Re: [Discuss-gnuradio] Simple GMSK encoder and decoder with very large bandwidth

[Marcus Müller]

As said, the bandwidth displayed is really just labeled that way because you configured the FFT sink to use the arbitrary number 32000 as full frequency range. If you instead put in 1000 in the same place, your signal will instantly look only 1/32 as wide :)

So, the takeaway here is: that GMSK signal takes up about 70% of the Nyquist bandwidth (i.e. your sampling rate). There's nothing wrong with that!

Point is: GNU Radio just pushes around sequences of numbers. Nothing more. They don't have any inherent frequency info. Something that happens with a period two samples simply has twice the frequency that happens to have a period of 4 samples!

So, what the FFT sink does is simply take things that take 4 samples and display them at one fourth of the x-range. The labels below that axis are just scaled to fit your display "as if" you were converting that signal to analog at the rate you set as bandwidth in the display. *the scaling of the x-axis says zero about what you do with the signal*, it's still just a sequence of numbers.

So, your actual GMSK bandwidth depends on what sampling rate you choose – and what BT and samples per symbol setting you use :)

Best regards,

Marcus

On 04/20/2017 08:01 PM, Fred Castello wrote:

Sure Marcus,

Here is an attempt to give you what you requested.  It appears to me that the fundamental peak is +/- 12 kHz which seems very wide to me.  Also included is a copy of the flow graph.

Best Regards, Fred

On 04/20/2017 01:44 PM, Marcus Müller wrote:

Hey Fred,

nice to have you around!

Could you maybe post a screenshot of the frequency plot so that we can evaluate what you mean with "huge"?

I think you might not be realizing that DSP like GNU Radio does it doesn't care about the number you set as sampling rate – i.e. a GMSK symbol in the configuration you're using of the GMSK mod will always be 2 samples long. And that is what defines the shape in the spectrum. So, the frequency axis in your frequency plot is really just -bandwidth;+bandwidth _labeled_.

Best regards,

Marcus

On 04/20/2017 07:19 PM, Fred Castello wrote:

Hi there,
My goal here is to ultimately develop a telemetry system but I am starting very simply by building a GMSK encoder using a file sync fed directly to a decoder.  It seems to be working as I can see the constellation diagram (which eventually goes away-dont know why) and on the scope sync I can see the phase transitions but in the FFT sync it appears that the bandwidth is huge.  From what I understand, I would not want to filter the results going from the modulator to the demodulator, but I may not understand that correctly.  I know I have a long way to go, but I am just trying to learn as I go and try to figure out the theory and see how it applies as I build using GNU radio.  Thank you anyone for your help!!  I am including the Flow Graph as well as the test file that I am using for the file sync. Much appreciated.
Fred

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio