FW: basic fuel metering equations

Dale Ulan ulan at ee.ualberta.ca
Sat Jul 2 00:23:39 GMT 1994


> Or how to do a look up table.  This way the project can be broken up into
> manageable chunks.  Below is my first pass at explaining my understanding of
> fuel injection systems.  Not as complete as yours, and mabe not as clear, but
> from a slightly different view point.  Damn!  I can't figure out how to 'cut
> and paste' text in Microsoft Mail, so I'll have to wait until tuesday to post
...
> do is scale the MAP and temp sensors to output 0-5V in the range of interest
>(simple inst. amp circuit for gain and offset, can be done in one quad op-amp).

Most automotive sensors are designed to run off of +5 volts or so and produce
a single-ended 0-5 volt analog signal. You can normally take this signal, run
it through a simple R-C right to the A-D port. The only sticky point is the
engine temp sensor (or other thermistors). They need a pullup resistor which
depends on the manufacturer. I have the calibration for a GM sensor with a
1k pullup on it. It's at home, though, and I'm no there...

>  The software black box will take MAP and Temp numbers from the A/D converter
> (of the range of 0->255) and convert them to a air density number of the range
> of 0->255 (arbatrary units).  The reson for not using calibrated units is to
> keep maximum resolution in the minimum # of bits.  It would be wastefull to
> measure temperature over the range of -50 -> 300 deg. F just because we can. 
> Similar with MAP.  The range of values I see for MAP/Temp are different for my
> engine/altitude than it is for your engine/sea level.

Normally, the range you see is no problem for anywhere in the world. MAP
will usually be between 110 kPa and 20 kPa. An arbitrary assignment of
2 units per kPa linearizes to 0 kPa to 127.5 kPa, which is quite reasonable
for natural aspiration. This can be trimmed in software quite easily.
For turbo/blower applications... you be the judge. Just scale everything
to be bigger.... :-)

>   The other way I see of doing things is the software scaling method.  What we
>can do is use a 12 bit A/D converter and measure the entire useful range of the
>sensors (-50 -> 300 deg F and all).  This 12 bit number can be scaled down to a

If you use automotive sensors, they have been designed with the correct
operating ranges already. Go with them! eg: Temp  -40C to +150C, MAP 20 to 110,
etc.

Hmmm... good idea, but 12 bits in a car??? I dunno. Maybe 10 if you're lucky.
If you're deciding on am MCU with on-board A-D, 8 bits is what you normally
get in a package, and I haven't had much luck trying to go more accurately
without expensive filtering. As it is, an 8-bit converter picks up enough noise
that it has to be filtered in software. This is especially true of O2, which
is a high-impedance signal. Also, MAP contains a lot of engine pulsations,
which must be filtered out intelligently to avoid aliasing effects but also
avoiding long lag times. Just filtering MAP is not trivial. Same with fast
MAF sensors, so that's not an easy way out... 8 bits is a nice trade-off,
as it also fits in half the memory that any other bit count takes...


-Dale



More information about the Diy_efi mailing list