Usrp_fft.py axis values and units

Hi,

I’m currently using USRP, GNU Radio and the file usrp_fft.py to find
the Signal-to-noise ratio, Noise- and Signal-power.
I use the ZigBee PHY from Thomas S. to generate the signals at
2.412GHz.

My questions are the following:

  1. In usrp_fft.py are the axis denoted as “amplitude [db]” and of
    course “frequency”. Now in order to deduce the signal strength or the
    noise strength respectively, I need the reference amplitude used to
    compute the db value. In the code there is the setting “–ref-scale”
    and it’s default value is 13490.0. Now is this reference Amplitude
    13490.0 mV? (look at units -> milli-volts)

  2. How can I get from here to the signal-power and noise-power? Or
    maybe the signal-strength isn’t bad either… (I need those values to
    find the transmitting power of the USRP, I guess it could be 1mW, but
    don’t know how to check that)

Thanks a lot for your help!
best,
Björn

On Wed, Jul 14, 2010 at 9:30 AM, [email protected] wrote:

value. In the code there is the setting “–ref-scale” and it’s default value
is 13490.0. Now is this reference Amplitude 13490.0 mV? (look at units →
milli-volts)

It sounds like you’re trying to figure out SNR based on an FFT. If
that’s the case, I don’t recommend that. There are plenty of in-band
noise sources that would be counted as signal when, in fact, they are
noise. There are other methods for estimating SNR if you indeed want
to get that calculation.

I don’t think the amplitude maps directly to milli-volts. The units
are literally the integer values after filtering. You need to take
into account any gains in the system before and after the digital
converter.

As a rudimentary example, if you had a converter which gave signed
14-bit numbers (+/- 8191), you may want to set the reference to be
84.28dB (6.02dB/bit * 14bits) and figure out how far down from full
scale input you are (this will be a negative number as 84.28dB should
be your top end). You can then add and subtract the overall system
gain from that point to figure out what power levels are at different
locations in the system. NOTE: This is something that requires you to
understand all the gains in the entire system.

Does that make sense to you?

  1. How can I get from here to the signal-power and noise-power? Or maybe the
    signal-strength isn’t bad either… (I need those values to find the
    transmitting power of the USRP, I guess it could be 1mW, but don’t know how
    to check that)

If you just want to find the transmitting power of the USRP, just
connect the antenna spigot to a 50Ohm terminated load, set your USRP
to output a vector of (1+0i) and measure the voltage. You now know V
and R as well as the fact it’s a single carrier tone - calculating
power should be easy. You can then compare that in ratio to 1mW to
calculate dBm.

Calculating the exact noise power, on the other hand, ends up being a
much more difficult problem.

Thanks a lot for your help!
best,
Björn

Good luck.

Brian

Quoting “Brian P.” [email protected]:

strength respectively, I need the reference amplitude used to compute the db
I don’t think the amplitude maps directly to milli-volts. The units
locations in the system. NOTE: This is something that requires you to
connect the antenna spigot to a 50Ohm terminated load, set your USRP
Björn

Good luck.

Brian

Hi,

Thanks a lot for the reply! But I have some follow-up questions:
About the statement above: “There are other methods for estimating SNR
if you indeed want to get that calculation.”:

       What I need to measure is the following: I need the

Transmit Power of the USRP (Pt) and the Noise Power at the Input of
the USRP (Pn), and if possible also the Received-Signal Power (Pr).
These values would then enable me to calculate SNR, distance between
Sender and Receiver (of course with an appropriate Path-Loss Model),
etc. Those values will then be used for some other calculations and
Measurements.

Of course it would be great if the measuring of Pn, Pt, Pr would be
possible using the USRP itself (e.g. using the usrp_oscilloscope or
usrp_fft), but if this is really that bad I’d be happy to get some
other suggestions. Any Ideas?

best and thanks a lot for your help!
Björn

Hi,

I was wondering if anyone found a datasheet of the USRP and/or
XCVR2450 which contain the Noise Figure of those devices, or at least
of a combination of the two!?

best and thanks a lot,
Björn

Hello,

The noise figure will in large part depend upon the receiver front-end
that is attached the USRP or in the daughter board that is used. In
that case that you would like to know the noise figure of the USRP
without any RF front-end then the results may vary whether you are using
a BasicRX board or one of the LFRX boards. For a 50 Ohm system, either
of these should yield accurate results. Mainly, the noise figure of the
USRP will be derived from the AD9862 digitizer. Since this IC contains
an amplifier, the noise figure will vary with the gain of the PGA.

I too am interested in measuring the noise figure of the USRP. If you
are unable to find anything, I will be able to supply measurement data
within the next few weeks for a USRP device and BasicRX daughterboard.

~Jeffrey L.

On 7/15/2010 9:15 AM, [email protected] wrote:

Hi,

I was wondering if anyone found a datasheet of the USRP and/or XCVR2450
which contain the Noise Figure of those devices, or at least of a
combination of the two!?

best and thanks a lot,
Björn


~Jeffrey L., K1VZX

On 07/15/2010 08:51 AM, Jeffrey L. wrote:

I too am interested in measuring the noise figure of the USRP. If you
are unable to find anything, I will be able to supply measurement data
within the next few weeks for a USRP device and BasicRX daughterboard.

The USRP + BasicRX will have a noise figure of around 35dB, since it has
no amplification. If designing a daughterboard, you can use that number
as the noise figure of the ADC for your calculations of the overall
noise figure.

Matt

Zitat von “Matt E.” [email protected]:

an amplifier, the noise figure will vary with the gain of the PGA.
Matt
Hi,

Thanks a lot for your help!

Can I also use those 35db as the total noise at the input of the USRP
to compute the signal to noise ratio?

For that I’d also need the transmitting Power Pt. Would you by any
chance also have an idea how i could figure out Pt?

best and thanks a lot!
Björn

The main chip of the XCVR2450 is the MAX2829. The data sheet it says
3.5dB at a high gains settings. I think I recall obtaining something
similar when I tested with sinusoids.

BR/
Per


From: discuss-gnuradio-bounces+perz=removed_email_address@domain.invalid
[discuss-gnuradio-bounces+perz=removed_email_address@domain.invalid] on behalf of Jeffrey
Lambert [[email protected]]
Sent: Thursday, July 15, 2010 5:51 PM
To: [email protected]
Subject: Re: [Discuss-gnuradio] USRP and XCVR2450 Noise Figure

Hello,

The noise figure will in large part depend upon the receiver front-end
that is attached the USRP or in the daughter board that is used. In
that case that you would like to know the noise figure of the USRP
without any RF front-end then the results may vary whether you are using
a BasicRX board or one of the LFRX boards. For a 50 Ohm system, either
of these should yield accurate results. Mainly, the noise figure of the
USRP will be derived from the AD9862 digitizer. Since this IC contains
an amplifier, the noise figure will vary with the gain of the PGA.

I too am interested in measuring the noise figure of the USRP. If you
are unable to find anything, I will be able to supply measurement data
within the next few weeks for a USRP device and BasicRX daughterboard.

~Jeffrey L.

On 7/15/2010 9:15 AM, [email protected] wrote:

Hi,

I was wondering if anyone found a datasheet of the USRP and/or XCVR2450
which contain the Noise Figure of those devices, or at least of a
combination of the two!?

best and thanks a lot,
Björn


~Jeffrey L., K1VZX


Discuss-gnuradio mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/discuss-gnuradio

On 7/15/2010 1:00 PM, [email protected] wrote:

Thanks a lot for your help!

Can I also use those 35db as the total noise at the input of the USRP to
compute the signal to noise ratio?

For that I’d also need the transmitting Power Pt. Would you by any
chance also have an idea how i could figure out Pt?

Not quite. Noise figure is a measure of the degradation of SNR, so if
you know your input SNR, you would know that the measurements from the
USRP will have a 35 dB worse (lower) SNR. Alternativley, you know your
input SNR is 35 dB better (higher) than what you measure “coming out of”
the USRP.


Patrick Yeon
ThinkRF
613-369-5104 x418

On 07/15/2010 10:20 AM, Patrick Yeon wrote:

Not quite. Noise figure is a measure of the degradation of SNR, so if
you know your input SNR, you would know that the measurements from the
USRP will have a 35 dB worse (lower) SNR. Alternativley, you know your
input SNR is 35 dB better (higher) than what you measure “coming out of”
the USRP.


Patrick Yeon
ThinkRF
613-369-5104 x418

No, that’s not how it works either. The SNR is only degraded by an
amount equal to the noise figure if the input signal has a noise floor
of the thermal minimum (i.e. straight from an antenna). For anything
coming through a device with gain or noise would not fit that criteria.

Matt

On Thu, Jul 15, 2010 at 10:51:58AM +0200, [email protected] wrote:

Measurements.

Of course it would be great if the measuring of Pn, Pt, Pr would be
possible using the USRP itself (e.g. using the usrp_oscilloscope or
usrp_fft), but if this is really that bad I’d be happy to get some
other suggestions. Any Ideas?

Hi Björn,

this gets asked so often on this mailing list–you should have another
browse through the archives. And like everyone else, I will tell you
it’s not that easy; measuring real powers means tedious calibration, and
even then, it will only work reliably once, and for one specific setup.

Even if you had Pt and Pr, calculating SNR still depends on many things.
E.g., the typically given equation for a BER for BPSK modulation in AWGN
depending on SNR assumes a very specific SNR value (i.e., the noise
distributed over the bandwidth related to the pulse shaping filter).

For digital signal processing purposes, you very rarely actually need
true powers (although of course I don’t know what you’re up to), but
there’s usually a way to estimate SNR (for a given definition of SNR)
for a specific application.

Cheers
MB


Karlsruhe Institute of Technology (KIT)
Communications Engineering Lab (CEL)

Dipl.-Ing. Martin B.
Research Associate

Kaiserstraße 12
Building 05.01
76131 Karlsruhe

Phone: +49 721 608-3790
Fax: +49 721 608-6071
www.cel.kit.edu

KIT – University of the State of Baden-Württemberg and
National Laboratory of the Helmholtz Association

Zitat von “Martin B.” [email protected]:

etc. Those values will then be used for some other calculations and
browse through the archives. And like everyone else, I will tell you
there’s usually a way to estimate SNR (for a given definition of SNR)
for a specific application.

Cheers
MB

An approximation of those values wouldn’t be bad either!

But thanks for the reply!