Inrush Current Limiting on Power Supply Boards

Switch on a power supply and the input capacitor is empty. For a few milliseconds it looks like a short circuit, and the current that flows is limited only by the source impedance and the resistance of the wiring. The surge can be tens of times the steady state current, and if it is not managed it will trip a breaker, weld a relay contact, or destroy a rectifier. Limiting inrush current is therefore a design requirement rather than a refinement.

Where the Surge Comes From

The energy that has to be delivered at start up is stored in the input capacitance, and the current is the rate at which that capacitance charges. With a low impedance source, the initial current is set by the total series resistance in the loop: the source impedance, the wiring, the connector, the rectifier, and the equivalent series resistance of the capacitor itself. In a battery powered product the source impedance is very low, which is why the surge is worst in those applications.

The duration of the surge is short but not negligible. A large capacitor bank charged through a few ohms of total resistance reaches its final voltage in a time constant that may be tens of milliseconds, and the peak current occurs at the beginning. Protection components must be selected for both the peak and the duration, because a fuse that survives the steady state current may still open on the surge.

Sizing the Capacitor and the Source

The first design decision is how much capacitance is really needed. Input capacitance is there to supply the switching current and to smooth the rectified voltage, and an oversized bank increases the surge without improving the performance of the converter. Where the capacitance is set by hold up time, the trade-off between hold up and inrush should be evaluated explicitly rather than accepted as a constraint.

The source impedance is the second half of the calculation. A product that can be powered from a bench supply with a long cable sees a higher impedance than one connected to a battery pack, and the surge differs accordingly. Where a product must operate on both, the design should be evaluated for the worst case, which is the lowest impedance source.

Passive Limiting with an NTC Thermistor

An NTC thermistor in series with the input is the simplest solution. Its resistance is high when it is cold, which limits the surge, and falls as the current heats it, so the steady state loss is low. The component is small and inexpensive, and it requires no control circuit.

The limitations are significant. A thermistor that is already hot when the supply is switched on provides no limiting at all, which happens whenever power is cycled quickly or the ambient temperature is high. The self heating also means the steady state resistance depends on the load current, so a product with a wide load range may have more loss than expected at full load and less limiting than expected when it is cold. Where the input is cycled frequently, a different approach is needed.

Precharge Resistors and Relay Bypass

A resistor in series with the input limits the surge predictably because its value does not change. The resistor then has to be removed from the circuit once the capacitor is charged, which is done by a relay or a semiconductor switch connected in parallel. The sequence is: close the main contact with the resistor in circuit, wait for the capacitor to charge, then close the bypass and take the resistor out of the path.

The relay bypass approach handles repeated cycling well because the resistor is always the same value when it is needed. It costs more than a thermistor and it requires a control signal with the right timing. If the bypass closes too early, the surge is not limited; if it never closes, the resistor dissipates the full load current and will fail. The control logic therefore belongs in the reliability analysis, not only in the start up sequence.

Active Limiting and Hot Swap Controllers

An active circuit uses a transistor in its linear region to control the charging current, which gives a defined ramp and avoids the delays of a relay. Hot swap controllers integrate the gate drive, the current sense, and the protection, and they are the standard solution in systems that plug into a live backplane. The transistor must be sized for the safe operating area during the ramp, which is where most design errors occur.

The advantage of active limiting is that it can be made independent of temperature and of previous power cycles, and it can report a fault. The cost is complexity and the need to verify the safe operating area under worst case conditions, including a short circuit at the output. A controller that limits current into a shorted load for too long will fail unless the energy is calculated and the device selected accordingly.

Protection Coordination

Limiting the surge changes the duty of the protection components. A fuse must carry the limited surge without ageing, and it must still open on a genuine fault. Where the limiting resistor is in the fault path, it may dissipate far more energy than its steady state rating suggests, and it should be selected for the fault case rather than for the normal one.

A varistor or a transient suppressor at the input protects against a surge from the mains or from a load dump, and its coordination with the inrush limiter should be checked. A thermistor whose resistance has fallen to a low value no longer limits anything, so it provides no help during a transient that occurs after the supply has been running.

Test and Verification

Inrush behaviour is verified with a current probe and an oscilloscope, capturing the first few milliseconds after switch on. The measurement should be repeated at the highest ambient temperature, after a rapid power cycle, and at the lowest expected source impedance, since those are the conditions where the limiting is weakest. A single measurement in a laboratory at room temperature proves very little.

The result should be compared with the ratings of every component in the path: the fuse, the rectifier, the connector, and the capacitor. Where the peak approaches a rating, the design has no margin, and the margin should be restored before the product is released. The design review and quality checks should both include the start up condition as a defined test case.

FAQ

Is an NTC thermistor enough for a product that cycles frequently? No. A hot thermistor provides little limiting, so a repeated power cycle sees nearly the full surge. A precharge resistor with a bypass is the better answer for that duty.

How large should the precharge resistor be? Large enough that the peak current is within the ratings of the rectifier and the capacitor, and small enough that the capacitor charges in the time allowed before the bypass closes. Both constraints should be calculated.

Does inrush limiting affect efficiency? Only while the limiting element is in the path. A thermistor adds loss that falls as it heats, while a bypassed resistor is out of the circuit entirely once the bypass closes.