Flush samples of blocks when changing centrer frequency

Hi,

I have a python flowgraph with several blocks.

One of them is a “filter.single_pole_iir_filter_ff” which does a moving
average. The last block is a Sink to file block.

I scan several bands, 10MHz each, in order to get the power of the
signals
in those bands. I use moving average to get an average of 5 readings in
each band.

In order to change the center frequency, I stop the flow-grahp,
reconfigure
the USRP center freq and then start the flow-graph again.

I guess samples from the previous scans remain in internal block buffers
since I get power in samples of BAND2 (BAND1+10MHz) which are only
present
in BAND1. That is, the moving average is effective between bands and I
only
want to be effective in the same band.

I thought that stop() and start() signals would reset the entire graph
but
I am afraid it doesn’t.

Is there a way to flush those samples so that everytime I run the
frequency
changed flow-graph I don’t get the samples from the previous band?

Many thanks in advance.
Jorge

On Thu, Feb 5, 2015 at 11:30 AM, Jorge G. [email protected] wrote:

In order to change the center frequency, I stop the flow-grahp,
reconfigure the USRP center freq and then start the flow-graph again.

You can use the message port[1] in uhd source/sink or stream tags [2] to
change the center frequency

frequency changed flow-graph I don’t get the samples from the previous band?

UHD source will give rx_time, rx_rate and rx_freq tags whenever there is
a
tune request to a new frequency. You can use these tags to get rid of
unwanted samples.

Many thanks in advance.
Jorge

[1] GNU Radio Manual and C++ API Reference: UHD Interface
[2]
gnuradio/gr-uhd/examples/python/freq_hopping.py at master · gnuradio/gnuradio · GitHub

Hi

I am not familiar with those solutions. Could you please give a little
more
detail?

How can I use msg ports/ stream tags? Are they available in python
flow-graph or do I have to look into C++ block programming?

Where can I see rx_time, rx_rate and rx_freq tags ?

Regards,
Jorge