AnsweredAssumed Answered

modeling PSA/receiver noise terms (not noise figure)

Question asked by achronon on Jun 24, 2009
Latest reply on Jun 30, 2010 by tabbott
Hi folks,
I'm trying to develop a model of a particular broadband attenuation measurement system.  I'm using an E4440A in measuring receiver mode as my linear detector.  I'm new to this RF game so I wonder if anyone can help me understand and/or solve some modeling questions.  It may take a while to explain myself but here I'll try give a brief synopsis of what I'm trying to do---you may need to interpolate or prompt my for clarification of my objectives.

I'd like to model the PSA/Receiver errors for relative TRFL measurements.  The E4440 does some sophisticated averaging and signal processing that I cannot imagine modeling, but supposing I can make do with a simple model.   The network sends the receiver some signal level (after accounting for reflections) = P_in.  This is mathematically just some number with an uncertainty (due to network noise, attenuation, leakage etc.).  The receiver processes this signal P_in somehow and I end up with a TRFL display value say =P_out (as just stated I'm assuming this includes linearity and range calibration error factors).  I'd like to model the receiver's contribution to the error in P_out, and ideally a simple multiplicative plus additive noise model seems logical.  The Agilent guides to relative TRFL accuracy (5989-8161EN) suggest three sources of uncertainty, linearity, range calibration and residual noise. As an example I've tried to use the Agilent "typical" uncertainties, given in dB, and convert them to linear units so that I end up with a model of the form

   P_out = (eta_lin + eps_rcal + esp_snr)*Cf* P_in

Here eta_lin = 1, eps_rcal=0, and  esp_snr=0, but each has a non-negligible uncertainty.  (Cf=cal factor is assumed constant for my purposes).  I believe one can make measurements to determine whether the linearity is slightly different from unity, and likewise for range calibration error, but I'm not sure about the noise term?  I can use the Agilent specifications for the noise uncertainty contribution (which is significant only for low SNR) but for completeness I'd like to figure out how to characterize the noise term for my particular receiver.  I haven't been able to work backwards from the information given in the reference Agilent App Note document "5989-8161EN" to see what I would need to do to directly measure my PSA/receiver's noise error term.  Can anyone provide some help/advice/suggestions for this?