Fred's firmware development diary

[Fred]

Comms bug is fixed and, more excitingly, a man in the US of A has just tested 6-2 missing tooth mode on an engine and had good readings. He just needs to wire up sensors and injectors and he'll have a runner. Cool stuff! :-)

[Fred]

Tested and fixed 60-2 yesterday, it has very low performance, need to increase that by at least 2.5x, but works well otherwise. There was a single character typo in a header that caused it to not work correctly. Tested 6-2 myself on the bench this morning and watched it work out to about 40k RPM. Pretty cool. I'm going to do a little hack on OLV so that I can make some cool videos using the fish and the hack and a decoder. It should be entertaining :-) I'll re-attack the issues list tomorrow and see what I can achieve towards 0.2.0 being complete.

Fred.

[Fred]

OLV upgraded slightly as a lazy Sunday arvo task: viewtopic.php?f=32&t=1552

Hopefully I can hack a fish file generator into it before bed and do an experiment with the log Sim did the other day :-)

Fred.

[Fred]

http://stuff.fredcooke.com/ScheduledFue ... Change.png

A nice log screeny showing how if the map changes between two cylinders the next PW is up to date and appropriate!

This is half fuel, half ign, both run through the scheduler. Pretty cool :-)

EDIT: FreeEMS/MTX labels are old/wrong

Fred.

[Fred]

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

That's right, 10 engines, and on the latest greatest code no less! Scheduled full 6 cylinder sequential EFI, baby, yeah! :-)

Congratulations to Peter for his fast work!

FUCKING BED!

Fred.

[Fred]

First drive of same car, Peter seems to like it! :-)

http://www.youtube.com/watch?v=eUWLKzSib_4

Awesome!

Fred.

[Fred]

http://stuff.fredcooke.com/FreeEMS.Installs.png

Something tells me that FreeEMS installs are on the rise! :-)

Fred.

[Fred]

I added a hack to OLV (temp branch) to generate RPM vs time output files for the hacked fish such that i can now replay real datalogs through my FredStim/FreeEMS bench test decoder. I'll wire up and setup some stuff in the morning to show it off, but for now here is OLV and GNUPlot side by side:

http://stuff.fredcooke.com/fish.output. ... ves.it.png

[Fred]

333am Presto the helpful bastard sent me some logs and some videos and I realised what the issue is.

Sched is done from the average speed which is nice and smooth, however the engine isn't nice and smooth, it varies. So if you schedule on average and reality is slower than average, then the event is effectively early (too advanced), and if reality is faster than average then your event is late (too retarded). Because the compression cycle is the bit that is slower than average, our carefully scheduled events are being nailed out too early for the reality of the dynamically changing crank speed.

The solution is to infer the effective RPM to base scheduling off, OR, to use hard coded scheduling below a certain RPM threshold, or something along those lines. I will prioritise this over the next few days so that the hotel can start happily. Right now I'm not sure what to do. I've thought about this before and I can think of various ways, but I need a clean way that works with the current impl or a way of changing the impl to work with whatever the solution is that I come up with.

http://issues.freeems.org/view.php?id=523

359am

Fred.

[Fred]

FINALLY, a small update to the "state of the project" thread:

viewtopic.php?f=41&t=530&p=23775#p23775

Fred.