Lead with the problem that mass production creates. A phased array steers its beam by precisely controlling the phase across many elements, but manufacturing tolerances, temperature, and aging all push those phases off their ideal values. An array that is not regularly calibrated drifts, and a drifting beam points where you did not intend. Calibration is what keeps a cheap array honest.
Amazon's grant US11804914B1, naming inventors including Alireza Mahanfar, classified in H04B 17/12 (calibration of transmission systems) with H01Q 3/26 (beam control) and H04B 17/21, claims calibrating the array using a probe antenna. The probe is the mechanism: a reference element that measures the array's actual behavior so the system can correct it.
This sits in a cluster of Amazon Project Kuiper antenna filings, including in-orbit calibration (US11322838B1, 2022), array-wall slot calibration antennas (US11876293B1, 2024), and peripheral calibration placement (US12040555B1, 2024). Read together, they show a company systematically attacking one problem: keeping a manufactured-at-scale phased array accurate across its life.
For a constellation that has filed comparatively little financial disclosure about Kuiper's economics, the patent trail is the clearest public read on where the engineering effort is going, and it is going heavily into the user-terminal and antenna-accuracy problem, the same cost-and-quality frontier SpaceX attacked from the steering-simplification angle.
The caution is that calibration patents describe a method, not a yield. They tell you Amazon takes array accuracy seriously enough to patent multiple approaches to it. Whether the calibrated arrays hold performance across a global, mass-produced fleet is a manufacturing-and-operations outcome the claims set up but do not settle.