Noise simulation problems

Hi,

I am getting a bit confused with the Noise Simulation with Spectre.

I have the following test set-up:

- Psin port as input, having R=50
- LNA_PB block from rfLib with G=2dB, Rin=50, Rout=50, IIP3=0dBm, NF=5dB
- Psin port as output, having R=50 as well. Output noise is taken from
this port probe.

Using this set up, with both Noise analysis and S-parameters I get the
NF to be 5.103dB which is differing from the expected value. Is this a
problem with my set-up, NF simulations, or perhaps the rfLib component?

One thing which I noticed was, that in noise summary the noise
contribution of a 50ohm port is different from 50ohm resistor and
voltage source in series. What is the reason for this? Could it explain
the difference?

What kind of set-up should I use for proper NF simulations?

Thanks in advance

 

One thing to be aware of is that the noise of the port is defined by the noise
temperature of the port - there is a parameter "noisetemp". This defaults
to 290K (it is specified in Celsius though) - which matches IEEE specifications
for noise figure. In other words it doesn't track with the simulation
temperature, and even the default is different from the default simulation
temperature. This is because the probe is part of your test equipment, and so
will not change temperature as the die heats up.

If you have access to sourcelink, check my solutions 11227334 and
11025268 which give worked examples.

Regards,

Andrew.