MAFs

[Chris Conlon]

A couple tangential comments...

At 07:07 AM 2/6/01 +0100, Axel Rietschin wrote:
>From: "Bruce Plecan" <nacelp at bright.net>
>
>> How about the shear fact that if the IAT compensation is internal to the
>> MAF, then the metering allows for heat saturation of the I/C.

What's the time constant on the MAF's temperature correction? The one
*circuit diagram* I saw did temp compensation via a temp sensor (RTD
I think) as one leg of the bridge. Whatever the constant is I bet (hope!)
it's a lot better than the 20sec value Axel gives for other sensors.

>
>What comes in, comes out, no? Less temp, more density, less speed but same
>amount of _air_, or did I miss something?

Yes, you missed a small point. The location of the MAF (before or after
the compressor), in a system that has a significant volume of plumbing
after the compressor *does* make a difference. The reason being the
"air through MAF = air into engine" relationship is only true in steady
state. When the pressure post-compressor is changing, up or down, the
airflow through the MAF is not equal to flow into the engine. Some is
being either pumped up into the plumbing, or delivered from the pumped
up volume into the engine.

I've measured this effect on my supercharged MR2 and it is very real.
It's a very small lag in terms of watching a boost gauge but in terms
of throttle response it's very very noticable... if you have a
differently-responding setup to compare it to anyway.


Questions about turbulence are very valid. An OEM engineer has reported
to me that they flowtest the entire intake system and meter as a unit
to generate each ECU's MAF voltage to airflow lookup table.


>to what extent? T63 constant of automotive air temp sensors I know of (Bosch
>& Marelli) is in the 20s range, meaning it takes about a full minute to get
>an accurate reading, which rule out most 'real-time' control strategies
>based on that value.

I'm guessing that those sensors are designed for measuring ambient temp,
not post-compressor temps?

One thing you can do to greatly sharpen up the response of temp sensors
is to make a first order slope correction. Measure temp (call it C),
calculate dC/dt, and add k * dC/dt to the raw value C. Use this sum as
your "actual" temp. You'll have to do some futzing to get k, but a couple
of datalogs plus a heat gun should get you close fast.

This is lots easier if your temp sensor is a bandgap or t-couple that
reads a voltage directly proportional to absolute temperature, eg. TMP3x
series. If you're using a thermistor you want to linearize the result to
actual degrees before doing the math. The more resolution you have, the
better this works, and the faster/more accurately you can compensate the
value.

A couple of MAP sensors now come with integrated temp sensors. I doubt
their tc is very fast but they might be useful for someone.


   Chris C.

----------------------------------------------------------------------------
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