USRP2 Noise figure and Phase noise measurement methods

Hi all!,

I am going to measure both parameters, noise figure and phase noise of
my
USRP2+XCVR2450 transceiver.
After some reading about different methods of measuring them I have some
questions.

Phase noise.
I read that in the lack of a Signal Source Analyzer (SSA) the most
method of measuring it is with the delay line technique. Is it the right
way
to measure the phase noise in the USRP? Has anybody of you done it
before?

Noise figure.
The standard way of measuring it seems to be the Y-Method using a noise
source connected to the DUT and then measuring the output power of the
DUT
when the noise source is on and off. However as I understand it, in this
case the USRP2 is not a two port device. It is just a receiver. Is it
valid
to measure the USRP2 power with software? (usrp2_fft.py for example, as
I
have read in another mailing list posts)

What are the range of values I should expect to understand I have made a
good measurements?

Jorge.

On Sun, 2010-10-03 at 15:30 +0200, Jorge M. wrote:

the right way to measure the phase noise in the USRP? Has anybody of
you done it before?

I just measure a CW and do an fft. Results: See the following post:

Noise figure.
The standard way of measuring it seems to be the Y-Method using a
noise source connected to the DUT and then measuring the output power
of the DUT when the noise source is on and off. However as I
understand it, in this case the USRP2 is not a two port device. It is
just a receiver. Is it valid to measure the USRP2 power with software?
(usrp2_fft.py for example, as I have read in another mailing list
posts)

What I do is to insert a CW with known power and then estimate the SNR.
The NF can then be calculated from

SNR = -174 + log10(BW) + Pin + NF,

where SNR and NF are in dB, Pin in dBm and BW in Hz. The BW is the noise
bandwidth. Roughly it is sample_freq/2.

What are the range of values I should expect to understand I have made
a good measurements?

Look at the data-sheet of the MAX2829. Note that at high Pin your SNR
will be dominated by the distortion and the NF will seem larger than in
actually is.

BR/
Per

Correction:

SNR = 174 - log10(BW) + Pin - NF,

BR/
Per