: [oc] DPLL
>
>
>>Hi,
>>
>>I take the liberty disturbing you.
>>I am a friend from Taiwan.
>>After reading your posts on Web, I know you are a professional
>>communication designer.
>>Now I am designing QPSK demodulator for wireless Lan with verilog.
>>Where can I find more helpful material, such as verilog code for DPLL.
>>Please kindly to give me some advice.
>>Thank you a lot!
>>--
>>To unsubscribe from cores mailing list please visit
>>
>http://www.opencores.org/mailinglists.shtml
>
>>
>
>
>--
>To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml
>
>
>
>
>--
>To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml
>



--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml