Hello,

I'm having troubles with the Pnoise time-domain noise analysis. The analysis I am setting up as benchmark is a simple BJT differential pair with an ideal tail current source. The differential pair is driven by a 0.5V sinewave signal at 2GHz. Using controlled voltage sources, I am plotting the time-varying differential output noise current of the differential pair. From theoretical analysis, it is possible to predict the output noise time domain variation which exhibits a maximum when the differential input signal is about about zero (both BJTs are driving). In my case the tail current source is 1mA and the expected peak noise voltage (1Mhz, 1Mhz+1Hz) is about 20pA/sqrt(Hz). [ref: mixer noise paper terrovitis, grey]

I've used Pss/Pnoise tdnoise analysis to validate the calculation. maxsideband parameter is usually set high so as to take in account noise folding effects. I obtained a peak current noise of 80pA/sqrt(Hz) which is not the expected value. Setting back maxsideband to 1 gives the expected noise current value of 20 pA/sqrt(Hz).

The theorectical analysis takes in account the time-varying large-signal behavior of the differential pair(cyclo-stationnary noise), so where is the fault ? I don't understand ...  

City