rusEfi: DIY ECU
 

DIY as in real Do-It-Yourself from scratch.

I am Andrey, I am involved with www.rusefi.com - that's a project to develop an open-source DIY engine control unit. I hope that this might be relevant - if you have full control over ECU logic and firmware you can potentially implement all kind of fuel-saving features, starting from alternator charging control to maybe shutoff @ stoplight?

Anyway, back to rusEfi. This is all still pretty alpha version, we only have two engines running this. We have our Frankenstein board we use for ECU hardware:

http://rusefi.com/images/Frankenstei...0306_state.jpg

And we have a wiki/forum website. Oh, and as I said - we have two engines running this, another one can join maybe later this week:

https://www.youtube.com/watch?v=M-qqHQ-quFg

https://www.youtube.com/watch?v=neGVmxol0Gc

So if anyone would be interested I would love to help with setting you up for rusEfi DIY ECU :)

Another thing - there is a KickStarter project behind this: https://www.kickstarter.com/projects/312898525/rusefi-gpl-automotive-engine-control-unit-ecu-firm

Would it also work with Diesel engines? A fully stand-alone ECU would caught the attention of the folks willing to use some of the newer-generation Diesels for swaps instead of the good old Cummins 4BT...

While gasoline and diesel have a 90% overlap, there is that 10% difference - so, rusEfi would support diesels only if someone would go and implement the missing functionality.

For 2014 focus is definitely on older gasoline engines: I know that GDI is too advanced for us, same issue with CDI.

Just wondering if you were aware of the megasquirt (open source also). This sort of thing was kickstarted a long time ago.

While I am definitely aware of megasquirt, I am afraid it is not open source.

AFAIK it was never "free open source" - older versions were "you can see the code but you have to use authentic hardware" and current version is plain closed sourced. That's why I believe there is space for a GPL open-source firmware.

None of these commercially-available stand-alone ECM setups are open-source.

Saw your project featured on Hackaday... very cool to see you join here too! Would also enjoy seeing a diesel engine implementation

After seeing so many Diesel conversions using all-mechanical Diesel engines into vehicles newer than '95, which is technically illegal since the engines should be compliant to OBD-2 for such application, there is an enormous marketing opportunity for a Diesel stand-alone ECM. Some folks consider using Mercedes-Benz engines out of Sprinters but the integration of all the vehicle's electronics is a major PITA.

Ah, I was meaning that anyone can build/modify/compile firmware for one for personal use without much concern (indeed there are clones). GPL would be better.

That's MS-2 then, 2005 design. It is 2014 here in Chicago.

MS-2: 24 MHz, 4 kB
rusEfi: 168MHz, 192kB

While you can run an engine even on 8MHz, the extra speed and RAM makes the source code much easier to modify & extend, it's just a different level of comfort for software developers.

Lol, well it is more expensive anyway :) I don't see all those SMD's being particularly DIY though.

edit: check that, the STM32F4Discovery is like $14!, I've only looked briefly at the ST processors, and would be eager to push it to its limits with asm blocks, 6ns per cycle?!? You might need that for CDI/GDI

FYI, someone made a diesel ECU (audi TDI) from an arduino :)
Arduino powered ECU for diesel engines - Arduino Forum

Why?
Through-hole is done, it's over. Nice components with good characteristics are SMD only now.

If you buy a $5 syringe of soldering paste SMD soldering is pretty doable. 1206 and SOIC is trivial, 0805 and MSOP is doable. 0603 is not an option for DIY.

One step at a time. Step by step.

Do you know C? Would you like to help with some coding? Eventually time would come to do CDI/GDI, I believe this platform has the potential to support it.

I've poked the Finn if he would be interested to try rusEfi on his test engine.

I'm just saying through hole is a lot easier in a diy kit, SMD is like soldering grains of sand at times. It is far from "dead". I'm sure there are plenty of folks who would appreciate ease of assembly and not really care about the ease of modifying the code.

I do, but I don't really understand the scope of this project. How are you going to deal with all the vehicle specifics to make it easier to get running?

settings/maps/sensors/hookups database by year/make/model or something? In some ways making it easier to code is the least of the issues. If you don't know the operating parameters of your car then you have to do like tdi guy and reverse engineer it from scratch. You don't even want to look at all the ecu connections on my prius :)

Wiring diagrams are sensor info is available for most of the vehicles. There are at least two options how to build a map for a vehicle, that's unless this info is already available:

You can start by making an educated guess about the base values (10 degrees BTDC is a common start for timing, and fuel amount directly correlates to engine output/displacement, and injector capacities are also known usually), or you can put the original ECU on a bench, stimulate it with fake signals and record the reactions - that's the way I did it.

If you think you might be interested in playing with all this you should check the http://rusefi.com/ wiki and rusefi.com • Index page forum.

I don't see an O2 sensor in the pinout, not sure this is terribly fuel efficiency related (already have kill switches and alternator cut-outs), especially with the current focus on forced induction. I don't really understand the goals of the kickstarter project here as it relates to fuel efficiency.

Do you think there is any overlap between a DIY ECU for forced induction and a DIY ECU for fuel efficiency?

Do you think that existing OEM and aftermarket ECUs are providing all the features beneficial for eco modding?

