Auditrigger on a inline 5 cylinder Audi engine,but now only with distributer,and 1 coil.
After 16 hours of hard work we tried to start the engine.Everything looked great,we had RPM signal during cranking,we had spark (at the right cylinders),and injectorclicking too.Unfortunately the engine did not fired up,after 4 hours of continous problem searching,we did not found the cause of our sadness.We tried different ignition and fuel settings.I also realised that the car dont really want to pull vacuum under cranking (in the vems,map sensor),tough i can feel it with my hand.(no plumbing,only throttle body).Note that the car was running in the past with dual K-jet without problems.We also checked Compression in the cylinders,and they are good.
- firmware 1.0.30
- [tables ]
- [config]
- mdd0c logs (zipped together) http://www.vems.hu/files/MembersPage/NanassyPeter/ClonInstallStartUp/mdd0c_logs.zip
I formatted the binary logs with /svn/firmware/utilities/vemslogger/auditrigger_logformatter <mdd0c_3.log >mdd0c_3.txt and watch:\nÿ1ÿ
The offending crankhome VR pulse typically comes after 1 crank rotation (==0.5 cam rotation, 2.5 events) instead of 2 crank rotations. So that means : check your cam-HALL ! that is responsible to mask every second crankhome VR pulses
- does the HALL get supply ?
- Is it connected at all ?
- It turned out that HALL was NOT connected electronically because the connector somehow did not reach the pins
- maybe cam-HALL position is rotated so the crankhome VR signal is not masked out ? The cam-HALL should be active (pull down to GND) at the same time when the crankhome VR comes! This is important to check
- what voltage do you measure on EC36pin13 ?
- when engine off, but power on: likely 4V or so (0V also possible if the HALL is just at the window) => goes in between when engine rotates. Measured mean (DC) voltage should change something like 4V => 3.9V when engine starts to rotate because the HALL pulls down 1/40 of the time.
- Don't try HALL 5-window yet. Make the 1-window CAM-HALL work.
- Don't invert crankhome VR polarity. It's likely good. When the CAM-HALL is working, we can tell for sure
It's good that you published config and mdd0c logs. After trigger hammered, don't forget
- take logs with MegaTune (remember to say bye after fiddling in Man mode, taking mdd0c log for example).
- we'll watch for fuel pulsewidth (possible fuelcut due to TPS>60% floodclear at cranking or kPa fuelcut, whatever...). I checked this,there is PW
- is the ignition right where you like ? Do you have a lamp? I remember this engine has a custom crankhome VR position, a few degrees off.
- if it's just 2 degrees or so, the normal procedure of varying trigger tooth or ign_tdcdelay should work well
- is the fuel system working ? fuel pump, fuel pressure ?
- Sure, its working,there is a LOT OF fuel i the cylinders
- Is it possible to just ditch the camsync HALL signal and map the ECU for only crankVR signals?
- of course NOT. It has 5 cylinders, so from crank it's not possible to tell when to fire (even if it has distributor). With CDI and COP it might work firing twice as much (like a 10 cyl), but a lame solution
- however, the cam-HALL (the 3d trigger signal, EC36pin13) would work with 5 windows just as well as with 1 window. Just make sure the HALL
- is surely high for one of the crankhome pulse and (traditionally the 134BTDC cyl5)
- and surely low for the other crankhome pulse (traditionally the 62BTDC cyl1)
- whatever it is at other positions does NOT matter
[ logs with working HALL] show:\nÿ2ÿ
This is perfect. (the initial x is before it catches sync).
In MT, during cranking that the RPM signal is silly, (goes from 0 to 30800, and the other signals fluctuates badly too, like MAP=61300 kPa ;-). This turned out to be noise in the RS232 cable (bad type USB-RS232 adapter anyway), not related to the cranking problem.
- check the grounds
- GND
- GND5
- check flyback
- check mcd config (trigger related part) against what we made according to InputTrigger/AudiTrigger ( maybe http://www.vems.hu/files/MembersPage/NanassyPeter/ClonInstallStartUp/mcd.txt ?)
After the trigger signal seemed to be hammered (the HALL simply didn't make it's way to the ECU first!), we tried to start the engine (with 1.0.30 firmware).
- the engine started immediately, when controlled from 5 window HALL (from a v3.x that had primary_trigger=HALL). This confirms the basic fuel and other settings to be roughly OK
- note: we didn't actually compare the config and tables of the 2 different v3.x. They should have been same, but ain't hurt to compare.
- the crankhome-VR polarity was verified in 3 ways:
- mdd0c log showed fine trigger
- hammer(/wrench) test. DVM DC voltage shows positive when approaching wrench and negative when moving away
- scope. Although the VR sensor distance is frightening 1.5 .. 1.7mm, the signal amplitude seems to be enough (with internal 43k pullup resistor on crankhome-VR)
- During the experiments, we happened to pull apart crankhome-VR (the backout plate in the EC18 was not there, so...). Although the mdd0c log showed this immediately, we happened to wast 1.5 hours to measure here and there (suspecting broken cores inside some wires of the harness), power the scope and things like that. When we brought the light to the ECM to reveal the EC18pin12 empty and wire with EC-receptacle waving ;-)
- pretty lame, but this was late at 2300 or so.
- we found that reset_engphase_after=F0 was different from Miska's so we set to 05. Didn't help
- one suspect is the cam-HALL phase. If the crankhome VR of 134 BTDC cyl5 (that we adjusted config for) sees high cam-HALL at some point, and low at other point, because the cam-HALL edge is near, that would suck. But when mdd0c log was good, it should have started, right ?
- note that the high/low state of cam-HALL only matters when the crankhome VR pulse comes. Therefore the 5 window HALL is also just as OK to use, as the 1-window. The high/low state is opposite 360 degrees apart, anyway (actually, the 5 window is harder to get wrong)
- also, the timing light showed that timing of cyl1 was OK. It just missed occasionally when losing sync, even when mdd0c showed "trigger fine".
- too bad that we didn't watch wheel-error counter at all. (but we took mdd0c log several times).
- maybe there is something around camsync that pulls engine.status_trig & _BV(engine_in_sync) low
- there was no TPS (==0%) but this shouldn't matter (we set overrun_fuelcut and fuelresume to very high RPM)
- megatune logs (when they were good, no noise!) showed that fuel pw suddenly falls to 0 when engine rpm > cranking_thres. The engine behaviour was also like this.
- the rumbling got less when we went from 399 to 299 RPM threshold
- we were focusing so badly to the trigger that we forgot about basic things like this. TODO: check the config, what could cause this. As an experiment, raise cranking threshold to 499 RPM to see how it changes.
- [config] used with 1.0.30]. wtf?! airden_ignore=00 instead of airden_ignore=62 ? Although MAT is not connected yet, shows -16C or so, this doesn't look good at all. I wonder how this got into the config ..
- iac_conf, knock_conf, als and shiftcut all seem bad
- als_retard=FF ??? why not 00?
- also, _channel=.. must be checked for possible collision. DONE, seems OK
We'll likely use a capacitor to inject HALL into the primary-trigger VR (EC36pin27). This allows easy retry with the original trigger system.