Suppose you were in the business of selling air. You could calculate the amount delivered by using a complicated formula that takes pressure, temperature, and the diameter of the valve into account, but that would be slow, unwieldy, and probably not very accurate.

You'd improve the efficiency of your operation considerably if you had a way of directly measuring the volume flowing into your customers' air canisters in cubic feet per minute. That would sure speed things up, but it wouldn't be perfect, either. On cold days, they'd be getting a big bargain because any gas is denser when it's cold than when it's hot -- in essence, you'd be giving them a baker's dozen and then some. Sales would go way up (and profits way down) in winter as soon as your patrons realized that fact of physical reality. And they'd always try to buy at sea level in dry weather because altitude and humidity affect density, too.

To keep from losing your shirt, you'd still have to use a formula that varied the price per cubic foot according to temperature, altitude, and moisture content. Wouldn't it be a lot easier, faster, and more accurate if you had a meter that read out in the actual mass instead of just the volume? Sure it would.

Historical perspective
And that parallels the evolution of EFI. While the mass of air being ingested by an engine can be calculated electronically from rpm, vacuum, throttle position, and intake air temperature sensor input (that's how speed-density systems do it), a direct reading aids swiftness and accuracy, which are both critical if optimum performance, high fuel efficiency, and low emissions are to be achieved. But since air/fuel ratios are by weight (stoichiometric is 14.7 lbs. of air to one lb. of gasoline -- in gallons it would be about 2,000 to one), measuring mass makes even more sense, so for over a decade we've had sensors that do just that.

The original Robert Bosch D-Jetronic EFI that appeared on Volkswagens way back in 1968 used a vacuum sensor (the "D" is for "Druck," which means "pressure" in German) to inform the electronics about the intake situation, but in '74 the rotating vane-type air flow meter showed up as the basis for L-Jetronic (the L stands for "Luftmengenmessung" auf Deutsch, meaning "air flow management"), also known as AFC (Air Flow Controlled) fuel injection. The jump to direct mass measurement occurred in 1984 when LH-Jetronic with its hot-wire sensor was introduced (to continue our foreign language lesson, the "H" stands for "Heisz," German for "hot"). Then came the GM hot film MAF (Mass Air Flow) sensor and other configurations from such outfits as Hitachi and Mitsubishi that may be of the wire, film, or Karman-Vortex variety.

Regardless of who makes it, if you see a sensor between the air cleaner and the throttle body (or in the case of some Mitsubishis and Hyundais, inside the filter housing) you'll know the electronics rely on direct volume or mass measurement (well, there is at least one exception: the AC Rochester speed-density Central Port Injection found on the Chevy/GMC 4.3 Vortec V6, which has the MAP sensor at the manifold inlet).

Flaps, wires, film, and vortices
Before I get to symptoms, I'd better quickly explain how the different types work. Vane air flow meters provide analog input to the PCM on the volume of air entering the engine by means of a flap that rotates against spring pressure as more and more air is ingested. This rotation turns a variable resistor that alters the amount of reference voltage returning to the computer. In most specimens, the greater the air flow, the higher the resistance, and the lower the signal, but some work just the opposite. Usually, an air temperature sensor of the NTC (Negative Temperature Coefficient) variety is included in the assembly. I should mention a variation on this theme that I've seen on certain Mazdas: a cone type sensor wherein the linear movement of the inner cone against spring pressure alters the position of the variable resistor.

Regardless of the type, a mass air flow sensor has some other advantages besides its ability to account for density -- speed, lack of moving parts and restriction, and the elimination of compensating sensors. A typical MAF has a wire or film element that's kept heated to a specified temperature above ambient (180 deg. F. in Bosch units) and is exposed to intake air. Through a Wheatstone bridge circuit and dedicated electronics, the amount of current required to maintain that temperature becomes the signal to the computer. High air flow obviously has a greater cooling effect than low, but so does the denser air of cold days and low altitudes, so the PCM gets the true data on mass it needs to provide the longer injector pulse width that extra oxygen needs to fire dependably. Bosch versions and GM 5.0/5.7L V8 units produce an analog voltage output, but Hitachi and most AC Delco MAF's send out a square-wave frequency.

