Possibility of Building an ECU

[ZDaddy]

I just finished my first REAL Computer Engineering class. We built our own
machines and next quarter I'll be taking a class where we build our own boards.
This is awesome, years of math and physics and finally I'm getting into what CE is
all about.

So, now that I read about ECU's and ECU tuning...its like I'm reading an assignment
specification for my class. Only thing is I need to know all the inputs and outputs to
the ECU. I can only imagine them...

Inputs: data from 02 sensor, MAF, CHTS ,throttle position sensor?....more?
Outputs: Injectors...err cant think of much else.

I havent done ANY ecu tuning nor do I have ANY experience with it. Dont really have
time, but I really should pick up one of those books Jason recommends. Still, I thought it would
be cool to start a discussion here. I'll be happy to share any knowledge I have/acquire.
It would be nice to produce an 88like ECU since they are getting more rare and expensive
as people catch on.

[MrwassmanNA2T]

sounds pretty kick ass, and fun at the same time.

[tmbg]

building an ECU is not all that difficult; these days it's just a microcontroller with some input and output conditioning circuitry.

The hard part is writing the software for it.

If you're really interested, look into Megasquirt. That's the quickest and easiest way to get your hands on an ECU that you can literally tear apart, both hardware and software.

[Stinky]

I would second the suggestion of looking into megasquirt. Making something that would start a car probably wouldnt be hard. Getting it to the point of being useful is the tricky part. The code and theory involved is time consuming. Megasquirt has been around for years and its still being improved.

I would suggest creating something a bit simpler if you really feel like building something. What about your own design for an electronic boost controller? Maybe your own version of a VPC (replaces the maf with a map). Or better yet, reverse engineer the hardware and code of an 88 Nissan ecu. Hehe.

[ZDaddy]

Stinky wrote: I would second the suggestion of looking into megasquirt. Making something that would start a car probably wouldnt be hard. Getting it to the point of being useful is the tricky part. The code and theory involved is time consuming. Megasquirt has been around for years and its still being improved.

I would suggest creating something a bit simpler if you really feel like building something. What about your own design for an electronic boost controller? Maybe your own version of a VPC (replaces the maf with a map). Or better yet, reverse engineer the hardware and code of an 88 Nissan ecu. Hehe.

Oh the possibilities...I love it.
In terms of reverse engineering the 88T/NA ECU, thats what I had in mind. It would then be nice to re-design it AROUND the romulator and I would do it for the Z31 community. Ooo...I can't wait until I graduate and have all the knowledge...it will be so awesome. :twisted: Muahahahaha...it's AHLIIIVE..or wait I'm getting ahead of myself. ops:

About Megasquirt...I'll look into it. I heard about it but had no idea what it was nor cared to look into it.
Thanks

[turbobluestreak]

how about a plug and play for the z31 ecu harness?






tbs

[Steve88t]

Like the others have stated, it is not the hardware that is difficult, it is the code. Even the original processor the MS used was more than enough for most people imho. The latest one is WAY overkill for most. But extra processsing power is NEVER bad.

[tmbg]

I went down the same road you're going down. At one point, I was dead set on building an electronic boost controller, and I actually got a lot farther than I expected. That's how my boost gauge was born (http://www.z-toys.com/).

The only reason I didn't get beyond that point was I never came across a good solenoid to use.

Nowadays, I'm planning on integrating boost control into megasquirt, again, just as soon as I can find a viable solenoid. Also, I'm expanding the boost gauge to be a boost and AFR gauge.

[ZDaddy]

Steve88t wrote: Like the others have stated, it is not the hardware that is difficult, it is the code.

This is why I'm in school.
This is what I will be getting paid for once I graduate.

