經過對嘉金源產線連續三週的SPI數據跟踪，我們發現0402和0603封裝電阻電容的錫膏體積CV值始終保持在9.2%以上，橋接缺陷佔單板缺陷的37%，而回流後墓碑率穩定在0.18‰，明顯超過工藝窗口上限。這表明當前的焊膏供應已經超過了焊盤的最大容量，幹預必須從源頭開始：鋼網設計。

多喜技術團隊檢索了過去六個月的鋼網使用記錄以及每批次相應的回流後AOI數據，確認了該問題是在改用新孔徑比為85%的30μm激光切割鋼網後出現的。這個問題在採用 FR-4 基板和沈金表面處理的 PCB 上尤其突出，因為這些板的焊料吸收能力較弱且熱膨脹係數較高，導致焊膏體積的耐受性較低。若孔徑比沒有相應調整，脫模時錫膏極易出現拖尾與堆積現象。

Rather than simply adjusting squeegee parameters to mask the problem, we went directly back to the stencil design stage. We uniformly narrowed the aperture dimensions corresponding to 0402 pads from the original 200×120μm to 190×115μm, reducing the aperture area ratio from 85% to 78%. Meanwhile, we fine-tuned the 0603 area from 260×140μm to 250×135μm to ensure that the Solder Paste release volume matches the pad wetting capacity. This modification does not require replacing the entire stencil; it can be completed through local re-engraving, with the production line changeover cycle controlled within 2 hours.

After reducing the solder paste supply, the SPI-measured average volume dropped from 6.8ng to 6.0ng, and the standard deviation was compressed from 0.72ng to 0.51ng. More importantly, bridging defects decreased to 14% of single-board defects, and the tombstoning rate fell back to 0.07‰, without causing new issues such as cold joints or insufficient solder joint wetting. This indicates that Duoxi's adjustment this time precisely hit the optimal threshold range for solder paste volume.

我們同時加強了對錫膏預熱和攪拌的控制，要求所有嘉金源錫膏在印刷前恆溫放置4小時以上，並使用離心攪拌機定速定時間進行60秒的預攪拌。這可以防止焊膏黏度的波動放大孔徑比調整帶來的敏感性；畢竟，無論模板設計多麼精確，它都無法承受不受控制的材料條件的影響。

現場驗證發現，當焊膏批次間黏度差超過150Pa·s時，即使最佳化孔徑比，仍會出現局部脫模缺陷。因此，我們嚴格要求每批錫膏入庫前必須附有黏度測試報告；如果資料偏差超過限制，則立即啟動隔離流程，確保異常材料永遠不會進入列印工作站。