Since the introduction of On-Board Diagnostics, Generation 2, or OBD-II, PCMs have done a pretty good job of monitoring oxygen sensors and setting diagnostic trouble codes when they’re faulty. Some of the things the PCM monitors for this are signal amplitude, sensor response time, rich limit, and lean limit.
But sometimes these monitors don’t pick out the fault with the vehicle accurately. If you replaced an oxygen sensor every time you saw an oxygen sensor code, you’s probably solve about 90% of the failures.
But what about the other 10%? A hasty diagnosis, or “shotgun” approach, based on codes in this manner can often result in comebacks, unhappy customers, angry employers, and could tarnish your shop’s reputation.
In this article, we’ll look at oxygen sensors and the proper techniques used to confirm a faulty sensor. Let’s start with this example: a 2002 Ford Escape with the MIL lit. Diagnostic trouble codes include P0171, P0174, P1131, and P1151.
The first two codes — P0171 and P0174 — are very familiar lean codes associated with both banks of the engine. The additional P1131 and P1151 pending codes — oxygen sensor not switching correctly; sensor indicated lean — are for bank one and bank two respectively.
The scan tool (figure 1) offers a lot of information, but before proceeding, ask yourself a few questions: First, can an oxygen sensor fail and set a false lean code? Second, could an engine run lean enough that it sets false oxygen sensor codes?
The answer to both questions is definitely yes. So which problem is the actual fault? To determine that, let’s step back and examine more of the data.
We’ll start with a little research: The Ford scan tool, the IDS, offers technical service bulletins and special service messages related to the codes.
In this case, all four DTCs refer to the same two bulletins. Generally they say that replacing the MAF sensor, oxygen sensor, or PCM won’t correct these codes if you have a vacuum leak.
That scares me a little, because if Ford issued these bulletins, it means technicians have been trying to correct these codes by replacing these components. It suggests they didn’t perform a proper diagnosis; they used the “shotgun” approach. 
According to these bulletins, a better choice would be to check for a vacuum leak.
The freeze frame data reveals positive fuel trim values. Positive fuel trim values could easily be the result of a vacuum leak, but under what conditions are vacuum leaks most likely to affect fuel trim? Idle speeds are when you can expect a vacuum leak to have the greatest effect on fuel trim.
But the freeze frame data shows engine RPM is 2160 and the engine load is 48.6%, which suggests the throttle is open and air is flowing. So, it’s probably safe to cross a vacuum leak off the list.
So, where should we look next? Since fuel trim numbers are based on oxygen sensor input, it only makes sense to start by testing the oxygen sensors. After all, we can’t trust the fuel trim numbers if we can’t trust the oxygen sensors. So how do we test them?
It’s important to understand that exhaust leaks can affect what an oxygen sensor reports. Small holes or cracks in an exhaust system can draw ambient air in while the engine is running and affect the sensors ability to operate accurately. Although the test method we’re going to use should work okay with minor exhaust leaks, it’d be a good idea to inspect the exhaust before proceeding.
Another concern is air injection: If air is being injected into the exhaust when it shouldn’t be, it could skew the oxygen sensor results. This vehicle isn’t equipped with a secondary air injection system, so we can move forward.
Once you’ve inspected the exhaust, testing oxygen sensors requires a few tools. Although a scan tool or a digital voltmeter can get us most of the way there, a scope is the best tool for this task.
We’ll also need a propane enrichment tool. The test itself is relatively simple and consists of three parts:
- How rich can the sensor report?
- How lean can the sensor report?
- How fast can it respond?
These are basically the same tests the PCM should be performing, but we have the advantage of forcing the engine much richer or leaner to achieve the desired result.
The first step is to see if the sensor can report a rich condition. An oxygen sensor causes a chemical reaction that generates a voltage when conditions are rich. As an oxygen sensor fails, it can lose its ability to generate this voltage.
You can check this by watching the sensor voltage on a scope while adding propane or an alternate fuel source into the intake manifold. The sensor voltage should exceed 800 millivolts. If the sensor voltage doesn’t go high enough, try adding more propane. If you add enough propane to make the engine run rough or stall, it’s safe to assume the engine ran rich enough that the sensor voltage should have reached 800mV.
IMPORTANT: You may have to increase the propane several times to get the system to run rich. That’s because, every time you add propane, the fuel injection system will try to lean out to compensate. You’ll need to continue increasing the propane until the injection system can’t compensate for it any more.
If the sensor voltage won’t rise to at least 800mV, the sensor is faulty. The red arrow (figure 2) indicates where propane was added while running the engine at around 2000 RPM. The oxygen sensor generates almost 1 volt; well over the 800mV threshold.
The second step is to make sure the voltage can drop low enough to indicate when the engine’s running lean. If the vehicle was in closed loop during the first part of this test, shutting the propane off is usually enough to drive the sensor lean. The voltage should drop below 175 millivolts, without dropping below zero.
If turning the propane off doesn’t allow the system to drop lean enough, you can try creating a vacuum leak, such as pulling off the brake booster vacuum hose. If the engine runs poorly due to the extremely lean condition and the voltage doesn’t drop below 175mV, the sensor fails step two. 
The red arrow (figure 3) indicates where propane was shut off and the oxygen sensor drops to about 80mV; well below the 175mV threshold.
IMPORTANT: The oxygen sensor voltage should never drop below zero. If you see oxygen sensor voltage below zero for even a moment, the sensor is faulty and needs to be replaced. Many scan tools won’t register negative voltages, so you can only check this with a scope or digital voltmeter.
The final step is to check the response time of the sensor. While the oxygen sensor is reading lean, add a quick pulse of propane to the intake while observing the scope. Sensor voltage should switch from 300mV to 600mV in less than 100 milliseconds.
This part of the test isn’t possible with a scan tool or digital voltmeter because their response times aren’t quick enough. If the voltage transition is too slow, the oxygen sensor fails. In our test we marked the 300mV and 600mV levels (figure 4). The transition between those levels was 19ms; again, well below 100ms.
We performed an exhaust inspection and all three oxygen sensor tests on the Escape and it passed all of them. In this case we were able to cross the oxygen sensors off our suspect list. If we’d have replaced these oxygen sensors there’d have been no change, the vehicle wouldn’t be fixed, and we’d be back at ground zero.
Since we eliminate the oxygen sensors as a possible problem, our next step was to move on to lean code diagnosis.
Further testing revealed a failed mass airflow sensor, which was causing the engine to run so lean the PCM would set false oxygen sensor DTCs along with the legitimate P0171 and P0174 lean codes. After we replaced the mass airflow sensor, cleared the DTCs, and performed a test drive, fuel trim numbers returned to normal and the vehicle was fixed.
Taking a little extra time and performing a few tests resulted in the correct diagnosis… and we didn’t even have to get our shotgun out!
