How charge pumps implement the technique of increasing or inverting DC voltage

A charge pump is a technique for increasing or inverting a DC voltage. For example, +5V can be converted to +10V or -5V (or higher/lower values). Compared to a boost converter, charge pumps require fewer components and less PCB space, and are less expensive; however, a limitation of charge pumps is that they can only provide relatively small amounts of current. Due to the low current limit, charge pumps are best suited for applications such as signaling (communications) or driving LCD backlights. Historically, a common application has been RS232 communication transceivers (such as the MAX232) that generate +3V to +15V and -3V to -15V.

theory

The working principle of a charge pump is very simple. One of the first things we learn about electricity is that if we connect batteries in series, their voltages add together.

Another electrical basic is that capacitors are like small batteries.

If we can charge a capacitor with a voltage source and then quickly reposition it in series with the voltage source, we can double the voltage (the same way adding batteries in series increases their voltage). In a circuit, repositioning can be accomplished with switches - either mechanical (physical switches or electromagnetic relays) or solid-state (transistors/diodes).

As far as generating a negative voltage is concerned, it is simply a case of repositioning the capacitor so that its positive charging terminal is connected to the negative terminal of the voltage source.

The principle can also be scaled, as any number of capacitors can be charged in parallel at a voltage source and then placed back in the stack.

Charge Pump Schematic

On a schematic, the above configuration can be done as follows:

(The screenshot was captured slightly after the switching moment, when the capacitor has been slightly discharged.)

The voltage reversal looks like this:

Of course, if there is any load the capacitor will immediately begin to discharge, so it is necessary to constantly switch the capacitor back and forth between parallel and series configuration in order to keep it recharged.

To keep the output voltage relatively constant while switching, we can add a capacitor to the output.

This smooths the output to some extent.

However, having someone flip a switch back and forth to run the charge pump is obviously not practical, and to maintain a relatively stable output voltage with a reasonably sized capacitor requires very fast switching; hence a fast clock signal to run the switching.

Clock Charge Pump

In the above circuit, the MC34063 buck converter is used to step down 25V to 5V. The switching transistor is integrated inside the controller, so no external transistor is required. The controller monitors the output voltage using feedback resistors R2/R3 and maintains a constant output voltage at the load.

Given a clock signal, we can connect that signal to the negative terminal of the capacitor and connect the positive terminal of the capacitor to a positive voltage source through a diode.

When the clock signal is low (0V), the capacitor will charge through the diode to the positive supply voltage (minus the voltage drop across the diode).

When the clock signal is high (the supply voltage, +1.5V in this case), the voltage stored in the charged capacitor is added to the voltage at its negative pin, resulting in a doubled output voltage.

(The diode prevents the capacitor from discharging to the supply voltage.)

The result is a doubling of the clock voltage.

To smooth the output voltage we can add another capacitor and a diode to the output to prevent it from reverse discharging during the low phase of the clock cycle.

We now have a very smooth output voltage.

Due to the voltage drop across the diode (1.5V is very low for the supply voltage, the voltage drop across the diode is relatively large, and relatively small for 5V/9V/etc supply voltages) and the non-ideal properties of real-world electronic components (such as internal resistance), the smoothed output voltage is not twice the input voltage, but it will be significantly increased above the supply voltage, and we can increase this further by adjusting the principle and adding more pump stages.

Dickson Charge Pump

Adding an extra pump stage requires an inverted clock. Clock inversion can be achieved using a simple N-MOSFET and a pull-up resistor:

However, this only works at higher supply voltages, as the gate threshold voltage of a typical N-MOSFET is around 2.1V, so at this point we will switch to a +5V supply.

We connect the inverted clock to the negative terminal of the second stage capacitor:

Let's analyze how this works (ignore the voltage drop across the diode/transistor for simplicity).

Initially, the clock is low and the capacitors in stage 1 are charged to the supply voltage (+5V). The capacitors in stage 2 are not charged yet because both its positive and negative pins have the supply voltage.

Next, the clock goes high and the second stage charges to +10V, just like before.

Now, the clock goes low again, causing the inverted clock to go high and boosting the now charged stage 2 capacitor to 3 times the supply voltage (+15V).

Again, due to the voltage drops across diodes and non-ideal real components, the output voltage is not exactly +15V, but it is certainly more than twice the supply voltage.

This process can be chained and scaled to produce arbitrarily high output voltages

This type of charge pump topology is called a Dickson charge pump.

Marx Generator

Another interesting design is the Marx generator:

In this design, spark gaps are used as switches. Spark gaps are conductors placed at a certain distance apart, which cause conduction once the voltage between them is higher than the breakdown voltage of the insulator (about 30kV/cm in air). Once all parallel capacitors are charged, a chain reaction across the spark gaps is initiated by triggering the first spark gap. Using this technique, voltages of hundreds of thousands of volts can be generated.

Turning our thoughts back to ordinary electronics, it's worth mentioning that there are some handy integrated circuits (ICs) that simplify the process of adding a charge pump to a design - requiring only a supply voltage and two capacitors - such as the industry-standard TC7660.

More advanced charge pump ICs are also available which can output relatively accurately regulated voltages by carefully controlling the clock that drives the charge pump while carefully monitoring the output voltage.

In summary, charge pumps offer an interesting, compact, and low-cost option for stepping up or inverting voltages that do not require high output current.