GRC FFT Block Question

Hi List,

I am trying to find the spectrum symmetry of a modulated signal.
That’s why i use the fft block on GRC. (The flow i used is attached).I
use AM modulation as an example. And i write a block that finds the
left and right side of the spectrum and make the calculation
P = (PL - PU)/(PL+PU).

But every time i run the flow the point of the carrier is changing. I
use 2048 point fft, hence i expect to find a carrier at the 1024th
point and every 2048 point after (3072, 5120, etc) until i stop the
flow.

But this value changes any time i run the code( some values i take as
an ex: 1010, 1012, 1018, 999, 1045, 1044, etc). I also attached the
code for the calculation. I couldn’t figure out what i did wrong.

Thanks for any help.

Mehmet.

On Wed, May 11, 2011 at 12:48 PM, mehmet kabasakal
[email protected]wrote:

point and every 2048 point after (3072, 5120, etc) until i stop the
flow.

But this value changes any time i run the code( some values i take as
an ex: 1010, 1012, 1018, 999, 1045, 1044, etc). I also attached the
code for the calculation. I couldn’t figure out what i did wrong.

Thanks for any help.

Mehmet.

Mehmet,
You are seeing the frequency drift of the oscillator in both the
transmitter
and receiver. They will never be perfectly accurate. You should probably
test your block using a simulation setup. That is, tie the transmitted
signal straight to the receiver in GRC and bypass the USRPs altogether
to
make sure your block is correct.

If you are looking for a symmetrical signal, you could then use your
calculation as the error value in a control loop to correct your
frequency
offset :slight_smile:

Tom

On 05/11/2011 07:57 AM, Tom R. wrote:

calculation as the error value in a control loop to correct your
frequency offset :slight_smile:

Tom
With the range of observed values given by Mehmet, it looks like a total
frequency error of about 12.5PPM, which is pretty darned good.


Marcus L.
Principal Investigator
Shirleys Bay Radio Astronomy Consortium

Hi Marcus,

May i ask you how you calculate the freq error i.e. 12.5 ppm ?
Is there a formula?

Mehmet.

2011/5/11, Marcus D. Leech [email protected]:

On 06/02/2011 03:01 AM, mehmet kabasakal wrote:

Hi Marcus,

May i ask you how you calculate the freq error i.e. 12.5 ppm ?
Is there a formula?

Mehmet.

Just take the error in Hz, divide by the “expected” frequency, and
multiply by 1.0e6


Principal Investigator
Shirleys Bay Radio Astronomy Consortium