Tuning Wiki

This is a historical revision. Read the current page or return to its history.

Page organisation in this revision
Page type
Technical article
Subject
Tuning
Parent
Section level
Applies to
No ECU-specific applicability
Tool groups
None
Manufacturers
None

What MAF calibration does

The mass airflow (MAF) calibration translates the sensor signal into an estimate of air mass for the installed sensor and intake. Depending on the ECU, this estimate contributes to cylinder charge, load, fuel, ignition and torque calculations. It is not simply a fuel adjustment. A replacement sensor, housing, airbox or pipework can change the relationship between signal and flow. Identify the exact ECU software, sensor and housing before selecting a curve. See airflow metering for sensor types.

A transfer curve may express voltage or frequency against grams per second, kg/h or a scaled value. Confirm axes, units, offsets, limits and any later corrections in the actual definition. Some sensors report frequency rather than voltage, and a frequency signal is not necessarily a mass measurement: a Karman vortex meter measures volume and needs the relevant density calculation. A different sensor or housing cannot be assumed to share a curve. The Haltech MAF documentation describes voltage and frequency inputs and sensor curves.

Examples across ECU strategies

SystemCalibration approach and qualification
Bosch ME7 petrolIn the documented ME7 examples MLHFM converts the MAF signal; MLOFS and additional correction functions affect the result. The effective air mass also affects load. Verify the exact software and calculation order before scaling; see ME7 fueling and MAF calibration.
GM example using frequencySome GM applications use airflow versus MAF frequency. EFILive's Calc.VE tutorial shows a frequency-indexed MAF calibration. Other airflow models may participate depending on the operating state and ECU; confirm the active strategy rather than assuming MAF is the sole input.
Subaru BRZ / Toyota 86 exampleEcuTek documents a voltage-to-g/s scaling map for the specified BRZ/FRS/GT86 family. Its tuning guide discusses changes for the housing or intake. This is an application example, not a table definition for every Subaru or Toyota ECU.
Nissan 350Z VQ35 exampleEcuTek describes a multi-stage conversion on the covered ROMs: voltage to a percentage, then a maximum-flow conversion and further airflow corrections. Check bank, ROM and the actual calculation rather than assuming a single direct voltage-to-g/s table. See EcuTek's VQ35 strategy guide.
Diesel air managementAir mass can be used for air-path and EGR monitoring and fuel-quantity limiting, depending on the ECU. Diesel operation is generally lean and petrol-style stoichiometric fuel-trim scaling is not a general diesel method. Compare specified and actual air mass under the manufacturer's test conditions and diagnose the air path before changing a curve. The Ross-Tech TDI procedure provides an application-specific logging example; see also Volkswagen TDI air-mass measurement.
Standalone ECUSome support a configured voltage or frequency input and a loaded sensor curve. Haltech documents these choices and a MAF fuel-load mode; other standalone systems may instead use speed density or another load strategy. Follow the specific controller's documentation and use data for the installed sensor and housing. Haltech MAF setup.

Worked examples: calculating and applying a correction

These figures are invented to illustrate the arithmetic; they are not recommended calibration values. First establish the fuel and sensor health, ECU strategy, exact table axes, units and correction sign as described below. A mixture error does not prove a MAF error. Retain the original file, make changes only in the evidenced region, and re-log the result.

1. Subaru BRZ / Toyota 86: voltage and closed-loop trims

The EcuTek guide describes this application's voltage-to-g/s table. Suppose a stable 2.00 V bin has a table value of 30.0 g/s and an established combined fuel correction of +6%: in this example positive means the ECU adds fuel. If diagnosis supports an underestimated MAF and the applicable correction acts proportionally, the starting estimate is:

factor = 1 + 6 / 100 = 1.06
new airflow at 2.00 V = 30.0 × 1.06 = 31.8 g/s

Apply 31.8 g/s to the corresponding voltage point, review neighbouring bins using their own adequate stable samples, and check interpolation and curve shape. Do not copy 6% across the entire curve. Log again at similar conditions and check the bank and learning state. This example assumes the displayed +6% is already the effective combined correction; do not blindly add STFT and LTFT. Their sign, representation, learning regions and combination depend on the ECU and logger. See EcuTek's BRZ MAF guide.

2. GM frequency-output example: measured versus commanded lambda

For a GM application with a frequency-to-airflow table, as shown in the EFILive Calc.VE tutorial, suppose the table gives 200 g/s at 8,000 Hz. During a stable point the final commanded lambda is 0.80 and a calibrated, time-aligned wideband reads 0.84. Assume the MAF is the active load input, injector and pressure data are correct, feedback and learned fuel corrections are accounted for, and the error is attributable to the MAF:

approximate factor = measured lambda / commanded lambda
                   = 0.84 / 0.80 = 1.05
candidate airflow  = 200 × 1.05 = 210 g/s at 8,000 Hz

The mixture is 5% leaner in this simplified ratio; the candidate adds 5% to reported air and consequently calculated fuel under the stated model. Apply only after confirming the table and active airflow model for that operating system. Inspect neighbouring frequency bins and re-test with a wideband, fuel pressure and load monitoring. If lambda becomes dangerously lean, stop the test and diagnose; do not continue a loaded pull to populate a histogram. The same calculation can suggest a reduction: 0.76 / 0.80 = 0.95, or 5% less, under the same assumptions.

