All three stages together must finish inside the withstand limit of the weakest component. That component is often a cable termination, a busbar support or a semiconductor die, not the device the designer was focused on.
During a short fault, conductor heating is adiabatic: there is no time for heat to escape. The energy a conductor can absorb before reaching its limiting temperature therefore rises with the square of its cross-sectional area, multiplied by a material constant that reflects conductor metal, insulation type and start and end temperatures. Damage follows energy, not instantaneous current. A very high peak that lasts a few milliseconds can be harmless, while a modest fault that lasts two seconds can destroy the same cable.
Short circuits. When a short circuit starts near voltage zero in an inductive circuit, a decaying DC offset adds to the symmetrical current. The first peak can then approach twice the symmetrical peak, and that first peak sets the mechanical forces and the breaker’s making duty. Any overcurrent protection device installed at that point must be rated for this asymmetrical peak, not just the symmetrical value.
Ground faults. These are current flowing outside the intended path. They are detected residually: the vector sum of the phase and neutral currents is zero unless some current returns by another route. This is why ground-fault elements sit alongside phase overcurrent protection rather than replacing it.
Arc faults. These are the blind spot of conventional overcurrent protection. A series arc at a loose terminal draws less current than the load itself. A parallel arc sputters in half-cycle bursts that a thermal element averages away and that never stay above an instantaneous threshold long enough to trip it. The arc’s signature is in its high-frequency content and irregular conduction, not its magnitude.
Legitimate transients. Several events are not faults but still must be ridden through: motor starting, locked-rotor tests, and inrush when capacitors or transformers are energized. Every overcurrent relay setting balances riding through these against tripping on real faults.
Overcurrent Protection vs Overload Protection Devices: Are They the Same?
No. Overload protection is the thermal subset of overcurrent protection; short-circuit protection is the other subset. Overload protection devices, such as thermal or electronic motor overload relays, model how the protected equipment heats up but have little or no ability to interrupt a fault. A motor overload relay only opens the contactor coil circuit. The contactor itself is rated to make and break locked-rotor current, not tens of kiloamps.
This is why IEC 60947-4-1 defines how a starter coordinates with its short-circuit protective device (SCPD):
- Type 1 coordination: the contactor and overload relay may be damaged by a short circuit, provided nobody is endangered.
- Type 2 coordination: only light, easily separated contact welding is allowed, and the starter stays serviceable.
Each part of the combination has one job. The contactor switches the load, the overload relay protects the motor against sustained overcurrent, and the fuse or breaker clears short circuits.
Types of Overcurrent Protection Devices Compared
Selecting an overcurrent protection device comes down to four questions:
- How fast is it at the fault level that matters?
- Is its interrupting rating above the available fault current?
- Can it be reset, and how much maintenance does it need?
- How accurate and repeatable are its settings?
| Over current protection device | Speed and interrupting capability | Reset, setting accuracy and typical use |
| Current-limiting fuse | Clears in under half a cycle; very high interrupting capability, commonly 100 kA or more | Must be replaced after operating. Fixed setting that can drift with ageing. Used at high fault levels and for semiconductors. |
| Thermal-magnetic breaker | One to a few cycles; interrupts up to its kA rating | Resettable. Wide tolerance and ambient-sensitive. Used for branch and building circuits. |
| Electronic-trip breaker | One to a few cycles, adjustable; interrupts up to its kA rating | Resettable. Tight, repeatable settings. Used for feeders and selective systems. |
| Relay plus breaker | Relay time plus 3–5 cycles of breaker time; interrupting capability set by the breaker | Resettable. Precise, programmable settings. Used for medium voltage, large low-voltage systems and transformers. |
| Solid-state breaker / e-fuse | Microseconds; interrupting capability limited by the semiconductor | Reset electronically. Precise settings. Used for DC systems, converters and batteries. |
Fuses: How Does This Over Current Protection Device Work?
Operation. A fuse element is a metal strip with notches: narrowed sections where resistive heating concentrates. Total clearing time is the pre-arcing (melting) time plus the arcing time. The arc lengthens across the notches and is quenched in silica sand filler, which absorbs its energy.
Current limitation. In its current-limiting range, a fuse melts within the first quarter-cycle and clears before the prospective first peak. One published peak let-through characteristic for a 30 A semiconductor fuse shows the peak held to about 3.25 kA with 100 kA rms available. Compared with a device that lets the first half-cycle through, let-through I²t falls by an order of magnitude or more. No other overcurrent protection device limits energy this effectively at high fault levels.
Fuse classes:
- Fast-acting fuses suit resistive loads and semiconductors.
- Time-delay fuses ride through motor inrush and are widely used for transformer overcurrent protection, where they must withstand energization inrush.
- General-purpose fuses protect cables across the full overcurrent range.
- Semiconductor fuses (partial-range aR and full-range types) exist because IGBT and SiC modules can withstand far less I²t than a general-purpose fuse lets through. The fuse’s total clearing I²t, at the real working voltage, must stay below the module’s I²t rating.
