How to Test an Oxygen Sensor with an OBD2 Scanner (Autel MaxiLink ML629 Review)

You don't need to remove an oxygen sensor to test it. Connect an OBD2 scanner, watch the live voltage for about ten minutes, and you'll have your answer on most cars. The Autel MaxiLink ML629 handles this test well, and—just as important for a busy shop—it gets you a diagnosis without a lift, without a multimeter, and without guesswork.

Here's the short version: warm the engine to operating temperature, connect the scanner, go to live data, and watch the O2 sensor voltage. A healthy sensor swings between about 0.1V and 0.9V. A dead one sits flat at around 0.45V. A dying one switches too slowly. That's the whole test, and it works.

I'm a mobile vehicle diagnostics specialist. Shops in my region call me when a car fails an emissions test or won't set its monitors, and the customer is running out of patience. Over the past three years, I've handled more than 200 of those calls. On a typical one, I pull in, connect an Autel scanner, and have a clear answer within 30 to 45 minutes.

One of the more memorable cases: March 2025, a local garage with a 2017 Subaru Outback that had failed its state emissions test because the O2 sensor monitor wouldn't go ready. The customer needed the car back in 36 hours for a road trip, and had it packed to the ceiling—right down to the Subaru's factory tire repair kit in the cargo area. The shop was slammed: three cars ahead, and a Genie GR-20 scissor lift still sitting in its crate because nobody had time to set it up. I'll get back to that one in a minute.

Why live data beats the old-school methods

People think you need lab-grade equipment or factory-level software to test O2 sensors. Actually, you need live sensor data you can see in real time. The old-school routine—back-probing wires with a multimeter, injecting propane, watching for voltage response—is slower, riskier, and genuinely less reliable than monitoring the sensor's own output under real driving conditions.

Here's the thing: a multimeter shows you voltage, but it can't show you fuel trim, engine load, coolant temperature, and what the other sensor is doing at the same time. That context is what tells you whether the sensor is lying or the engine is actually broken.

How to test an oxygen sensor with an OBD2 scanner

This procedure relies on the standard OBD2 live data stream, defined by SAE J1979 (Mode 01) and supported on every OBD2 vehicle in the United States since 1996. It works the same on an Autel ML629, a high-end platform, or any mid-range scanner that can display live data. The ML629 is a good example here because it's affordable, refreshing fast enough for O2 sensor work, and doesn't bury the important values in menus.

Step 1: Get the engine to closed-loop

O2 sensors don't contribute anything while the engine is cold. The ECM runs in open-loop, ignoring the sensors until they reach operating temperature. So get the engine fully warm. Take the car for a five-minute drive if you can. If you're at the shop and can't road-test it, hold it at about 2000 to 2500 RPM until the sensor starts switching.

Step 2: Connect and navigate

Plug into the OBD2 port—under the dash, usually within reach of the steering column. On the ML629, tap Live Data. On most four-cylinder cars, you're looking at B1S1 (Bank 1, Sensor 1), which sits before the catalytic converter and controls fuel trim, and B1S2 (Bank 1, Sensor 2), which sits after the cat and monitors converter efficiency. A V6 or V8 adds Bank 2, so you'll see B2S1 and B2S2 as well. The ML629 can show several of these at once, which makes comparisons easy.

Step 3: Watch the voltage behavior

For a conventional zirconia sensor, a healthy reading looks like this: the voltage swings roughly between 0.1V (lean) and 0.9V (rich), and it crosses the mid-point around 0.45V at least once per second when you hold the engine at 2000 to 2500 RPM. Under load, the pattern gets more active.

Common failure patterns:

• Flat at about 0.45V. That's the ECM's reference voltage showing through. When the sensor stops generating its own signal, the ECM reads its own bias voltage. Dead sensor, disconnected wiring, or a damaged wire.
• Stuck high, around 0.7V. The sensor consistently sees a rich mixture. It could be the sensor, but it could also be a fuel pressure problem or a vacuum leak. Check fuel trims to tell them apart.
• Lazy switching. It crosses the mid-point, but only every few seconds. The sensor has degraded. It might not set a code yet, but fuel economy and emissions will suffer.

