sotaku

iTrader: (4)

Crazy idea, curious your guys' thoughts on it. Is it possible? Worth the time? Blah?

For the past few years I've been tracking and marginally involved in a rapid prototyping platform known as Arduino. Cheap, easy to use, and fairly capable for small tasks. My recent frustrations in getting the ole megasquirt drivable started to make me wonder if it'd be possible to track, and map the stock injector pulses to the megasquirt's fuel map. This is possible since my car is otherwise stock and the the stock ecu actually means something still and there are already projects out there that track pulse widths, RPMs etc using the Arduino (OBDuino, and MPGduino or instance).

Even better, it'd be great to have the stock ecu drive the parts of the map that it knows and as soon as we hit boost switch over to the Megasquirt map.

One big problem though, as soon as I change the injectors the Stock ECU really doesn't matter any more since it's looking for 265's and there'd be 460s in there. So I'm curious, would it be possible to convert pulses in real time? I don't see how since we never know how long the pulse is going to be until it ends. So we'd always be a pulse behind at best - which is sure to mess things up I would imagine. But then again you see things like the Zoom3 and Xede doing stock ecu pass through and modifying signals so they work on larger injectors. So maybe it's possible some other way?

Joe Perez

iTrader: (8)

This is exactly how the EMU works. It keeps track of how long the previous input INJ pulse was, calculates a corrected duration off of that, and uses that value to initiate a corrected output pulse on the falling (leading) edge of the next input pulse.

Seems to me that you're looking for complexity here, though. Not only has it proven entirely adequate to let MS generate the fuel pulses for the whole map, but in doing so you gain the ability to do things like lean cruise which, while beneficial from an economy standpoint, were unavailable to Mazda owing to emissions requirements.

sotaku

iTrader: (4)

Very interesting. So since the time is so small delaying by one pulse isn't even noticeable and won't cause a problem. I imagine you'd want to delay the spark as well? Or is it a matter of once again that a one cycle delay is such a small difference that it doesn't matter of the spark is the right one?

Also how fast does a chip have to be running to record, convert and spit out the next pulse without missing a beat? While the Arduino platform is cheap and handy I don't believe it's terribly fast. It uses the AtMega8 (8mhz?) but can run the 168 (16mhz?). Though I'd think eight million cycles a second would do fine when you need to do something ~117 or ~234 times a second - but my electronics skill is uh, lacking so I could be very wrong there.

Yeah, mostly I'm looking to tinker. Would be nice to have a meaningful project on the platform. I've done small electronics projects here and there but never anything with much reason. I think the logging the stock ecu may be a better project to start but my head started to wander to some possibilities. =D

However, that said it would be nice to give some more options with the Megasquirt. I like it as a platform but it seems that it's very difficult to ever get to the drivability of the stock ecu. Would be nice to combine the two and maybe give a different upgrade path for those who aren't quite ready for the MS to run the show but may in the future. (Not to mention Xede is just crazy expensive)

patsmx5

iTrader: (16)

I'll admit getting stock driveability is tough. I'm still not there. But doing a parallel install will help as now MS only does fuel and spark, while the stock ECU does fast idle for temp, idle compensations for turning P/S alternator load, A/C compressor, etc. My driveability is probably 96% ok. I have a random "off for .2 seconds" problem that is MS2 specific and I have found a solution to it, just need to fix it. Other than that, it's better than stock in drivability.

sotaku

iTrader: (4)

That's really impressive. I'm quite jealous. Right now Autotune's adjustments put me a little rich or very lean. Each of these are +/- 1% too. Not sure if I can make them smaller. MLV seems to be often worse, though similar. Each logging session seems to swing between A bit too rich to far too lean (18+). That's with it on "Very Hard" too. Surprising.

Not wanting to get off track, but is this common? Things like this is why having something definite like the stock ecu to help dial in my map is very very appealing.