[Top][All Lists]

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

[Discuss-gnuradio] Fwd: [USRP-users] USRP, GNURadio Companion, Polyphas

From: Ali Riaz
Subject: [Discuss-gnuradio] Fwd: [USRP-users] USRP, GNURadio Companion, Polyphase Channelizer Issue
Date: Sun, 1 Mar 2015 16:49:47 -0600

Hello everyone,

Hope you're all doing well. I have a grc design related question, and it deals with the polyphase channelizer. I've tried looking into the documentation as had been previously suggested to see how everything works, and I got things working to an extent, but I'm confused as to the result that I'm getting. Here's what I have so far:

I want to divide the input spectrum into 8 equal channels using the channelizer, so the mapping should be as follows correct: (4 - 5 - 6 - 7 - 0 -1 - 2 - 3 - 4)?

And I'm sending the 4th channel to a null sink, and working with the remaining 7 channels with a channel mapping of [0,1,2,3,5,6,7,4]. To see if I was doing things correctly, I decided to use the WX GUI Scope Sink. The end result that I get is somewhat confusing as you'll see in a moment. I'm attaching the pictures for your reference.

1) Firstly, the channels are not being divided equally, is it perhaps because of the taps of the prototype filter? The way I'm calculating them was actually suggested by a really helpful person at the USRP-Discussion group, so I've also attached a picture for that as well for your convenience. Would you suggest something else?

2) Sometimes, when I'm comparing the individual channels with the spectrum of the combined channels, there seems to be this phase delay. Do you think it's a hardware issue (because of all the computations and processing going on) or does this usually happen when using the channelizer?

Just an fyi, this is not an optimized version of my design (Thus, the number of blocks lol); but I'm pretty sure some sort of vectorization can be done to optimize the code and reduce the total amount of blocks. But will get to that once I figure out why I'm not getting the desired output.

P.S My apologies if this is a redundant email, I just realized that I hadn't subscribed to the mailing list earlier, so sending it again just in case you didn't receive it.

Thank you in advance for your help.

Ali Riaz

On Thu, Feb 19, 2015 at 2:42 PM, Ali Riaz <address@hidden> wrote:
Awesome, thank you so much, both of you! I'll take a look into the documentation and Tom's blog, and will come back if I face any problems.

Best Regards,

On Tue, Feb 17, 2015 at 8:31 AM, Tom Rondeau <address@hidden> wrote:
On Tue, Feb 17, 2015 at 7:22 AM, sreeraj r <address@hidden> wrote:

You can also look into Tom's blog on channelizers [1]. Some of the grc files in the blog might not work out of the box (especially those ones with QT-GUI frequency sink + bus ports). You can simply toggle the busports and connect it properly and make it work.

[1] http://www.trondeau.com/examples/2014/1/23/pfb-channelizers-and-synthesizers.html

Sreeraj Rajendran

On Mon, Feb 16, 2015 at 11:02 PM, Marcus D. Leech <mleech@ripnet.com> wrote:
On 02/16/2015 04:54 PM, Ali Riaz wrote:
Hi Marcus,

I was the one who posted a question on the usrp-mailing list regarding the gnuradio companion channelizer a couple of days back, and was wondering if you could help me on something. I tried using the filter as you suggested, but I get a very strange result. I tried to get two partitioned channels, but they're being partitioned in a very strange manner. I've attached a photo for a visual for what I'm getting as a result. Any idea why that's happening?

P.S My apologies, I know you pointed me to the GNUradio discussion mailing list, but it's been a while, and nobody has responded yet, and so far all I have is your clue to go on.

Thank you so much for all your help, really appreciate it.


This is very squarely in the domain of the discuss-gnuradio mailing list.

My understanding is that the polyphase channelizer behaves that way for even-numbers of channels.  Tom Rondeau on the discuss-gnuradio mailing list
  is the best person, probably, to explain this.  I've copied that list.

Also, look at the documentation for the block:

Specifically, the set_channel_map function. It shows the mapping of the channels. With an even number of channels, the N/2 channel wraps around fs/2 and -fs/2 while channel 0 spans 0 Hz. You just have to set up your input bandwidth, sample rate, and tuning to account for this.


Attachment: Channelizer_Configuration.png
Description: PNG image

Attachment: Channels_Phase_Issue.png
Description: PNG image

Attachment: Channels_Width_Issue.png
Description: PNG image

Attachment: Taps_Calculation.png
Description: PNG image

Attachment: Channelizer and Detector GRC Plot.png
Description: PNG image

reply via email to

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