Upscale applications of the Karman-Vortex air mass measurement principle can be found on the Lexus LS 400 and '87 and up Toyota Supras. The speed of vibration of a thin metal mirror in the intake vortex is monitored optically to generate a frequency signal that varies according to the mass of air entering the engine. Various Mitsubishi units use sound waves to exploit the vortex idea.

Out of whack
A variety of symptoms can be expected when one of these sensors or its wiring or ducting goes awry. GM says the engine will crank up and die. Bosch talks about starting problems both hot and cold, hesitation, stalling (especially under load), rough idle, and low power output. Nissan gives stalling, poor idle, black smoke, and switching to the fail-safe mode as evidence of air flow meter problems (in some models, this mode will be manifested by the inability to exceed 2,000 rpm). Generally, contamination of the sensing element, which slows response, will result in stumble.

Regardless of those corporate opinions, the most logical effect of a bad signal (or no signal) is stalling, sagging, or missing at transient throttle. If this is far out of range, it may cause the computer to go into its limp-in mode, also known as LOS for "Limited Operating Strategy." In other words, driveability and performance will be pretty bad.

Diagnosis isn't so easy because lots of other things can cause those same symptoms -- problems with ignition, compression, fuel pressure, etc. Something that's often overlooked is ripped duct between the sensor and the throttle body, which admits unmeasured or "false" air and leans out the mixture. A PCV system with a valve that's stuck wide open or a hose that's cracked can do the same thing, and a plugged air filter can mean trouble, too. So, at the risk of repeating what I've preached for years, don't automatically blame the engine management electronics or fuel injection set-up just because it's there. Actually, the trouble is most likely elsewhere because these systems are generally quite dependable.

That's not to say you'll never encounter a failed air flow/mass sensor. For instance, older GM MAF's have a poor reliability record (the higher-frequency 10kHz Hitachi unit used on late model GM cars has a much lower failure rate).

Codes and beyond
In cases where the basics have checked out and you suspect the EFI, use the OBD (On-Board Diagnostic) function as your first step in determining if the air flow/mass meter is the culprit.

Until OBD-II regulations take affect, fault codes will vary from make to make, so you'll need the specific service info for the car you're working on. I'll give you a couple of examples, anyway. On a Ford EEC-IV, Code 26 tells you that the VAF (Vane Air Flow) or MAF is out of self-test range, Code 56 means the signal is above max voltage, and Code 66 indicates signal voltage below minimum. Hyundai, on the other hand, has just one code for the air flow sensor -- 12 -- and the manual says that if you see that code and the harness and connector are both okay, go ahead and replace the sensor.

Relying on codes alone can get you into trouble. They should just point your diagnostic efforts in a general direction. Don't consider them the final word on what's wrong. You'll need to do some more troubleshooting and rely on your experience before you can be sure.

Scan tools are expensive, but they sure streamline troubleshooting. Errors in sensor calibration are magnified as air flow increases, so being able to test during road load can be helpful. In some cases, both the air flow value the PCM is using and the actual signal from the sensor can be displayed. If these two numbers don't match, the computer is probably reverting to a substitute value from memory because the MAF info is faulty. By the way, GM units should produce a signal that results in a reading of 4-7 grams per second at idle, or 100-240 gps at WOT (naturally, depending on the displacement and horsepower of the particular engine).

You're so vane
On the vane type, reach inside the air box and move the flap through its range by hand. You should feel no binding or roughness. If it's a version that incorporates a fuel pump switch, make sure you hear the pump start when you push the vane (key on).

Look for reference voltage input (that's 5V) at the connector. Switch to the output contact and you should see the reading change smoothly as you push the flap to the fully-open position (in most cases, this will be a falling reading, but on Ford EEC-IV units it will rise -- look for .25V closed and 4.50V open). Some carmakers give you direct resistance specs, too. Just as with a TPS (Throttle Position Sensor), you're checking the condition of the resistive strip or track, and any jumps in either voltage or ohms readings mean it's time for a new sensor.

