The output side is explained on MembersPage/MarcellGal/EngineSwap/Ignition
- 60-2 Opel Astra X16SZR factory crankwheel pulley (diameter=140mm) had to be mounted on the Daewoo engine so the air conditioner and alternator are driven via V-belt as before (the Daewoo engine had multirib belt, and it seemed easier to mate them this way). However, because of this, the 60-2 toothwheel was not centered any longer. Therefore:
- we fabricated a [stainless steel mounting adapter] so the VR sensor is more central related to the wheel. This has been slightly adjusted since than, for smaller gap.
- VR sensor (resistance=480 ohm)
- VR sensor mounted with appr 40..43 mm, so it sees the missing tooth earlier than it would from the factory VR sensor location. This is a side effect of the mounting: the reason for the adapter was to make the sensor align (better) with the crankwheel (read above).
- The factory mark was 4 teeth (24 degrees) off (compared to what we expected) because of the Daewoo-Opel hybrid. We checked TDC through an open sparkplug hole (much after the engine had been started :-)
- engine rotates clockwise when viewed from the timing belt side (To avoid missunderstandings engine rotation described as looking at the end opposite to the output shaft). This is verified in 5th gear, moving the car a bit. So the engine rotates in the same direction as the wheels in gear 1..5.
Sloppy VR mounting problem - gone
The gap between 60-2 crankwheel and VR sensor (with the home-made adapter) was originally a bit too big. At cranking the signal (low amplitude because of low RPM) was susceptible to noise.
- It usually didn't start on first crank
- wheel_error++ at cranking
- I captured InputTrigger/TriggerLog and was horrified how random the trigger pattern at cranking 150..200 RPM
- when I modified multitooth.c so that significantly_bigger() macro detects >2.5 teeth distance instead of original 1.5 it did not start (no RPM) at all: likely there are random mistrigs that (in lucky case) kick the engine over to RPM range where the VR trigger signal amplitude is good enough for normal operation.
Since the gap had been made reasonable, it start OK.
Signal quality and polarity - VR fixed
After hooking up LEDs and notebook soundcard it became obvious that trigger syncron is lost soon after it would fire up (ign and inj output not activated for some periods).
Even though MegaTune logs revealed that there were no resets, GND5 turned out to be very weak (and clearly interfered with GND through internal protecting diodes that I used between GND and GND5). 3 powerful 1mm2 wires used for GND5 fixed it.
With mde40 logging (see InputTrigger/TriggerLog) it became obvious immediately that trigger signal was reversed. Instead of one 3x long pulse (60-2 wheel) there were 2 longer teeth (appr. 1.5 .. 2x long). We connected it according to the documentation (neglecting the result of simple but effective "hammer" test). Confusingly, usually there was trigger signal, it was just unstable and lotsof wheel-errs showed up (eg. on LCD).
Runout - InputTrigger/RunOut
I could reproduce the runout very easily on the table (I got a flue anyway incar testing ain't work). I hooked up LEDs to injection and ignition outputs.
- Pull up a free injector FET output with 270 Ohm resitor to 12V. Position 4 in my h table activates the given channel.
- connect the output to primary trigger input through a 10nF capacitor
After playing with mstBA (60-2 toothwheel OutputTrigger) and msp01 .. msp06 (mostly msq00) I noticed that ignition and injection LEDs flash unevenly. There are full crankrotations left out. This is typical for loss of inputtrigger signal. I hooked up a soundcard to check signal with audacity and saw that signal amplitude is significantly lower for the first tooth after missing tooth. So I inverted the stimulator output from 40 => 80:
so after the missing tooth it starts with a (neglected) rising pulse, than comes the (used) falling pulse.
I could discover the same pattern in the sound recording of the real VR log, and LED flashing in the car behaved in a similar way.
I'll have to patch the LM1815 circuit with the resistor that lowers the adaptive threshold (or jumper to nonadaptive mode).
Position - solved
I didn't have time to hook up and log the factory ECM, but the changed block changed position anyway.
The way it was found
After a good start from photos (see below), I just simply sweeped through the tdc_delay using the directly attached keyboard. It's very easy to do so :-) Apparently best results:
- trigger_tooth=10 (decimal 16)
- with ign_tdcdelay=78 (60 degrees)
- and cranking advance=32 (12.5 degrees)