And actually, if you want to talk about "software" it is incredibly easy compared
to dealing with hardware design, I know from first hand experience. Especially if you are dealing with verilog.
I could go the easy but clumsy route. I could buy myself an FPGA, which would be overkill,
and create an entire ECU on that. For $99 I could do this, and even with an old,cheap outdated
FPGA the possibilities would be limitless. Heck, I could even hook up a monitor and keyboard to
it if I wanted. :lol: That would be funny.

[tmbg]

The software is harder than the hardware. Trust me. I'm a programmer by trade, also an EE, and I have experience with both microcontrollers and FPGAs.

It's not that writing software is hard, because it's not. Its that there's a LOT MORE to running an engine under all conditions than you realize. Yes, it's pretty trivial to build a device that says "when this line goes high, shoot the injectors. When this line goes low, fire the spark". However, actually tuning that into a drivable ECU is way more difficult.

I'm going to reiterate--if you want to learn how ECUs work and maybe even eventually design and build one of your own, I would suggest you get a megasquirt kit. You will very quickly learn everything you need to know about the inputs an ECU needs to run, and you will learn about the algorithms it uses to control an engine. If you really intend to build your own, get some experience with a working standalone, and megasquirt is ideal because the internals are transparent, and it forces you to understand them in order to make it run.

[Jason84NA2T]

tmbg wrote: I'm going to reiterate--if you want to learn how ECUs work and maybe even eventually design and build one of your own, I would suggest you get a megasquirt kit. You will very quickly learn everything you need to know about the inputs an ECU needs to run, and you will learn about the algorithms it uses to control an engine. If you really intend to build your own, get some experience with a working standalone, and megasquirt is ideal because the internals are transparent, and it forces you to understand them in order to make it run.

I agree.

IMO trying to "build your own ECU" is biting off a lot more than even experienced EE's can manage. It takes entire teams of people and thousands or millions of $ in development to build things like the MS or ECCS ECU (in this case JECS which is a major corporation).

I've worked with MS a little bit before. It's quite the platform for the technically inclined to build almost anything they want that is EFI and engine control related. I would have gone for a MS if an easier, cheaper and potientially more reliable solution (modification of stock ECU parameters) had not hit me in the face. It would be somewhat foolish to start "building" anywhere else when there's such a huge established base of information to learn from. Why reinvent the wheel?

[ZDaddy]

Why reinvent the wheel?

Not trying to do that. I dont know if you guys got me all wrong or I totally came off on the wrong foot.
Look, all I was trying to bring up is that I gained a little bit of experience in CE finally after years of study, and I'll be a full fledged Computer Engineer in the next 1.5 years. I see that it is possible to, and would like to open disccusions regarding the
Z31 ecu and recreating the 88-89 ECU (reverse engineer if you will) which would have all it's advantages maybe more. No NOT trying to reinvent the wheel. Yes There are fully working systems out there, and thank god, because we wouldnt be going into this blindly. And I would like to bring up.

1. 88-89 ECU's are getting old
2. 88-89 ECU's are becoming scarce, especially since your discovery, Jason.
3. Because of 1 and 2 they are becoming more expensive, and are approaching the price of MegaSquirt.

So, yes I'll look into Megasquirt but guess what, I will not afford it before I graduate. Yeah fine, I'll go pick up a
book on ECU's, sure I'll read about Megasquirt as suggested. But I guess we cant have a decent discussion here or anywhere else about such matters accordingly. Seems like message boards are good for troubleshooting only, not brainstorming.

I'll come back here when I have trouble with MS, or have Romulator tuning issues..... and thanks for the advice.

[Bernardd]

`

[G-E]

and the 87 ecus are just as mod friendly now that nistune gives us instructions how to rely solely on the 28pin chip, with the only drawback that the voltage regulation for the afm hasn't been documented and therefore limits you to 2-5v

does anyone know if the z32 ecus boost the voltage at all? is the vqmap accurate without any voltage adjustment (as in removing all amps)?

[grahfz]

I was thinking about building a 2v pull-up for AFM's... in theory, if you can build one, it'll allow you to not have to figure out which resistors to cut out.