If you're using a DSO (Digital Storage Oscilloscope, also known as a lab scope), set it for two volts and 200 milliseconds per division, then tap into the sensor's signal wire. Any spikes or jagged areas in the pattern as you move the vane are cause for replacement.

With a Bosch hot-wire air mass sensor, make sure you have battery voltage at the appropriate terminal, then measure output. A reading of 2V at idle that rises to almost 3V at 3,500 rpm is typical. If you blow compressed air through it, you should see the voltage change.

Beating on it, and Hertz
The basic GM MAF test is nothing if not straightforward: When tapped with a tool (a screwdriver, not a sledgehammer) at idle, a bad sensor will not only produce a dramatic change in frequency, it may also cause the engine to stumble or stall. This is certainly a convenient check, and one GM expert I know says it's almost 100% accurate. You might even want to try it before pulling codes.

It makes sense to follow that with another great quick check. Unplug the MAF's harness connector, then start 'er up. If the engine runs appreciably better now, the sensor's bad.

If your DMM can measure frequency, you can use that mode to check AC, Hitachi, and any other unit you run into that produces a frequency signal. Set the meter to read Hz or kHz, and connect its leads to the sensor's signal and ground wires. An ordinary AC MAF as found on a 2.8L Chevy V6 should show you about 45 Hz at 1,000 rpm and 72 Hz at 3,500, whereas the high-frequency type of a late-model 3800 will read 2.9 kHz and 5.0kHz at those same speeds. Record the readings at various rpm and compare them to specs. You should see a linear frequency rise with no dips or jumps as speed increases.

But since these are high-frequency sensors, you might not catch a glitch with your DMM, so using a lab scope makes sense. Start out by setting amplitude to 5V per division. Changing the amplitude to one volt per division will give a different view of this signal. The timebase setting varies according to the range of the MAF. Use 0.1 milliseconds per division for a 10kHz Hitachi unit, for example. You should see a square waveform with even frequency pulses. As engine speed and load is varied, the voltage and pulse width frequency should change smoothly and evenly. If you see any gaps in the pattern while tapping the sensor or driving the car, a new part is in order.

Several service bulletins have been issued on the power and burn-off relays used with '86-'88 GM 5.0L/5.7L V8's. Dig them out if you're presented with a hard start, rough running, surging, or stalling complaint.

You can measure a Toyota/Lexus Karman-Vortex signal using a lab scope set at 1V and 10 milliseconds per division. At idle, you should get a nice even rectangular waveform. Another test, which Lexus gives in the 1990 service manual, is that of resistance compared to temperature. Unplug the sensor from the harness, then check the resistance between terminals THA and E2 of the meter's connector. At 68 deg. F., you should see 2-3 ohms. This should fall to 0.9-1.3 at 104 deg., and 0.4-0.7 at 140 deg. At the other end of the scale, 10-20 ohms is specified at -4 deg., and 4-7 ohms at 32 deg.

A Logical VAF (Vane Air Flow)/MAF (Mass Air Flow) Diagnostic Sequence
I've boiled down the basic troubleshooting procedures into half a dozen steps. Follow these when stalling, hesitation, missing on transient throttle, or poor idle quality causes you to suspect the air flow/mass sensor:

Pull any trouble codes from the self-diagnostics, then erase the codes.
Check the basics -- not only compression, vacuum, fuel pressure, ignition, and wire and connection condition, but also the air-tightness of the duct between the VAF/MAF and the throttle body.
If MAF, tap the sensor to see if the idle changes. Disconnect the MAF (key off) and see if the engine runs better.
If VAF, move the flap to feel for binding, roughness.
VAF or MAF, use a scan tool, lab scope, or DMM to catch any steps or jumps in the grams-per-second reading, or voltage or frequency signal.
Make the repairs, clear any new codes, and take a test drive to verify that the symptoms have been eliminated.