Greetings,

Can somebody please explain why magnitude offset on NF trace results in a change of only one half of setting? When magnitude offset is set to -0.2dB, there is only -0.1dB change...

Is this intended?

N5245A opt.029 and N5244A

A.10.25.02 and A.10.40.12

Edited by: rfelectek on May 14, 2015 4:40 PM

Edited by: rfelectek on May 15, 2015 11:50 AM -

Can somebody please explain why magnitude offset on NF trace results in a change of only one half of setting? When magnitude offset is set to -0.2dB, there is only -0.1dB change...

Is this intended?

N5245A opt.029 and N5244A

A.10.25.02 and A.10.40.12

Edited by: rfelectek on May 14, 2015 4:40 PM

Edited by: rfelectek on May 15, 2015 11:50 AM -

The Noise related parameters (Noise figure, noise power, noise power density) all use as an underlying data representation the fundamental units of "degrees, Kelvin". This is from a choice made long ago and unfortunately propogated to the PNA that noise should be represented as noise power. All other powers used in VNA are represented as waves (which is the square-root of power) and so the lowest level representation has a power of 2 offset.

Since this definition is used, the internal formatter uses some special checking code to say "is this a noise-related parameter? If yes then formate 10 log (reading), if not then format 20 log (reading).

But, the offset function doesn't actually know about this so if it offsets a parameter by 3 dB, it actually doesn't shift the internal number in dB, but rather in linear (that is it multiplies by 2). But for the formatter, it will result in a shift in dB for a noise-related thing as half the expected dB change. oops. I will add it to our defect data base, but in the mean time just know that you have to use 2x the offset until this is fixed.

Alternatively, you can use equation editor to acheive the offset, but remember the same rule applies: 10 vs 20.

In the equation editor, the formatter uses the base parameter to decide. So if you have an S11 trace and create and equation that is MyNoise=NF, you will see the wrong dB value. If you do MyNoise=sqrt(NF) then it will be correct. OR, more common, if you want to compute noise figure from noise power and gain, you set up an noise power parameter, and you want to create NF=Noise_power/Gain, you need to create NF=Noise_power/(mag(S21)*mag(S21). ; if your "container parameter" (starting parameter) is S11, then you need the equation to be NF=sqrt(NoisePower)/S21.

sorry for the confusion but it is mathematically rigorous (NF and S21 are not commensurate values).