[USRP-users] DC value

Sanat Gulvadi sanatgulvadi at gmail.com
Thu Jun 14 13:01:51 EDT 2012


Good Evening,

I have an OFDM based communication system in which I use a UHD application
to communicate between 2 USRP2 devices. I notice from my received data that
there is a small attenuation in the subcarriers which are relatively in the
middle of the band. The QAM points in the constellation are drawn towards
the point (0,0) of the constellation. Could this be because of a DC value ?
How do I avoid this or maybe push the DC out of my band of interest ? In
one of my previous questions, it was suggested that I use offset tuning.
How do I do this ?
Also, each of the 4, 4-QAM "clouds" have are "spread" to an extent.
Sometimes this spread is very less and almost unnoticeable, sometimes quite
large but still enough to keep all the points in their correct quadrant,
but quite often it is so large that they tend to have some points in the 2
two neighbouring quardrants and this leads to symbol errors. To illustrate
what I mean by this spread, I have a link to a rough sketch I made, since I
don't have an actual plot with me right now.
link : http://imgur.com/7g9yl
I should add that I have no external reference to synchronize the 2
devices. I have all my synchronization algorithms in the receiver. However,
when I did use a 10Mhz square wave at 10dBm from a function generator
connected to the REF inputs of the 2 devices, and run
usrp->set_clock_source("external",0), I get the E LED on the front panel to
glow, but there is no significant change. I don't have any other better
external source to generate a PPS signal simultaneously.
I understand the second part of my email is on a different topic. I'd be
grateful to get some response for both, but even just a suggestion for the
DC stuff would be great.

Have an awesome one.
Best Regards,
Sanat
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20120614/c2fae3e4/attachment-0002.html>


More information about the USRP-users mailing list