mmWave beam steering - analog vs hybrid vs digital tradeoffs in production

Hybrid beamforming is the textbook compromise, but production cost and calibration effort differ wildly. For a 28GHz fixed-wireless product, which architecture did you ship and what would you change?

Analog vs hybrid vs digital beamforming in production is usually decided by IC cost, cal complexity, and power, not by peak array gain on a slide.

What tends to ship at mmWave:
1. Pure analog (phase shifters / vector modulators per element or subarray): lowest DC and BOM when beams are few; calibration is RF-path heavy and multi-beam is awkward.
2. Hybrid (RF beamformers + digital on subarrays): the common production compromise for FR2 / fixed wireless; you buy enough digital DoF for MU-MIMO or nulling without an ADC/DAC per element.
3. Full digital: best flexibility and cal in the baseband, but front-end power, data-converter count, and thermal usually win the argument against it below ~64–128 elements unless you already own a high-volume SoC.
4. Whatever you pick, lock a factory cal plan early (coupled vs OTA sample) — hybrid arrays often fail schedule on cal time, not on silicon.

If you share element count, target beams, and whether the beamformer is discrete or integrated, people can be more specific on which branch actually hits cost.

— rfforum 编辑部 / Editorial Team