"""🔗 量子傳輸線 — 多導體傳輸線高頻自動化分析儀(互動區塊)。 改寫自 ~/軟體/代理人/pages/03擴充自動化.py(舊架構獨立頁面): - 移除 set_page_config 與模組級副作用(改由 app.py 統一配置) - 參數輸入由 sidebar 改為主畫面左欄(sidebar 為全站共用) - 蒙地卡羅改為純函式 + @st.cache_data 快取 - 新增「📐 內容模板」頁籤(量子傳輸線視頻腳本模板) 物理模型(與原始版本一致): - 2 導體 RLGC 矩陣導出(邊緣場微擾近似) - 電報方程矩陣指數解(scipy.linalg.expm)→ NEXT/FEXT 串擾 - 高斯製造公差蒙地卡羅 → 阻抗良率(50Ω ±10%) """ import os import matplotlib.pyplot as plt import numpy as np import streamlit as st from plotly.graph_objects import Box, Figure, Histogram from scipy.linalg import expm from src.views.doc_browser import REPO_ROOT DK_NOMINAL = 4.3 # FR4 介電常數名義值 DF_NOMINAL = 0.02 # 損耗角正切 # ============================================================================== # 核心電磁計算引擎(多導體矩陣) # ============================================================================== def derive_rlgc_matrices(w, h, t, s, dk, df, freq): """導出 2 導體系統的 2x2 RLGC 矩陣(2D 場求解近似)。""" w_m, h_m, t_m, s_m = w * 1e-6, h * 1e-6, t * 1e-6, s * 1e-6 omega = 2 * np.pi * freq # 互容與自容矩陣(考慮邊界場微擾) c_self = 8.854e-12 * dk * (w_m / h_m + 1.2) c_mutual = 8.854e-12 * dk * (0.5 / (s_m / h_m + 0.1)) C = np.array([[c_self + c_mutual, -c_mutual], [-c_mutual, c_self + c_mutual]]) # 互感與自感矩陣(光速不變性與空間微擾) L = 1.256e-6 * 8.854e-12 * 1.0 / C L[0, 1] = L[1, 0] = L[0, 0] * (0.2 / (s_m / h_m + 0.5)) # 高頻趨膚效應電阻 R 與介電損耗導納 G r_skin = 1.0 / (5.8e7 * t_m * w_m) * np.sqrt(freq / 1e9) R = np.array([[r_skin, 0.0], [0.0, r_skin]]) G = omega * C * df return R, L, G, C def sqrt_matrix(mat): """矩陣開平方根(特徵值分解)。""" evals, evecs = np.linalg.eig(mat) return np.dot(evecs, np.dot(np.diag(np.sqrt(evals)), np.linalg.inv(evecs))) def analyze_crosstalk(R, L, G, C, freq, length): """電報方程矩陣指數解 → NEXT/FEXT 串擾(dB)。""" omega = 2 * np.pi * freq Z_mat = R + 1j * omega * L Y_mat = G + 1j * omega * C prop_mat = expm(-sqrt_matrix(np.dot(Y_mat, Z_mat)) * length) next_xtalk = 20 * np.log10(max(1e-6, abs(prop_mat[0, 1] * 0.3))) # 近端串擾 fext_xtalk = 20 * np.log10(max(1e-6, abs(prop_mat[1, 0]))) # 遠端串擾 return next_xtalk, fext_xtalk @st.cache_data(show_spinner=False) def run_monte_carlo(w, h, t, s, w_sigma, h_sigma, dk_sigma, freq, length, samples): """高斯製造公差蒙地卡羅掃描 → 阻抗與串擾分佈。""" z0_list, next_list, fext_list = [], [], [] w_samples = np.random.normal(w, w_sigma, samples) h_samples = np.random.normal(h, h_sigma, samples) dk_samples = np.random.normal(DK_NOMINAL, dk_sigma, samples) for i in range(samples): R, L, G, C = derive_rlgc_matrices( w_samples[i], h_samples[i], t, s, dk_samples[i], DF_NOMINAL, freq ) z0 = np.sqrt(L[0, 0] / C[0, 