Make your own mechanical AFC

[ciper]

I came up with this idea when I planned to put large injectors in a vehicle and not use fuel management. I wanted to get everyones opinion.

Ill write it simple to start.

You will need a ball cock valve and some hoses plus fittings to fit the hoses.

Make a hose fitting on the air box on the FILTERED side just before the AFM.
Make a hose fitting on the intake pipe AFTER the AFM.
Install the ball cock in between and adjust the amount of unmetered air that is passed.

I got this idea from meters that actually have a bypass screw built in! When I owned a Supra a custom made bypass screw was even available.

As you open the bypass it will lean the mixture out. The goal is to run large injectors on the stock engine computer that will idle and drive normally and will continue to function at high RPM/LOAD

[vrg3]

Done right, that would work. As you say, many stock airflow meters have valves built in for this. On late 80s/early 90s Mazda 626/Mazda MX-6/Ford Probes, there's actually a plug you can drill out to expose an adjustable needle valve underneath. :)

The only concern I would have is whether this would decrease the airflow reading by the same fraction at all airflows. Given the design of Subaru's intake system, it seems like that would be hard to guarantee with an external bypass.

It would be really easy and cheap to electronically scale down the MAF signal, too... A very simple sub-unit gain amplifier (simpler than the FCD!). But one problem there is the MAF sensor's output is not linear with voltage:

http://www.graphics.cornell.edu/~v/pics.cgi/aa694.gif

I can't say for sure if it would be a big deal, though, since something this ghetto would never work 100% anyway.

Seems like either one would be easy enough to try...

[vrg3]

Wow!

I got curious, so I wanted to see how much error you'd get if you just divided the input voltage by a constant value.

For various divisors, I computed the voltage signal in and out you'd need to divide the actual measured airflow by the divisor. For example, for a divisor of 2, I figured out the relationship between MAF signal and the signal for half the airflow.

I was surprised by what came out:


So, if I got this all right, simply dividing the signal should work pretty well! You'd just use this table to look up how much to divide the voltage by:



(This is all based on the Hitachi MAF sensor's transfer function, which seemed reasonable since most people who want bigger injectors have Legacy Turbos, which all use Hitachi sensors.)

[vrg3]

I've been thinking about this some more, and I'm no longer sure that either method is a great idea... Looking again at the ignition maps, you can see that at higher loads the maximum permissible ignition advance is actually less than the base ignition advance at lower loads. What that means is that if you trick the ECU into thinking there's a lot less air coming in than there really is, it may advance the timing too much. The knock sensor would still be active, but timing control outside the maps is not very sophisticated.

Maybe if you were sure that more advance was okay (if you went very far in improving knock resistance) or if you had other ignition controls it would work okay... It's possible that just using a more efficient turbo and intercooler would be enough, but I dunno.

[Dizzy]

What that means is that if you trick the ECU into thinking there's a lot less air coming in than there really is, it may advance the timing too much. The knock sensor would still be active, but timing control outside the maps is not very sophisticated.

Mmmm. Given that our cars seem to run rich, I think there is a little room for timing advance. However, trusting my motor to an ECU "thinking outside the map" doesn't give me a warm fuzzy. I would prefer to know how much advance over the stock map I was getting at a given RPM (ala ITC).

[vrg3]

An ITC doesn't let you know exactly how many degrees advance you're getting given RPM and load. The only way to specify that is to map the ECU yourself.

All an ITC does is trick the ECU into sparking a fixed amount earlier or later than it thinks it is, and the amount is determined solely by RPM, and determined by interpolation along a 5-value table. The ECU will still try to advance the timing up to its limits if the knock sensor lets it.

If you're serious about managing your engine you wouldn't be using an AFC in order to run big injectors to begin with.

[Dizzy]

Good point. I re-read my post & realized I had the "and load" attached to my mention of the ITC.