3. Bosch ME7: include the MLOFS offset

The ME7 guide describes a documented relationship in which effective air mass equals MLHFM minus MLOFS. Suppose a particular point has MLHFM 120 units, MLOFS 10 units, and a justified factor of 1.05. Use the actual definition's units and confirm that this relationship applies to the software:

old effective air mass = 120 − 10 = 110
new effective air mass = 110 × 1.05 = 115.5
new MLHFM value        = 115.5 + 10 = 125.5

Write the calculated 125.5, rounded according to the editor's precision, into the applicable MLHFM point; retain MLOFS in this example. Directly multiplying 120 by 1.05 gives 126 and changes the offset's contribution as well. This arithmetic illustrates the offset, not a claim that every ME7 has a value of 10 or that every correction is applied at this stage. Verify calculation order, corrections and load effects, then re-log. The MxT ME7 guide documents one software-specific way to group and analyse trim data.

4. Nissan 350Z VQ35: a percentage stage

EcuTek's VQ35 guide describes voltage-to-percentage and percentage-to-g/s stages on covered ROMs, followed by further corrections. For arithmetic only, suppose a particular point reads 40% and a separately established 1.05 factor is appropriate at that stage. The candidate is 40% × 1.05 = 42%. This does not by itself establish the final g/s: trace the subsequent conversion, bank and corrections, and confirm the editing tool's representation and recommended adjustment location. Do not assume a percentage table is a direct airflow table or change a shared maximum-flow scalar to fix one local bin.

5. Standalone ECU: choosing the correction location

Haltech documents a configurable sensor curve and, in MAF tuning mode, a Fuel Base correction factor that starts at 1.000. Suppose a valid MAF curve point reads 50.0 g/s and an independent flow reference for the same installed sensor and housing establishes 52.5 g/s: 52.5 / 50.0 = 1.05, so the candidate sensor-curve point is 52.5 g/s. Alternatively, a base-fuel factor of 1.050 would apply a 5% correction in the documented fuel calculation, but the resulting effects and conditions are not necessarily equivalent throughout the ECU. Do not apply both 1.05 changes for the same error: 1.05 × 1.05 = 1.1025, or 10.25% combined. Use the appropriate correction location and verify the resulting fuel and load behaviour. See Haltech MAF setup and Fuel Base documentation.

6. Diesel: use a known flow reference

Suppose an independent, appropriate flow-bench reference measures 440 kg/h for the complete sensor and housing at a stable signal, while the ECU conversion at that signal is 400 kg/h. If the definition is a direct, proportional flow value at that signal, 440 / 400 = 1.10, giving a candidate of 440 kg/h. Repeat at multiple flow points and verify the sensor, units, setup and reproducibility before changing the curve. In a vehicle, a requested 440 and measured 400 kg/h are not an independent 1.10 calibration factor: EGR, boost, leaks, restrictions and control conditions can explain the difference. Follow the vehicle-specific test, such as the Ross-Tech TDI logging procedure, and retain emissions and diagnostic functions.

General validation workflow

  1. Record the original calibration, ECU identification, sensor part number and intake arrangement. Establish whether MAF, speed density or a blend is active in each test region. A speed-density estimate is another model and is useful for comparison only when independently validated.
  2. Correct faults first: unmetered-air and exhaust leaks, sensor supply and grounds, contamination or damage, fuel pressure, injector characterization, fuel composition and lambda sensor or logger errors can all distort the result. Check both banks separately where available.
  3. Use a documented transfer curve for the complete sensor and housing as the starting point. A bore-area ratio is only a rough starting estimate: placement, bends, screens, pulsation and flow distribution matter. Check signal headroom at idle and peak flow; a saturated sensor cannot reveal additional airflow.
  4. At stable, warm closed-loop petrol operating points, log raw voltage or frequency, calculated airflow, RPM, load, requested and measured lambda, short- and long-term corrections and control state. Group samples by signal, reject transients and allow for sensor and exhaust delay. Interpret correction signs and the way learned and immediate trims combine for that ECU; a trim is evidence of a mixture discrepancy, not proof that the MAF caused it.
  5. For regions where feedback is absent or cannot measure the commanded rich mixture, compare final requested lambda with a calibrated, suitably installed wideband while monitoring fuel pressure and engine safety. Feedback and learned corrections can mask an error; establish their actual state and follow the ECU-specific procedure. Do not infer high-load accuracy from idle trims.
  6. Make small, justified changes in well-sampled regions, preserve a smooth physically plausible curve, and re-log. Under a simplified open-loop petrol model with correct injector and pressure data and no active fuel correction, measured lambda divided by commanded lambda is an approximate multiplier for the reported airflow: for example 1.05 / 1.00 suggests about 5% more reported air. This is a diagnostic approximation, not a universal formula: offsets, filtering, blending and ECU correction conventions can change how the stored table must be edited. Investigate large or inconsistent errors before applying it.
  7. Restore any temporary test settings and retain diagnostic and engine-protection functions. Verify idle, part load, transitions and safely controlled higher load, both banks, fuel trims, final lambda, calculated load, plausibility and headroom. Recheck under representative temperatures and conditions. Do not fill unmeasured regions with invented certainty.

See ME7 MAF and lambda log review for a worked logging approach and EcuTek's BRZ intake rescaling guide for its specific MAF and speed-density comparison. Modifying emissions-related calibrations may be restricted by local rules; retain applicable monitoring and verify the requirements for the vehicle's use.