Real time FX correlator

[David Lonard]

Hello everyone,

I've been working on a real time correlator that doesn't require the need to save files on sold state drives.  This will allow for continuous observation and will greatly increase the observing throughput for my three element wide band interferometer.  It does come with the trade off though, that while real-time observation will soon be possible with BladeRFs that are sample stream synchronized, it is not 100% duty cycle like my previous SSD method.  This means that for pulsar work, this approach will not work.  It will work well though, for long, highly integrated observing that should allow for good image synthesis of steady sources or sources that vary over longer time frames.

dualwater.grc contains a Gnuradio flowgraph that I use to stream samples over a TCP socket.  

tcp2.py is a Python client script that reads data from the TCP socket and does the FX correlator step much like my previously posted code.

The PDF file shows some results from using some ATSC digital TV signals at 505 MHz for testing. Even though the BladeRFs are not synched, enough correlation exists for these digital signals to see that phase information remains intact over a ten minute period.  The waterfall spectrograms are showing signal phase information as different colors (see earlier presentations for more background).

David

[Mario Cannistrà]

Hi David.

Streaming correlation on GPUs !!

...also really good to know that python can be so fast on tcp, i didn't know...

 

Every time I read your posts I'm amazed by the innovation and improvement steps you add.

I guess you are leaving all of us without words every time...  :-)

Regards,

Mario

[Iban Cardona]

Woow, amazing work, congratulations.

Currentlly I have two ettus n210 with mimo with PC with a i7 and a gtx680. I hope try to test it in few weeks

73! Iban

eb3frn

--
You received this message because you are subscribed to the Google Groups "Amateur radio interferometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to amateur-radio-interf...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[David Lonard]

Mario,

Thanks for the encouraging words.  I am very eager to get out of the engineering phase to the place where I am getting real data. I think I am almost there.

David

[David Lonard]

Iban,

Thanks.  Let me know how it works, I am interested to see how your n210's work out.  

David

[John Yu]

Truly interesting project. Glad to see some deep dive into this area.

Could you please share your findings? if you have multiple I/Q stream collected by your phase synchronized BRF (I guess you got 3 x 15MSPS) IQ files, I can make an analysis and discuss the findings.

BR

John

[David Lonard]

John,

I don't have the raw IQ files from my BladeRFs any longer. However, I have some integrated filterbank IQ data that is much smaller and should be suitable for analysis at the link below.

This link includes some programs and results that should come in handy also.

David

https://groups.google.com/g/amateur-radio-interferometry/c/AUzPQAZcUDQ