discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Polyphase Clock Synchronization block add some no


From: Tom Rondeau
Subject: Re: [Discuss-gnuradio] Polyphase Clock Synchronization block add some noise to constellation
Date: Thu, 8 May 2014 09:52:32 -0400

On Thu, May 8, 2014 at 6:07 AM, Irfan Ullah <address@hidden> wrote:
Hi all,
   I am using Polyphase Clock Synchronization block of GRC for symbol timing synchronization.currently I am doing simulations of QPSK using channel modal block. Ployphase clock synchronizer block work properly means it will synchronize the timing offsets in the signal remove the ISI from constellation but the problem is that it will add some noise to the constellation so could some one tell me how to solve this problem.
The picture of my flow graph is attached with this email you can see the parameters of polyphase clock synchronization I have used the firdes.root_raised_cosine_filter() for filter design in this block.

A few things. First, yes, there will be some noise added to the received signal. The PFB clock sync block very specifically produces noise based on the settings. The default value of using 32 filters in the filterbank is designed such that the aliasing noise it creates is below the quantization noise of a 16 bit number. The length of the RRC filter per filterbank arm will determine how much ISI you will accept. You can trade these issues off between computational performance and signal fidelity.

From your flowgraph, my guess is that you have something wrong with your filter taps and settings. The value of 44 for the number of filters is strange. It feels like you think that that's the number of taps per filter instead of the number of filters in the filterbank. The RRC prototype filter (rrctaps, I think, in your flowgraph) should be the normal RRC filter upsampled by the number of filters you use. So rrctaps will be a really long filter, but the filterbank partitions it among the N filters, so each arm becomes a reasonable filter again.

Also, you are using 2 for the output samples per symbol. The Costas loop works better off 1 sps as will your constellation.

Check out the documentation for this block to try to understand better what it's doing and how it works:

http://gnuradio.org/doc/doxygen/classgr_1_1digital_1_1pfb__clock__sync__ccf.html

Tom


reply via email to

[Prev in Thread] Current Thread [Next in Thread]