Custom Ignition system project

[Taff]

So I have been cruising around with my 18RG and bouncing ideas with my housemate and we have decided to have a go at a completely custom ignition setup. Keeping the carbs to look after the fuel but going to delete the dissy completely and use our own custom PCB and software on an avr. He is a mechatronic engineering student and I am doing mechanical so between us we should hopefully be able to get some decent results. The key is budget though so our goal is $200 - this will be complete sequential programmable ignition. The idea of this is to learn so we have no interest in programmable controllers like megajolt ect.

The plan is 36-1 trigger wheel on crank running a Hall effect and a MAP sensor for engine loading. Will be run by an atmega and coil on plug with inbuilt ignition drivers.

Currently we have
4x Hall effect sensors off late model Hyundai ($10)
2x MAP sensors from A series corollas. ($5)
36-1 trigger wheel from the UK ($30)

Using an arduino development board to begin with. Going to get the sensors mounted up on the car and log some data then progress from there - initial setup will run wasted spark and then later on progress to sequential with a sensor on either a cam or the old distributor drive.. Will update this thread as we go along as I am sure there will be a few people interested in this sort of thing. Today he worked out a way to balance the carbs with the two MAP sensors - digital carb balancer for $5, cool!

[Flex]

I wouldn't suggest using arduino, as this type of project you will need precious timing, like 1usec. You will need to use C programming language. Also a micocontroller which is design for engine control will make your project a lot easier. Check our Freescale automotive products.

[Taff]

arduino is just for development. Coding is no issue at all. Will check out what you mentioned cheers.

[thechuckster]

to save you designing code from scratch, go thru the MS-Extra code with regards to ignition, interrupt and input handling and if you feel brave, dig thru the wayback archives for info about the MegaJolt Lite Jr system (before it went closed source). Will give you some pointers on code structure and methods.

The more powerful Arduinos will have enough timing resolution but am not sure about how they work as a real-time device (as in interrupt handling). The megasquirt fraternity use freescale devices, and for the next generation of ECUs are possibly going down the MPC5634 (Qorriva) route (also designed for automotive/powertrain use).

note: the big shift in these ECUs is shifting from dedicated hardware handling/transformation of input data (e.g. VR or HE interfaces, A-D interfaces, etc) to DSP and using simpler input interfaces. If you have the ECU grunt, it seems to be the way to progress as it frees you up from the hassles of dedicated hardware interfaces.

Would also suggest you design a engine simulator (or just copy/use/adapt the JimStim) in hardware for development & testing.

Am curious about why using Hall Effect instead of VR? no need for a magnet in the trigger wheel if using VR.

[Taff]

Hall effect gives a cleaner signal and it will interface well with what we have planned. Afaik the trigger wheel just needs to be ferrous - we use the same sensors on an FSAE car for wheel speed and the wheel is just plain steel, they work fine. Will have a look for that megajolt code, it could be interesting. By some rough calcs at 8krpm we should still be able to maintain 0.2degree resolution based on the avr but of course this is best case.

Not aiming for race car tunability or performance just reliability and a clean spark by doing something different.

Thanks for the tips - appreciate any input as long as it's not "Go buy a megasquirt" as that isn't the goal of this project.

[thechuckster]

sounds interesting about the hall effect - I had assumed they always needed a magnet to trigger.

This seems to the be the new code the megajoltJnrLte project has released (current firmware version only available as binary):
http://www.autosportlabs.net/MJLJ_V3_Downloads
(see Source Code section)

This project worked in conjunction with an EDIS module but the code/concept might be useful.

I bought one of the very early versions to use with my 18RTE but that project took so long to come about that MSExtra came along negating the need for it so it never got used. Was thinking of using it for the shift-light & thermo-fan functions on my Soarer until I get it running an MsExtra ECU.

btw, Any pics/data on the Hyundai HE sensors?

[George]

Hall effect would trigger with as little as a piece of steel. But coil and magnet is more reliable than Hall sensor and amplifier. All modern toyotas use the simpler sensors for cam/crank timing.