Limitations. Elements fatigue under cyclic loading and can drift. Every operation means a replacement, and fitting a wrong-class replacement silently changes the overcurrent protection. Fast clearing also cuts arc-flash incident energy, but only when the arcing current falls inside the fuse’s current-limiting range.
Circuit Breakers: Thermal-Magnetic and Electronic Trip Units
Thermal-magnetic breakers have two elements:
- A bimetal gives the inverse-time overload response, the same principle used in many overload protection devices. It is sensitive to ambient temperature and must be derated in hot enclosures.
- A magnetic armature gives instantaneous tripping.
Electronic trip units measure current through internal current transformers and provide four functions, with repeatable settings that do not depend on ambient temperature:
- Long-time (L): overload.
- Short-time (S): delayed fault tripping, used for selectivity.
- Instantaneous (I).
- Ground fault (G).
Curve classes. IEC 60898-1 fixes the instantaneous trip bands for miniature breakers:
- Type B: 3–5 times rated current.
- Type C: 5–10 times rated current.
- Type D: 10–20 times rated current.
North American practice describes the same idea as a fixed or adjustable instantaneous pickup, expressed as a multiple of the rating.
Interrupting ratings. These are stated in kA. IEC 60947-2 separates two values:
- Icu (ultimate breaking capacity): the breaker interrupts the fault but may not be fit for further service.
- Ics (service breaking capacity): stated as 25, 50, 75 or 100% of Icu; the breaker remains serviceable afterwards.
Never install a breaker where the available fault current exceeds its rating. Series ratings let a lower-rated downstream breaker be protected by a tested upstream device, but only for the exact combination that was tested and listed. A series rating cannot be created by calculation, and it usually gives up selectivity.
Overcurrent Relays: 50, 51 and Inverse-Time Phase Overcurrent Protection
An overcurrent relay separates sensing and decision from interruption. That separation is why relays dominate medium-voltage and large low-voltage systems.
The trip chain. Primary current passes through a current transformer (CT) with a 1 A or 5 A secondary. The CT feeds the overcurrent protection relay element, whose trip contact energizes the breaker trip coil, which opens the breaker. Trip-circuit supervision monitors the coil and its wiring continuously, because a relay that operates into an open trip circuit protects nothing.
ANSI device numbers that matter:
- 50: instantaneous phase overcurrent.
- 51: time-delayed (inverse or definite) phase overcurrent.
- 50N/51N: residual ground overcurrent, calculated from the summed phase CTs.
- 51G: ground overcurrent from a CT in the neutral or ground connection.
- 46/50Q: negative-sequence overcurrent, for unbalance and phase-to-phase faults.
- 67: directional overcurrent.
Phase overcurrent protection uses 50/51 elements on every phase to detect three-phase and phase-to-phase faults. On solidly grounded systems it also backs up ground-fault protection.
Definite-time and inverse-time elements. A definite-time over current relay trips after a fixed delay once current exceeds pickup. An inverse-time element operates faster as current rises. Pickup is set in primary amps and converted to secondary amps through the CT ratio. The time multiplier setting (TMS, in IEC terms) or time dial (in IEEE terms) moves the curve up or down.
The standard inverse curves. In the IEC 60255-151 inverse characteristic, operating time equals the TMS multiplied by a curve constant, divided by one less than the multiple of pickup raised to a curve exponent. IEEE C37.112 uses the same structure plus an added constant.
| Curve (standard) | Constant and exponent | Added constant |
| Standard inverse (A), IEC 60255-151 | Constant 0.14; exponent 0.02 | None |
| Very inverse (B), IEC 60255-151 | Constant 13.5; exponent 1 | None |
| Extremely inverse (C), IEC 60255-151 | Constant 80; exponent 2 | None |
| Moderately inverse, IEEE C37.112 | Constant 0.0515; exponent 0.02 | 0.114 |
| Very inverse, IEEE C37.112 | Constant 19.61; exponent 2 | 0.491 |
| Extremely inverse, IEEE C37.112 | Constant 28.2; exponent 2 | 0.1217 |
Worked example. A 400 A feeder has a 600:5 CT and a standard-inverse element with pickup at 500 A primary, which is 4.17 A secondary. A 3,000 A fault is 6 times pickup (25 A secondary). At that multiple the standard-inverse curve gives a base time of about 3.85 s, which the TMS then scales:
| Setting | Operating time at 3,000 A |
| Standard inverse, TMS 0.1 | 0.38 s |
| Standard inverse, TMS 0.3 | 1.15 s |
| Very inverse, TMS 0.1 | 0.27 s |
| Extremely inverse, TMS 0.1 | 0.23 s |
Time scales linearly with TMS. The roughly 0.77 s gap between TMS 0.1 and TMS 0.3 is far more than any coordination interval requires, so in practice the upstream relay can be made faster.
