Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
As I said before the best way and practical and easy way is using Preamp.
because spectrum analyzer noise floor is about 20 dB, (For a PSA Agilent)
this preamp is basically an LNA with NF=6-7 dB and gain=23dB at 3.5 Ghz.
so it is impossible to measure the noise figure of LNA which is in order of 0.3-1 dB
more over you dont know what exactly is the SA noise figure.
another methode is using noise source if you got one.
well, all of you only gave half the answer ...yes you should use an LNA if you want to measure a DUT with low NF. after you use it how will you measure the NF of your amp?
well, you can either use a signal (cw) with a known BW at the spectrum and measure the SNR after the amp, or you can just measure how much the noise rised when you connect the DC (without any signal).
1.Please read vfone's link carefully, especially method 3.
2.You can measure with 10 dB Noise Fig. Sp.Analyzer amplifiers with 3 dB NF, if the amplifier's gain is high enough. Important only, that the noise power at the analyzer input, from the amplifier, is substantially higher, than the noise power of the analyzer.
I tried to use Agilent PSA to do noise measurement. The problem is that it seems to have an internal LPF that actually cuts off the 1/f noise starting roughly below 10kHz. I think the PSA frequency range is from 3Hz to 26GHz.
Is there an option to disable the filtering or is the filter built-in? It seems to me if the PSA can operate down to 3Hz, it doesn't make sense to filter out low frequency noise.
In my opinion, the question has an economical answer rather than technical one.
Best accuracy = need specific instrument.
In noise figure measurement, the largest difference between using NFA rather than SA is the residual uncertainty.
Very often uncertainty related to ENR is dominant, but using SA, especially if economic or not accurate, the overall uncertainty will rise a lot.
This may be accepted if tipically your DUT is in the range NF≥5dB, but if You try to measure very low noise DUT (say NF≤0.7 dB) the NFA is needed (and not enough...)
this measurement needs basic knowledge of noise measurement.
with LNA NF=7 dB you can make a good measurement, as long as you are using
a good PSA with low noise floor.
the LNA noise figure doesn't have to be very low , all matters is to be able to identify the noise change in PSA due to DUT.
the basic methode is to measure noise density on PSA which most of new spectrum analyzers do automaticaly in dB/Hz unit.
let say without DUT you have -147.0dB/Hz noise floor readout on PSA.(NF=7dB for preamp) given the fact that the gain for the additioanl preamp(LNA) is 20 dB at frequency of measurement, the resulting NF before Preamp is -167.0dB, now when you insert DUT, let say the noise density reading wil going to be -153.5dB on PSA.
you will decrease 20dB (preamp gain) the resulting number is -173.5dB , it means the NF of DUT is about -174.0dB +173.5dB=0.5dB, your done!!!!