EDN Access--01.16.97 Switching current pump makes better charge

-January 16, 1997

 Design IdeasJanuary 16, 1997

Switching current pump makes better charger

Alexander Belousov, Rego Park, NY

Many currently proposed battery-charger designs are simply voltage regulators redesigned to be current sources. Unfortunately, this design approach is far from optimal because it leads to unnecessary hardware redundancy and complexity. This type of charger also raises a stability problem because of the additional active device in the feedback loop.

The alternative design in Figure 1, which is based on a switching-inductive current pump, is much simpler and easier to build because it uses common discrete devices. Also, the primary design goal of a switching voltage regulator is to keep the constant output voltage under the different load conditions, but a current pump is designed to maintain constant load current under varying terminal voltages. Thus, a current pump is inherently suited to the charging requirements of NiCd and NiMH batteries, and even the constant-current requirements of Peltier Coolers.

 FIGURE 1 Combining a switching-inductor current pump with a temperature-sensing circuit produces a battery charger that's simpler and more inherently suited for charging requirements than are switching-voltage-regulator designs.

The design in Figure 1 combines a switching current pump with a temperature-monitoring design formerly proposed in Kerry Lacanette's Design Idea, "Temperature sensor doesn't discharge battery"(EDN, Aug 1, 1996, pg 106). The current pump in Figure 1 aims to keep the current constant while the battery voltage varies. The circuit uses a typical switching inductor similar to the boost topology in voltage regulators; thus, the design formulas are similar. The main design formula for discontinuous mode is

where ILOAD is the average load (charge) current; VBAT and VCC are the battery and supply voltages, respectively; D is the duty cycle equal to TON/TON+TOFF; FSW is the switching frequency; and h is the overall efficiency. The efficiency term is the source of most uncertainty in this equation. The efficiency of switching regulators based on currently available n-channel MOSFETs is typically 85 to 95%. So, using an efficiency of 0.9 when solving this equation produces acceptable results.

Note that the equation assumes full inductive discharge during TOFF, which implies discontinuous-mode operation. The mC must provide the proper duty cycle to ensure that the energy stored in the inductor completely dissipates during the TOFF interval. This requirement means that

where VF is the forward voltage of catch-diode D1, approximately 0.5V. Continuous-mode operation is not recommended for this application.

Note also that for this current-pump topology, VBAT has to be greater than VCC to prevent dc flows directly through the forward-biased catch diode. A VCC of 5V well suits the charging of camcorder and other batteries. For lower voltages, you can modify the design in Figure 1 for 3.3V operation or change the topology of the current pump (Figure 2). In Figure 2b, for boost-mode operation, the control circuit has to turn on MOSFET Q2 continuously; for buck mode, the control circuit has to turn off Q1.

In Figure 1, zener diode DZ protects the output from the short-term voltage spikes that may occur if the load disconnects during the operation. The digital control algorithm should include a subroutine that checks the terminal voltage and stops the current-pump operation when the terminal voltage rises above the predetermined threshold. In this design, CH0 of the ADC10732 performs continuous monitoring of the terminal voltage. Output capacitor C1 improves stability and reduces ripples, which is an important criterion for driving a Peltier Cooler.

Use the following equations to select the inductor:

and

where IL(PEAK) is the peak inductor current and E·T is one of the inductor's basic specifications. The inductor, D1, and Q1 must be able to carry the inductor's peak current. You can digitally adjust ILOAD by controlling D, FSW, or both. It is a good design practice to keep FSW within 50 to 500 kHz because both analog and digital parts of the circuit can easily handle frequencies in this range, and the inductor does not "buzz."

Two control algorithms are possible for switching-inductor circuits: pulse-width modulation (PWM) and pulse-frequency modulation (PFM). For certainty and simplicity, this circuit uses the PFM control algorithm and discontinuous operation. First, the A/D converter samples VBAT, and the mC calculates the moving average (10 to 100 samples are reasonable). The system also checks the load temperature. If there is no reason to terminate charging, the mC then calculates and adjusts FSW to produce the necessary load current. The system also must check that, for the calculated frequency, the inductor's E·T and IL(PEAK) are still in the safe area.

The component values in Figure 1 produce a 6V/0.1A battery charger with D=0.2 and FSW=50 kHz. For higher currents, such as 0.25A, you can choose a smaller inductor, such as 15 µH, or reduce FSW. You can increase D, but keep DMAX, or you will convert the system into a continuous-conducting mode. You can upgrade the 10-bit ADC10732 to the 12-bit ADC12032 for higher resolution. Q1 features low RON providing higher efficiency under operation to VCC as low as 2.7V. Any similar MOSFET is also applicable. This circuit uses VCC and battery temperature for charge-termination criteria. If you use this circuit to drive a Peltier Cooler, you can use the LM235 temperature sensor to measure the surface temperature of both sides. (DI #1975)

 FIGURE 2 For low-voltage operation, you can modify the current pump to include a transformer (a) or create a universal buck/boost design (b).