Directional elements. On feeders with distributed energy resources (DER), and in ring or parallel networks, fault current can flow in either direction. A 67 element uses a voltage polarizing quantity so that it trips only for forward faults. This prevents a healthy feeder from tripping on its own backfeed into a neighbouring fault.
CT saturation. High current with DC offset, or excessive burden, saturates the CT core and collapses part of each secondary cycle. The relay then measures less current, inverse-time elements slow down, and coordination margins shrink exactly at the highest fault levels. Check the CT accuracy class and knee-point voltage against the maximum asymmetrical fault current and the real burden of the leads plus the relay.
Overcurrent Protection Circuit Design Inside Converters and Power Supplies
An overcurrent protection circuit inside a converter has to act in microseconds, far faster than any external overcurrent protection device can respond.
Current sensing options:
- Shunts: accurate and cheap, but lossy and not isolated.
- Hall-effect sensors: isolated, but limited in bandwidth and subject to offset.
- Rogowski coils: wide bandwidth, but AC-only and in need of an integrator.
- Integrated sensing: built into power modules and gate drivers.
Detection and turn-off:
- A fast comparator trips against a reference. Leading-edge blanking masks the turn-on spike caused by diode recovery and capacitance charging, so it is not mistaken for a fault.
- Cycle-by-cycle limiting in peak current-mode control ends each switching pulse when current reaches the limit.
- Desaturation (DESAT) detection in IGBT and SiC gate drivers watches the on-state voltage. If the device leaves saturation while commanded on, the driver turns it off with a soft shutdown, slowing the turn-off to keep overvoltage within the device rating.
Protective components:
- E-fuses combine a switch, a current limit and a fault timer.
- PPTC devices trip thermally, and are slow but resettable, which makes them compact overload protection devices for low-power rails.
- Solid-state circuit breakers interrupt DC in microseconds without an arc, making them a fully electronic over current protection device.
Fault-response policies:
- Constant-current limiting holds current at the limit. It suits chargers and bench supplies, where a laboratory supply crosses over from constant voltage to constant current at the set limit.
- Foldback reduces the current limit as output voltage collapses, which cuts dissipation in linear pass devices.
- Hiccup shuts down and retries periodically. It suits transient faults in DC/DC converters.
- Latch-off stays off until reset. It is the right policy wherever a repeated fault is hazardous.
Why the layers are not interchangeable. A SemiEngineering technical analysis of SiC short-circuit ruggedness puts the SiC MOSFET short-circuit time and resulting power loss “in the range of 2–3 μs”, and a 2023 IET Power Electronics paper notes that gate-driver short-circuit protection for SiC MOSFETs must act “within 3 µs, while for Si IGBT it is less than 10 µs.” The fastest mechanical device needs milliseconds. The upstream breaker protects the cable; only the gate driver protects the die.
The same layered approach applies inside Impedyme real-time Hardware in the loop platforms:
- Analog inputs are filtered by low-pass filters before the ADCs on an UltraScale+ FPGA/SoC.
- In parallel, the inputs feed hardware comparators with high (L.HIGH) and low (L.LOW) thresholds. Any violation is ORed into a fault manager.
- The fault manager gates the PWM outputs through AND logic, independently of the CPU running the user application.
- A watchdog covers a stalled real-time execution.
- A separate safety FPGA, with its own power supply and an interlock connector, adds a final independent layer.
Impedyme describes this as a dual-layer architecture. A configurable, time-delayed software overcurrent protection algorithm sits on top of factory-defined, non-configurable analog comparators and hardware logic gates, which act as the last line of defense.
GFCI, RCD and AFCI: Protection an Overcurrent Device Cannot Provide
A 20 A breaker will never trip on a 30 mA shock current, because overcurrent protection is sized for conductors, not people. Residual-current devices fill that gap by passing all live conductors through a core-balance CT; any imbalance means current is returning by another path.
Thresholds. Personnel protection uses 30 mA in IEC practice, and North American GFCIs trip at 4–6 mA. A general-type 30 mA RCD must trip within 300 ms at rated residual current and within 40 ms at five times rated.
RCD types:
- Type AC detects sinusoidal AC only.
- Type A adds pulsating DC and tolerates up to 6 mA of smooth DC.
- Type F adds mixed-frequency residual currents and tolerates up to 10 mA of smooth DC.
- Type B also detects smooth DC.
Type B matters for EV chargers, solar inverter test and three-phase variable frequency drives (VFDs). In that equipment a fault behind a rectifier can produce smooth DC residual current, which saturates the core of a Type A device and blinds it. IEC 62955 defines a 6 mA DC detecting device (RDC-DD) built into Mode 3 chargers, which allows Type A protection to be used upstream.
AFCIs detect arcing by analysing the current waveform’s high-frequency content and irregular conduction intervals. This lets them separate real arcs from the normal switching signatures of loads such as brushed motors and dimmers, and catch faults that never cross the pickup of a conventional over current protection device.