A SiGe HBT low-noise amplifier on IHP SG13G2 on IHP SG13G2, IHP's open-source 130 nm SiGe BiCMOS PDK — designed by AI agents driving klayout-tools and the open-source xschem + ngspice flow.
Status: just opened. Nothing is designed yet. The first work is the S-parameter and noise testbench methodology — proving what ngspice can and cannot measure before any transistor is sized.
Built agent-native. Every specification, decision record, testbench, and line of documentation here is produced by AI agents working from a ratified spec and an append-only evidence trail — not human-authored work that agents merely assisted with. Verification is the product: every claim traces to a recorded result under PVT corners. Where the agents hit friction with the open-source tooling — most often klayout-tools — that friction is filed as a public issue against the tool itself, so the fix benefits everyone using this PDK, not just this repo.
A low-noise amplifier is the other half of the RF flank (with sg13g2-vco): small-signal, noise-figure-driven, and impossible to fake with a topology port from the CMOS canaries. SG13G2's HBTs are the reason to build it here — sub-dB noise figures at low GHz are textbook SiGe territory, and no open-PDK open-tool LNA with a committed, reproducible NF evidence chain is publicly available. That gap is the repo's reason to exist.
The honest centerpiece is noise figure and S-parameters with open tools.
ngspice's .sp S-parameter analysis and noise analysis carry assumptions
(port definitions, source impedance) that must be stated with every number.
Inductive-degeneration matching leans on the same PDK passive models the VCO
canary characterizes — coordinate through klayout-tools issues, not by
copying numbers across repos without their derivations.
The detailed table — band, gain, noise figure, input/output match
(S11/S22), stability, IIP3, and supply/power, each row source-cited and the
50 Ω port convention stated explicitly — lives in
spec/target-spec.md. This section stays
deliberately thin: rows are ratified per
spec/decision-records/0002-target-spec-first-ratification.md
(issue #19; the IIP3 numeric target and the former stretch columns are
explicitly unratified), and rows get numbers only from
committed benches. See spec/README.md for the full
spec-directory index, including the porting plan and decision-record
process.
Apache-2.0.