I just finished testing the Sharp GP2Y3A001K0F Multi-Beam Ranging Sensor. If you don't already know, its 5 Sharp IR units in one single super unit.
The datasheet was seriously lacking, so I did some experiments to characterize it. Hopefully this data is useful for someone . . .
There are two power lines, one for digital circuitry, and one to power the LED.
The LED output line requires 0.190mA (yeap, incredibly low!). This is stable, and can be powered right off a mcu digital output pin no problem.
The digital circuitry power line requires ~20mA average, but fluctuates quite a lot at high frequency. I suspect its electrically noisy like the other Sharp IR units, and probably needs a ceramic cap across it and ground to improve the power draw stability . . . I didn't further test this theory.
If the object is within 25cm, this sensor is accurate within ~1cm. If the detected object exceeds this distance, the sensor will acknowledge that further distance in its measurements. However, this output is both unstable and unreliable by a large percent, and becomes more inaccurate as the object distance increases.