CVS-head firmware now, at 2004-09-18 is experimental, as you see on GenBoard/UnderDevelopment/FirmwareChanges there is a pending development
After some cleanup in dispatcher I had perfect results with ign + inj + WBO2 + stimout at the same time. Many will try the CVS head in production. Anyone overcautios, will probably wait a little before applying in an engine.
We had 2 issues during refactoring:
- the interrupt handler was ifdefed out (all deaf behavior)
- with the readTMR_f() killed, scheduling OCR..=PAST was a problem - now really fixed
Here's my testing:
Latest testing - I let it run for ~60 minutes. Here's the Benchstats output. Looks good from first glance! :)
0=0000 1=0000 2=0000 3=0000 4=0080 5=0048 6=0001 7=0000 8=003B
B0=0001 B1=8D29 B2=137B B3=D5A8 B4=0000 B5=0000 B6=0000 B7=0000 B8=000B B9=0001
BA=0000 BB=0000 BC=8D2D BD=0004 BE=094F BF=0101 B0=0000 B1=0000 B2=0000 B3=0C2A
B4=0000 B5=0000 B6=8D33 B7=001E B8=0000 B9=8D35 BA=0000 BB=F500
E0=0000 E1=0000 E2=0000 E3=0000 E4=0000 E5=0000 E6=0000 E7=0000 E8=0000 E9=0000
EA=0000 EB=0000 EC=0000 ED=0000 EE=0000 EF=0000 E0=0000 E1=0000 E2=0000 E3=0000
E4=0000 E5=0000 E6=0000 E7=0000 E8=0000 E9=0000 EA=0000 EB=0000 EC=0000 ED=0000
EE=0000
I0=0061 I1=0013 I2=00F7 I3=009F I4=0086 I5=0085 I6=0081 I7=0051 I8=0017 I9=0001
IA=00CC IB=0000 IC=00BA ID=0000 IE=0046 IF=0000
KB: 0001 0000
Changing config.divider doesn't change the ouput anymore (tested up to 0x15), except the frequency of events of course. Looks great! Great work mcell and mik!
My configs are here:
http://www.vems.hu/files/MembersPage/JasonRussell/test_config.txt
- Exceptions to the above config
- primary_trigger=03 (coil type this time)
Generic advice for measurements
If you see any strange behaviour,
- check the config and tables. Stupid values in h[2] can screw up ign outputs badly. (or ..._channel variables)
- if you see anything suspicious, it is essential to measure how behaviour changes as you increase RPM, change dwell, etc...
- try with input from external source. Try at leas 2 different RPM.
- try with stimout: without trigger input, with external trigger and optionally with stim output fed back
- provide benchstats, copy-paste the output of mdkffmdfff after doing some OutputTrigger (eg. mst3fmsp01msq3b).
- try with even-type trigger (and optionally with multitooth)
- as always, specify what config was used (referring to your page can be perfect)
- provide scope cutnpaste or snapshots if possible
- make sure you make clean; rm *.o before compiling firmware
- it is also good practice to show the output of mcd and mct (just in case: to confirm that make mtt and config upload was successful ).
Always check benchstats.unknownevent, eg. sg. like this is suspicious:
BA=0000 BB=F100 (interestingly: even right after powerup ??? Fortunately it ain't change after that.) coming from dispatcher.c:\nÿ1ÿ
Formal Testing
This is an important topic, well deserves a page of it's own: OnlineCourse/OffEngineTesting/Regression.