【Editorial Discussion】

A pulse or TDD burst that looks clean in the middle of the slot can still have a leading-edge overshoot. It is easy to treat that spike as a DUT problem, or as “just the way the generator looks on a wideband detector.” Before you retune the DUT or change the test waveform, it is worth writing down whether the overshoot is already present at the generator, is being shaped by the cable or the path, or only appears after the DUT.

1) Look at the same burst at a plane in front of the DUT. A coupler, a power splitter, or a short, known cable into a fast detector or an analyser in zero span will tell you whether the leading edge is already high before the DUT sees it. If the overshoot is already there, the first suspect is the source — often the automatic level control (ALC) fighting the pulse, a levelling loop that is too slow or too fast for that pulse width, or a burst mode that was left in a CW-ALC setting. Record the generator’s ALC or open-loop / burst setting, the pulse width, and the detector or video bandwidth you used to see the edge.

2) Cables, pads, and the levelling sensor path can invent or hide an edge. A long cable, a mismatch, or a levelling coupler that is not at the same plane as the DUT will make the generator correct a different waveform from the one you measure. If the overshoot changes when you only change cable length or move the levelling point, it is a path story, not yet a DUT story. Keep the same detector settings when you compare planes, or you will only be comparing two instrument responses.

3) Confirm with a setting that does not rely on ALC at the edge, or with a known-good source mode for pulsed work, before you blame the DUT. If the edge cleans up when ALC is held, frozen, or replaced by a manual / open-loop level, and the mid-slot power is still the intended value, the generator loop was the term. If the edge is clean at the DUT input and appears only at the output, then the DUT — a PA ALC, a gate transient, or a bias transient — is the more honest suspect.

4) Do not read a slow detector or a gated average as proof that the edge is fine. A thermal sensor or a narrow video bandwidth will hide a short overshoot that a peak detector or a wideband I/Q capture will show. Write down how you looked at the edge, not only the mid-slot power.

Open question: the last time a burst showed a leading-edge overshoot, what actually closed the gap — the generator ALC or burst mode, the cable and levelling plane, or a transient in the DUT? Please share the kind of waveform and how you proved it; no customer names or unverified numbers needed.