Anyone to waste time on DIY engine control unit? (Page 5 of 5)

i'm running my own 2nd gen chassis controller, installed in my Rambler Roadster. it does everything *except* timing and fuel -- i have a Weber 32/36 carb, and EDIS-6 ignition controlled by a Megajolt Lite Jr.

it's partitioned into chassis controller and interface controller. it works, now, realtime control, error checking, inter-processor communication, reset recovery, gauge (sensor) processing and decision-making, full lighting control ("single filament" large LED tail and nose lights, separate headlight circuit).

at some point i'll handle EDIS PIP/SAW and set the MJPJ aside, but i've had it for some time and it's working. and i plan a TBI FI system. no schedule for that.

the hardware and software both are designed for drop-dead reliability and realtime operation. i'm currently assembling a 3rd-gen chassis controller, increased conservatism in the design and redesigned electrically and physically. multiple internal thermal sensors and thermal strategies. load-dump resistant, etc.

it's in use now. rock solid. code on github.

hardware is designed around Arduino MEGA 2560. software is a cooperative multitasking strategy, about 10% or less CPU utilization. plenty of headroom for EDIS and FI. code uses the (awful) Arduino IDE simply because it's ubiquitous.

the code sorta-kinda-almost adheres to some of the MISRA-C:2004 coding standard for embedded control.  not the testing and institutional part, but the C and C++ code methodologies.

most recent feature set is closed-loop cooling system using an electric pump, which turns out to be more complex that it seems it ought to be!

haven't quite got around to putting tech info up on my project website, but i will if bugged.

the code is partially described at http://sensitiveresearch.com/1961-Rambl … index.html . scroll down to the software description (states and such). i'm not using that LED-heavy panel, i've moved to a very tiny box at the wheel.

the github links are on that page.

the reason for the lack of web docs is that i need to first get across the idea of cooperative multitasking. too many folk get bogged down in looking for realtime OSs for embedded controllers without really considering the overall strategy and problem-solving approach. that, plus the messaging scheme are key here. if anyone cares i'll gladly help out.

as far as the EFI part goes, the crappy 16MHz Atmel processor has more than enough guts to run the fuel calcs in realtime, coded in portable C/C++, project done on the IDE. the solution is understanding the problem thoroughly, planning out the approach and not brute force it.

SWEET!!

https://www.facebook.com/TeamEmoticons33/ did it again! Second Lemons race on rusEfi.

I kind of need to start using my own ECU as well...

blasphemy. turbo taxi did it and look how it ended!
(PS: actually MS was not at fault, just kidding)

Sorry didn't know this was a brand specific thread...

rusEfi allows this via http://rusefi.com/wiki/index.php?title= … ible_Logic

RPM is the trickiest input of the above mentioned: first you have the challenge of analog part in case of VR signal (hall signal much easier) than you have the trigger decoding code. If you are interested in an Arduino-based ECU check out http://speeduino.com/ - I believe their board is around $100 same as rusEfi Frankenstein

Might be looking for this to move away from CSI-E on my Dictatormobile ( Mercedes 560SEL).  Just repairing the vac leaks could end up costing a few hundred so been looking at Megasquirt before this popped back up.

https://motoiq.com/rusefi-diy-open-sour … -your-car/

Same story here - I'm a software guy with very little EE knowledge. A few smart EE people on rusEFI slack see https://rusefi.com/forum/viewtopic.php?f=13&t=1198