OFDM behaviour of recent UHDs

Hi everybody,

I have been experiencing some difficulties in obtaining
the same “signal clarity” performance that I used to have with

uhd_003.004.000-f500b92

when I tried to transmit the same samples of the same signal
(also same usrp B100, same XCVR2450 front-end, same frequency =5.8GHz,
same
receiving device, same tx/rx antenna positions)
with both:

uhd_003.005.001
and
uhd_003.005.002

at the following links you can find images of a DVB-T modulation
analyzer
exposed to both transmissions.

www.pellegrini-radio.it/uhd_003.004.000-f500b92.jpg
www.pellegrini-radio.it/uhd_003.005.002.jpg

If you look at the MER value, a much poorer constellation quality is
evident when operating with the newer UHD and its matched images.
(no performance difference was noticed in this respect between
uhd_003.005.001 and uhd_003.005.002).

Has this been experienced before?
As somethin changed in between that I should be aware of?

my best regards to everybody

vince

On 04/15/2013 03:39 PM, Vincenzo P. wrote:

with both:

If you look at the MER value, a much poorer constellation quality is
evident when operating with the newer UHD and its matched images.
(no performance difference was noticed in this respect between
uhd_003.005.001 and uhd_003.005.002).

Has this been experienced before?
As somethin changed in between that I should be aware of?

If 3.4.0 worked, I would definitely give 3.4.5 a try to see if any of
the patches or fixes could have inadvertently caused the issue. That
will at least narrow down which versions are causing the difference.

There was a fix to DSP scaling to prevent clipping. Its possible that
the power level is simply lower? What signal amplitude are you using,
and does doubling it solve the issue?

http://code.ettus.com/redmine/ettus/projects/uhd/wiki/ChangeLog#003004005

-josh

hi Josh,
thanks for the very prompt answer.
Will run the tests you suggest tomorrow morning (CET) and let you know.

As to the power, the receiver measures the very same received power
value
in both cases. You can see this from the (first from top) bar indicator
on
the instrument in the two pics.

regards

vince

2013/4/15 Josh B. [email protected]