niconiconi<p><a href="https://mk.absturztau.be/tags/openEMS" rel="nofollow noopener" target="_blank">#openEMS</a> simulation of 0402 AC-coupling capacitor parasitics. On a 4-layer board, with 0.2 mm from signal layer to plane. Interestingly, the parasitic capacitance of the mounting pads and the parasitic inductance of the capacitor forms an accidental matching network and compensated each other in this stackup, making it usable even at 20 GHz! You can see similar results without a full-wave solver in lumped circuit simulators, by modeling the mounting pads as two short microstrips, and inserting an 0.1 - 0.2 nH ESL in series. But I won't trust this result unless I can borrow a 20 GHz VNA somewhere, since it's extremely sensitive to parasitic modeling. Also, the magical effect disappears for a better 0.1 mm dielectric, often preferred for lower loop area and lower crosstalks. <a href="https://mk.absturztau.be/tags/electronics" rel="nofollow noopener" target="_blank">#electronics</a></p>