What Are Charge Pumps?

Charge pumps are switched-capacitor DC-to-DC converters that use capacitors as the primary energy-storage elements to produce an output voltage higher, lower, or inverted relative to the supply. Unlike inductor-based switching regulators, charge pumps contain no magnetic components, making them well suited to monolithic integration in CMOS processes where spiral inductors are impractical. The operating principle was formalized by J. F. Dickson in 1976 and remains the basis for the majority of high-voltage generation circuits in integrated memory and power management devices.

Charge pumps occupy a distinct position in power electronics: they achieve efficiencies as high as 90 to 95 percent, produce low electromagnetic interference compared with inductor-based converters, and require only a small number of external capacitors. Their practical current output is limited relative to inductor converters, so they are most common in applications requiring modest currents at voltages well above or well below the supply rail.

Operating Principles

A charge pump cycles through two phases controlled by non-overlapping clock signals. In the first phase, a flying capacitor is connected across the supply and charged to the supply voltage. In the second phase, the circuit is reconfigured so the charged capacitor is placed in series with the supply and the output, delivering approximately twice the supply voltage to the load. Cascading multiple such stages multiplies the voltage by additional integer factors. Fractional multiplication ratios are achieved by varying the timing ratio between phases or by connecting stages in more complex topologies. The MDPI Electronics analysis of CMOS charge pump design topologies for low-power applications surveys the principal clocking strategies and efficiency trade-offs across these variants.

Voltage Multipliers and Dickson Topology

The Dickson charge pump is a diode-capacitor ladder that cascades voltage-doubling stages along a single signal path. Each stage adds approximately one supply-voltage increment to the output, so an N-stage Dickson pump delivers roughly (N+1)·V_DD. In practice, threshold voltage drops across the diodes reduce the gain, a limitation that has driven the replacement of diode-connected transistors with clock-bootstrapped NMOS switches in modern implementations. Voltage multipliers built on this principle are embedded in EEPROM and flash memory ICs to generate the 10 to 20 V programming and erase voltages required by floating-gate cells, voltages that would be impossible to supply directly from a 1.8 V or 3.3 V system rail.

Circuit Integration and Performance

Charge pump circuits in integrated form typically consist of a ring oscillator, a charge transfer network, and an output regulation loop. The oscillator frequency, commonly 1 to 10 MHz in CMOS implementations, determines how quickly the flying capacitors can transfer charge and therefore sets both the output current capability and the output ripple voltage. Regulation is achieved by modulating the oscillator frequency or by switching stages in and out of the cascade. Negative-voltage pumps, which drive the substrate or body terminal of NMOS transistors below ground, suppress latch-up in dense CMOS logic and were among the earliest charge pump applications in microprocessors. For detailed circuit-level guidance, the IEEE Press text on power management integrated circuit design covers sizing, stability analysis, and layout considerations.

Applications

Charge pumps have applications in a wide range of fields, including:

• Non-volatile memory programming voltage generation in flash and EEPROM
• LCD bias supply generation in portable displays
• RF switch driver circuits requiring above-rail gate voltages
• Substrate biasing in CMOS processors to suppress latch-up
• LED driver circuits in battery-powered devices
• Capacitive MEMS sensor interface circuits