MAF theory and facts
by: Richard MacCutcheon
The following is a compilation of information about MAF
sensors and their function from mostly WWW and mail list information.
It is biased towards the Ford style two wire MAF used on mostly any Ford
since the early 1990s gasoline engines. The MAF sensors used on many of
today's cars primarily use the two-wire, hot/cold wire, setup to detect
the mass of the air passing through it. These wires are designed to give
a feedback voltage to the engine control system (EEC, PCM, ECM, whatever)
using an electronic function wired into the sensor. Many misconceptions
have arisen in the realm of modifying the MAF to gain power on the vehicle.
Here are some facts and information to help you better understand the
use and function of the system.MAF electronics, black areas are carbon
ink resistors that are grooved to tweak the outputs.The MAF is a sensor.
It cannot take an active role in the operation of your vehicle, which
is what the PCM does. The PCM uses all of the various inputs and runs
the engine according to its program. The MAF transfer function not only
tells the computer how much air is coming into the motor, but the function
is used to calculate load using other inputs which profiles spark and
fuel. As far as sensors go, it is a somewhat complex unit. It basically
uses two wires within its body to create a current draw and output based
on the different characteristics of the wires. But the way these wires
act in the sensor body itself can be somewhat problematic.
The sensor wire at high magnification. Ceramic bobbin
with wire wound coated with glass material.First, there is the flow of
air entering the sensor. The air stream can have many influencing characteristics
that can affect the sensor's ability to give accurate output. Turbulence
and velocity variations can give false output based on the original flow
profile of the sensor. So let's talk about the design for a second. When
the engineers at the factory incorporate the sensor into an intake system,
they must not assume it to be a perfect flow of air. The air filter, air
box, sensor placement and many other factors affect the true sensing capability
of the sensor. This is why when a MAF transfer function, the data stored
in the computer about the sensor, is derived for one vehicle it may not
be identical to the transfer function for a similar vehicle with the same
exact sensor. The load calculations will also vary. Likewise, the downstream
inputs that can affect the sensor vary from one design to another.
A running engine creates impulses that flow upstream,
intake growl noise, that are waves of air that are now impinging on the
sensor from behind. In an effort to eliminate the effect of various intake
feedbacks the designers use a backflow-preventor in the form of a bar
across the MAF sensor cavity using a backing plate to deflect these impulses
away from the sensor elements. When the engineers develop the transfer
function for a system they measure the flow of the sensor in a real-life
Designers actually use a laser Doppler system to precisely
measure the flow characteristics. Designing a single basic design that
can be used across many applications is the goal so costs can be minimized.
Suffice it to say that a MAF from identical cars will not be identical.
The variations in the electronics that drive the sensor have some influence
on the way the car runs. As a model year progresses other variables, like
part lot numbers and such can cause one car to run much better than the
next. If all of the variation of the sensors and devices like injectors
are taken into account, you could have a marginal car to begin with. This
is why some people see great gains with slight mods and others see nothing
or it gets worse.
Modifying the MAF has risks as well as potential benefits.
Changing the flow characteristics can cause great problems. First, let's
look at the post removal issue. Removing the post does increase the flow
capacity of the meter, but now has the potential for noise from the intake
causing other problems. This is more important at idle and low throttle
settings. Why? The MAF transfer function is not linear.
The function is flatter at low flows and increases at
an increasing rate as flow increases. If there is airflow noise at the
flatter portion of the curve, the noise voltage is a much greater percentage
of the total voltage being sent to the computer and the computer may balk.
Thus you end up with rough or poor idle. At high flows the noise can be
greater but the curve is exponential and the noise is less influential.
Also, the velocity of the air in the sensor with no post
has dropped at the same load. This affects the load calculation that causes
the injector pulses to be off the intended design. They may be better
they may be worse. Cars with closed loop operation capabilities can detect
the lean or rich condition caused by the change and make adjustments on
the fly. Car may start rough but smooth out as you drive. As you drive
in the Wide open throttle condition, open loop, the car is now only relying
on the data stored in the computer and modifications, which are not "good"
or beneficial, will now affect the performance of the motor. The one perceived
increase in performance has to do with the fact that the original programming
of the PCM is on the rich side, extra fuel for reduced wear and tear on
This modification effectively leans out the mixture and
provides more efficient combustion. Over the long run though, the PCM
will use it's adaptive capabilities to make the mixture correct as read
by the Oxygen sensors. But having changed the flow characteristics of
the sensor, the flow across the elements is also changed, possible reduced.
Thus the voltage would be lower and the effective flow would cause even
leaner conditions. Also, load calculations will probably be off their
original curve and injector pulses may be affected. One trick to provide
for the change in flow of the no-post MAF is to make a proportional change
in the sample tube that contains the sensor wires. Assuming you get the
hole size just right, the transfer function of the MAF is still not likely
to match that of the original function map in the computer.
The curve could be steeper sooner or flatter later or
shifted completely. At a given MAF voltage the air coming into the meter
is known through the transfer function table. Knowing that most people
drive around in closed loop situations these variations are quenched by
other inputs. Just unplug the MAF all together and you will see that it
will run however poorly, but it is not dead. The problem with the sample
tube issue is that the overall flow characteristics of the MAF are just
different than before. If you were to compare the two curves against each
other, they will be very close. And it seems for the most part that the
increased airflow, especially at WOT, is of greater benefit than the slight
miscalibration of the sensor.
