The charge pump is a circuit topology that may be used to increase a power supply’s voltage or provide a complement (negative) source. The term “pump” is applicable as a digital circuit is used to push and pull against a “flying” capacitor. The energy stored in a capacitor is transferred from one place to another using a series of passive and active switches. In some ways, this is like pumping water with the activation of the pump’s handle (digital pumping action) along with one-way valves (diodes) to control the flow. This is an excellent learning opportunity involving state-based circuitry leading to a better understanding of more complex circuits such as the boost and buck regulator. The resulting circuit and schematic are shown as Figures 1 and 2.
The charge pump featured in this article is designed to provide gate voltage to an N-channel MOSFET. We assume a 24 VDC system with the MOSFET in a source-follower configuration. The desired voltage is about 10 to 15 VDC above the 24 VDC rail allowing the MOSFET to be driven into a low resistance state thereby fully activating the associated load. This application does not need a sustained current as the MOSFET is a voltage driven device. However, it does require a pulse of current to charge the gate to channel capacitance. This pulse of energy is provided entirely by the 100 uF C6 capacitor.
The associated pulse-by-pulse energy calculations will wait for another day. Instead, we will characterize the charge pump power using a steady state analysis. Based on these data we can properly characterize this charge pump as a 34 VDC supply with a 4 mA current limit. While this may seem inconsequential, it is more than enough to provide power to the MOSFET driver.
Tech Tip: The bootstrap topology is a closely related circuit. Both the charge pump and the bootstrap may be used to provide an N-channel MOSFET with “above the rail” voltage to properly turn on its gate. The difference is related to how the MOSFET is operated. The bootstrap circuit is preferred when the MOSFET is repeatedly turned on and off as when driven by a PWM signal. The charge pump is preferred when the MOSFET is held on for an extended amount of time. In both cases, there is an on-off attribute that charges a “flying capacitor.” We could say that the bootstrap is a self-priming charge pump.
Figure 1: Picture of the charge pump providing 34 VDC from a 24 VDC source.
Figure 2: Schematic of the charge pump featuring a 555 timer IC to provide 24 + 11 VDC for use by a MOSFET driver (not shown).
Square Wave Generator
The digital section of the charge pump is the quintessential 555 timer operating in an astable configuration. The 555 timer’s power is provided by a 15 VDC regulator. This 7815 linear regulator is in the same family as the popular 7805. This is a good match for the 555 timer’s 18 VDC absolute maximum voltage as described in the datasheet.
The duty cycle and frequency of the charge pump are not critical, although the results can certainly be improved as suggested at the end of the article. The resistor and capacitor value were chosen out of convenience being common values of 1 kΩ, 10 kΩ, and 0.1 uF. The resulting 700 Hz signal has a duty cycle of about 65 % with the on-time being longer than the off-time. You may see the results as shown in Figure 3 using DigiKey’s 555 timer calculator.
Figure 3: DigiKey’s 555 Timer Calculator showing the values used in this article.
Flying Capacitor
The heart of the charge pump is a capacitor C5. Observe that C5 is connected to the output of the 555 timer via the 150 Ω resistor R3. The operation of this “flying” capacitor is best understood by considering discrete the states of the 555 timer:
-
Off State: Let’s assume the C5 is initially uncharged, and the 555 timer is in the off state. This will cause C5 to charge. The charging path includes the 24 VDC supply, D1, C5, R3, and the 555’s internal pin-3 to ground transistor. The current is limited to 200 mA by R3 as both the 555 timer and the 1N4148 diodes have a 200 mA limit. This 150 Ω value may appear incorrect. Rather than using the anticipated 120 Ω resistor, we assume a worst case 30 VDC is applied to the 24 VDC rail. Also note that the current associated with the capacitor is not constant and is instead governed by the exponential curve of the RC circuit.
/
The 2.2 uF capacitor along with the 150 Ω resistor have a time constant of 0.33 ms which 2 times faster than the 0.67 ms low time for the 555 timer. The result is that the flying capacitor C3 charges to approximately 20 VDC.
/
V_C(t) = V_{Final}(1-e^{-\tau/RC} )= (24-0.7)(1-e^{-0.67/0.33}) \approx 20 \, VDC -
On State: The 555 output now toggles to a high state. The lower leg of the flying capacitor (C5) is pulled up to the 555 timer’s 15 VDC rail. Since this capacitor is already charged to approximately 20 VDC, the voltage on the capacitor’s positive terminal is now 15 + 20 VDC. This reverse biases D1 and forward biases D2. Current now flows from the flying capacitor to C6 with current limited by R3.
-
Repeat: This process is repeated continuously with the flying capacitor constantly being charged and discharged. As the name implies, the digital signal from the 555 timer pumps energy from one location to another. This process is best illustrated in the startup waveform showing the voltage measured across C6 as shown in Figure 4. The steady sate waveform shown in Figure 5 is also of interest and the blue waveform shown the ripple. In this example, the charge pump is powering a 10 kΩ load.
In an ideal system, the output voltage may be approximated as:
V_{Final} = 24 + (15 – 0.7 – 0.7) \approx 38 \, VDC
Where the 0.7 VDC values are associated with the voltage drops across D1 and D2.
This is indeed the unloaded output voltage. However, the circuit is quickly loaded down dropping to 34 VDC with a 10 kΩ load and 32 VDC with a 5 kΩ load.
Tech Tip: Initial startup conditions must be considered. One challenge with this circuit is the large value of C6. Observe that the capacitor will be charged via D1 and D2 as soon as power is applied. This is an unacceptable condition for these small-signal diodes as they will not handle the high inrush current. To mitigate this situation the 1N4001 D3 diode was added to provide the initial capacitor charge. This diode may handle a non-repetitive peak current of tens of Amperes without damage.
Figure 4: Start up waveform showing the voltage developed across the output capacitor C6 while driving a 10 kΩ load.
** Figure 5**: Ripple waveform on C6 (Ch2 blue) and 555 pin 3 drive waveform (Ch 1 orange).
Next Steps
This circuit is not optimized. It was built as a quick easy to assemble circuit. As an educational design challenge you could:
-
measure the efficiency for a variety of load conditions
-
utilize Schottky diodes instead of the general purpose 1N4148 diode
-
optimize the frequency, duty cycle, and capacitor(s)
-
eliminate the linear regulator
-
devise a method to turn the charge pump off when the circuit is fully charged
-
identify sources and mitigate EMI generated by the circuit
-
replace the 555 timer with a microcontroller including the ability to monitor and adjust for optimal performance including the ability to change duty cycle
We look forward to your feedback.
Best Wishes,
APDahlen
P.S. Here is the circuit incorporating the charge pump: The MOSFET Active Clamp: The Case Against a Relay’s Parallel Flyback Diode