Confusion, help please

GNU Radio
#1

Hello all,

I just had a quick question regarding the polyphase channelizer. As a
start, I’m trying to divide an input spectrum into 6 channels of equal
width. My sample rate for now is 5 MHz and the way I’m calculating the
taps
as follows:

firdes.low_pass2(1.0, samp_rate, samp_rate/(nchan2.15),
samp_rate/(nchan13.8),attenuation_dB = 10, window = firdes.WIN_HAMMING)

The channels are being divided equally now, but the first channel
behaves
very different from the rest, it becomes super wide for a moment, and
then
shrinks back to normal just like the rest of the channels and this
happens
repeatedly. I’m honestly stumped as to what the reason could be, so
would
really appreciate if anybody can point me in the right direction or at
least clarify why this is happening.

You can find the images here: http://tinypic.com/a/3e8so/4

Thank you,

Best,
Ali

#2

Hi Ali,

that’s interesting.
Would you mind sharing the flow graph that you use?

Greetings,
Marcus

PS: Although using adblock plus, before I see every other picture on
tinypic, I get stuff like casino advertisment, and similar.
Could you use any image hoster that doesn’t block view on the pictures
with advertisement? I found imgur.com to work fairly well.

#3

Hey Marcus,

Thank you for replying back! Absolutely, I’ve posted my flowgraphs on
imgur.com now.

Here’s the link: http://imgur.com/a/a0KCX

Also, for some reason, I think it could be because I’m using the
Spectrum
analyzer’s tracking generator?

P.S I’ve done a couple of changes since my last email:

  1. My sample rate is now 10M
  2. My filter taps for the channelizer are being calculated as you told
    me
    earlier: firdes.low_pass(1.0, samp_rate, samp_rate/(nchan2.15),
    samp_rate/(nchan    4.5))
  3. Assuming that the tracking generator is at fault, and everything else
    in
    the design is good, it works perfectly fine for a 2M sample rate, but
    when
    I try using a higher sampling rate say 10MHz, I get strange results
    which
    leads me to believe that there’s something wrong in my design, though I
    just can’t put my finger on it. One of the pictures is when I’m asking
    the
    channelizer to make a total of 8 channels, and the other one is where
    I’m
    asking the channelizer to create 20 channels. (Only 6 of which are being
    plotted in both cases). Is the way I’m calculating my filter taps at
    fault
    here or is it perhaps something else?

Best,
Ali

Date: Tue, 03 Mar 2015 09:42:19 +0100
From: Marcus M?ller [email protected]
To: [email protected]
Subject: Re: [Discuss-gnuradio] Confusion, help please
Message-ID: [email protected]
Content-Type: text/plain; charset=“windows-1252”

Hi Ali,

that’s interesting.
Would you mind sharing the flow graph that you use?

Greetings,
Marcus

PS: Although using adblock plus, before I see every other picture on
tinypic, I get stuff like casino advertisment, and similar.
Could you use any image hoster that doesn’t block view on the pictures
with advertisement? I found imgur.com to work fairly well.
On 03/03/2015 12:27 AM, Ali R. wrote:

Hello all,

I just had a quick question regarding the polyphase channelizer. As a
start, I’m trying to divide an input spectrum into 6 channels of equal
width. My sample rate for now is 5 MHz and the way I’m calculating the
taps as follows:

firdes.low_pass2(1.0, samp_rate, samp_rate/(nchan2.15),
samp_rate/(nchan13.8),
attenuation_dB = 10, window = firdes.WIN_HAMMING)