Why the vane meter is replaced
A vane air-flow meter measures volume with a spring-loaded door and a wiper on a resistive track. A mass-air sensor measures the mass directly, with a heated wire or film, and it has a clear bore instead of the door, the compensation plate and the sharp edges inside a vane meter. On a naturally aspirated engine the larger, smoother bore is worth power: one 2.5 litre BMW E30 M3 with 284/276 camshafts and a freer exhaust gained 14 hp when its vane meter was replaced by a larger mass-air sensor, and the torque curve sat higher across the rev range. On that same comparison the vane-meter run was near 14:1 for most of the pull, and the mass-air run was close to the 12.5:1 full-load target.
The control unit still expects the old voltage. Three jobs follow from that. The mass-air voltage has to be reshaped so it looks like the vane-meter voltage at each flow. The mass-air sensor answers faster than a door, so the converted signal has to be slowed to the door’s response. And because a mass-air sensor follows density, a climb in altitude moves its reading much more than it moved the vane meter, so the calibrator has to compensate for elevation.
Which vane meter is fitted
Early meters run from battery voltage and swing from 0 V toward about 9 V. Later meters run from a 5 V reference and swing 0 to 5 V. Some swing the other way, 5 V at rest falling toward 0 V as the door opens. The calibrator has to match that direction: one arrangement for 0–5 V, one for an inverted 5–0 V, and one for 0–9 V.
Identify the signal with the intake duct off. Ignition on, engine stopped, negative probe of a DC voltmeter on chassis ground. Watch each wire while the door is pushed open and shut by hand. One wire changes. Record it fully closed and fully open. That pair of voltages is the range the converted signal must stay inside.
Tracks wear. A groove in the resistive element makes the voltage drop out at particular door angles, which is a flat spot, a stutter, a loss of power, or a failed emissions test. The mass-air conversion removes that track. It also removes the temperature sensor that lived in the vane meter. A calibrator can substitute a fixed temperature on a spare wire, which is enough to run. Cold start and warm-up improve if a real intake-air temperature sensor is wired in its place.
Fitting the mass-air sensor
The sensing element looks at a sample of the bore, through a sampling tube, and infers the total flow. That inference holds while the flow is smooth. Turbulence makes the sample unrepresentative and the engine undriveable. Mount the sensor on an airbox, or feed it with at least 9 inches of tube. A cone filter clamped on the inlet is the arrangement that produces the turbulence. The rest of the intake is worth opening at the same time: many airboxes breathe through a small tube whose only job is to keep intake noise down, and that tube can be replaced with a larger one. On the E30 M3 the elbow between the meter and the plenum was opened up along with the meter.
Mass-air sensors run from 12 V and regulate internally, so the reading holds as battery voltage moves. Outputs in general use are 0–5 V, 0–5 mA or 0–10 kHz. The sensors used for this conversion are 0–5 V, and with enough flow they will go above 5 V. Do not drop the sensor. Oil from an over-oiled reusable filter coats the element. A cleaner made for these sensors will remove it. A general electrical cleaner will not.
Put the calibrator somewhere cool and dry, typically behind the dashboard or in the glovebox. Its ground goes to the vane meter’s ground, not to a convenient screw. Its 12 V feed has to stay up while the engine is cranking. Accessory circuits that drop out during cranking will make the car hard to start, because the calibrator collapses with them.
The shape that has to be matched
A vane-meter signal rises steeply from zero, so small flows at light load make a large voltage change, then flattens as the door hits its stop. In one comparison that stop is 4.8 V. A mass-air signal is straighter: voltage rises more slowly as flow rises, and it keeps rising after the vane meter has already saturated. At zero flow both are 0 V and no offset is needed. As flow rises, an increasing offset is added to the mass-air voltage so it catches the vane curve. The offset peaks in the middle of the load range, then falls, and is zero again where the mass-air voltage meets the vane meter’s 4.8 V maximum. Above that, an offset is subtracted so the control unit never sees more than the vane meter could produce. A voltage past the old maximum is read as a flow the fuel map does not contain, and the mixture goes lean.
