mmWave packaging - wirebond inductance at 60GHz

Wirebonds at 60GHz look like open circuits. Flip-chip is the answer but assembly cost is 3x. Anyone making wirebond work at V-band with compensation structures?

At 60 GHz a 0.5–1 nH bond looks like tens of ohms of XL, so "make wirebond work" really means absorbing that reactance into a designed discontinuity, not hoping the bond is short enough.

Approaches that show up in published V-band work:

1. Multiple short bonds in parallel with a CPW/GCPW launch and a series C or radial stub to cancel L — yields are better than a single long bond, but bond-length control becomes a process KPI.
2. Ribbon instead of round wire for lower L per mil and more predictable current sheet.
3. Embedded compensation in the package leadframe / interposer (stepped Z0, open stub) so the bond is part of a matched section rather than a lumped L.
4. If assembly cost of flip-chip is the blocker, some teams use copper-pillar / stud-bump with a reflowable underfill as a middle ground — still not wirebond, but cheaper than full flip-chip on every SKU.

If you can share die pad pitch, package type (QFN / embedded wafer / metal lid), and whether you own the laminate stackup, people can say which compensation geometry is realistic for your assembly house.

— rfforum 编辑部 / Editorial Team