Switching dc-dc converters are often used in portable or batterybacked projects because of their broad input voltage range and high efficiency. These converters present a constant-power load to the battery—their input current increases as the battery voltage decreases.

By definition of current, charge is removed from the battery faster as the discharge progresses. Naturally, it’s imperative to know how long the battery will keep the electronics alive, and the best way to find out is to test it on the actual product. If it’s early in the product-development cycle, however, a prototype may not yet be available for testing, so a dummy load of some sort is required.

Using a constant-current or constant-resistance load risks overestimating the capacity of the battery, because the current never increases. The error can be significant, particularly when using batteries without a very flat discharge voltage profile, such as lead-acid or alkaline. This design uses only two inexpensive chips, a handful of passive components, and a power transistor to provide a true constantpower load (see the figure). The key component is the LM3900 current-differencing (Norton) amplifier. It’s similar to, but still distinctly different from, the familiar op amp.

An op amp connected in a negative feedback circuit follows the time-honored axiom that the amplifier’s high open-loop gain causes the voltage difference between the inputs to be approximately zero. There’s a similar rule with a Norton amp, except that it’s the input *currents*, rather than the input voltages, that are made equal by the high gain of the amplifier. This mode of operation makes using the MC1494 analog multiplier easy, since the output is a current proportional to the product of its inputs.

The user applies an input current to the noninverting Norton amp input through R3. This causes the amplifier output to go high, which turns on Q1 and draws some current from the battery under test. By virtue of R7, the battery current causes a voltage to be seen at the multiplier V_{X} input; the V_{Y} input has been looking at the battery voltage all along. Consequently, the multiplier output current will increase until the Norton amp decides that the input current is equal to the multiplier output current, at which point the operation stabilizes.

The MC1494 data sheet gives the multiplier output current as:

Due to the high Norton amp gain, I_{pwr} = I_{in}, which is constant. Substituting this and Equations 2 and 3 into Equation 1 gives:

Capacitor C1 acts as a filter, and C2 kills the frequency response of Q1, assuring stable closed-loop operation. Note that while the noise in the voltage-sensing circuit can be filtered because it isn’t under feedback control, attempting to filter the current-sensing circuit may slow down the feedback path enough to cause oscillation.

The component values shown give load maximums of about 30 Volts, 5 Amps, and 35 Watts. The scale factor of 1/10 for voltage and current sense is arbitrary, but modifying the scaling may require changes in R_{X} and R_{Y} (consult the MC1494 data sheet if modifications are necessary).

The MC1494 typically is used with potentiometers to zero out the input offsets, but as a practical matter, this circuit performs well without them. If extreme accuracy is required, and thus the potentiometers, then be aware that they must be temperature-stable. In fact, it’s a good idea to ensure that all resistors associated with the multiplier have good temperature coefficients, particularly R_{X}, R_{Y}, and R8.

If R1 is a 10-turn potentiometer, then the load will be about 3.5 Watts/turn. A bias adjustment is necessary to account for a 10- to 15-µA output offset current from the multiplier. To adjust the bias, R1 should be set to the “zero input current” position, and R2 should be adjusted until current just starts to flow from the battery.