Fuel injection design. Your Thoughts

[1965ford]

Hello Guys. Over the years i have had a lot of different cars and hotrods Mini's,  fords, chryslers. escort to vw's.  One of the things that has perplexed me is why no one has not bridged the carb vs fuel injection gap. My favorite hotrod was a 65 ford econoline van It was converted by a friend of mine.  He started with a ford econoline then he added a air ride suspension, corvette brakes, Dodge engine box. Then he put a complete 300 I6 from a 89 van Aod transmission. Including the entire fuel injection. That truck was awesome the coolest thing was starting it up. Fired eveytime and without touching the gas pedal smooth idle and just got stronger the longer you pressed the go pedal. Most of that was due to fuel injection. (Van was demolished by drunk driver hitting it in front of my house.) Anyway i have often wanted to fit fuel injection to other engines. But the Megasquirt system is just to complicated to this non computer literate gearhead.  I had a bit of hope a couple of years ago when Fast came out with their system as it seemed possible then to fit fuel injection to just about anything. But the problem is they have concentrated on the big chevy and 500 hp and up. if you look at their system, Each part costs an arm and a leg.  I have often thought that a perfect system is something that used standard sensors chevy, ford or Bosch then the sensors would be wired to a self learning ecu. i don't need a laptop or a dyno to tell me when something is running right. i am willing to pay a bit more for a ECU but 1500 dollars is just too much. I bet you have thought the same thing. but getting from here to there has been just too hard. 
   So if you could have a say in what is needed in the marketplace list your wants here. I will start.
#1 Self learning
#2 pick from a 2 through 16 cylinder.
#3 pick from sensor type/brand
#4 how it is controlled TPS only or map or maf, open or closed loop
#5 Additional sensors  example, knock sensors or air temp, idle air control 
#6 NA to start then add boosted systems later 

[v8ian]

www.emeraldm3d.com/

Does what you want....(not sure about the 16 tho, Ran my engine without any problems after initial parameter set up. and my system was the earlier M3dK. that was on a 215 buick/rover

[welshpug]

Dta does it too, however you wont get any ecu that will do a decent job of self learning that will beat half a day with a good mapper.

[SamR380]

To be honest, I think you'll struggle to beat megasquirt. It was designed as a DIY system for the home mechanic, although it can stick it to the big boys of the ecu world if you do want to get complicated. There is endless free internet based support too, if you've got a question its already been asked, posted online and answered.

You can download the programming software for free (or a voluntary donation iirc), its called TunerStudio. Have a look at that and see how you feel about using it, on the whole it is self explanatory to someone like you who knows their TPS from their MAP sensor.

Edit: Oh you meant the ECU should be self learning, not you learning the ECU! Megasquirt can do that too

[welshpug]

There is free sotfware and support for dta, omex, emerald etc too

[retrowrongun]

Something that's easy peasy to map and understand ,with as many plot points as you can give

[bmwv8capri]

I think there is a gap in the market for a more basic plug & play system that will appeal to people who are normally scared of electronics. But if a supplier does a comprehensive kit that provides complete injector manifolds, fuel rails, fuel regulators, all sensors, coils, etc... it can't be done cheaply with new components.
If you don't need to drive a lot of coils then microsquirt v2/3 or pre-assembled MS2 is cheap and isn't really that complicated to wire in, & you can use pretty much any sensors & ignition setup.

[jrevillug]

I like the idea, but unfortunately from the ECU maker's POV there's little incentive to develop something like that.

If it'd only do fuelling self-learning there's little point as so much of the gain from modern engine management is on the ignition side. The only way to really do self-learning is a combination of ion-sensing ignition and a wideband lambda. Complex, and far from intuitive to program. Ion-sensing ignition needs lots of extra hardware too.

I'd love to develop a system that would do it, but it'd need vast amounts of R+D work; even if one had the backing to work on it full time, I can't see it coming in at less than a year or so of time with a dedicated team of skilled engineers (both automotive and electronic) - and that'd only be to get it to a beta-test stage...

So it'd need huge investment to get it to market, the market is limited (be honest, how many people would even consider something like this), and so the price will be high. A laptop interface is *required* to set up certain things, even if it self-learns the required maps.

In addition, simply stating that a system is a perfect self-learning design means that people will fit it, and then drive. If, for whatever reason, it goes wrong and melts the engine, the ECU maker will be seen as liable for the damage. I guess this is a big reason why self-learning hasn't taken off in a bigger way.

Finally, how many moderns have you seen running badly because the EFI is confused by a fault sensor etc.? Who would you take this system to in order to diagnose a fault?

To be honest, EFI is a complex system; either stump up the cash for something like an Emerald or Omex, and get it mapped on a rolling road, or learn to do it yourself and go megasquirt. Both are a big investment, but there's no shortcut to these things.