What has to take place is the voltage should still represent
a given flow, the modification just shifts the curve to provide flows
at lower throttle levels. Again this can affect load. Other modifications
can affect even a stock sensor. The airbox, inlet tube and intake tube
to the throttle body will cause the system as a whole to be slightly different.
Let's not get too freaked out though. Again, the benefits of these mods
are often marginally good. One key thing to remember is that the closer
the MAF is to the throttle body the higher the effect the intake feedback
has on the meter's accuracy. The fewer restrictions you have between these
parts also can cause problems, usually rough idle and poor performance
at low throttle positions.
Ideally get the MAF farther away from the throttle body.
For the greatest benefit from any modification to the MAF, its flow characteristics
must be input into the transfer function table in the PCM. Recalibrating
the MAF for larger injectors works somewhat but load calculations will
be wrong. Most MAFs are limited by their size as to how much they can
flow, but even then you can peg the electronics. Adding a supercharger
can cause the meter to not function to its full potential if flows above
the transfer function table range occur. The sample tube size can be adjusted
to give a broader range than the .5v to 5.0v the current Ford sensors
have. Below are two side by side lists of MAF transfer function for two
different setups on the 1994 Mustang GT. The first number in the parenthesis
is the voltage and the second is the airflow in KG/hr.
Notice that there is a maximum flow of 932 kg/hr on the
first and 882 kg/hr on the second. Model year changes or different options
in the system caused the difference. Also, the curve for the first one
is very similar to the second all the way up to 4.0v and then the first
one climbs rapidly. So even though these are the same model and year,
just a MAF swap will cause a discrepancy in the way WOT is computed.
This swap is a full swap. The sensor bodies have been
very standardized. Swapping just the bodies wouldn't make as much a difference
as swapping the electronics. Put the first one in the second car and at
the high end the car will get a voltage that represents a flowrate less
than the actual flow which is higher, it will run leaner and possibly
cause detonation. Now these are just examples and to specifically say
what will really happen depends on the other variables. Dirty sensor elements
gives less voltage at same flow. Dirty injectors may cause a lean condition
as well as fuel pressure and volume.
# Mass Air Transfer Function
( 15.9998, 932.145 ) ( 15.9998, 882.085 )
( 5, 932.145 ) ( 5, 882.085 )
( 4.75, 808.577 ) ( 4.6001, 717.327 )
( 4.5, 697.683 ) ( 4.19995, 568.729 )
( 4.25, 598.512 ) ( 3.80005, 443.577 )
( 4, 510.114 ) ( 3.5, 362.466 )
( 3.80005, 446.745 ) ( 3.30005, 313.989 )
( 3.6001, 389.397 ) ( 3.1001, 270.265 )
( 3.3999, 337.118 ) ( 2.8999, 230.66 )
( 3.19995, 290.86 ) ( 2.69995, 195.491 )
( 3, 249.037 ) ( 2.5, 163.173 )
( 2.80005, 211.65 ) ( 2.3999, 148.281 )
( 2.6001, 178.381 ) ( 2.30005, 134.974 )
( 2.3999, 149.232 ) ( 2.19995, 122.301 )
( 2.19995, 123.568 ) ( 2.1001, 110.894 )
( 2.1001, 112.162 ) ( 2, 100.122 )
( 2, 101.389 ) ( 1.8999, 89.9828 )
( 1.8999, 91.2501 ) ( 1.80005, 80.7944 )
( 1.80005, 82.0617 ) ( 1.69995, 72.2397 )
( 1.6001, 65.2692 ) ( 1.6001, 64.3187 )
( 1.5, 57.9818 ) ( 1.5, 57.0313 )
( 1.30005, 44.9914 ) ( 1.3999, 50.3777 )
( 1.19995, 39.2882 ) ( 1.30005, 44.3577 )
( 1, 29.4662 ) ( 1.19995, 38.9714 )
( 0.75, 19.961 ) ( 1, 29.1493 )
( 0.600098, 15.2084 ) ( 0.899902, 25.0304 )
( 0.399902, 10.4557 ) ( 0.800049, 21.2283 )
( 0, 8.87154 ) ( 0.600098, 14.5747 )
( 0.5, 11.7231 )
( 0, 11.7231 )
Above Graph of '94-'95 Mustang GT To modify the MAF sensor
is in all practicality a bad move without telling the computer that the
flow has increased via the transfer function table. Also, modified intakes
will not be as effective as presumed unless the flow changes are calculated
The best way to increase the flow to the motor and let
it know is to get the system tested or get a system that has been flow
profiled and includes the ability to program the PCM. You can't just "recalibrate"
the MAF. Load will be incorrect. This can affect the durability of the
motor especially when WOT driving is done. These are issues that you must
take into account when debating a change to the system. There are some
slight risks involved and you'll have to decide.I hope this information
proves useful. Any discrepancies are my own and you can e-mail me about
it. These are only the facts presented as I see them with the information
I have gathered.