More USBDM hell. Now it won't work on the EEE at all, either. I did find a fix for the small monitor thing, though. It's not devilspie, which the USBDM app ignores the commands of, rather this:
IE, give myself a scrollable view port through which to look at an imaginarily big desktop.
Success! Looks like the cable between the BDM and the device is too long. Having it straight makes it work. Any kinks = fail. Can't blame that on Mr BDM Australia. My single biggest reason to not like that country right now.
Working on improving the RPM calculation setup and using Sean's R1 which is 4 cylinder with a 720 degree decoder current calcs have a step of 69 RPM (from 12042 to 11973 taken from his log 0.2.0-SNAPSHOT-173-g19e741f-DEV-SEANKR1 / Sat Sep 15 10:42:15 MST 2012). New code on the other hand, correctly configured, will have a step size of just 4 by default (@ 12100 RPM), and a step size of just 1 if I add an optimisation. The optimisation will only really help large cylinder count engines that turn high RPMs. At lower RPMs the step size will be that of the variable, IE, 0.5 RPM. The new way that I'll be doing it is optimal, period. To do better requires a finer timer resolution, which isn't a good idea, and would be largely pointless. This is a nice improvement in performance, and will later be supplemented by improved sched calculations too. The main benefit of the new calculations is NOT the accuracy, but rather the smoothness of data. Current code has values which jump around a lot, this doesn't matter much in practice as they are typically used from a consistent location/value. But the logs look very ugly... and that's not good for publicity :-p This post typed in very very slowly with one hand... :-/
If we assume our main loop is horribly slow @ 10ms, then we have a 273 RPM difference from one calculation to the next, and a potential error, if we're not taking rate into account, of:
273 / current RPM % in our timing calculations.
On the brighter side, under acceleration, timing will be retarded, only under deceleration of the same rate will it be advanced more than the tune. The only time this is a real problem is when the clutch hits hard while under load. Be warned.
I have about 500 lines added for the RPM calculation change, and I'm about half way through. Hopefully I can get it usable/testable tomorrow some time.
Also from the above mentioned log: 96 milliseconds from closed throttle to wide open throttle. This indicates TPS readings need to be taken every 10ms or less to be effectively useful.
