Yes, we calibrate every 15 min or so against blackbody loads to estimate sky temperature when doing long term measurements.
However we need to find a method to make sure the USRP is in it's optimal dynamic range.
In other systems we look at the time domain signals but I'm not sure how I can extract it (with sign) from I/Q samples in Gnuradio.

Den 2015-09-25 kl. 20:13, skrev mleech@ripnet.com:

If you're trying to derive input powers strictly-mathematically, but "modelling" the gains/losses in the hardware, you're going about it the wrong way.

The ONLY way to be certain is to use a calibration process with a source of known power. And use that at various gain/sample-rate/frequency settings to make a determination of what the input power actually is, vs what you're measuring.

USRP hardware isn't a calibrated device, so simply using the gain setting numbers and making assumptions about what that actually means in terms of observed power at the RF input terminals is not reliable. Further, the signal flow will do things to your numbers, and while most of those things are linear, you still need to calibrate against an external known signal source.

On 2015-09-25 14:03, Simon Olvhammar via USRP-users wrote:

Hi, thank you for your answer.

For our application gain settings are crucial due to the need of very precise calibrations.
How are you handling gain setting in the gr-ra_blocks? Is it fixed or have you implemented an automatic gain control?

Den 2015-09-25 kl. 03:24, skrev Marcus D. Leech via USRP-users:
On 09/24/2015 06:55 PM, Simon Olvhammar via USRP-users wrote: Hi How are gain settings usually handled in USRP devices? Is there any function provided so I can make sure the A/D converter is not saturated? If not, how can I extract the time domain data in a simple way? Best regards Simon _______________________________________________ The UHD API provides a mechanism for adjusting analog gain. Most daughtercards have a maximum RF input (for safety) of about -15dBm or so. The level where clipping may occur is usually a bit lower than that--keep in mind these receiver cards are intended to be connected directly to an antenna, so their sensitivity and dynamic range is skewed towards the lower end of power levels, unlike a standard laboratory spectrum analyser, which tends to be skewed in the other direction. The actual amount of RF gain adjustment range varies from daughtercard to daughtercard. The "usual" data stream from the devices is (usually decimated) time-domain complex

In Gnu Radio, samples are typically carried in complex-float format,
scaled to {-1.0,+1.0}.  If you want to occasionally take a peek at samples
flowing past, you can use a probe block, then a function-probe to call you
own function.

The dynamic range of USRPs is quite large, so if you start out in the
optimal part of the dynamic range, you're unlikely to drift outside of it
very much.

On the X310, the ADC is 14 bits, which gives you an SFDR somewhere around
80dB of dynamic range at the ADC.

On 2015-09-25 14:31, Simon Olvhammar wrote:

Yes, we calibrate every 15 min or so against blackbody loads to estimate
sky temperature when doing long term measurements.
However we need to find a method to make sure the USRP is in it's optimal
dynamic range.
In other systems we look at the time domain signals but I'm not sure how I
can extract it (with sign) from I/Q samples in Gnuradio.

Den 2015-09-25 kl. 20:13, skrev mleech@ripnet.com:

If you're trying to derive input powers strictly-mathematically, but
"modelling" the gains/losses in the hardware, you're going about it the
wrong way.

The ONLY way to be certain is to use a calibration process with a source
of known power. And use that at various gain/sample-rate/frequency settings
to make a determination of what the input power actually is, vs what you're
measuring.

USRP hardware isn't a calibrated device, so simply using the gain setting
numbers and making assumptions about what that actually means in terms of
observed power at the RF input terminals is not reliable.  Further, the
signal flow will do things to your numbers, and while most of those things
are linear, you still need to calibrate against an external known signal
source.

On 2015-09-25 14:03, Simon Olvhammar via USRP-users wrote:

Hi, thank you for your answer.

For our application gain settings are crucial due to the need of very precise calibrations.
How are you handling gain setting in the gr-ra_blocks? Is it fixed or have you implemented an automatic gain control?

Den 2015-09-25 kl. 03:24, skrev Marcus D. Leech via USRP-users:

On 09/24/2015 06:55 PM, Simon Olvhammar via USRP-users wrote:

Hi How are gain settings usually handled in USRP devices? Is there any
function provided so I can make sure the A/D converter is not saturated? If
not, how can I extract the time domain data in a simple way? Best regards
Simon _______________________________________________

The UHD API provides a mechanism for adjusting analog gain. Most
daughtercards have a maximum RF input (for safety) of about -15dBm or so.
The level where clipping may occur is usually a bit lower than that--keep
in mind these receiver cards are intended to be connected directly to an
antenna, so their sensitivity and dynamic range is skewed towards the lower
end of power levels, unlike a standard laboratory spectrum analyser, which
tends to be skewed in the other direction. The actual amount of RF gain
adjustment range varies from daughtercard to daughtercard. The "usual" data
stream from the devices is (usually decimated) time-domain complex samples.
Normally, without RFNoC, you get a stream of complex samples at whatever
rate you command. You then "do stuff" with those samples on the host PC.
There isn't a clipping indicator that is exposed, although some hardware
(USRP is a broad product line now), may have that on the ADC hardware, it's
not exposed anywhere that I know of.
_______________________________________________ USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

USRP-users mailing listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Would you consider looking at the I/Q samples level (calibrated) a
good approach to make sure I'm in the optimal part of the dynamic
range? Or is there some other preferred method using GnuRadio/UHD?