First, most application use a continuous current to energize the IrED, although the sampling of the phototransistor is fairly infrequent. Although the phototransistor-IrED pair does have a lag time due to physical characteristics, the lag time is often in low tens of microseconds. This means the IrED only needs to be switched on about 50 before the phototransistor is sampled.
Let us consider a situation in which a sensor is read 100 times per second. If the IrED is energized for only 50 for each sample, the power consumption is reduced to of that of the continuously energized design.
Another trick commonly done to phototransistor-IrED pairs is to connect IrEDs in series. Because each IrED only has a forward voltage drop of 1.7V (at 20mA), two can be connected in series when powered by a source of 5V. Obviously, the current limiting resistor value must be adjusted. Given a voltage source of 12V, up to 6 IrEDs can be connected in series.
Be careful not to connect too many IrEDs in series. Although some current will flow at a forward voltage of less than 1.7V, the amount is too small to emit any infrared.
The last, but most important, trick is to use a high pulse current to energize the IrEDs. Because the pulsing technique reduces the duty cycle of the on-time, we can now use a larger current and stay within the average power dissipation parameters. This turns a dimly lit light into a strobe. It helps to improve the signal-to-noise ratio (most noise comes from background lighting), and reduce the value of the pull-up resistor (to improve recovery time). The one drawback of this approach is the loss of repeatability. The pulsing circuit introduces uncertainty because it is difficult to guarantee a constant current during a pulse.