two-stage-opamp - SKILL.md Agent Skill

name: two_stage_opamp description: "Two-stage CMOS operational amplifier simulation skill. Use when the user wants to simulate or size a Miller-compensated two-stage op amp with PMOS differential input pair, NMOS current-mirror first-stage load, NMOS common-source second stage, PMOS current-source load, bias mirror, Cc compensation, and CL load. Supports ngspice DC operating point, AC gain/phase sweep, pole-zero analysis, and output-noise analysis."

Two-Stage Op Amp Skill

Miller-compensated two-stage CMOS op amp using PTM 180nm models.

Topology

M7 + Ibias -> PMOS bias mirror
M5        -> first-stage PMOS tail current source
M0, M1    -> PMOS differential input pair
M3, M4    -> NMOS current-mirror active load
M2        -> NMOS common-source second stage
M6        -> PMOS current-source second-stage load
Cc        -> Miller compensation from first-stage output to Vout
CL        -> output load capacitor

Ports: INP, INN, OUT, VDD, VSS.

Running

cd two_stage_opamp/scripts
python run_opamp.py

This runs:

DC operating point and basic bias currents.
AC open-loop gain and phase versus frequency.
Pole-zero extraction for dominant pole, non-dominant pole, and zeros.
Open-loop output-referred noise, input-referred noise, and unity-gain closed-loop output noise.

Outputs go to WORK/plots, WORK/logs, and WORK/netlists.

Main Scripts

Task	Script
Full simulation	`run_opamp.py`
DC operating point	`run_opamp_dc.py`
AC gain/phase	`run_opamp_ac.py`
Pole-zero	`run_opamp_pz.py`
Noise only	`run_opamp_noise.py`

Design Knobs

Edit scripts/opamp_common.py:

IBIAS controls the PMOS bias mirror current.
CC and CL set Miller compensation and output loading. The default CC=12pF is intentionally conservative so the extracted non-dominant pole and first RHP zero clear the GBW-ratio checks.
W_M0_M1_UM, W_M2_UM, W_M3_M4_UM, W_M5_M7_UM, W_M6_UM set transistor widths.
VCM sets input common-mode voltage.