Upgrading a 2.6L to CFI

[markmakeitso]

Good day all,
I've been threatening to tackle adding fuel injection to a carbed engine for awhile now, and it might be time to put up or shut up. Recently put some time and money into getting the ball rolling, and I'm kind of hoping the threat of internet ridicule will spur me to long term action.

The plan is to use a Motorcraft 2 barrel CFI throttle body, adapter plate, and stock intake manifold. The throttle body was $20 on Craigslist, and the seller claimed it came from a 3.0L V6. I've lined up an extra intake manifold so I can fabricate the adapter without tearing apart my functional if unimpressive 87 Sport.

On the nerdier side of things, current top contender for the ECU is the PIC18F46K80. I have some experience with the PIC series already (sensors, outputs, character LCDs, etc), and individually most of the tasks are pretty straightforward. The 18 series should have more than enough speed, memory, and peripherals to meet any plausible need. Plus I have a couple on hand already. Might be overkill, but I'd rather overspec it at the beginning than change platform later. Right now I'm also considering a frequency to voltage adapter for RPM signal from the coil pickup, and an IC peak and hold driver for each injector.

That's about where the project is at. I'd welcome questions, critiques, cat pictures, etc.
Let 'er rip,
Mark

[geezer101]

It should work pretty well. I had thought about attempting to adapt the ECI from a 4.0 litre Ford 6 cylinder engine. We here in Australia got the G54B engine with MPI but only on FWD configured engines.

[dash]

sounds like alot of adapting and mix-matching. Didn't mitsu already do this for u on the 2.6 starions.... pretty simple system

[jamesw]

sounds like pics are in the future of the upgrade

[pennyman1]

The US Starquests are 2 stage throttle body injection - not a great system. The upgrade for them is the same Magna MPI intake Geezer is talking about.

[markmakeitso]

sounds like alot of adapting and mix-matching. Didn't mitsu already do this for u on the 2.6 starions.... pretty simple system

Hi Dash,
That is a fair question. Group with "why not just use a Megasquirt" or "buy a newer truck." Half the reason is for improvement of my current vehicle, since the fueling is by far the least satisfactory part. Half is a basis for further projects, like injecting motorcycles, or even a bargain platform for FSAE cars. The last half is for professional development, as I'm employed in an industry where a project like this would raise some eyebrows and prove a level of competency in automotive electronics hardware and software.

Some pics soon.

[pennyman1]

get a set of hayabusa throttle bodies and graft them to a stock intake. Use Megasquirt to run the injectors.

[dash]

most starions today still get around on TBI. Testament of a very good system. Quirks r minor
Available still, so I figured the low cost TBI a decent/natural donor for a 2.6. That is all.
Magna MPI only an upgrade(over TBI) when seeking fuel & timing control...... with a final goal of 4,5,600 hp
Yeah, ITBs and tuning would wake up a n.a. 2.6. Then it's camshaft time

[pennyman1]

Here is another idea - uses a weber adaptor to the intake: http://www.ebay.com/itm/322132723258...%3AMEBIDX%3AIT . Then run with Megasquirt.

[geezer101]

^ Ran into those when looking at injection and ITB's. They look cool but the price is a little scary...

[BradMph]

Little scary? By the time your done I think a turbo could be installed for less. What computer would you use with this? modified or some customized one that runs a little in the higher prices.

[geezer101]

Little scary? By the time your done I think a turbo could be installed for less. What computer would you use with this? modified or some customized one that runs a little in the higher prices.

An ECI would be very efficient compared to carburettion but it still falls short of EFI. At least you get the midrange torque band that a carb has to offer. Bike throttle bodies are a lot cheaper than a quad rack of carbs but then you have to factor in the management system as well. You'll be lucky to make 150 hp in NA tuned form - the heads are the stumbling block on the 4G54 engines. They are too restrictive. You'll be able to make worthwhile gains in economy and response but the fact you're using a lightweight truck means a nice leap in the power to weight ratio.

[markmakeitso]

Some decent progress over on this end.

First of all, my spare intake manifold arrived from James, thank you sir. The plan is to use two layers of aluminum plate, the lower which matches the intake bolt pattern and air passages, the upper will fit the larger Ford pattern. I ordered up some 1/4" aluminum plates from eBay, about $15 shipped. Hopefully will be able to machine those with my meager abilities and tools.

