Measuring the Noise figure of a Direct conversion receiver
Regarding the measurement of a N.F of a direct conversion receiver with signal generator and spectrum analyzer.
It can be defined that the NF is the ratio of the SNRin to SNRout...So you enter a signal from the S.G to the RF input and assume thermal noise of -174dBm/Hz and measure the SNR at the output between the signal and the noise (which has overgain relative to the Spectrum analyzer's noise). The difference yeilds the NF.
The question is:
Does this measurment gives the DSB noise figure or SSB noise figure?
(for measurment purpose,the signal from the signal generator is converted to be 5MHz at Baseband i.e., LO=2.450GHz, RF=2.405GHz ==>IF=5MHz).
(Direct conversion receivers usually specifies DSB noise figure and not SSB noise figure)
Thanks,
>Does this measurment gives the DSB noise figure or SSB noise figure?
Has there been any image cancellation in the down conversion?
If yes then the measurement is SSB noise figure ( ignoring imperfect image cancellation ).
If not then it is a DSB measurement.
This is not a particularly accurate measurement method but it will give you an idea of the noise figure.
I would recommend using an external attenuator to reduce the signal level from the generator to reduce any unwanted noise.
Peter
The DSB noise figure is lower then 3dB from the SSB noise figure due to the fact that you get the negative frequency falling on top of you and thus you earn extra 3dB in the noise figure (since the signal is now doubled).
If you measure the NF a DC RX, or any other RX, that has no image rejection with a noise source then you will get a measurement that is 3dB optimistic.
If you measure it as you suggest with a signal generator then this is not the case.
You will be measuring the actual system noise figure including the contribution from the image noise.
The diagram below, if it comes out OK shows the RF input to the direct conversion receiver. I have assumed a signal on the low side of the local oscillator.
If the receiver has image rejection then the image noise will be suppressed and the output will be signal + noise.
If there is no image rejection then the output will be signal + noise + image noise.
Assuming noise and image noise are equal then the system noise figure will be 3dB worse than it should be. The nose level at the output is 3dB higher, but the signal is the same.
. !
. ! !
. ! !
^^^^^^^^!^^^^!^^^^^^^^^
. ^ ^ ^
. ! ! !
signal + noise ! !
. ! !
LO ------------------- !
. !
Image noise ------------- ! this gets mixed to baseband along with the wanted signal.