DIYEFI.org Forum

Hi,

I have an engine, a mitsubishi 4g15 dohc efi engine 16 valve. stock sensors include:
- tps
- coolant temp sensor
- map sensor
- cam angle sensor (built-in the distributor)
- 1 o2 sensor on the headers

It is distributor based, and does not have a crank angle sensor. the stock ecu gets engine speed (rpm) from the distributor and sends the approprote "trigger" signal to the dizzy to send the spark on the cables to the plugs. In your opinion, operating in sequential injection, does having just a cam angle sensor w/o a crank sensor enough here? - its the stock anyway.

there are 4 injectors that "sequentially" deliver fuel (according to the manual). idle up is achieved by a pair of bipolar stepper motor on the throttle body (6 pins). other controls involved in my engine are power steering pump pressure switch (for compensation during turning), a/c controls, a/c condenser and radiator fan on/off.

So much for the background of my project engine. As for myself, I can pretty much understand embedded systems (microcontrollers, microprocessors, etc) because I dealt with it during college days. as for programming, i can pretty much handle any language, c, assembly, name it. Given these, I guess building, tuning would be a bit easier on my end.

Now, I am interested in making a stand alone for this engine of mine. I'd like to ask, where do I begin if I wanted to make my own ecu? I myself have seen the drawbacks of the megasquirt - its a good system though in fairness, but there are mishaps. Plus, I hate the idea of having a "forest" of wires as jumpers inside...

Initially, what I want to achieve is to make this stock system work... distributor ignition, sequential injection, idle up compensation during warm up and a/c engagement.

A later project assuming i've done the above objective well, is to get rid of the distributor - either a cop or a wasted spark is fine with me.

Again, I ask, where do I begin? or rather how do I begin?

Please, don't tell me to read... because i've done and always doing it. I go over the megasquirt manuals over and over again.

You've already identified your i/o's of interest .. so go find a platform to fit those expectations and start writing some code.

To step back a bit more and jumpstart your efforts for minimal investment, rewrite the elements of the megasquirt code you don't like. That should give you a good 'taste' of the hurdles you'll need to clear in the future.

EssEss wrote:You've already identified your i/o's of interest .. so go find a platform to fit those expectations and start writing some code.

by "find a platform", are you saying get a known working ems such as ms and modify it according to my needs?

EssEss wrote:To step back a bit more and jumpstart your efforts for minimal investment, rewrite the elements of the megasquirt code you don't like. That should give you a good 'taste' of the hurdles you'll need to clear in the future.

any samples or threads here?

spyghost wrote:by "find a platform", are you saying get a known working ems such as ms and modify it according to my needs?

by platform, I mean a 'microcontroller platform' which has a suite of peripherals in addition to a core you are familiar with or prefer. for example, my platform of choice would be an mpc55xx/mpc56xx series or anything nxp arm7 or cortex based (lpc23xx/lpc17xx/lpc11xx/lpc13xx). What platforms did you use in college ? how much money do you have ? some platforms are powerful but have a high cost barrier to get started. there are LOTS of options out there, and you'll have to narrow them down based on your resources.

spyghost wrote:any samples or threads here?

not that I know of.



If you want to put off some of those decisions until later, your best bet may be to use your existing ms platform and use the free assembler to feel around for where you want to end up. Technically, you could start tonight for no more than what you've already invested. You may find in short order that it might not fit your goals, but at least you've gained a little bit of knowledge to help converge on a final solution. You're going to 'throw away your first' anyways ...

Sounds like you have the basic IO plan. Like EssEss, I'm curious about your parameters. For example what's your ball park $ limits? How much hardware or sofware work would you consider tollerable, ect. Basically are you looking for a sytem that is plug and play, or are you willing to roll your own from scratch? Sounds like you don't want either, but I'm not sure where you stand inbetween those extremes.

My guess is that you are interested in FreeEMS , or 5554. Fred has run the FreeEMS on a vehicle before, however, there is a lot of work before this can be considered a viable system to use on a motor. 5554 is even farther off and more work.

Perhaps you can help develope either FreeEMS or 5554? Do you have interest in writing / debugging code ect on a very young piece of hardware?

well, frankly speaking, building things from scratch is a real deal of fun for me, but if i take that route, when will i ever finish? i have to design the circuit, make a prototype, layout the pcb, make the pcb, etc... etc... i like it but it might not be too practical - maybe when i retire after 20 years...

in terms of $, i'd like to keep things minimal for the moment. once i get the hang of it, or once my engine runs, maybe i can deal with the addons then

anyway, something that is pre-built but fully customizable is i think what will work best for me. like what jharvey said, something in between building from scratch and plug and play. i like the idea of megasquirt, but i hate the customization part of it. at the end of the day, what happens is the board piles up with a huge number of wires - and i don't like it that way.

i'll be laying out a quantitative approach to my parameters later on (resistance, volts, etc...). i like to deal with the high level overview/concept first before dwelling in the math.

helping you guys out here would be a wonderful opportunity for me...

I know these forums contain a lot of information and they take a long time to read / absorb. I've tried to keep the wiki reasonably up-to-date. If you haven't already browsed that, you may want to. Check out freeems.org and click the wiki link.

Sounds like you might want to reproduce Freds engine. He did that with a TA card and components tossed on a proto board. It was fuel only. You may want to find and browse Fred's videos.

can you show me a link to that?

http://www.youtube.com/watch?v=EGWIOMh5kTo

Also check this thread.

http://www.diyefi.org/forum/viewtopic.p ... 4&start=10