The injectors are marked E3VE-AA, and according to the turboford site below are rated at 37 lb/hr. Some more reading says that the regulator is set to 39-40 psi, the 2nd site claims this setup will theoretically support up to 127 crank horsepower with 90% duty cycle, which is plenty good for government work.

The injectors are 2.25 ohm across each, aka low impedence injectors, which requires peak and hold drivers. I'm leaning towards the LM1949 chips from TI right now, link below. A few dollars each, and needs some support components, but nothing too exotic.


Speaking of chips, the planned ECU is a PIC18F46K80. An EE friend got me into microcontrollers, which are small chips which run custom program code when they are powered up. I did some programming on the PC in highschool and college, but didn't really get into it. Microcontrollers are often focused on relatively simple input/output tasks rather than on huge piles of RAM and super fast processors like a modern consumer computer. There's also usually no operating system to wrestle with, just your lines of code whizzing around, and they really got me back interested in programming. Prices for the chips themselves start under a dollar for the super basic ones, and don't get real spendy even when you get to the higher end models. At any rate, the 18F46K80 has pretty generous RAM, more than adequate speed, and necessary peripherals like an analog to digital converter (mainly for measuring sensor outputs). It also comes in a 40 pin PDIP package, which means it's easier to prototype with. That's the microchip link below, the 46K80 runs about $4 even for quantities of 1.


The bikeboy link is a quick description of the basic tasks of the ECU. That shows the lookup table for injector on time during operation. Hopefully I'll have an analog voltage input for both RPM (taken from the coil primary side by another IC) and also from the TPS on the throttle body. That should determine the 4 cells to interpolate for total on time, and that value is then modified for coolant and intake temperatures, as well as accelerator pump function. Alpha-N might not be the best bet, so implementing a MAP or MAF setup might happen at some point, we'll see.

Last link is to my Flickr album showing the throttle body and intake.



http://www.turboford.org/faq/injectors.shtml

http://fuelinjectorclinic.com/flow-calculator

http://www.ti.com/lit/ds/symlink/lm1949.pdf

http://ww1.microchip.com/downloads/e...Doc/39982a.pdf

http://www.bikeboy.org/fuelinjection.html

https://www.flickr.com/photos/354657...57675834762935

Sorry for the long, blathering post. Getting rolling here, maybe next time I'll have new pics and some details on individual circuits.

Have fun,
Mark

[geezer101]

If you can add to resource material for the site - that's awesome. If you come up with something different and it works well - extra awesome.

[jamesw]

Some decent progress over on this end.

First of all, my spare intake manifold arrived from James, thank you sir.

your quiet welcome sir pics looking good

[markmakeitso]

Good day,
No, haven't given up already. I made a mounting board for all the prototyping parts for easy storage and transport, but haven't finished attaching everything just yet. It has a terminal strip for adding larger connections, fuse holder, 2 breadboards for smaller components, the throttle body itself, and eventually my PIC programmer. Have a relay on hand to simulate controlling the fuel pump also, but haven't wired it up. Hit a slight hiccup with my programmer, as the older ICD2 doesn't support the correct lower programming voltage for my chosen chip. I've played dirty pool and just used a voltage divider previously, but couldn't get that to work with the 18F46K60 so far. Adapter is on hand now, although I haven't tried it out. One new picture on Flickr of the prototyping bench setup if anyone cares.

On the programming side I have a basic framework in place. So far the chip cycles through the various analog input channels, starts the sample, and stores the value once measurement is complete. Currently the plan is for the TPS input voltage and RPM input voltage to primarily control the fuel injected by interpolating in a 10x10 or 12x12 array of injector on durations. Other channels are primarily used to modify this process, for example, on time is bumped up when measured intake air or coolant temps are cold.

Fairly soon in development I'd like to get a serial port working so measured values can be transmitted to a laptop in real time, and perhaps some rudimentary commands can be given to the chip also. Laptop side might be handled in Hyperterminal or LabVIEW if I'm feeling ambitious.

