Tr2LC 网络测量工作台LC tank measurement workbench
谐振、损耗,
起振裕量。Resonance, loss,
startup margin.
把 N-port、电容阵列和 gm 核心放上同一画布。一次原生运行,逐码查看 f0、并联损耗电阻、端口 gm/gds、负电导和起振裕量,再追到各模块的损耗占比。任何电容阵列也能逐码查看 C、Q、C 步进与 INL/DNL。Place an N-port, capacitor array and gm core on one canvas. One native run reports f0, parallel loss resistance, terminal gm/gds, negative conductance and startup margin for each code, with loss shares by module. Any capacitor array also gets C, Q, C step and INL/DNL per code.
演示片Film
从画布到起振指标。From canvas to startup metrics.
电容阵列 C/Q 与 8 位大码字线性度,Resonance / tank 的起振指标与损耗,最后是与独立有源仿真的精度对比和大码字耗时预测。中英文版各有配音,随页面语言切换。Capacitor-array C/Q and 8-bit linearity, Resonance / tank startup metrics and losses, then accuracy against an independent active simulation and a large-code time estimate. Narration follows the page language.
流程Workflow
连接、测量、逐码核对。Connect, measure, inspect each code.
Resonance / tank 与 Capacitance 两种测量模式共用测量画布和工作簿。The Resonance / tank and Capacitance modes share a canvas and workbook.
搭好测量网络Build the measurement network
放置 Touchstone N-port 或电感、寄存器驱动的电容阵列,以及可选的 gm 核心。用命名网络连接端口与偏置,寄存器定义扫描码。Place a Touchstone N-port or inductors, a register-driven cap array and an optional gm core. Assign named nets for terminals and bias, then choose register codes.
确认角色与端口,运行Assign roles and ports, then run
勾选画布包含的模块,核对每个实例的角色。包含 gm 核心时分析有源起振条件,未选 gm 时分析无源谐振与损耗。角色调整可重新分析保存的数据。可选第二端口:两个管脚设为 P2/N2 即可得到 Z21,栅极接在端口 2 的变压器耦合核心按 gm·Z21 计算环路增益。Select the modules on the canvas and check each instance role. Including a gm core enables active startup analysis. A passive tank yields resonance and loss. Role changes re-analyze saved data. An optional second port (pins P2/N2) adds Z21; for a transformer-coupled core with its gates on port 2 the loop gain is gm·Z21.
读指标,追到损耗来源Read metrics and trace the losses
选中控制码,查看 f0、Rp、G_neg、起振裕量、环路增益 gm·Z 和 Q。图表页可自选纵轴量和横轴:逐码曲线、多码 |Z11| 与环路增益随频率变化、模块损耗占比,点选曲线与表格联动,深浅主题一键切换。需要更细的谐振定位时,用 Refine sweep at each f0 在每个码的 f0 附近加密扫频点后再运行。Select a code to inspect f0, Rp, G_neg, startup margin, loop gain gm·Z and Q. Plot tabs take any quantity and x axis: values per code, multi-code |Z11| and loop gain versus frequency, and loss share by module; clicking a curve selects its table row, and the theme switches between dark and light. For finer resonance placement, Refine sweep at each f0 adds dense points around every code’s f0 before the next run.
电容阵列:C/Q 与线性度Capacitor arrays: C/Q and linearity
Capacitance 模式适用于任何电容阵列:逐码比较浮置 P–N 端口的 C/Q,可切换差分与单端(P、N 对地)分量;Show 选项再给出 C 步进与 INL/DNL(LSB,首尾端点拟合)。观察频率在已保存频谱内调整时不必重新仿真,设置和频谱存入工作簿。Capacitance mode works for any capacitor array: it compares floating P–N C/Q across codes, with differential and single-ended (P or N to ground) components, and Show adds C step and INL/DNL (LSB, endpoint fit). Observation frequencies inside the saved spectrum need no rerun; setup and spectra stay in the workbook.
指标Metrics
每个数都有测量口径。Each metric has a definition.
所有 f0 处的量由包围谐振的原生样本插值得到。端口 gm/gds 采用核心端口导纳定义。Values at f0 are interpolated between bracketing native samples. Terminal gm/gds use the core terminal admittance definition.
