Build Your Own Fuel Injection Computer

renaissance-redneck writes "I was searching for details on assembling a fuelie setup from junkyard parts when I stumbled across this: 'MegaSquirt is an Experimental Electronic Fuel Injector Controller for Internal Combustion engines. Its main goal was to create a simple fuel-only controller that could be easily reproduced with cost-effective components.' If you've got the time and the skills, it beats spending $2000."

Comments

Interesting...

[wowbagger]

This raises an interesting question - namely, when are we going to see more hacking done on modern engine control computers?

I don't mean the current crop of "chips" that increase horsepower by overriding emissions control code, or remove the rev limiter code, I mean real, creative hacks.

For example, a friend of mine and I were discussion this possible hack:
On an engine with sequential multipoint injection (one injector per cylender, near the intake valve), could one modify the ECU so that, when the engine is under minimal load, instead of firing the injector on every intake stroke, it would fire on every other intake stroke - in effect changing a four stroke engine into an eight stroke engine?

In theory, this would allow you to run a leaner mix (leaner = hotter burning) without burning up the cylinder, and you could save fuel. When the ECU detected the engine being placed under more load, it would start firing on every intake stroke (and remap to a richer mix).

Of course, I suppose the reason we don't see this is because hotter combustion = more carbon monoxide produced instead of carbon dioxide.

Now, if I could just get anoncvs access to Ford...

Re:Interesting...

[Juggle]

Actually that has been brought up already on the MegaSquirt list - and Chevy is doing something similar on a vehicle in the next year or two with an engine that dynamically changes from 6 to 4 cylinders. It was done before (by Cadillac IIRC) but didn't work out well because the technology to pull it off just wasn't available.

Oh - and it's also a technique used in some motors alreeady. The Ford Focus will do exactly what you described if it starts to overheat in an attempt to "Limp home".

Re:Interesting...

[DonFinch]

Fuel flow curves and mix balencing is fairly complicated stuff. Most of the chips you see just alter the full throttle intermix curves. and that is about all you can do for power with injection control, and that only gets you a small gain. The stock computers at non WOT (wide open throttle) are designed to keep the engine at the most effective mix, by very quickly qhanging the mix from lean to rich. The best way to get more fuel into the engine is to get more air to the engine. You need more air to make more fuel burn effectivly, simply increase the fuel, you wont have the air to reap the rewards. Get more air to the engine, and the air sensor will compensate by injecting more fuel to the chambers. This is the basis behind forced induction.

Now, at WOT, you want lotsa fuel, since there is the max amout of air getting to the motor. Engineers know this so when your gas pedal hits the floor, the computer quits computing the best mix. It reverts to pre-programed mix curves from the factory., now if your motor is getting more air than it was originally designed to get, you need new WOT mix curves. this is what your aftermarket chip does. Now, you could program these curves yourself, but you better KNOW your shit. Get this fuel mix wrong, and manage your engine poorly, or make a typo, and you'll be debuging by rebuilding your cylinder head, or worse, replacing pistons and rods.

if you want to see great examples of automotive stuidity, go here www.overboost.com

Here's why...

[Anonymous Coward]

The reason you would want to do this: If you do significant engine modifications, generally to increase engine output, you'll need to tell the injectors to supply a non-standard amount of fuel. Add a turbo, for example, and you can easily double your engine's horsepower -- assuming the engine's internals are up to it -- but you'll need to remap fueling and timing or the engine goes boom.

It's a pretty cool way to get supercar performance from lesser (read: cheaper!) vehicles. My daily-driver Miata has the factory 1.8L engine, force-fed 15psi boost by turbo. It puts out 250hp/250lb reliably (up from factory 120hp or so) and, thanks also its low weight, it outperforms 'Vettes -- except for top speed, and excluding the awesome Z06, of course. But then I paid about $20k for my car, not $50k... (Check out Flyin Miata for more info, no affiliation except as a happy customer.)

I'm not using the MegaSquirt (using Flyin Miata's Link ECU instead) but the concept is the same. There is a HUGE need for an affordable programmable ECU among us automotive tinkers.

Holy lack of engine tuners!

[xtal]

Ok; It seems that engine tuning isn't the biggest geek topic. That suprises me, since a few years ago when I got my first car, I learned everything I possibly could about every nook and cranny in it, engine included. There's a lot of interesting things.

This project isn't of interest to most big engine tuners right now, but it could be. The Megasquirt - which only has fuel injection controllers, you still need to time your ignition spark, a much more difficult problem. Replacing your cars computer right off will not get you much - it will probably cost you HP, because cars have many, many R&D hours into them and are already running at close to maximum efficiency. You can trade a little emissions for power, but in a modern Honda engine, it isn't much.

The big power comes when you use something like this in combination with a supercharger or turbo charger, then use additional control software to monitor your engine. Think of it like overclocking - you trade safety margin for speed, or in an engine's case, power. Instead of upping the voltage; You shove more air into the engine, allowing you to burn more fuel - added by a device like this, or something that's missing - a true, open source ECU replacement. It is a very difficult problem.

I love to irk people with big V8s - they are silly in the age of forged engine internals, ECU control, and modern design. You haven't lived until you've seen the expression on a Z06 corvette owner's face when he gets is ass handed to him by a built and tuned 1.8L Acura Integra. My current project is an engine buildup for my 2.2L Prelude, and I have ~300HP as a reasonable target. It's amazing what a little thought and hacking can do. You can't run much more power than that on the street feasibly, so penis issues aside - if you can make 300HP+ from a 1.8L engine - why bother with more? Troll in the extreme - you can design 1500+ HP V8's, but they're not driveable. IMHO. YMMV. YHBT. etc.

Engine tuning is a great hobby - if you can assemble a PC, with proper tools and the manuals, you can probably assemble a car engine. It might even run!

OBD-III scares the hell out of me. I don't plan on owning an OBD-III vehile, though.

There is another sister project or sorts to this one, called MegaSpark. I'll let you find the links, because I don't want to slashdot the poor guys. Ignition timing is a very difficult problem because you need very, very precise control. This is difficult to do. I have played around with this, but the approach I favour is to piggyback a microcontroller onto the computers stock signals and modify them how you want. Complete standalone systems can be very, very difficult and expensive to dyno tune and test.