0]) nxt, fxt = analyze_crosstalk(R, L, G, C, freq, length) z0_list.append(z0) next_list.append(nxt) fext_list.append(fxt) return np.array(z0_list), np.array(next_list), np.array(fext_list) # ============================================================================== # 檢視 # ============================================================================== def _render_calculator(): left, right = st.columns([1, 2]) with left: st.markdown("#### 📁 傳輸線 DSL 與製造公差配置") st.radio( "結構類型", ["Microstrip (微帶線)", "Stripline (帶狀線)"], horizontal=True, key="tline_type", ) st.slider("線寬 W (μm)", 50, 300, 150, key="w_nominal") st.slider("介電厚度 H (μm)", 50, 500, 100, key="h_nominal") st.slider("導體厚度 T (μm)", 15, 70, 35, key="t_nominal") st.slider("線間距 S (μm)", 50, 400, 150, key="s_nominal") st.divider() st.markdown("#### 🎲 製造公差波動(Monte Carlo σ)") st.slider("σ_W 線寬 (μm)", 0.0, 10.0, 3.0, 0.5, key="w_sigma") st.slider("σ_H 厚度 (μm)", 0.0, 10.0, 2.5, 0.5, key="h_sigma") st.slider("σ_Dk 介電常數", 0.0, 0.2, 0.05, 0.01, key="dk_sigma") st.divider() st.markdown("#### ⚡ 模擬條件") st.number_input("目標頻率 (GHz)", 1.0, 30.0, 10.0, 1.0, key="freq_ghz") st.number_input("線長 (m)", 0.01, 1.0, 0.1, 0.05, key="line_len") st.radio( "蒙地卡羅抽樣次數", [100, 500, 1000], index=1, horizontal=True, key="mc_samples", ) # 讀取 widget 狀態 w_n, h_n, t_n, s_n = ( st.session_state["w_nominal"], st.session_state["h_nominal"], st.session_state["t_nominal"], st.session_state["s_nominal"], ) w_s, h_s, dk_s = ( st.session_state["w_sigma"], st.session_state["h_sigma"], st.session_state["dk_sigma"], ) freq_hz = st.session_state["freq_ghz"] * 1e9 line_len = st.session_state["line_len"] mc_samples = st.session_state["mc_samples"] # 名義值計算 R_n, L_n, G_n, C_n = derive_rlgc_matrices( w_n, h_n, t_n, s_n, DK_NOMINAL, DF_NOMINAL, freq_hz ) z0_nominal = np.sqrt(L_n[0, 0] / C_n[0, 0]) nominal_next, nominal_fext = analyze_crosstalk( R_n, L_n, G_n, C_n, freq_hz, line_len ) # 蒙地卡羅掃描 z0_s, next_s, fext_s = run_monte_carlo( w_n, h_n, t_n, s_n, w_s, h_s, dk_s, freq_hz, line_len, mc_samples ) yield_rate = np.sum((z0_s >= 45) & (z0_s <= 55)) / mc_samples * 100 with right: m1, m2, m3, m4 = st.columns(4) m1.metric("特徵阻抗 Z₀", f"{z0_nominal:.2f} Ω") m2.metric("近端串擾 NEXT", f"{nominal_next:.2f} dB") m3.metric("遠端串擾 FEXT", f"{nominal_fext:.2f} dB") m4.metric("阻抗良率 (50Ω±10%)", f"{yield_rate:.1f} %") c_a, c_b = st.columns(2) with c_a: st.markdown("##### 🖼️ 雙導體截面幾何") fig, ax = plt.subplots(figsize=(6, 3)) ax.set_facecolor("#f8f9fa") ax.add_patch( plt.Rectangle((-500, 