All too often, a mass airflow (MAF) sensor failure on an OBD I vehicle begins to reveal itself when the driver complains of intermittent poor cold driveability and loss of power. This month's Diagnostic Dilemma, a 1993 Buick Century 3.3-liter with about 65,000 miles on the odometer, illustrates the point perfectly. According to the owner, the Buick had been developing an increasingly severe cold driveability complaint "for quite a while." A previous repair shop retrieved a code 44 from the diagnostic memory. Of course, I don't know what process was used to arrive at the diagnosis of a faulty oxygen sensor, but I certainly wouldn't challenge replacing an oxygen sensor at 65,000 miles. Replacing the oxygen sensor, however, didn't resolve the complaint.

Before we begin our MAF diagnosis, however, let's review the basic logic that OBD I systems use to detect a sensor failure and set a diagnostic trouble code (DTC). In most cases, the engine must enter a closed-loop fuel control mode before a DTC can be recorded. On early OBD I applications, closed-loop fuel control depends heavily upon the engine reaching normal operating temperature while later OBD I applications depend more upon the oxygen sensor generating a return signal. A code is usually set when OBD I systems detect an interruption in current flow through the sensor's reference voltage, ground circuit, or return signal circuit caused by an open or grounded wire. The duration of the interruption must also be long enough to meet the PCM's built-in criteria for setting a DTC.

The trickiest part of sensor diagnosis, however, is detecting no-code faults caused by sensors that return too little or too much return signal voltage to the PCM. Some OBD I systems use a process called rationalization that detects out-of-calibration sensors by comparing that sensor's signal voltage with the return voltage of other sensor signal voltages. For example, the PCM may compare the throttle position sensor signal and the engine speed signal with the MAF signal to detect a faulty mass airflow sensor.

Hypothetically speaking, let's say that airflow should be about 5 grams per second (gps) at idle (700 rpm) and 50 gps at full throttle (4,000 rpm) for a particular application. At a MAF of 5 gps, the PCM would look for 0.50 TPS voltage and 700 rpm engine speed. At a MAF of 50 gps, the PCM would "look" for 4.56 TPS voltage and 4,000 rpm engine speed. If the MAF numbers don't match the TPS voltage and engine speed, the PCM may rationalize this data to set a DTC for a faulty MAF.

Now, in the real world, why doesn't rationalization set a MAF-related DTC when the MAF is out of calibration? Simply put, many MAFs develop return signal faults that trick the PCM into "thinking" that the amount of fuel entering the engine is within normal parameters when this isn't actually the case. Depending upon the return signal's waveform pattern, the PCM may therefore reduce or increase fuel flow, which, in turn, will cause driveability or emissions complaints. In my experience, most defective MAFs will reduce fuel flow, which will cause a lean driveability complaint.

Before beginning a MAF diagnosis, it's a good idea to put the MAF sensor in a default mode by disconnecting it and starting the engine. The MIL should illuminate and an appropriate diagnostic code set in the diagnostic memory. If this doesn't happen, it's time to break out the diagnostic flow charts and check for a PCM failure. But, in most cases, the PCM should substitute a default airflow value (ex. 6 gps) in place of the normal MAF signal. In the default mode, the engine performance should improve.

Remember, too, that many MAF designs accumulate dirt on heat-radiating surfaces that reduce their sensitivity to airflow. Sometimes using an aerosol electrical cleaner to remove grease and dirt can restore the MAF's metering accuracy that, in turn, may solve a no-code driveability complaint.

Now that we better understand how a defective MAF can radically disturb fuel mixtures without setting a DTC, let's apply this information to our Buick. I began the diagnosis by checking the oxygen sensor voltage, mass airflow in grams per second, and engine speed or rpm appearing in the serial data stream*. The Buick's data stream immediately showed zero O2 sensor voltage above idle speed. Maximum WOT injector pulse width was 5 ms, with an indicated airflow of 30 gps.

Obviously, the Buick's MAF was developing a signal that tricked the PCM into believing that intake airflow was low, even during a WOT, 4,000 rpm condition. I might mention at this point that this testing was done at an 8,000-foot altitude. To adjust these readings to sea level values, you can increase intake manifold vacuum and mass airflow by about 25 to 30 percent.

The most revealing technique in any MAF diagnosis, however, is analyzing the O2 sensor data stream. At idle, the Buick's O2 data appeared normal, which indicated that the O2 sensor circuit was operating normally. As soon as the vehicle was accelerated, the O2 voltage dropped to zero. To confirm the data stream, I road-tested O2 sensor by connecting a four-channel lab scope equipped with a "flight recorder" feature. I used the second lab scope channel to record the injector pulse width. Both of these direct readings confirmed the accuracy of the data stream.

When gathering direct readings or data, remember that the indicated pulse width in many data streams is a calculated, rather than actual value. It's therefore possible to encounter a discrepancy between the pulse indicated on the functioning data stream and that indicated by the lab scope. In many cases, this is perfectly normal.

Variable frequency MAFs like those on our Buick generate a square-wave signal that is easily diagnosed with a lab scope. I recorded the MAF's waveform pattern during various driving conditions and the waveform patterns were completely erratic, with no single, well-defined pattern failure. I probably recorded a half-dozen different signals from this particular MAF with two different lab scopes, so we're not going to split hairs on frequency and voltage values. Coupled with the previous data, these irregular waveforms were enough for me to condemn the MAF sensor.

After installing the new MAF, I verified that the serial data had come back into line, which it did with WOT, 4,000 rpm pulse width rising to 8 ms and WOT, 4,000 rpm air flow rising to 50 gps. Although I cautioned the customer that the increased fuel delivery might reduce his fuel economy, he returned a week later from an extended trip reporting the Buick had delivered more power and fuel economy than it had since he had owned the vehicle.

* When interpreting serial data, keep in mind that road-testing in a diagnostic mode may change the PCM's operating parameters. If your scanner indicates that road testing shouldn't be performed in a particular diagnostic mode, then you should obtain a direct reading of circuit values by tapping into sensor or actuator circuits through a break-out box or by back-probing the PCM connector