Build an ECU porject

[mk e]

Arrhhhh !! ...... then all I can say is ' hurry up Mk e '

There's a lot to think about isn't there?

What I'm working on (with other more talented than I) will be in the "science project" category for at least another year or 2. It should be able to run an engine today, but that is a lot more it than just that. Baby steps though......and it will get to where it's useful I think.

[Warpspeed]

I did something like this myself back in the 1980's.
The ECU part is pretty easy, as long as you only want to map fuel and spark just to run a basic engine, with maybe some minimal simple corrections.
Its definitely a great learning exercise from all aspects.

But today, the average user expects a dozen high speed data logging channels, and to be able to control his airconditioner and sound system, as well as control several different types of variable valve timing , staged injectors, odd fire engines, turbo antilag, and a host of other off beat features, all instantly available with just a few key strokes.

The actual ECU part will take up about five percent of your software and hardware development time.
The other ninerty five percent will be the graphical user interface that is needed to be able to tune the thing in real time with the engine running.

If it is a fun project just for yourself, the user interface can be pretty crude.
My ancient system just used a green screen terminal running DOS.....
It was fine for just me, but it was quirky, and had bugs that could cause problems, there would have been a huge learning curve for anyone else, and totally impossible as a commercial product.
It was never intended for anyone else, just for myself as an interesting personal challenge.

But these days, unless you provide very professional high resolution on screen graphics, and dozens of features, the average "joe" will say your system looks like crap, is too difficult to use, and is just not worth the hundred bucks you are asking.

Megasquirt have already been down this road, and you will need to work pretty hard to compete.
But I suggest you look at how Megasqirt has evolved over the years, the problems they had, and how they were overcome.
Plenty to learn from that as a case study.

[upinthehills]

Why not choose a suitable high quality oem ecu, matching transducers, and wiring complete with connectors from a donor vehicle rather than an aftermarket system almost certainly utilising obsolete chips, plus shag around sourcing transducers and connectors, then soldering and braiding wiring?

It's easy to be ahead of the OEMs in terms of chips. New ones come out almost every year and they don't update their ECU designs nearly that often. Building a quality board and using state of the art parts is pretty easy.

If Tuner Studio suffices for the user interface, than this seems pretty doable.

[Leftcoaster]

It's easy to be ahead of the OEMs in terms of chips. New ones come out almost every year and they don't update their ECU designs nearly that often. Building a quality board and using state of the art parts is pretty easy.

If Tuner Studio suffices for the user interface, than this seems pretty doable.

Please be more specific with your claims

For instance how much to build a substitute ecu to replace the oe system running a 2010 Ford Fusion with 2.5 16v C.O.P. engine with twin cam phasers and rated at 175hp 170 ft/lbs in std form - - please, within reason itemise the costs to that of hardware, software, and time to assemble the unit

Now compare that against the $500 odd to purchase the likes of SCT's flasher and license http://www.sctflash.com/whatwedo.php which can reprogram the oe ecu to all but race applications

I just can't see how assembling your own system, or buying an aftermarket ecu, can compete

[RednGold86Z]

It's easy to be ahead of the OEMs in terms of chips. New ones come out almost every year and they don't update their ECU designs nearly that often. Building a quality board and using state of the art parts is pretty easy.

If Tuner Studio suffices for the user interface, than this seems pretty doable.

Please be more specific with your claims

For instance how much to build a substitute ecu to replace the oe system running a 2010 Ford Fusion with 2.5 16v C.O.P. engine with twin cam phasers and rated at 175hp 170 ft/lbs in std form - - please, within reason itemise the costs to that of hardware, software, and time to assemble the unit

Now compare that against the $500 odd to purchase the likes of SCT's flasher and license http://www.sctflash.com/whatwedo.php which can reprogram the oe ecu to all but race applications

I just can't see how assembling your own system, or buying an aftermarket ecu, can compete

I make a dual cVVT, Electronic throttle body, coil on plug, knock, pre & post cat O2 ECU (OEM in China - for Geely, and it's bi-fuel, so it's got 8 injector drivers and 2 pump relay drivers, two spark tables). Hardware is cheap and easy. Software is hard. Calibration is hard.

[Leftcoaster]

I make a dual cVVT, Electronic throttle body, coil on plug, knock, pre & post cat O2 ECU (OEM in China - for Geely, and it's bi-fuel, so it's got 8 injector drivers and 2 pump relay drivers, two spark tables). Hardware is cheap and easy. Software is hard. Calibration is hard.

Thank you, R&G

North American or offshore agent? Approximate cost? English manual? Transducer brand? Whose software & what price?

Apologies to keep harping on about costs, but that's got to be the key to increasing aftermarket sales - - and I cannot fathom why it's being ignored

[upinthehills]

Please be more specific with your claims

I tried to be specific. The hardware is not that expensive and gets cheaper on a schedule. So my claim was it's easy to use state of the art electronic parts. These parts are much better then the ones used a few years ago and maybe even currently in OEM ECU's. They have safety features now which were missing and/or unused according to trial evidence during some of the stuck throttle cases a few years ago.