0), 1000, h_n, color="#dcd6be", label="Substrate") ) ax.add_patch( plt.Rectangle((-500, -t_n), 1000, t_n, color="#b87333", label="GND") ) ax.add_patch( plt.Rectangle( (-w_n - s_n / 2, h_n), w_n, t_n, color="#ff4b4b", label="Trace 1 (Aggressor)", ) ) ax.add_patch( plt.Rectangle( (s_n / 2, h_n), w_n, t_n, color="#1f77b4", label="Trace 2 (Victim)" ) ) ax.set_xlim(-400, 400) ax.set_ylim(-50, h_n + 80) ax.set_xlabel("Width (μm)") ax.set_ylabel("Height (μm)") ax.legend(loc="upper right", fontsize="small") st.pyplot(fig) plt.close(fig) st.markdown("##### 📋 導出的 RLGC 矩陣(@ 目標頻率)") st.markdown("**單位長度電感矩陣 L (nH/m):**") st.dataframe(L_n * 1e9) st.markdown("**單位長度電容矩陣 C (pF/m):**") st.dataframe(C_n * 1e12) with c_b: st.markdown("##### 🎲 蒙地卡羅阻抗良率預測") fig_hist = Figure() fig_hist.add_trace( Histogram(x=z0_s, name="阻抗分佈", marker_color="#ff4b4b", opacity=0.75) ) fig_hist.add_vline( x=45, line_dash="dash", line_color="red", annotation_text="下限 45Ω" ) fig_hist.add_vline( x=55, line_dash="dash", line_color="red", annotation_text="上限 55Ω" ) fig_hist.update_layout( title="隨機製造抽樣阻抗分佈", xaxis_title="Impedance (Ω)", yaxis_title="Count", template="plotly_white", height=320, ) st.plotly_chart(fig_hist, use_container_width=True) st.markdown("##### ⚡ 公差引起的串擾波動") fig_xtalk = Figure() fig_xtalk.add_trace( Box(y=next_s, name="NEXT (近端)", marker_color="#ab63fa") ) fig_xtalk.add_trace( Box(y=fext_s, name="FEXT (遠端)", marker_color="#00cc96") ) fig_xtalk.update_layout( title="串擾波動範圍 (dB)", yaxis_title="Crosstalk (dB)", template="plotly_white", height=320, ) st.plotly_chart(fig_xtalk, use_container_width=True) with st.expander("💡 多導體分析備忘錄"): st.markdown( """ 1. **互感與互容矩陣**:自導體的邊緣場(Fringing Field)是引起多導體串擾的主要物理成因,當線距 `S` 縮小時,`L[0,1]` 與 `C[0,1]` 將顯著增大。 2. **NEXT (近端串擾)**:主要受電磁耦合總量影響,在長度達到飽和長度後不再隨長度增加。 3. **FEXT (遠端串擾)**:受微帶線不均勻介電質環境(空氣與 FR4)影響,高頻下損耗會隨傳輸線長度呈現線性惡化。 """ ) def _render_template(): tpl_path = os.path.join(REPO_ROOT, "data", "docs", "quantum_tline_template.md") if not os.path.isfile(tpl_path): st.warning("找不到 data/docs/quantum_tline_template.md") return with open(tpl_path, encoding="utf-8") as fh: st.markdown(fh.read()) st.divider() st.caption( "模板來源:~/軟體/代理人/content_templates/quantum.md — " "量子傳輸線主題的視頻腳本結構、關鍵字與目標平台" ) def render_tline(): st.title("🔗 量子傳輸線 — 多導體高頻分析儀") st.caption( "幾何 DSL 輸入 → RLGC 矩陣導出 → 電磁計算 → 蒙地卡羅良率掃描 → 近/遠端串擾分析" ) tab_calc, tab_tpl = st.tabs(["⚡ 互動分析儀", "📐 內容模板"]) with tab_calc: _render_calculator() with tab_tpl: _render_template()