Noise Spectral Density Curves

[yrr...@gmail.com]

Hi, 

I'm having a hard time trying to simulate the noise spectral density curves in Xyce. It seems like the output I am getting is the running integral over frequency and not the spectral density at each frequency.

The simulation I use looks like this:

.options device temp=27

.lib /pdk/sky130_xyce/models/sky130.lib.spice tt

XM vd vg 0 0 sky130_fd_pr__nfet_01v8 w=1 l=1

VVg vg 0 0.8 AC 0.0

VVd vd 0 1.8

HH1 out 0 VVd 1

.noise v(out) VVd dec 10 1 1G

.print noise inoise onoise

.end

Using the spice model from here: sky130_reveda

The output is flat for most of the frequency band with a very slight increase at the high end when I expect to see the normal FET noise response with 1/f noise blending into white noise.

I've run the same netlist through ngspice and got the result I expected.

It's probably some command I'm missing but any help would be greatly appreciated.

Thanks,

Thomas

[J Jacob Wikner]

i have similar issue but with other pdks. Xyce 7.2 linux.

[J Jacob Wikner]

Looks like flicker noise is not supported for the transistor models yet, unfortunately. They can be used in ADMS interface, and some of the work has already been done prepping the models for flicker noise.

Do developers have some updates on if we can enjoy flicker noise in the near future?

[J Jacob Wikner]

Actually it seems like a hack is needed to the N_DEV_MOSFET_B4.C. Initialize the noise density[FLNOIZ] and recompile 

Got quick and excellent help from the maintainers.