3
$\begingroup$

Really bizarre issue here. Here is an image of a signal taken from SDR Sharp:

enter image description here

Freq: 926.365M Bandwidth: 30k

Now, using the above measurements, here is this same signal observed in GNU Radio with the QT Waterfall Sink at 5M sample rate:

enter image description here

Here is the same signal at 2M sample rate:

enter image description here

Here is the same signal at 10K sample rate:

enter image description here

And at 1K:

enter image description here

This cannot be possible?!

My GNU Radio flow is as simple as possible, but here's what it looks like:

enter image description here

Can anyone explain this? Surely as I close in on the signal, it should be filling up the Waterfall much more.

$\endgroup$

2 Answers 2

3
$\begingroup$

Most likely, the sample rates you are asking for are unsupported by the hardware. The osmocom source is just a wrapper around a bunch of specialized device drivers, and generally you're not going to notice if you don't get the sample rate you asked for unless you're also e.g. using an audio sink so you hear the audio skipping.

If you're programmatically interacting with the source you can get a list of supported sample rates, but this isn't easy within GNU Radio Companion; the usually convenient thing to do is look at the source code within gr-osmosdr for the interface for the particular hardware you're using.

In order to actually get a lower sample rate, you must start with a supported sample rate and then use a decimating filter block (there are many) to convert to the one you want.

Note that having a high sample rate is not necessarily completely wasted; the information in the extra samples, as processed by the filter, allows you to get more dynamic range as if your ADC had more bits per sample.

$\endgroup$
0
$\begingroup$

Answered my own question. It appears the SDR I am using (RTL2832U) has a minimum sample rate of 900k, which is why I was only able to observe the signal at that sample rate.

The solution was to record the signal at 900K, and then pass it through a Freq Xlating FIR Filter, with a decimation value set above 1.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .