Skip to main content ホーム クリエイター deanyou virtuoso-cli spectre-netlist-template
spectre-netlist-template Add stimulus and analysis templates to a Spectre netlist that has no excitation.
Use when: (1) Ocean/ADE-generated netlist has no testbench wrapper (no vsource,
no analysis statements), (2) you need to measure a specific figure of merit
(GBW, PM, PSRR, CMRR, noise, load regulation, offset, propagation delay, etc.),
(3) you are setting up standalone Spectre simulation from a bare subcircuit netlist.
Identifies circuit type from port names / device types, then inserts the matching
stimulus + analysis block from the canonical templates below.
インストールへ移動 Skills Marketplace コミュニティが作成したAIスキルを発見・探索
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/deanyou/virtuoso-cli --skill spectre-netlist-templateコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Zipをダウンロード ダウンロード中... このリポジトリの他の Skills Maestro (ADE Assembler) session management and simulation. Use when: running simulations via Maestro, configuring tests/analyses/outputs, updating design variables, reading results.
Amplifier design copilot — topology selection, sizing via gm/Id lookup tables, PVT corner validation, and process-portable design. Use when designing amplifiers (OTA, opamp, comparator), selecting topology from specs, sizing transistors, or characterizing a new process node. Also triggers on keywords like amplifier, OTA, opamp, gain-bandwidth, CMRR, PSRR, slew rate.
Explore Virtuoso cellviews - list libraries, cells, instances, nets, layers, and hierarchy. Use when browsing a design, understanding circuit topology, or inspecting layout/schematic contents.
name spectre-netlist-template description Add stimulus and analysis templates to a Spectre netlist that has no excitation.
Use when: (1) Ocean/ADE-generated netlist has no testbench wrapper (no vsource,
no analysis statements), (2) you need to measure a specific figure of merit
(GBW, PM, PSRR, CMRR, noise, load regulation, offset, propagation delay, etc.),
(3) you are setting up standalone Spectre simulation from a bare subcircuit netlist.
Identifies circuit type from port names / device types, then inserts the matching
stimulus + analysis block from the canonical templates below.
author Claude Code version 1.0.0 date "2026-04-19T00:00:00.000Z" source /opt/cadence/IC231/doc/spectremod/Chap1.html (vsource/isource/port syntax) /opt/cadence/IC231/doc/spectreref/chap3.html (analysis statements) argument-hint ["circuit type","e.g. \"OTA\" or \"LDO\" or \"bandgap\" or \"VCO\""] allowed-tools Bash(python *) Bash(spectre *) Read Write
Spectre Netlist Template — Stimulus & Analysis
All syntax verified against Cadence IC231 / Spectre 20.1 reference documentation.
Step 0 — Read and Classify the Netlist
Before adding any template:
Read the .scs file; locate the top-level subcircuit declaration:
subckt CELL_NAME port1 port2 ...
Identify circuit type from the port list + device inventory:
Circuit type Identifying signals Dominant devices OTA / Opamp VIN+ / VIN- / VOUT / VOUTP / VOUTNMOSFETs, tail current source Fully-differential OTA Two differential outputs (VOUTP VOUTN) MOSFETs + CMFB LDO / Regulator VIN / VOUT / FB / ENpass FET, error amp Comparator VIN+ / VIN- / VOUT / CLK (if clocked)strong-arm / regenerative latch Bandgap reference VREF / VBG / VBGR / IREFBJTs, poly resistors Current mirror / bias IBIAS / IOUT / VBIASstacked MOSFETs Active filter VIN / VOUTOTA + C (no large devices) Ring oscillator / VCO VCTRL / VOUT / OUT + odd-number inverter chaininverters LNA / RF amp RF_IN / RF_OUT / port elementsLC matching, RF MOSFETs
If unsure, inspect instance types: nmos4 pmos4 → analog amplifier; mostly inverters →
oscillator/digital; bjt /
bsim3v3
Syntax Reference (from IC231 documentation)
vsource parameters // DC bias supply
VVDD (VDD 0) vsource dc=<V>
VVSS (VSS 0) vsource dc=<-V> ; for negative rail
// AC small-signal (MUST use mag=, NOT ac=; ac= is SPICE-only → SFE-30)
VIN (VIN 0) vsource dc=<bias_V> mag=1 ; single-ended AC input
VIP (VIP 0) vsource dc=<bias_V> mag=0.5 ; differential AC input (+)
VIN_ (VIN_ 0) vsource dc=<bias_V> mag=-0.5 ; differential AC input (-)
// PSRR supply perturbation
VVDD_psrr (VDD 0) vsource dc=<V> mag=1 ; inject on supply
// Pulse (for tran)
VPULSE (NODE 0) vsource type=pulse val0=<V0> val1=<V1> \
period=<T> rise=<tr> fall=<tf> width=<pw> delay=<td>
// Sine (for tran / distortion)
VSINE (NODE 0) vsource type=sine sinedc=<bias> ampl=<A> freq=<Hz>
// PWL (arbitrary waveform)
VPWL (NODE 0) vsource type=pwl wave=[<t0> <v0> <t1> <v1> ...]
isource parameters // DC current bias / load
ILOAD (VDD VOUT) isource dc=<A> ; load current sink (+ flows from VDD into VOUT)
IREF (VREF 0) isource dc=<A> ; reference current
// AC perturbation
IIN (VIN 0) isource dc=0 mag=1 ; AC current input
// Pulse (load step for LDO)
ISTEP (VOUT 0) isource type=pulse val0=<I0> val1=<I1> \
period=<T> rise=<tr> fall=<tf> width=<pw>
port element (S-parameter / RF) // Port for sp analysis — resistance sets reference impedance
PORT1 (RF_IN 0) port r=50 num=1
PORT2 (RF_OUT 0) port r=50 num=2
iprobe (current probe — zero-voltage series sense) IPROBE0 (VOUT_INT VOUT) iprobe ; zero-voltage drop, senses current at VOUT
Analysis Statement Syntax (from IC231 Spectre Reference Manual) // ─── DC operating point ───
dcop dc oppoint=rawfile save=allpub
// ─── DC sweep (parameter) ───
dcswp dc param=<param_name> start=<val> stop=<val> step=<val> \
oppoint=rawfile save=allpub
// ─── DC sweep (voltage/current source) ───
dcswp dc dev=<vsource_name> start=<V> stop=<V> lin=<N> oppoint=rawfile
// ─── AC frequency sweep ───
ac1 ac start=1 stop=100Meg dec=50
// ─── Transient ───
tran1 tran stop=<time> maxstep=<step> errpreset=moderate
// ─── Noise ───
// outputport: element name (NOT node name) → use 1TΩ parallel resistor
// inputport: driving vsource element name
noise_an noise start=1 stop=100Meg dec=50 \
outputport=<Rprobe_name> inputport=<Vsrc_name>
// ─── Stability (loop gain, PM) ───
// probe: iprobe element OR element:p/n for current measurement point
stb1 stb start=1 stop=100Meg dec=50 probe=<iprobe_or_element>
// ─── Transfer function ───
xf1 xf start=1 stop=100Meg dec=50 probe=<output_node>
// ─── S-parameters ───
sp1 sp start=100Meg stop=10G dec=20 ports=[PORT1 PORT2]
// ─── Periodic Steady State (oscillator / VCO) ───
pss1 pss fund=<Hz> harms=20 errpreset=moderate maxacfreq=10G
// ─── simulatorOptions global ───
simulatorOptions options temp=27 tnom=27 scale=1e-6 \
audioverilog=no gmin=1e-12 rforce=1 noisefloor=1e-11 \
save=allpub dc_pivot_check=yes pivtol=1e-13 vabstol=1e-6 \
iabstol=1e-12 compatible=spice2
Circuit-Type Templates
1. OTA / Single-Ended Opamp Ports expected : VDD VSS VIN+ VIN- VOUT (± VBIAS, VCMFB)
// ── Supplies ──────────────────────────────────────────────
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS ; 0 if single-supply
VCMFB (VCMFB 0) vsource dc=VCM ; if CMFB input exposed
// ── Common-mode bias ──────────────────────────────────────
VICM (VCM_NODE 0) vsource dc=VCM ; adjust to mid-rail
// ── Differential AC input: ±mag/2 for 1V differential ─────
VIP (VIN_P 0) vsource dc=VCM mag=0.5 ; positive half
VIN_ (VIN_N 0) vsource dc=VCM mag=-0.5 ; negative half
// ── Load ──────────────────────────────────────────────────
CL (VOUT 0) capacitor c=CL ; load cap (parameter)
Rload (VOUT 0) resistor r=Rload ; if resistive load
// ── Noise oprobe (parallel, NOT series) ───────────────────
Rprobe (VOUT 0) resistor r=1T ; 1TΩ parallel sense element
// ── Analyses ──────────────────────────────────────────────
parameters VDD=1.8 VSS=0 VCM=0.9 CL=1p Rload=1T
dcop dc oppoint=rawfile save=allpub
ac1 ac start=1 stop=1G dec=50
stb1 stb start=1 stop=1G dec=50 probe=Rprobe
noise1 noise start=1 stop=1G dec=50 outputport=Rprobe inputport=VIP
Measurements from results :
GBW: AC magnitude crosses 0 dB → read frequency
Phase margin: phase at GBW frequency
Loop gain (stb): loopGain signal in PSF; PM from phase of loop gain
Input-referred noise: noise analysis → in (input-referred) signal
2. Fully-Differential OTA Ports expected : VDD VSS VIN+ VIN- VOUTP VOUTN VCMFB
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS
VCMFB (VCMFB 0) vsource dc=VCM ; ideal CMFB for open-loop test
VIP (VIN_P 0) vsource dc=VCM mag=0.5
VIN_ (VIN_N 0) vsource dc=VCM mag=-0.5
// Differential output load
CLp (VOUTP 0) capacitor c=CL
CLn (VOUTN 0) capacitor c=CL
Rprobep (VOUTP 0) resistor r=1T
Rproben (VOUTN 0) resistor r=1T
parameters VDD=1.8 VSS=0 VCM=0.9 CL=1p
dcop dc oppoint=rawfile save=allpub
ac1 ac start=1 stop=1G dec=50
stb1 stb start=1 stop=1G dec=50 probe=Rprobep
noise1 noise start=1 stop=1G dec=50 outputport=Rprobep inputport=VIP
3. LDO / Voltage Regulator Ports expected : VIN VOUT FB GND (± EN, VBIAS)
VVIN (VIN 0) vsource dc=VIN
VCMFB (FB 0) vsource dc=VFB ; if FB is a direct pin (no resistor divider)
// or: external resistor divider R1/R2 from VOUT to FB to GND
// Load: nominal + step
ILOAD (VOUT GND_NODE) isource dc=ILOAD_DC
ISTEP (VOUT GND_NODE) isource type=pulse \
val0=ILOAD_DC val1=ILOAD_MAX delay=5u period=30u rise=100n fall=100n width=10u
// Loop-break iprobe for Bode plot
Ibrk (VOUT_SENSE VOUT) iprobe ; insert in feedback path
// Noise probe
Rprobe (VOUT GND_NODE) resistor r=1T
parameters VIN=3.3 VREF=1.2 VFB=VREF ILOAD_DC=1m ILOAD_MAX=100m
dcop dc oppoint=rawfile save=allpub
// Line regulation
linereg dc dev=VVIN start=2 stop=5 lin=31
// Load regulation
loadreg dc dev=ILOAD start=0 stop=ILOAD_MAX lin=51
stb1 stb start=1 stop=100Meg dec=50 probe=Ibrk
tran1 tran stop=100u maxstep=10n errpreset=moderate ; load step
noise1 noise start=1 stop=10Meg dec=50 outputport=Rprobe inputport=VVIN
4. Comparator (static / strong-arm) Ports expected : VDD VSS VIN+ VIN- VOUT CLK (clocked) or VIN+ VIN- VOUT (static)
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS
// Static comparator: ramp one input, hold the other at VCM
VPOS (VIN_P 0) vsource type=pwl \
wave=[0 VCM 1n VCM 101n VSTOP] ; slow ramp to find threshold
VNEG (VIN_N 0) vsource dc=VCM
// Clocked comparator: differential pulse, external CLK
VCLK (CLK 0) vsource type=pulse val0=0 val1=VDD \
period=TCLK rise=100p fall=100p width=TCLK_H delay=0
VDIFF (VIN_P VIN_N) vsource dc=0 type=pulse \
val0=-VDIFF_NEG val1=VDIFF_POS rise=1p fall=1p width=TCLK delay=TCLK_H
parameters VDD=1.8 VSS=0 VCM=0.9 VSTOP=VDD TCLK=2n TCLK_H=1n \
VDIFF_NEG=10m VDIFF_POS=10m
dcop dc oppoint=rawfile save=allpub
tran1 tran stop=10n maxstep=10p errpreset=moderate
// Hysteresis (static comparator)
hysteresis dc dev=VPOS start=0 stop=VDD lin=101 hysteresis=yes
Propagation delay: tran → time from CLK edge to VOUT 50% crossing
Offset: DC sweep → VOUT switches midpoint vs VCM
Hysteresis: DC hysteresis sweep → forward/reverse threshold difference
5. Bandgap Reference Ports expected : VDD GND VREF (± IREF, VSTART, VTRIM)
VVDD (VDD 0) vsource dc=VDD
// Supply ramp for startup check
VRAMP (VDD_RAMP 0) vsource type=pwl \
wave=[0 0 100n 0 1u VDD_MAX] ; slow ramp on supply rail
// Noise probe
Rprobe (VREF 0) resistor r=1T
parameters VDD=3.3 VDD_MAX=3.6 TEMP_START=-40 TEMP_STOP=125
dcop dc oppoint=rawfile save=allpub
// Line regulation: VREF vs VDD
linereg dc dev=VVDD start=1.5 stop=VDD_MAX lin=51
// Temperature coefficient: requires corner sweep (use parametric or Monte Carlo)
// tempco dc param=temp start=TEMP_START stop=TEMP_STOP lin=166
// PSRR
VVDD_psrr (VDD 0) vsource dc=VDD mag=1 ; AC inject on supply
psrr ac start=1 stop=100Meg dec=50
noise1 noise start=1 stop=10Meg dec=50 outputport=Rprobe inputport=VVDD_psrr
tran1 tran stop=10u maxstep=10n errpreset=moderate ; startup check
TC [ppm/°C]: (VREF_max - VREF_min) / (VREF_nom * ΔT) × 1e6
PSRR: AC analysis → 20log(VREF_out / VVDD_in)
Line regulation: DC sweep → dVREF/dVDD [mV/V]
6. Current Mirror / Bias Generator Ports expected : VDD VSS IBIAS_IN IBIAS_OUT (± VBIAS, VCASC)
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS
// Reference current source
IREF (VDD IBIAS_IN) isource dc=IREF_DC
// Output compliance sweep
VCMP (IBIAS_OUT 0) vsource dc=VSTART ; sweeps to check compliance
// Small-signal: output impedance
VTEST (IBIAS_OUT_AC IBIAS_OUT) vsource dc=0 mag=1 ; inject test voltage
parameters VDD=1.8 VSS=0 IREF_DC=10u VSTART=0.1
dcop dc oppoint=rawfile save=allpub
// Mirror ratio check
compliance dc dev=VCMP start=0.1 stop=VDD lin=51
// Output impedance via AC
ac_rout ac start=1 stop=100Meg dec=50 ; measure ITEST/VTEST at IBIAS_OUT
7. Active Filter (OTA-C / Gm-C / Sallen-Key) Ports expected : VDD VSS VIN VOUT
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS
// AC input (singled-ended)
VIN_ac (VIN 0) vsource dc=VCM mag=1
// Step input for tran
VSTEP (VIN 0) vsource type=pulse val0=0 val1=VSTEP_AMP \
delay=100n rise=1p fall=1p width=10u period=20u
// Output sense
Rprobe (VOUT 0) resistor r=1T
parameters VDD=1.8 VSS=0 VCM=0.9 VSTEP_AMP=100m
dcop dc oppoint=rawfile save=allpub
ac1 ac start=1 stop=1G dec=100 ; frequency response (fine resolution)
tran1 tran stop=50u maxstep=1n errpreset=moderate
noise1 noise start=1 stop=1G dec=50 outputport=Rprobe inputport=VIN_ac
8. Ring Oscillator / VCO Ports expected : VDD VSS OUT (± VCTRL for VCO)
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS
VCTRL (VCTRL 0) vsource dc=VCTRL_NOM ; VCO control voltage
parameters VDD=1.8 VSS=0 VCTRL_NOM=0.9 Tosc=2n
// Tran: measure oscillation period
tran1 tran stop=100n maxstep=1p errpreset=moderate skipdc=no
// PSS: periodic steady state → spectral purity
pss1 pss fund=1/Tosc harms=20 errpreset=moderate maxacfreq=100G
// VCO gain (KVCO): DC sweep of VCTRL
vco_kvco dc dev=VCTRL start=0.2 stop=VDD-0.2 lin=51
Measurements from tran : zero-crossings of OUT → period → frequency
PSS : phase noise, KVCO from pss/pac
9. LNA / RF Amplifier (S-parameters) Ports expected : VDD VSS RF_IN RF_OUT (± VBIAS, IBIAS)
VVDD (VDD 0) vsource dc=VDD
VVSS (VSS 0) vsource dc=VSS
VBIAS (VBIAS 0) vsource dc=VBIAS_NOM
// S-parameter ports (50Ω reference)
PORT1 (RF_IN 0) port r=50 num=1
PORT2 (RF_OUT 0) port r=50 num=2
parameters VDD=1.2 VSS=0 VBIAS_NOM=0.6
dcop dc oppoint=rawfile save=allpub
// S-parameter sweep
sp1 sp start=100Meg stop=10G dec=20 ports=[PORT1 PORT2]
// Noise figure (input-referred noise)
noise1 noise start=100Meg stop=10G dec=20 \
outputport=PORT2 inputport=PORT1
// Large-signal tran (1-dB compression point — vary ampl)
tran1 tran stop=50n maxstep=1p errpreset=moderate
S11 (return loss), S21 (gain), S12 (isolation), S22 (output match) from sp analysis
Noise figure from noise analysis in signal
IIP3: two-tone tran → FFT
Workflow
Read netlist → identify subcircuit name + port listMatch circuit type from table in Step 0Choose template from the matching section aboveInstantiate the subcircuit at top level:
simulator lang=spectre
global 0
include "input.scs" ; ADE-generated subcircuit
parameters VDD=1.8 ...
// ── DUT instantiation ──
XDUT (VDD VSS VIN_P VIN_N VOUT) CELL_NAME
// ── Stimulus from template ──
...
// ── Analyses from template ──
...
Adjust node names to match actual DUT port orderVerify mag=1 on all AC excitation sources (not ac=1)Run : spectre input.scs -raw psf -format psfascii
Automated Script: inject_stimulus.py For batch or scripted use, the skill ships a Python helper that auto-detects circuit
type and generates a complete testbench file — no Virtuoso running required.
python ${CLAUDE_SKILL_DIR} /scripts/inject_stimulus.py bandgap.scs
python ${CLAUDE_SKILL_DIR} /scripts/inject_stimulus.py amp.scs --type ota
python ${CLAUDE_SKILL_DIR} /scripts/inject_stimulus.py amp.scs -o amp_testbench.scs
python ${CLAUDE_SKILL_DIR} /scripts/inject_stimulus.py chip.scs --cell Bandgap
python ${CLAUDE_SKILL_DIR} /scripts/inject_stimulus.py chip.scs --list
spectre bandgap_tb.scs -raw psf -format psfascii
The script requires no third-party libraries (stdlib only).
After generation, review and adjust parameter values (VDD, CL, temperatures, etc.)
to match the target PDK and operating conditions.
Common mistakes : see /spectre-netlist-gotchas for SFE-30 (ac=1 → mag=1), SFE-1997, oprobe wiring, global 0, port order, and noise parameter errors.
