rfstab_examples

rfstab — two-port stability & gain report (Enhancement-253)

The first of the RF design-aid additions on top of the existing S-parameter machinery (.sp, n-port, Touchstone, stb, loadpull, noise figure). After a .sp analysis publishes the scattering parameters S_1_1 … S_2_2 versus frequency, the rfstab command post-processes them into the standard linear-two-port figures of merit, one value per frequency:

quantity	formula
determinant Δ	S11·S22 − S12·S21
Rollett K	`(1 −
stability μ (load)	`(1 −
stability μ' (source)	`(1 −
max stable gain MSG	`
max available gain MAG	`

A two-port is unconditionally stable at a frequency iff K > 1 and |Δ| < 1 (equivalently μ > 1). The results are stored as real vectors k, magdelta, mu, mu_src, gmax, msg, stable versus frequency in a fresh rfstab plot, and a summary (stability verdict, worst-case K/μ, gain range) is printed.

.sp lin 101 1meg 10g 1
.control
run
rfstab              * or: rfstab S_1_1 S_1_2 S_2_1 S_2_2  (e.g. a Touchstone plot)
plot k mu           * stability factors vs frequency
plot gmax msg       * gain circles / MAG-MSG
.endc

It only reads vectors, so it is analysis- and solver-independent.

Verification

verify_rfstab.py (both solvers, no numpy — Python's built-in complex is enough):

1. a passive T-attenuator (R1=R3=25, R2=100, Z0=50) has hand-computed metrics K=2.125, μ=2.33333, |Δ|=0.212121, MSG=0 dB, MAG=−6.02060 dB, stable, which rfstab reproduces exactly;
2. for a common-source MOSFET amplifier (a non-reciprocal, well-conditioned two-port), the four S-parameters are read back and K/Δ/μ/MSG/MAG are recomputed independently in pure Python, matching rfstab to ~3e-7.

Scope

New frontend command (frontend/com_rfstab.c); the ngspice binary is rebuilt. No solver, analysis, or numerical change — rfstab is pure post-processing of the S-parameters a .sp (or Touchstone) run already produced.