The cases I have seen cluster in three places, and the fix depends on which one you are in. First, adjacent-channel and out-of-band leakage from a nearby 5G base station into a satellite earth-station receive band, usually where the satellite downlink sits just above the terrestrial allocation. The symptom is a raised noise floor that tracks base-station traffic load, so a single swept trace can miss it; compare a max-hold trace over a busy hour with one taken when the cell is lightly loaded. Second, blocking and intermodulation in the satellite low-noise front end when a strong terrestrial signal is present even outside the wanted band. The receiver passes a single-tone bench test, then degrades in the field because two or three strong carriers mix in the first stage. A two-tone test at realistic field levels, plus a band-pass filter ahead of the low-noise amplifier, separates a linearity problem from a filtering problem. Third, uplink interference in the other direction, where a satellite terminal transmits into a band a neighbouring service is receiving in. Here antenna pointing, sidelobe level and transmit duty cycle matter more than the spectrum mask, because a brief burst in a sidelobe can break a link. In every case I log the same evidence before blaming either side: the exact frequency pair, the level at the victim antenna, the traffic or transmit cycle at the time, and whether the degradation disappears when the suspected transmitter is off or reduced by a known amount. Without that time-correlated pair of measurements, the argument stays a standards debate instead of an engineering one.