You do not see O2 in the pinout because well, it's not a product on a shelf but a kickstarter project :)

Efficiency folks and performance meat-heads don't usually overlap much in personality is my point, and since O2 sensor wasn't even listed in the priorities while speed density is, my gut tells me this is targeting the performance crowd. I can't help them.

Oh, I did not realize that this is all about personalities :)

Sounds to me like you are more interested in revealing the sins of the performance crowd then in a DIY ECU then. I do not know why O2 is not on the pinout you are looking at, but we both know that a O2 (Wideband O2 usually) is an essential tool for engine tuning, and it does not matter which goals you have. You would find a configurable WO2 decoder in ego.c

Once again, please excuse me for daring to post about this project here. Since I am a noob here and you are a regular, I submit myself to your judgement.

You are asking for $ without much of a plan, at least not one that you have documented. Have you done an open source project before? Especially a vehicle one? Do you know what the phrase "herding cats" means?

You are gonna need a plan, especially for something this involved.

Love all your hard work that you have put into this.

As for the Performance and Efficiency people butting heads its very true. lol

I have seen this first hand with my project (600hp/50mpg turbo talon). Most people don't understand what or why I do it?

On your system can you run a wide-band in narrow band simulator mode?
Load/TPS switch over?

There is no build-in WBO controller yet, for now we are relying on external controllers. Anyway, if I am a ECU why would I need to simulate a narrow band - who would consume this signal?

You can switch between modes but only manually via tuning software, right now there is no dynamic switch. If you are referring to a dynamic switch, when is it useful?

That's for sure. Good news I have a plan.

I believe the goals for the https://www.kickstarter.com/projects...-unit-ecu-firm are listed there.

As for herding cats, trust me - I am already enjoying this. There are so many people with opinions and ideas and not so many people who can focus and develop a particular feature, actually that's exactly the reason why I need funding to hire developers & technical writers for the boring stuff. So far I am failing to locate enough capable software developers and realistically 95% of the code is written by me. At least I was lucky to figure out how to focus some electrical engineers into developing the current board :)

I read that and didn't recognize it as a plan :)

The real goldmine here is the database of vehicle specifics. Anybody can twiddle mosfets in the right order these days and read ADC/frequency/etc.

Are you going to set the expectation that users contribute their vehicle specific data and that said data will be publicly available (i.e. also open)?

:thumbup:

Here is an open source wideband monitor with narrow band output:
https://github.com/14Point7/SLC-Free

Speaking about source code:

http://i.imgur.com/QVxW3uQ.png

well, if it works. didnt do a code review. though something like (char)Base10+'0' would be more succint.

Exactly.

I know you are concerned with the whole performance thing, but maybe you would be available to have a glance @ wiki/index.php?title=Development:Source_code_intro and maybe review rusEfi code?

too many projects (and microcontroller manufacturers), got a three phase controller kit to assemble (through hole BTW), and a motorcycle to convert to EV. Plus developing a distributed and cheap BMS, still need to sort out a cost effective charger. The gas stuff is gonna have to wait, though I could do a custom ECU on my 95 civic without worrying too much about the environ police, maybe, someday.

Oh, one other point about through-hole, breadboards. I imagine if the EE guys did all the design they might not think about the end users of a kit much. They probably just say "hey joe, make this pcb for me and throw it in the reflux oven" everyday. Or they send it to the in-house bot.

Someday everyone will probably have to go all SMD, but not this day :)

Oh, btw, for something like a vehicle with all kinds of kookie sensors and controls, cypress has some interesting offerings, it is sort-of FPGA(called psoc), but gobs of op-amps/comparators/logic gates/com etc surrounding an M3 core, with up to 62 20 bit ADC (4 in parallel). I know you are going the modular route (which is another kit detractor, did I buy all the right stuff?), but with PSoc I would think the number of external components could really be trivial, and modules become firmware/configuration with enough spare drivers on the mainboard. It isn't as communication intense as that stm, but who needs ethernet in a moving car? It does have can/usb/etc, and lots of "real world" interfacing options that might be better suited to the task at hand though. Just fyi, I don't know how much of a PITA they are to configure or anything, but glancing the datasheet is interesting.

Apparently you have to roll your own btoa though :) But I'm not sure adding an operating system is the better option either ;)

New fancier board:

http://i.imgur.com/Yn3LAUS.png

make more mileage with an ECU
 

Nice plan a DIY ECU.Maybe you could test some fuel saving tips with it like controlling the water temperature with it and mapping out the fuel enrichments due to accelerating and load.
For that you'd have to take out the thermostat , replace it with a valve controlled by a servo and temperature sensor connected to the ECU.
Please read my post on the foolish burning of gasoline

have not posted in a while, this project is still quite alive

http://www.youtube.com/watch?v=CD2lb1KpI3M

http://www.youtube.com/watch?v=3xz66oR95F8

Any update on this? I've been running MS/2 Extra in my 1988 BMW 325iX for about four years now. I built it from parts and put it in the original ecu case.

We had version 1.0 release recently! One of the users has put a 8 hours continues road trip on this and we've decided it means realibility.

https://www.youtube.com/watch?v=a_0XbZutufU

Now working on a nicer more convinient packaging - want SD card accessible without taking the case apart. So short answer - still playing this this, I believe it's quite useable now.

http://i.imgur.com/FhgxSoZ.png

and you use TunerStudio to tune?

yep, TunerStudio to tune