Mass airflow sensors.

[Andrew Brownsword]

>Well, it seems to me that you could hook the MAP sensor up to your car and
>hook up a data collection system that included information about the "VAF",
>MAF and tach inputs which should give you the necessary transfer functions.
>Once this info was available it should be a simple matter to create a
>computer that would take the MAF and tach inputs and output the simulated
>VAF.  AAMOF, one could design a computer that "learns" the transfer
>functions by accepting the VAF, MAF and RPM inputs in a learn mode.  This
is
>probably why Link is tight lipped about what they do.


Yes, no doubt... but I'm still a little annoyed with them.  I offered to
sign an NDA or whatever... and all I want is the rough shape of their output
curve, i.e. "linear" "parabolic" "1/x^2".  From there I could tune using
their interface.

>The thing that makes me wonder why you would even want to do this is that
>the ECU to which the VAF is attached is calibrated to the characteristics
of
>the VAF.  It is expecting that small signal change for a given change in
>flow at high rates.  What do you expect to accomplish with this mod?
Please
>don't take this message as being a flame ... I'm working with limited
>information here ... I don't even know what kind of car we are discussing
>here.

No problem, I appreciate any help you can give.  The car is a 1993 Ford
Probe GT, 2.5L DOHC V6 with an aftermarket turbocharger.  The initial
problem is that the stock injectors are rather anemic and can't support over
about 3psi of boost using an FMU (spray pattern goes completely to kak at
>80psi).  To solve this I bought a set of high flow injectors from VENOM --
unfortunately they might be a little TOO high flow (350 cc/min, I bought
them while under the delusion that the stock rate was 220 cc/min, now I have
strong reasons to believe they are closer to 180).

So... I've reverse engineered enough of the ROM to figure out how to
recalibrate for the injector flow rate.  The main sticking point could be
getting the car to idle nicely, but I won't know that until I try and it
depends on what the minimum practical on-time for the injectors is.

Now, assuming that I get the injectors working properly, I need a plan to
get the pulse widths up when confronted by a higher airflow.  Problem #1 is
that ECU's programming has an explicit limit built into it based 1
atmosphere.  Problem #2 is that the VAF generates a voltage based on (1 /
airflow^2 )... which results in a nearly vertical slope at high airflows.
I've followed the logic from the reading of the VAF input voltage through to
the fuel logic, and there don't seem to be any other issues.  Problem #1 I
can just remove -- I'm pretty sure the reason it is there is because of the
slop in the VAF at high airflows.  Problem #2 is solved by a new airflow
meter, hence my question to the list.


>  If you were talking about reprogramming the maps in the ECU then this
>sort of swap might make a lot more sense ... and might make it possible to
>simplify things.  Many ECUs already have an independent RPM input that
>becomes one of the input variables to the fuel/timing maps.  If you know
how
>the inputs are processed to access the maps it seems to me that it should
be
>completely possible to hook the MAP sensor directly to the input to the ECU
>and account for the differences in the behavior of the two signals by
simply
>programming the map appropriately.  You will probably still need to do some
>experimentation to determine the different behavior of the VAF vs MAP
>sensors.


The RPM is factored into the equation in the ECU because the VAF measures
CFM and the ECU wants essentially "airmass per revolution".  I do want to
simplify as much as possible, and ROM space is limited... which is why I
have essentially ruled out using a MAP sensor directly.  A speed/density
system needs to have a pressure/RPM map, and I don't have the space to
implement that, nor the debugging facilities to make it work!  A MAF sensor,
on the other hand, generates an input similar in nature to the VAF -- except
that the curve is replaced by some kind of almost-linear ascending function
that already incorporates air temperature and humidity.

Whew.

So... now that I've given you the whole spiel, do you have any thoughts?  Am
I completely insane, or is this just the kind of challenge you guys love.
:)


  Andrew


PS:  we've been around and around and around on getting the fueling on this
car right, and this is the best direction we've got.  Aftermarket parts for
it are fairly rare, and a replacement ECU is very expensive and doesn't have
all the nice features of the stock ECU (i.e. emissions control and 1 O2
sensor feedback loop per bank of 3 cylinders).




----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org