The calibrator applies that offset from a table of 50 columns and 16 rows. Columns are mass-air voltage, 0.1 V apart, from 0 to 5 V. Rows are engine speed, 500 rpm apart, from 0 to 8000 rpm. The neutral cell value is 10.0. The offset in volts is (cell − 10) / 4. A cell of 11.0 adds 0.25 V. A cell of 14.0 adds 1 V. A cell of 11.5 adds 0.375 V. A cell of 14.3 adds 1.075 V. Start with the same value down a whole column, so the first job is 50 numbers rather than 800. Idle and wide-open throttle are the regions that later need a cell to differ from its column.
Closed loop and open loop
With a narrowband sensor the control unit trims injection to 14.7:1 at idle, cruise and moderate acceleration, and it stores that trim. The memory is volatile: disconnecting the battery or unplugging the control unit for 10 minutes clears it. Clear it before tuning, or the old trim fights the new curve. A scan tool shows the trim as a multiplier. 1.00 is neutral, above 1.00 the control unit is adding fuel, below 1.00 it is removing fuel. Manufacturer-specific diagnostic plugs, round or rectangular and usually under a cap in the engine bay, cover the late 1980s through 1995. From 1996 the plug is the standardised one.
Above the load where the control unit wants enrichment it leaves closed loop. About 12.5:1 is the high-load mixture. The changeover can be seen as the enrichment flag on a scan tool, or as the wide-open-throttle switch on the throttle sensor. Tuning the light-load region to 14.7:1 is most of the driving the car will do. A wideband meter in a bung ahead of the catalyst is the measurement. A tailpipe clamp usually agrees within 0.3 AFR. A narrowband-based meter can watch the closed-loop dither and can sketch the rich region, and the rich region is then checked with a wideband, because a narrowband is only accurate near 14.7:1.
Street, then the dyno
The calibrator software used for this conversion does not tune while the engine is running. Change the table, stop the engine, write the file, restart, and look again. There is no separate save command: closing the file writes it. Start the next session by reading the calibrator back into a new file. Note the fuel pressure in the file if the regulator is adjustable.
Warm the engine. Set the idle columns, usually 0.8 to 1.2 V, for 14.7:1. A surge at idle can be a low cell in that region, or a large step between neighbouring idle columns. Holding the same cell across several idle columns settles it. Free-rev to walk the active cell across the next columns and set those to 14.7:1 as well. On the road, two people: one drives, one watches the mixture and the highlighted cell. Stay out of traffic. A light cruise in third at about 30 mph lands near 1.5 V. If that cell reads 16:1, raise the cells there. The mixture will not change until the file is written, so collect several columns, up to about half throttle, then pull over and write. Fill gaps with the autofill once two columns five apart are both correct. The whole range up to about two-thirds throttle stays at 14.7:1. Above that, set 12.5:1, and keep the output under the old vane-meter maximum, about 4.8 V. Past that voltage the mixture leans out, including at high rpm.
One worked map starts at 10.0 in the 0 V column, so zero in is zero out, then rises. It sits on a plateau of 11.5 through the 0.9, 1.0 and 1.1 V columns for a stable idle, peaks at 14.3, and from the 3.6 V column trends back down. Above 4 V the cell falls by 0.4 per column, which subtracts 0.1 V for each extra 0.1 V of input and holds the output just under 5 V.
If the narrowband sensor was unplugged for open-loop tuning, plug it back in and drive long enough for the trim to settle. The dither should spend equal time rich and lean, and it should speed up as the throttle moves from a light cruise toward the enrichment point. A dither that sticks to one side is not finished.
The dyno comes after the street curve is complete. Use it to flatten wide-open throttle across rpm. Record the pull so rpm on the table lines up with the measured AFR. At high rpm a pull often stays inside a narrow band of voltage columns, so the remaining correction is by row, down the rpm axis, rather than by column.