I'm using a USRP B200mini SDR card for my GNU Radio energy detector project. I wanted to measure its Receiver Sensitivity ($RS$) experimentally and compare the experimental value to the theoretical value. Assume the minimum needed Signal to Noise Ratio ($SNR$) threshold for detection is $0\ dB$ and assume the bandwidth ($BW$) to monitor is $20\ MHz$, then using the $RS$ equation: $RS=N_{Floor}+N_{Figure}+10\times \log(BW)+SNR$,
where, $N_{Floor}$ is the noise floor, and $N_{Figure}$ is the noise figure.
We have: $N_{Floor}=10\times\log(kTB)=10\times\log(1.38×10^{−23}×290×1)=−204\ dB=−174\ dBm$,
where, $k$ is Boltzmann's constant, $T$ is room absolute temp in kelvin, and $B$ is $1Hz$ bandwidth.
$N_{Figure}<=8\ dB$ according to the Ettus website.
Hence, $RS=−174+8+10\times\log(20,000,000)+0=−93\ dBm$.
The problem now is that I don't see a match between the theoretical value which is $−93\ dBm$ and the experimental value which is $−63\ dBm$. Note that I found the experimental value based on the definition of receiver sensitivity, which is a measure of the minimum signal strength that a receiver can detect. Based on this definition, my energy detector can start detecting at a signal power around noise power level which is around $−63\ dBm$ (see the attached flow graph). There is around a $30\ dB$ difference between the theoretical value and the experimental one! Note that I didn't connect any antenna to the SDR card and kept the SDR gain at $12\ dB$.
Notes:
I put the SDR card gain at $12\ dB$ in order to calibrate the average power calculations from the flow chart to match the average power value provided by a Keysight power sensor U2000A. I used a sine wave source to do the calibration.
Note that the Keysight power sensor U2000A also gives a noise power level of around $-63\ dBm$ without an antenna connected.
The "Receiver Gain (dB)" label is given by GNU Radio itself and I don't know how to change it, but it should be "Power Spectral Density (dBm)" for this flow chart. But anyway, the problem remains the same, there is a big gap between the calculated $RS$ and the measured one.
What am I missing here? I'm not sure how to explain this difference, any help is appreciated. Thank you very much.