Fred's firmware development diary

[Fred]

It's 4am and I must be lonely! Or soldering.

My fuel injection drivers should flash LEDs while running now. Tomorrow I'll make the ignition channels do the same.

Tonight has been mostly celebratory. I guess I've earned it.

Fred.

[Fred]

LEDs are sweet. I'll probably tune the ign table statically next. Then take it for a cruise/low load tune with a guy who is interested in getting one (that's everyone these days all of a sudden).

Fred.

[Fred]

I experienced a strange loss of sync or sched issue tonight that didn't seem to be present before. I saw something similar during cranking too. It may be because of what I'm currently doing in the decoder comparing things too close together with insufficient resolution, or it may be some math error, who knows. I'll look at the logs tomorrow and sort it out, I hope.

[Fred]

And today my number one rod ripped the bottom off the piston and the cap off the rod and left the unmarked crown sitting in the bore smooth to move up and down. IE, FreeEMS made me enough power to blow my engine into pieces, again. Fcuk! At least I verified the ground work was good before it exploded. Who's next to run FreeEMS? :-)

[Fred]

http://stuff.fredcooke.com/f8.rod.failure/1280x1024/

On the small end, 30 next to forging seam, 31 on it - for f8 rod, and 33/34 for the fe rod, f8 shank is 14.5x20 and fe shank is 15.5x22 + taper to 27 before curving out to big end, f8 are straight, no taper... mm for all units!

I have a plan, stan!

Fred.

[Fred]

OK, Gufi got OLV fixed up and I was finally able to review the BIG log from the wet tuning session we did the other night before the engine exploded. It clearly shows that sync was NOT being lost and the CAS decoder is MINT, which is good. On the other hand, it also shows lean spikes which are the result of the dropped spark that we were seeing in the car with the functional LEDs on the unit. IE, the sched code needs more work, something is rolling over wrongly or similar, but only at specific RPMs. I can check this in two ways, 1 by spinning a spare CAS with a drill, 2 by writing a unit testing framework to run against the live code on a CPU over serial. With the former I'll get real worldish results, with the latter I can generate detailed results in a specific and targeted way. It might pay to do some work on that when I get a chance. If only the truck was still running.. I could easily get good results by taking it for short drives with fresh code :-( I hope those hw guys get their shlt together soon so that various people can run various engines with various FreeEMS boxes.

[Fred]

http://www.sump.org/projects/analyzer/

Someone needs to buy me one of these, and some sort of scope, too.

Fred.

[Fred]

Did these:

• Finish sched code which is partly incomplete - 4 hours
• Calibrate my LC1 - 1 hour
• Add data struct size checks to preprocessor directives - 1 hour
• Tune timing table with suitable base figures - 1 hour

Don't need to do these anymore:

• Adjust TPS to be sensitive at tip in time - 1 hour
• Adjust FreeEMS to understand where tip in and wot are - 1 hour
• Generate proper curves for my temp sensors with marcos' component values - 1 hour
• Correct and test the battery V calcs/calibration - 1 hour
• Fix my exhaust leak - 2 hours
• Tune lambda table for suitable goal figures - 1 hour
• THEN tune ute properly - 8 hours+

Which still leaves these:

• Add proper RPM calc/adc sample timing to the mitsidecoder - 2 hours
• Partly split init of timers off into per decoder function for seank/nissan/lt1, most decoders can have empty function - 4 hours
• Add prime pulse code and data - 4 hours
• Add post start enrichment code and data - 4 hours
• Setup MTX bar graphs to be more useful and check scaling of vars in realtimemaps file - 2 hours
• Rework some of the corrections such that they are more useful in MTX

Fred.

[Fred]

FreeEMS runs its third car today, things are looking up ;-)

The issue I was facing, if Spudmn is reading this, was LOW V dropout! IE, ECU was dying while cranking. this SMD fucking regulator, IT IS LDO, right?????? I fucking hope so, or hope there is an equivalent... if not, it is UNACCEPTABLE :-) Car wouldn't start...

Powered it from an external battery separate from cranking batt and away it went, beautiful.

Fred.

[Fred]

OK, I've now tested the device with low V and it performs admirably! If you start it on 4.5v it works, code runs, etc. D1 drops 263mV or something like that. Marcos, leave D1 in place, it works fine. I think the issue I was seeing was related to dwell time due to decreased voltage and some code I took out before running the ute. Now it's back in, everything is working great, or so it seems. I want to do more testing and some more code inspection too, and verify the scheduling stuff a bit better.

I want to get a lot more debug out of the device too, so I'll work on that in the morning. Hopefully I can get the scheduling and output code rock solid using this platform.

I also need to put it back to stock and see where it idles at and where it has timing under load at higher revs. If I lock the dizzy I'll need that info, and in any case, I need that info to set the engine offset variable correctly.

FreeEMS Cars:

1) Volvo fuel only
2) Ute boost, cop, bank injection
3) Hyundai, dizzy ign only
4) ???? YOU :-)

Fred.