| 指标Metric | 定义Definition |
|---|---|
| f0f0 | 完整电路 Im(Z) 首次过零,包括核心电容,对应小信号起振频率。The first Im(Z) zero of the complete circuit, including the core. The small-signal startup frequency. |
| f0(仅 LC)f0 (LC only) | 电感与阵列电纳之和的零点,去掉核心电容后的估计。The zero of inductor-plus-array susceptance, estimating resonance with core capacitance removed. |
| Rp(LC)Rp(LC) | 1 / (G_inductor + G_array),在 f0 处计算。1 / (G_inductor + G_array), evaluated at f0. |
| Rp(LC+loading)Rp(LC+loading) | 1 / G_loss,计入 LC 损耗、其他负载和核心加载。1 / G_loss, including LC loss, other loads and core loading. |
| gm / gds @ f_refgm / gds @ f_ref | 每侧端口值,分别由核心 P/N 导纳矩阵的交叉项和对角项提取。Per-side terminal values from the cross terms and diagonal terms of the core P/N admittance matrix. |
| G_negG_neg | 按参考频率的端口 gm,在 f0 实际电压下抵消跨导得到。对称管对为 gm/2。The conductance supplied by reference-frequency terminal gm under the actual f0 voltages. It equals gm/2 for a symmetric pair. |
| 起振裕量 / 环路增益 gm·ZStartup margin / loop gain gm·Z | 裕量 = G_neg / G_loss,大于 1 满足线性起振条件。环路增益 = gm(f_ref) × Rp(LC+loading);栅极接在端口 2 的变压器耦合核心用 gm 抵消后储能腔的 Re Z21。Margin = G_neg / G_loss. A value above 1 meets the linear startup condition. Loop gain = gm(f_ref) × Rp(LC+loading); for a transformer-coupled core with its gates on port 2 it uses Re Z21 of the gm-cancelled tank. |
| Q / C_eff / L_effQ / C_eff / L_eff | 由谐振处电纳斜率提取,C_eff = ½·dB/dω,同时给出 LC 与加载后的 Q。Extracted from the susceptance slope at resonance, with C_eff = ½·dB/dω and both LC and loaded Q. |
| G_loss 占比G_loss shares | 逐模块或耦合绕组组列出电导与损耗占比,核心加载包含 gds、输入损耗和 gm 的射频滚降。Conductance and loss share by module or coupled-winding group. Core loading includes gds, input loss and RF roll-off of gm. |
| Z21(可选端口 2)Z21 (optional port 2) | 端口 1 驱动、端口 2 开路时的差分转移阻抗;端口 1 的各项结果都按端口 2 开路计算。Differential transfer impedance with port 1 driven and port 2 open; all port-1 results are taken with port 2 open. |
| C 步进 / INL / DNLC step / INL / DNL | Capacitance 模式逐码计算:C 步进为相邻码的电容差;LSB 取首尾端点拟合,DNL = 步进/LSB − 1,INL 为偏离端点直线的量(LSB)。Per code in Capacitance mode: C step is the change from the previous code; with an endpoint-fit LSB, DNL = step/LSB − 1 and INL is the deviation from the endpoint line, in LSB. |
MaestroMaestro
继承当前窗口的设置。Inherit the invoking window’s setup.
从源 Maestro 窗口启动时,工具捕获该窗口的测试与运行设置。多测试工程需明确选择来源测试。Launch from the source Maestro window to capture its test and runtime setup. Select a source test explicitly in a multi-test setup.
私有测量设置Private measurement setup
保留源测试的非测量选项,并继承运行分配、作业策略和 Save Options。Maestro settings 中可勾选字段,对私有测量设置单独覆盖。需要继承未保存的修改时,先在 Maestro 保存。Retain source-test options outside the measurement and inherit run distribution, job policies and Save Options. Checked fields in Maestro settings override the private setup. Save pending Maestro edits before launch when they must be inherited.
沿用层级与绑定Preserve hierarchy and bindings
从所选测试的 config 层级放置原有 symbol,私有 config 保留实际实现与下层绑定。测量自主管理模型列表、分析、变量、温度与必要的保存选项。桌面启动使用独立工作簿。Place original symbols from the selected test’s config hierarchy. The private config preserves implementations and descendant bindings. The measurement manages its model list, analyses, variables, temperature and required save options. Desktop launch opens a standalone workbook.
界面Gallery
展开查看当前工作台。Inspect the current workbench.
结果页收起设置面板,让指标、曲线和损耗表使用更多空间。The results view folds away setup panels to leave more room for metrics, plots and losses.
安装Install
打开工作簿,开始测量。Open a workbook and measure.
源码以 MIT 许可证发布。工作台使用 Python 标准库与 Tk;装有 Pillow 时曲线抗锯齿,深色主题带荧光效果(可选)。原生测量需要用户已有的 Virtuoso、Maestro、Spectre、库映射、模型与许可证。Released under MIT. The workbench uses the Python standard library and Tk; with Pillow installed the plots are anti-aliased and glow in the dark theme (optional). Native measurements require your own Virtuoso, Maestro, Spectre, library mappings, models and licences.
# 打开 Tank Scan Lab(Resonance / tank 模式)# open Tank Scan Lab (Resonance / tank mode) $ ./tank-scan-lab.sh # 打开已保存的工作簿# open a saved workbook $ ./tank-scan-lab.sh --project workbook.json # 合成示例:理想储能腔;8 位温度计码阵列# synthetic examples: ideal tank; 8-bit thermometer array $ ./tank-scan-lab.sh --project examples/ideal_tank.json $ ./tank-scan-lab.sh --project examples/thermo8_cq.json
C/Q 的观察频率在已保存频谱范围内调整时,可直接复用数据。保存工作簿会同时保存页面、配置和已完成的频谱。C/Q observation frequencies within the saved spectrum can reuse the data. Workbooks retain pages, setup and completed spectra.
测量边界Measurement scope
- Tank scan 在线性化 DC 工作点上计算起振条件。稳态振幅与增益需要 HB 或瞬态分析,自偏置核心应核对 DC 电流与稳态的差异。Tank scan evaluates startup at the linearized DC operating point. Steady-state swing and gain require HB or transient analysis. Check DC versus steady-state current for a self-biased core.
- f0 与损耗占比受扫频间隔影响。插值不会增加原生样本分辨率,用 Refine sweep at each f0 加密后重新运行可缩小谐振括区。Touchstone 频段外的低频或 DC 点可能需要仿真器外推。The sweep bracket affects f0 and loss-share estimates. Interpolation adds no native resolution. A rerun after Refine sweep at each f0 reduces the bracket. Low-frequency or DC points outside a Touchstone band may require simulator extrapolation.
- 截图来自当前公开版应用与合成示例(含 8 位温度计码阵列),使用通用工程名称。影片中的精度对比来自真实仿真;大码字耗时为按实测比例推算的预测值。Screenshots use the current public application with synthetic examples (including the 8-bit thermometer array) and generic project names. The film’s accuracy comparison comes from real simulations; the large-code time is an estimate scaled from measured ratios.
其他工具Other tools