Bandpower measurment USRP2

Dear all,
I would like to receive as many suggestions as possible on how to
accurately
measure bandpower with a USRP2 + WBX setup.
I know I should use a block that does the square magnitude (FFT squared)
and
sum the resulting coefficients but after this I don’t really know what
the
correct procedure is. I would like to have the value in dBm and I know
I’m
suppose to use 30 dBm + 10log10, but the resulting value is not the
correct
one.
Please help me by offering your insight on this matter.
Thanks!

Vlad.

View this message in context:
http://old.nabble.com/Bandpower-measurment-USRP2-tp30323488p30323488.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 11/28/2010 08:38 AM, Vladutzzz wrote:

Vlad.

The basic flow for power measurement is:

source–>bandpass_filter->complex-to-mag-squared–>

single-pole-iir_filter–>calib_multiplier–>calib_offset–>log10*10

You’ll need to determine your calibration constants by experiment, and
you’d need to determine
what those should be for any given
bandwidth/center-frequency/gain-setting.

You can’t simply apply a fixed formula–there are too many uncertainties
in the analog realm
to make precision power measurement work without caibration
experiments. The power
seen by your detector (complex-to-mag-squared+filter) will be
proportional to:

             GAIN*(system-noise+signal-power)

There’s substantial uncertainty in the precise value of GAIN, due to
inevitable, expected
part-to-part variability. If you, for example, command the GAIN on
your daughtercard to
65dB gain, the actual gain may vary by up to about 2dB, and such
uncertainties are
generally frequency dependent. RF amplifiers usually have somewhat
more gain at their
“bottom end” than at the “top end” of their frequency range. Further,
you don’t know
how much system-noise there is, at least, not precisely, which means
that for measuring
very small signals, the total-power seen by the detector may be
dominated by system-noise.

So, you have calibrate, through experiment. If you’re trying to make
precision power measurements,
you’re going to have to calibrate for each variation in your system
setup (gain settings,
frequency settings, bandwidth settings).


Principal Investigator
Shirleys Bay Radio Astronomy Consortium

Marcus,
First of all thank you for your reply!
I have some questions about your very much appreciated explanations:
After using the complex-to-mag-squared block, should I consider the
coefficients as being in W or mW (should I use 10log10 or 30 + 10log10
to
get the power value in dBm)?
What is the function of the single-pole-iir_filter? What does it actualy
do?
I’m asking this so I will know how to calculate its parameters. Does it
give
the frequency bin based signal band, a more rounded appearance?
Again thank you for your help!
I hope other people will also find this useful.

Vlad.

===================================================================

Marcus D. Leech wrote:

correct procedure is. I would like to have the value in dBm and I know

You can’t simply apply a fixed formula–there are too many uncertainties
part-to-part variability. If you, for example, command the GAIN on
dominated by system-noise.
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org


Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page


View this message in context:
http://old.nabble.com/Bandpower-measurment-USRP2-tp30323488p30326643.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 11/28/2010 06:53 PM, Vladutzzz wrote:

I hope other people will also find this useful.

Vlad.

You expressed a desire to measure power over a finite bandwidth (at
least, that’s what I think
you were asking).

The values after the complex-to-mag-squared block are unscaled total
power estimates over
the input bandwidth. The IIR filter is used to smooth the result–in
a hardware power detector,
it would be a simple R-C network, here we use a simple single-pole
IIR filter, although you could
use a simple FIR low-pass as well, but the single-pole IIR is cheap,
and has adequate
behaviour for total-power detection–it’s what I use in radiometry.

In order to convert to your favourite units (like dBm), you’ll have to
take some
experimental measurements, and scale the output values
accordingly. Please re-read
my original reply, I’m not going to repeat it here.


Marcus L.
Principal Investigator
Shirleys Bay Radio Astronomy Consortium

Thanks Marcus.
You did manage to tie up several loose ends, and I really appreciate it!
All the best.

Vlad.

Marcus D. Leech wrote:

do?
least, that’s what I think
behaviour for total-power detection–it’s what I use in radiometry.

Discuss-gnuradio Info Page


View this message in context:
http://old.nabble.com/Bandpower-measurment-USRP2-tp30323488p30335394.html
Sent from the GnuRadio mailing list archive at Nabble.com.