At 3.5 GHz with 6–8 dB average backoff, both ET and Doherty can win — the deciding factors are usually modulator bandwidth / tracking error vs combiner loss and load-pull bandwidth, not the textbook peak PAE.
What measurements tend to show:
1. Doherty: simpler TX chain if you already have a good 90° / uneven splitter and can live with the combiner's bandwidth and AM–PM. Average efficiency at 7 dB OBO is often excellent when the peaking amp is biased cold and the main is load-modulated cleanly. Pain: PCB combiner iteration and memory effects under wideband LTE/NR.
2. Envelope tracking: higher ceiling on average efficiency when the supply modulator is fast enough for your channel BW and the PA is designed for supply modulation (not a CW-biased part with ET bolted on). Pain: modulator efficiency, supply noise into ACLR / EVM, and calibration across temperature.
3. Hybrid (Doherty + slow average power tracking) is what a lot of macro designs actually ship when full ET BW is too expensive.
For a first spin, if your instantaneous BW is modest and you have load-pull / EM capacity, Doherty is usually the lower-risk path to a publishable average-efficiency number. If you already own a supply-modulator IP and the PA process is ET-friendly, ET can beat Doherty at the same OBO.
Share channel BW, PAPR target, and whether the PA is discrete LDMOS/GaN or an integrated MMIC — that usually picks the winner faster than a generic "ET vs Doherty" debate.
— rfforum 编辑部 / Editorial Team
