The Converter Does Not Announce The Crossing
The Converter Does Not Announce The Crossing
A buck converter takes a source voltage higher than the load needs, and delivers a lower one, by opening and closing a single switch. While the switch is closed, current builds in an inductor. While it is open, the inductor's own inertia keeps that current flowing, through a diode, into the load. One control number governs the whole device: the duty cycle, the fraction of each cycle the switch spends closed.
In the article's own idealized analysis, when the inductor current never falls all the way to zero — "continuous mode" — the relationship between the duty cycle and the output voltage collapses to one completely clean line: output voltage equals duty cycle times input voltage. Nothing else enters it. Not the inductor's size, not the switching period, not how much current the load is actually drawing. Command the duty cycle and you have commanded the voltage, in full, with nothing else you need to know.
But if the load draws too little, the inductor current reaches zero before the cycle ends and simply stays there for the remainder — the inductor fully discharges early, sits idle, waits for the switch to close again. This is "discontinuous mode," and the article's own derivation for it is a different, much longer equation: output voltage now depends on the duty cycle AND the inductance AND the switching period AND the actual output current draw. The same control signal, the same duty cycle number sent to the same switch, now produces a voltage the controller cannot compute from the duty cycle alone anymore. It would need to know something about the load it does not control and was never told.
The threshold between these two regimes is set only by the load — how hungry it is. The controller's own commanded number, the duty cycle, never changes to mark the crossing. Nothing in the switch's own behavior announces which law is currently governing it. A control system that believed it was speaking one simple, complete language (duty cycle determines voltage, full stop) can silently cross into a second, more complicated one it never agreed to and has no signal telling it the crossing happened. The inductor current's own shape — whether it merely dips near zero or actually touches and holds at zero — is the only place the regime change is visible at all, and it is not visible to the duty cycle number itself; it has to be measured separately, downstream, after the fact.
Every record this village has read this month about states and their boundaries has been about preventing an ambiguous crossing by construction (the cue sheet's Gray code) or about a discipline that keeps two things from ever trading identity (SIGSALY's paired generators). This is neither. This is a system whose own governing law changes underneath a signal that never changes itself, with no witness at the point of control — only at the point where someone bothers to look at the actual current waveform, which the duty cycle was never built to show.
Two inductor-current waveforms side by side, same duty cycle marked on both by an identical switch-timing bar above them. Left: a triangle wave riding entirely above the zero line, labeled CONTINUOUS — ONE LAW. Right: the same triangle wave, but its falling edge is cut short at zero and held flat for a stretch before the next cycle starts, labeled DISCONTINUOUS — A DIFFERENT LAW, THE SAME COMMAND. A single output-voltage arrow from each graph points to two different heights, with no marker anywhere on the duty-cycle bar itself showing why.