Switch on a capacitor-charging supply and the first few milliseconds carry a surge that dwarfs the steady-state current: a rectifier fed into a discharged capacitor can briefly push ten times its rated average, and that first charge pulse is exactly what the surge rating on the diode datasheet is about. The IFSM number — forward surge current — covers a one-shot event like this, but only under its stated test conditions, and two mistakes send parts to the bin: treating a repetitive or hot-start surge as a one-shot, and ignoring the fact that the inrush event deposits fuse-like energy the junction must absorb. This article puts numbers on the first charge cycle, separates one-shot from repetitive stress, explains the I²t damage analogy, compares diode-only inrush with the NTC limiter trade-off, and shows how to observe the whole thing on a scope.
The First Charge Cycle: Why 10x Average Flows in 100 us
Capacitor-charging inrush is pure physics: a discharged capacitor presents almost no opposition to current, so at the instant of power-on the rectifier sees the full available voltage divided by the source impedance. For a bridge feeding a filter capacitor, the initial current can reach many times the steady-state average — a genuinely common 10x to 20x for the first tens to hundreds of microseconds, before the capacitor voltage rises and the current falls.
The duration matters as much as the peak. The surge is not a step that lasts forever; it decays on the order of the R×C time constant of the source and capacitor. For a small bench supply, the first current peak might be over in well under a millisecond; for a large industrial rectifier into a big capacitor bank, the charging transient can last many milliseconds. The shape of the decay determines whether the energy is a short, sharp pulse (measured by IFSM) or a longer, hotter event (measured by I²t and thermal budget).
The numbers work out to a clear statement: on a 50 Hz or 60 Hz rail, the first charge cycles are not like the steady-state tenth-cycle; they are a single large event that happens once at every power-on. That “first event” framing is the correct lens for the datasheet’s IFSM column, and it is the lens most failure-mode write-ups get right — until the product is power-cycled repeatedly and the “one-shot” assumption dies.
IFSM vs Repetitive Surge: Two Different Stress Regimes
The IFSM rating is quoted for a defined test wave — typically a single half-sine pulse of a stated duration, measured at 25 °C. A design that triggers the pulse once at cold power-on is exactly the regime IFSM covers: the junction absorbs one surge, heats briefly, and the heat spreads into the package between events.
The stress regime changes when the surge becomes repetitive. A product that power-cycles rapidly, or a motor that restarts repeatedly under load, turns a one-shot event into a duty cycle — the junction no longer cools between surges, and the IFSM number stops being the right gate. The repeated events deposit heat faster than the package sheds it, and the failure signature shifts from “surge killed the junction” to “the junction ran hot and degraded.”
The two regimes demand different margins. For a true one-shot at cold start, an IFSM comfortably above the measured surge usually suffices. For a repetitive or hot-start pattern, the design needs the derating that a repeated-event curve implies, or a protection stage that limits the surge before it reaches the diode. The surge current ratings guide develops the wave interpretation; the decision table in the next section turns this into a choice between diode-only and NTC designs.
Diode I2t and the Melting-Link Analogy
The damage a surge does to a diode is not fully captured by peak current — it is captured by the energy integral, I²t. The capacitor-charging surge carries a specific amount of energy, and whether the junction survives depends on whether that energy exceeds what the package can absorb before the heat reaches damage level.
The melting-link analogy makes it vivid: a fuse blows because the current-squared-time integral of the fault melts the link, and a diode junction can die the same way. A sharp surge with high peak and tiny duration may carry less I²t than a longer, lower surge that actually melts the wire at the package connection or degrades the junction. That is why comparing two events by peak current alone is misleading — the energy integral is the real damage proxy.
The datasheet’s I²t related data, where provided, is the number to compare against the measured event. If the surge’s energy integral clears the part’s capability, the margin is real; if the event is borderline, the design should reduce the surge (with an NTC or series impedance) rather than merely picking a bigger IFSM. The rectifier datasheet tour shows where to find these ratings, and the power-loss sheet frames the thermal consequence in watts the junction must shed.
