The O2 sensor works like a miniature generator and produces its own voltage when it gets hot. Inside the vented cover on the end of the sensor that screws into the exhaust manifold is a zirconium ceramic bulb. The bulb is coated on the outside with a porous layer of platinum. Inside the bulb are two strips of platinum that serve as electrodes or contacts.

The outside of the bulb is exposed to the hot gases in the exhaust while the inside of the bulb is vented internally through the sensor body to the outside atmosphere. Older style oxygen sensors actually have a small hole in the body shell so air can enter the sensor, but newer style O2 sensors "breathe" through their wire connectors and have no vent hole. It’s hard to believe, but the tiny amount of space between the insulation and wire provides enough room for air to seep into the sensor (for this reason, grease should never be used on O2 sensor connectors because it can block the flow of air). Venting the sensor through the wires rather than with a hole in the body reduces the risk of dirt or water contamination that could foul the sensor from the inside and cause it to fail.

The difference in oxygen levels between the exhaust and outside air within the sensor causes voltage to flow through the ceramic bulb. The greater the difference, the higher the voltage reading.

An oxygen sensor will typically generate up to about 0.9 volts when the fuel mixture is rich and there is little unburned oxygen in the exhaust. When the mixture is lean, the sensor’s output voltage will drop down to about 0.1 volts. When the air/fuel mixture is balanced or at the equilibrium point of about 14.7 to 1, the sensor will read around .45 volts.

When the computer receives a rich signal (high voltage) from the O2 sensor, it leans the fuel mixture to reduce the sensor’s reading. When the O2 sensor reading goes lean (low voltage), the computer reverses again making the fuel mixture go rich. This constant flip-flopping back and forth of the fuel mixture occurs with different speeds depending on the fuel system. The transition rate is slowest on engines with feedback carburetors, typically once per second at 2500 rpm. Engines with throttle body injection are somewhat faster (2 to 3 times per second at 2500 rpm), while engines with multiport injection are the fastest (5 to 7 times per second at 2500 rpm).

The oxygen sensor must be hot (about 600 degrees or higher) before it will start to generate a voltage signal, so many oxygen sensors have a small heating element inside to help them reach operating temperature more quickly. The heating element can also prevent the sensor from cooling off too much during prolonged idle, which would cause the system to revert to open loop.

Heated O2 sensors are used mostly in newer vehicles and typically have 3 or 4 wires. Older single wire O2 sensors do not have heaters. When replacing an O2 sensor, make sure it is the same type as the original (heated or unheated).

O2 sensors are amazingly rugged considering the operating environment they live in. But O2 sensors do wear out and eventually have to be replaced.

The performance of the O2 sensor tends to diminish with age as contaminants accumulate on the sensor tip and gradually reduce its ability to produce voltage. This kind of deterioration can be caused by a variety of substances that find their way into the exhaust such as lead, silicone, sulfur, oil ash and even some fuel additives. The sensor can also be damaged by environmental factors such as water, splash from road salt, oil and dirt.

As the sensor ages and becomes sluggish, the time it takes to react to changes in the air/fuel mixture slows down which causes emissions to go up. This happens because the flip-flopping of the fuel mixture is slowed down which reduces converter efficiency. The effect is more noticeable on engines with multiport fuel injection (MFI) than electronic carburetion or throttle body injection because the fuel ratio changes much more rapidly on MFI applications.

If the sensor dies altogether, the result can be a fixed, rich fuel mixture. Default on most fuel injected applications is mid-range after three minutes. This causes a big jump in fuel consumption as well as emissions. And if the converter overheats because of the rich mixture, it may suffer damage.

One EPA study found that 70% of the vehicles that failed an I/M 240 emissions test needed a new O2 sensor.

The only way to know if the O2 sensor is doing its job is to inspect it regularly. That’s why some vehicles (mostly imports) have a sensor maintenance reminder light. A good time to check the sensor is when the spark plugs are changed.

You can read the O2 sensor’s output with a scan tool or digital voltmeter, but the transitions are hard to see because the numbers jump around so much. An analog voltmeter is better for viewing transitions, but may not respond quickly enough on systems with higher transition rates. So the best instrument for observing the O2 sensor’s voltage output is a digital storage oscilloscope (DSO). A scope will display the sensor’s voltage output as a wavy line that shows both it’s amplitude (minimum and maximum voltage) as well as its frequency (transition rate from rich to lean).

A good O2 sensor should produce an oscillating waveform at idle that makes voltage transitions from near minimum (0.1 v) to near maximum (0.9v). Making the fuel mixture artificially rich by feeding propane into the intake manifold should cause the sensor to respond almost immediately (within 100 milliseconds) and go to maximum (0.9v) output. Creating a lean mixture by opening a vacuum line should cause the sensor’s output to drop to its minimum (0.1v) value. If the sensor doesn’t flip-flop back and forth quickly enough, it may indicate a need for replacement.