Happy New Year to everyone on the board.
Have fun,
Mark

[jamesw]

i was kind of wondering if you had won the lottery and went to an island somewhere lol sounds like you are still on track

[markmakeitso]

Long delay again, had to push on a school project to finish my degree, that took almost all of my free time. I'm hot to trot on this project even more than before.

Paged through a great book the other day that a friend had:
https://www.amazon.com/gp/offer-list...ds=steve+hatch

It has pretty good descriptions of operations of a bunch of earlier OEM ECU setups, seems like mainly late 80s to mid 90s. Newer versions of the book may have even more, but the older version is a fair bit cheaper on Amazon. It described different setups, including sensor information and also control strategies implemented, that isn't super commonly available. Pretty useful, mine is en route already.

One of my employers has a new engine project, with take off fuel rail and injector assemblies from a Ford 4 cylinder. They have a small pile already and it appears likely I'll be able to sneak a few out the side door. Haven't looked at fuel rate or measured cylinder spacing yet, but that might be a convenient source for a port fuel injection setup in the future. Still looking at a throttle body injection setup for now though.

Alright, that's all for now. Will post again when I have some further technical updates rather than just tertiary rambling. Have fun out there.

[Law Dog]

My 89 Jeep 2.5 has a CFI set up on it terrible mileage for a 4cyl it's a Renault setup in that year Jeep.

[pennyman1]

that was an Renault cfi on an old iron duke GM motor - no wonder it sucks! The Iron duke would be much better with a carb than that abortion on it.

[markmakeitso]

Oh hai.
Some progress over here.
The book linked above is very good, I'd highly recommend it for anyone interested in a project like this. Not exactly something to curl up with by the fire with some hot cocoa, but technical books seldom are.

Found another interesting webpage:
https://github.com/javiermuniz/tunin...jector-Scaling
Although it's aimed at tuning a certain car it has good information about injector compensation for a variety of effects. Also good stuff.

Lastly, had an Arduino Uno knockoff (an alternative to Microchip PIC above) sitting next to my laptop happily blinking for a couple days. At some point I though "what the hell, the IDE is only one click away, I could get started on the ECU in 15 seconds or so." That might be the biggest appeal of an Uno, not horsepower or gaggles of pins, but immediate accessibility. So, whipped up some code for some basic functions, tests alright so far. I may add that to the development board described above and keep rolling. If nothing else code ports fairly easily from Arduino to PIC C, so it shouldn't be wasted development even if I later change over.

If anyone is interested in following along or even contributing, or forking development even my very preliminary code is attached. An Uno can be had for just a few bucks all over the place, and the software for development is free.

ECU_experiment.zip
Not sure that attachment will work.

[jamesw]

just wondering once you get this done is this going to be a show truck/daily driver/ racer? with all the work your putting into it i was figuring a show truck.

[markmakeitso]

Nah, my truck will never go to shows, or be a racer, alas. I'd just like some slight brag factor, learn a bit, and document a setup that other people could duplicated relatively easily. We'll see.

[pennyman1]

have you thought about the TBI off of a 90's GM truck, like a 4.3 or 5.7? Holley even made a performance one for these trucks. They would be dirt cheap at a upullit yard.

[markmakeitso]

Hi pennyman1,
Thought I responded to this, but maybe it failed and I didn't notice. Yes, I have thought about that actually. I used to have a TBI 2.8L V6 S10, and a 4.3 TBI kinda fell into my lap a month or two ago. That is another option.

It's been mostly research lately. Couple phone calls with a friend who does GM tuning as a side gig. He had good advice, but also pointed me towards more work:
-Recommended a fairly large fuel map, much more than a 10x10 grid. Programming is not much more work, but tuning it will take a fair bit longer.
-Recommended non-linear divisions on the map (i.e. not strictly 500 RPM or 10% TPS steps for example). Said this will allow better precision on the areas that see the most usage.
-Said alpha-n might be adequate, but would recommend planning for a MAF setup in addition fairly soon.
-Besides the programming he recommended next working on a temperature curve for coolant and air intake temp sensors

Fair enough, will get to chopping on that. I'd like to get it swapped over before winter rolls in, not 100% confident in that kind of schedule though. We'll see.