For instance how much to build a substitute ecu to replace the oe system running a 2010 Ford Fusion with 2.5 16v C.O.P. engine with twin cam phasers and rated at 175hp 170 ft/lbs in std form - - please, within reason itemise the costs to that of hardware, software, and time to assemble the unit

I'll think of this as a typical case I guess. I don't have answers yet, really. Pricing is hard to figure but the cost of an ECU is something like a tank of gas. Here's an example of difficulties though. A company makes a new series of chips for automotive use, they sell a sample board you can play with for $25. The connector it could plug into costs about $16. One of these things has millions and millions of transistors measured in billionths of an inch on it, it's assembled onto a multi layer printed circuit card where each layer is individually made and then stack to thousandth's of inch accuracy etc. But the simple socket costs almost as much.

The software is expensive and time consuming, but people have been doing this for decades. The software doesn't wear out or get corroded or something. So some people are making good money on these expensive units.

The trick now is how easy it is to buy parts and build boards. Now we get to see if the effort of people interested in this subject and modifying their software can make some progress. Some of these ECU's are full of nightmares of code. They don't need to be that way though. We'll see if the code can be made safe, simple and reliable.

[upinthehills]

Apologies to keep harping on about costs, but that's got to be the key to increasing aftermarket sales - - and I cannot fathom why it's being ignored

No need for apologies, cost is important. You should keep harping on about it. I think a cheap ECU could be designed, if I do this I will try to make it open and publish the schematics and even the board files. Realistically I don't think there is a lot of benefit to being proprietary now with this stuff. Anyone can just take a look and see what you've done anyway. If they make the card cheaper, they are doing you a favor. We'll see how that goes...

[Sir Yun]

I can understand doing this for the learning experience but why does it have to cost $200? People here drop 2k on a crank or a set of heads and burn $400 of race fuel and tyres in a few weekends and drive car that does 12MPG to commute.

MS has taken years to come where it now.


IMHO you should do it because you want to ( a good reason, that is why I port heads) but it has F all to do with cost if you are honest.

when this costs $1100 and has support ( and does COP, VVT ect ect etc ) would you really be wanting to spend 2-4 years coding software and scavenging hardware ?

http://www.emeraldm3d.com/

or something like this.

http://www.specialist-components.co.uk/ ... anagement/

[mk e]

I can understand doing this for the learning experience but why does it have to cost $200?

have you seen this thread:
viewtopic.php?f=1&t=42577

Where the same people who drop $2k on a crank are screaming a $2k priced ECU is ridiculous?

I think that getting an easy to use system out there working with open, free software and a modern processor has value and is very different from MS which has neither open SW or modern HW.

It is a very large undertaking though.......

[upinthehills]

So questions have come up in other threads regarding how accurate sensors used in our ECU systems are. In particular at the moment the discussion is about VR vs. Hall effect for measuring crankshaft position. SInce it's a thread on calibrating a MAF sensor maybe we can continue some of it here. viewtopic.php?f=1&t=42642

So how would a person do a good job of checking these sensors on and off an engine? Does a lathe have too much magnetic interference from the motor or stray magnetism to be useful? Do you need a 3 phase motor on your lathe? Can you ruin your competition's crankshaft by pointing a magnet at it from your car while they pass you?

One tenth of a degree crankshaft rotation would be a fraction of the face of a flywheel tooth. Has anyone tried scribing teeth and taking video with a strobe?

In a related question has anyone here figured out how much an engine can accelerate or decelerate between teeth on their crank sensor? Do good ECU's do filtering on their signals to account for this acceleration and predict the next tooth? Is that in the Freescale eTPU code?

[dieselgeek]

I'd love to see what a "modern" hardware ECU for a few hundred bucks can do at a place like Engine Masters' Challenge, or Bonneville.

I don't mind you being "anti megasquirt" but you better have a REALLY good alternative, ready to run. Let's see it.

[mk e]

I've never figured out a good to test crank sensor accuracy on a running engine....I think highspeed camera might work but I don't have any way to do that.

On the bench (lathe, mill, ect) it's pretty easy. The error can really propagate in the ecu past the next tooth so what matters is the consistency tooth to tooth to tooth, and that it very easy to see/measure on a scope. I normally check 10 consecutive teeth and call it representative.

From what I'm told some ECUs use predictive code, but I think that is the exception., the canned eTPU code does NOT do that....so there is some error as a result.

[lada ok]

Ive decided to blast on with a delco hack
Looks like I'll get plenty of help from the PCM site , which is Aussie based.
The particular set up I'm attempting is based on the GM delco ' 808 ' ecu I just have to find one with the 4 cyl chip( not the 6 or 8 cyl )

[Circlotron]

^^ That 4-6-8 chip for the '808 only contains a resistor or two and a jumper. Nothing special at all. I would post the details except I am using my tablet not my PC. Their $12P tune is a good starting point; I have one in my 5L VP.