Hi.

DSB_NF calculations are different between ADS and SpectreRF.
For example, see "ne600" mixer circuit well known in SpectreRF.

For this mixer, ADS and CDS(Spectre Shooting & Flexible_Balance)
give following NF results.
     ADS_NFssb=15dB, ADS_NFdsb=8dB
     CDS_NFssb=15dB, CDS_NFdsb=12dB

ADS calculate DSB_NF like following.

              vn^2/Rl + kTo(G1+G2+...+Gn)
 Fdsb = -------------------------------------
                 kTo(G1+G2+...+Gn)

kTo is the noise power at the input termination.
G1 is the primary frequency conversion from RF input frequency to IF output frequency. G2 is the other frequency conversion from RF input image to the same IF output frequency.
G3 and higher are the higher order, low efficiency conversions
that occur as other input frequencies mix with harmonics of the LO to
appear at the IF output frequency.

As far as my understanding, Spectre calculates DSB_NF like following.

              vn^2/Rl + kTo(G1+G2+...+Gn)
 Fdsb = -------------------------------------
                 kTo(G1+G2)

In most mixers, G1 ~= G2 and G3..Gn << G1. So ADS and Spectre could give almost same NFdsb value.

But this is not true for "ne600".
This mixer have large direct feed through(G3) from RF input to IF output.
   G1=Gain(900MHz->100MHz)=10dB
   G2=Gain(1100MHz->100MHz)=10dB
   G3=Gain(100MHz->100MHz)=15dB

This is the reason why NFdsb of ADS is lower than Spectre.

Measurement using actual instruments is well correlated with ADS_NFdsb not Spectre.

How do you think ?