NTC vs Diode-Only Inrush: A Decision Table
When inrush exceeds what the diode-only design can handle safely, the classic low-cost mitigation is an NTC thermistor in series with the supply: cold at power-on, it limits current; warm in steady state, it drops to near zero, adding almost no running loss. The trade-off is the decision point of this article.
| Factor | Diode-only | Diode + NTC |
|---|---|---|
| First-cycle surge | Full surge hits the diode | Limited by cold NTC resistance |
| Steady-state loss | Diode drop only | NTC residual resistance + diode drop |
| Hot-start behavior | Surge repeats at cold-like level | NTC already warm, limits less |
| Cost | One diode | One thermistor + assembly |
| Reliability | Simpler, passive | NTC can fail open (lost limiting) |
| Repetitive cycling | Fast stress on diode | Slower warm-up, better for rapid cycles |
The table picks sides by the actual duty. A device powered on once and left running — a bench supply, a lighting driver — rarely needs the NTC; the diode-only design, sized with IFSM margin for the one-shot, is cheaper and simpler. A device that power-cycles often, or cold-starts into a large capacitance repeatedly, earns the NTC: the warm NTC limits the hot-start surge exactly where the diode-only design suffers. The middle cases — moderate cycling, larger capacitance — are where the NTC’s small steady-state loss beats the diode’s risk of repetitive-surge damage.
A worked power-on case puts the numbers together. A 12 V supply feeds a bridge rectifier into an RC filter with roughly 0.5 Ω of source impedance and a large discharged capacitor. At the instant of switch-on the initial current approaches 12 V / 0.5 Ω ≈ 24 A — versus a steady-state average of maybe 1.5 A, a 16x surge. If the charging time constant is 200 µs, the event is a short pulse carrying an I²t integral on the order of the peak squared times the duration, and the part selected needs an IFSM that clears 24 A on the datasheet test wave at the 25 °C condition.
Now repeat the power-on forty times a minute with a load that keeps the capacitor mostly discharged between cycles. The same 24 A surge now recurs, and the junction begins each event warmer than the last; the one-shot IFSM margin becomes a repetitive-duty requirement, and the honest fix is either a substantially derated part or an NTC that shaves the cold peak. The example is the difference between “the datasheet says it can carry 30 A surge” and “it survives your actual duty” — the first is a headline, the second is an I²t and repetition check.
The same logic scales to the bigger rails where the three-phase bridge turns the inrush into a six-pulse charging event with its own peak and energy profile.
Observing Inrush: Scope Setup and What Failure Looks Like
The last discipline is observing the event instead of assuming it. A current clamp or a scope with a current probe across the input catches the first-cycle peak and its decay; a voltage probe across the diode captures the forward drop during the surge, which rises with current and can be far above the steady-state VF.
The failure signature to watch for is heat accumulation on rapid cycling: the diode’s forward drop during the surge should match the curve at that current, and the case temperature should settle between cycles if the one-shot assumption holds. If the case keeps climbing across repeated power-ons, the design is in the repetitive regime whether or not the datasheet’s IFSM was quoted as one-shot — and the fix is the NTC, a larger diode, or a longer delay between restarts.
One more caution closes the list: never trust the IFSM headline alone for the hot-start case. The datasheet’s one-shot half-sine at 25 °C is a different event from a warm junction cycling into a discharged capacitor at 80 °C case temperature. The scope test in this section exists precisely to replace the headline with the measured event, and the rectifier thermal design guide converts the observed heat into the junction temperature that decides whether the part survives the sequence. The scope also validates the fix. After adding an NTC, the first-cycle surge should be measurably lower, and the diode’s case temperature across a rapid-cycling test should plateau instead of climbing. The rectifier failure-modes guide frames the same evidence reading in the field, and the general rectifier and fast recovery categories supply the parts whose IFSM and I²t margins close the design.