Four skill packages that give an Agent the ability to design and simulate analog circuits: ngspice basics / gm/ID design / PTM model library / Sky130A PDK workflow.
If you are human: the examples below include images that show each skill's output at a glance.
If you are an AI Agent: skip the images and go directly to Installation. The full instructions for each skill are in its own
SKILL.md; runnable scripts and model files are in theassets/directories.
| Skill | Positioning | Features |
|---|---|---|
| ngspice | Beginner | 9 standard simulation examples (DC / AC / Tran / Noise) for learning SPICE from scratch |
| gmoverid | Advanced | gm/ID characterization simulation + design API, with automatic lookup of W, Id, Vgs, fT, and gm*ro |
| transistor-models | Model library | Full PTM model files (bulk silicon 65-180nm, HP/LP 22-45nm, FinFET 7-20nm) |
| sky130-pdk | Open PDK workflow | Install/locate Sky130A with Volare and run ngspice PVT + Monte Carlo smoke tests |
9 simulation examples covering the core concepts of introductory analog circuits:
| # | Type | Description | Output |
|---|---|---|---|
| 1 | Tran | RC charging voltage and current | tran_rc_charging.png |
| 2 | DC | NMOS Id-Vds family of curves | dc_nmos_iv.png |
| 3 | AC | RC low-pass filter frequency response | ac_rc_bw.png |
| 4 | Noise | RC filter output noise spectral density | ac_rc_bw.png |
| 5 | Tran | Sample-and-hold switch comparison | sample_hold_compare.png |
| 6 | Tran | kT/C noise time-domain statistics | tran_ktc_noise_hist.png |
| 7 | DC | NMOS current mirror output characteristics | dc_current_mirror.png |
| 8 | AC | Common-source amplifier Bode plot | ac_cs_amp_bode.png |
| 9 | DC | Transmission-gate on-resistance | dc_tgate_ron.png |
See ngspice/SKILL.md for usage. The Agent will automatically deploy the assets and run the examples.
NMOS Id-Vds Family of Curves
RC Low-Pass Filter Frequency Response
RC Charging Voltage and Current
kT/C Noise Time-Domain Statistics
Each process node generates three sets of standard plots:
IV Characteristic Plot (2x2)
- Id vs Vov in linear scale: clearly shows the threshold and saturation current
- Id vs Vov in log scale: subthreshold slope and an approximately 7-decade dynamic range
- Id vs Vgs (full sweep from 0 to VDD)
- Output characteristics, Id vs Vds (multiple curves at fixed Vgs values)
gm/ID Four-Quadrant Characteristic Plot (2x2)
- gm/ID vs Vov: full weak-inversion to strong-inversion behavior, including the BJT limit q/kT = 38.6 V^-1 and the 2/Vov asymptote reference
- Id/W vs gm/ID (log Y-axis): current density as the bias point changes, spanning approximately 3 decades
- fT vs gm/ID: cutoff frequency; PTM 180nm peaks at about 50 GHz, while PTM 22nm HP exceeds 600 GHz
- gm*ro vs gm/ID: intrinsic gain distribution versus bias point; 180nm reaches about 40-42 in weak inversion (gm/ID ~= 20), while 22nm HP is only 2-4 because of significant short-channel effects
Gate Capacitance Plot
- Cgg / Cgs / Cgd / Cgb vs Vgs: shows capacitance distribution and transitions from cutoff to threshold to strong inversion
Comparison Plots
- Channel-length comparison (L = 180 / 360 / 1000 nm): longer channels significantly improve gm*ro (up to ~140), with a corresponding reduction in fT
- Cross-node comparison (180nm SVT vs 22nm HP): clearly shows the speed-gain tradeoff across technology generations
Design API: given a gm/ID target, automatically looks up W, Id, Vgs, gm, fT, and gm*ro
from design_gmoverid import GmIdTable, print_op
tbl = GmIdTable('nmos180', W=10.0, L=0.18, vds=0.9)
op = tbl.size(gmid=15.0, Id=100e-6) # Fixed gm/ID and drain current; solve for W
op = tbl.size_from_ft(5e9, W=20.0) # fT >= 5 GHz; choose the lowest-power operating point
print_op(op)The first call automatically runs the ngspice simulation and caches the result. Later calls read directly from the cache.
The package includes three built-in PTM models: 180 / 45 / 22 nm. Once installed, it is ready for simulation. If you need more process nodes, install the transistor-models skill.
PTM (Predictive Technology Model) is a public SPICE model set maintained by Arizona State University (ASU), intended for process exploration and teaching or research when no PDK is available. This skill packages all models from mec.umn.edu/ptm:
- Traditional bulk silicon: 180 / 130 / 90 / 65 nm
- Bulk silicon HP/LP: 45 / 32 / 22 nm
- PTM-MG FinFET (multi-gate): 20 / 16 / 14 / 10 / 7 nm, HP + LSTP
It is independent of gmoverid. gmoverid already includes commonly used nodes; install this skill only when you need additional nodes such as 32nm LP or 7nm FinFET.
Copy the required .lib files from transistor-models/assets/models/ into your project's models/ directory as needed:
cp transistor-models/assets/models/bulk_cmos/ptm32lp.lib <project-dir>/models/
cp transistor-models/assets/models/finfet/nmos7mg_hp.lib <project-dir>/models/File naming rules:
bulk_cmos/ptm{node}{hp|lp}.lib- bulk silicon HP/LP, including NMOS + PMOS (model names:nmos/pmos)bulk_cmos/ptm{node}.lib- traditional bulk silicon, including NMOS + PMOSfinfet/{n|p}mos{node}mg_{hp|lstp}.lib- FinFET (model names:nfet/pfet)
For the detailed parameter table, see transistor-models/references/model_params.md.
Sky130A is a real open PDK workflow, not a PTM model file. This skill does not vendor the PDK payload; it teaches the Agent how to install or locate Sky130A with Volare/open_pdks and how to run ngspice PVT + Monte Carlo smoke tests.
Included smoke examples:
- NMOS Id-Vgs across
tt/ff/ss/fs/sfplus MC current spread. - Three-stage CMOS ring oscillator frequency across process corners and MC.
- Five-transistor OTA low-frequency gain across process corners and MC.
- Standalone
.scsand ngspice.spiexamples forring_oscillator,ota-5t, andamp-2s-miller, plus a small Sky130 SCS-to-ngspice-Sky130 converter for user circuit netlists.
Typical install:
python3 -m pip install --user volare
export PATH="$HOME/.local/bin:$PATH"
volare enable --pdk sky130 c6d73a35f524070e85faff4a6a9eef49553ebc2bRun a smoke test from a cloned repo:
git clone https://github.com/Arcadia-1/gmoverid-skill
cd gmoverid-skill/sky130-pdk
python3 assets/run_sky130_mos_iv_pvt_mc.py --mc-runs 3
python3 assets/run_sky130_ringosc_pvt_mc.py --mc-runs 3
python3 assets/run_sky130_five_transistor_ota_pvt_mc.py --mc-runs 3Run a direct ngspice example:
export PDK_ROOT="$(volare path)/volare/sky130/versions/$(volare output --pdk sky130)"
cd gmoverid-skill/sky130-pdk/examples/ota-5t
ngspice -b tb.spiTypical model entry:
.lib "$PDK_ROOT/sky130A/libs.tech/combined/continuous/sky130.lib.spice" tt
.lib "$PDK_ROOT/sky130A/libs.tech/combined/continuous/sky130.lib.spice" mc
See sky130-pdk/SKILL.md for usage.
The model files are copyrighted by the Arizona State University PTM project and are free for academic research. Please cite the following when using them:
-
Bulk CMOS nodes:
W. Zhao and Y. Cao, "New Generation of Predictive Technology Model for Sub-45 nm Early Design Exploration," IEEE Transactions on Electron Devices, vol. 53, no. 11, pp. 2816-2823, Nov. 2006. doi: 10.1109/TED.2006.884077
-
PTM-MG FinFET nodes (only the 7-20nm models in
transistor-models):S. Sinha, G. Yeric, V. Chandra, B. Cline and Y. Cao, "Exploring sub-20nm FinFET design with Predictive Technology Models," DAC 2012, pp. 283-288. doi: 10.1145/2228360.2228414
git clone --depth 1 https://github.com/Arcadia-1/gmoverid-skill /tmp/gmoverid-skill \
&& cp -r /tmp/gmoverid-skill/{ngspice,gmoverid,transistor-models,sky130-pdk} ~/.claude/skills/ \
&& rm -rf /tmp/gmoverid-skillgit clone --depth 1 https://github.com/Arcadia-1/gmoverid-skill /tmp/gmoverid-skill \
&& mkdir -p .claude/skills \
&& cp -r /tmp/gmoverid-skill/{ngspice,gmoverid,transistor-models,sky130-pdk} .claude/skills/ \
&& rm -rf /tmp/gmoverid-skillRun the following in Claude Code:
/skills
The output should include ngspice, gmoverid, transistor-models, and sky130-pdk, indicating that installation succeeded.
If you are an Agent, perform the installation steps above on behalf of the user. Choose either the global or project-level installation path according to the user's needs; no manual user action is required.
- ngspice (installed globally on the system)
- Python 3, with dependencies:
numpy,matplotlib,scipy;volareis recommended for Sky130A PDK installation
