Re: Questions on Polyphase Clock Sync block

[Dave Miller]

George,

No deep reason for the gain block between the Polyphase Clock Sync and the Costas Loop.  I just used the gain block for my convenience when testing the flowgraph with an external modulator and different settings to see what worked best.

Regards,

Dave

On Wed, Mar 10, 2021 at 12:12 PM George Edwards <gedwards.eng@gmail.com> wrote:

Hi Dave,

Thanks again for the help!

I followed your example to set the firdes filter parameters and my GRC flowgraph is now working,...Thank you! I must mention an observation, the I/Q constellation points after timing synchronization are around +/-0.8 on the real and imaginary axes and not +/-1. I know the filter gain scaling you use is correct. The gain must be equal to the number of arms in the polyphase timing synchronizer and I am using 32 arms like you so I set it to 32 (Fred Harris' work makes that clear). At any rate, I am not worried about this because I increase filter gain to a value greater than 32 to get +/-1 on the I/Q constellation axes. 

I have a question for you: Is there a reason why you scaled the values leaving the Polyphase Clock Sync before they entered the Costas Loop?

Thanks for the great help!

Regards,

George 

On Tue, Mar 9, 2021 at 4:11 PM Dave Miller <dave.todd.miller@gmail.com> wrote:

George,

I would recommend that you start with the below referenced example symbol sync flowgraphs and parameter settings and tutorials first, get them up and running well, and then tweak the symbol sync block settings to your desired parameters/settings for your scenario:

• .grc examples: You can find some symbol sync example flowgraphs with gnuradio version 3.8.2 installed on usr/share/gnuradio/examples/digital/demod
• GRCon 2019 Slide Presentation: I developed the tutorial presentation at the following link in order to help get others up and running quickly with the Polyphase Clock Sync block including the RRC filter taps. See slides #8 and #10 for the exact parameter settings that may answer your question. It was done in gnu radio version 3.7.11, but the flowgraph and all its parameter settings still work in version 3.8.2.

              https://www.gnuradio.org/grcon/grcon19/presentations/NASA_Space_Communications_Network_Modem/

• GRCon 2020 Presentation Video: Daniel Estevez’s video tutorial uses the other symbol sync block in gnu radio with the Gardner TED selection. You can pause his video at about 1:01:24 with the symbol sync “properties” box open and see all the exact settings.

            https://www.youtube.com/watch?v=RDbs6l4rMhs

You may also want to check your symbol rate frequency offset setting between your modulator and the symbol sync block and verify that the frequency offset is not too large for the loop bandwidth of the symbol sync.

Hope the above is useful,

Dave

_________________________________________________________________________

From:	George Edwards
Subject:	Questions on Polyphase Clock Sync block
Date:	Mon, 8 Mar 2021 16:21:22 -0600

---

Hello,

My goal is symbol timing using the Polyphase Clock Sync block. I generated a QPSK waveform which I input to the Channel Model block with epsilon set to 1.0005 (small signal timing offset) and pass the signal to the  Polyphase Clock Sync. The Polyphase filter is designed to run with 32 arms and receive the tap coeff's from the RRC Filter Taps. The parameters for RRC Filter Taps block are are set with defined variables as follows: 

Gain =1

Sample_rate (Hz): sample_rate

Symbol_rate (Hz): sample_rate/sps        (where sps is samples/symbol and

                                                                 defined)

Excess_BW: excess_bw

Num Taps: 32*(sps*6*2)           (designing for overlap of +/-6 symbols with "sps"                                                      samples/symbol)

1) When I run the GRC and look at the constellation plot at the output of the Polyphase Clock Sync, all I see is a dot at the origin of the I/Q plot.

2) So I tried changing the number of taps to see if things change and only when I try the number of taps equal to 32*11 do I get something that makes some sense. I get data over the 4 quadrants between +/-.04 +/-0.4j, spread out. Of course, it keeps smearing into one. 

Will appreciate any suggestions on how to use the Polyphase Clock Sync block in conjunction with the RRC Filter Taps for symbol timing.

Thank you!

Regards,

George