The lazy one is sneakier than people expect. No check engine light, no rough idle—just an O2 sensor that takes a little too long to react. If you're chasing a drivability complaint and the sensor is slow, that's often your money fix.

Autel MaxiLink ML629 OBD2 scanner review: is it the right tool?

I've used the ML629 on everything from Subarus to Ford vans to BMWs. Here's the practical review, rather than a spec sheet echo.

What it does well: live data refreshes fast enough to catch O2 sensor oscillation—some cheap scanners update too slowly and make a healthy sensor look lazy. The screen shows multiple PIDs at once, so you can watch both O2 sensors and fuel trims side by side. It reads and clears all generic OBD2 codes, and it's built for shop use. The housing is rubber-protected; I've dropped mine more times than I care to admit, and it still works.

What it doesn't do: it's not a bi-directional scan tool. You can't command the ECM to run an injector test or force a regeneration cycle. That's not a knock—that's the product category. It also won't interpret manufacturer-specific enhanced codes the way a dealer-level platform would. For generic live data, it's plenty.

As of mid-2025, the ML629 typically lists between $150 and $220 depending on the seller. Don't hold me to the exact price—bundles and promos change. In that range, it's a solid value for O2 sensor diagnostics.

A real case: the value of reading the data

Back to that Subaru. I connected the ML629, warmed the engine to closed-loop, and watched the live data. B1S1 was switching like a healthy sensor should. B1S2 was not. The downstream sensor sat at a flat 0.45V—the ECM's bias showing through, which is the textbook signature of a dead O2 sensor.

The surprise wasn't that B1S2 had failed. It had 180,000 miles on it. The surprise was that B1S1, with the same mileage and the same exhaust stream, was still switching perfectly. You'd normally expect the upstream sensor to die first because it takes the brunt of the exhaust heat. But in this case, the downstream sensor had been killed by years of moisture and thermal cycling, sitting low in the system. It went first.

We replaced B1S2, cleared the code, and the O2 sensor monitor went ready on the next drive cycle. The customer hit the road on time, and the shop owner later admitted he'd have replaced both sensors "to be safe" if I hadn't shown him the data. There's something satisfying about that—ten minutes of live data saved him a couple hundred dollars of unnecessary parts.

When the scanner test isn't enough

Look, I'm not saying this test covers everything. There are cases where you put the scanner down and reach for other tools:

• Wiring problems. A sensor can be fine but the signal never reaches the ECM due to chafed wiring or a bad connector. The scanner will show a flat 0.45V, same as a dead sensor. You need a multimeter and wiring diagrams to check continuity.
• Exhaust leaks. Air getting into the exhaust upstream of the sensor will skew the reading lean. The sensor will faithfully report it, and you might replace a good part. Inspect the exhaust before you buy a sensor.
• Wideband (air-fuel ratio) sensors. Newer vehicles—Toyota, Subaru, VW, and others—use wideband sensors that don't swing between 0.1V and 0.9V the way zirconia sensors do. The ML629 can show calculated A/F ratio data, but you need to know the expected range for your specific vehicle. Manufacturer documentation matters here.
• Intermittent faults. If the code comes back after a few days, a ten-minute live data session can miss it. I usually suggest a drive test with the scanner connected, and for signal dropouts, an oscilloscope is the definitive tool.

This approach worked for me because of the kind of work I do: independent shops, ordinary consumer cars, emissions-focused diagnostics. If you're working on heavy trucks, high-end Euro cars with complex wideband systems, or heavily modified vehicles, the calculus might be different.

Bottom line

Testing an oxygen sensor with an OBD2 scanner is a ten-minute job that saves you from replacing a part that wasn't the problem. The Autel MaxiLink ML629 is a genuinely good tool for this task. Not because it's flashy, but because it's fast, clear, and affordable. It has limits, and I've told you what they are. For the most common O2 sensor questions, it's the right tool.