Trigger related subpage of MembersPage/MarcellGal/PowerAndTraction
The trigger is now setup well, including the HALL sensor. HALL sensor in the distributer works now, and position set, so startup is fine now, without kickback caused by bad initial phase (which happened with the bad HALL-sensor).
Analyzing a cranking triggerlog (captured with 1.1.92) shows:
- after more than 1 but less than 2 crankrot, the sectrig pulse finally arrives.
- 41.4 msec / 40 deg means ~161 RPM
- 16 * 8 degrees later the triggerlog shows a spark event
- measurement starting 40 degrees later, 36.2 msec /80 deg means 368 RPM (so the RPM more than doubled !)
- Actually, the most extreme change in only 40 deg seems to be: measuring 37.2msec/40 deg (before 20ATDC), and 20.4 msec/40deg (after 20ATDC) so time is 55% (not 182%) of the previous period - quite impressive (with factory flywheel)!
- in a few more events the engine is up at 8.46 msec /40 deg = 788 RPM: the engine is running.
- cannot start the engine much quicker than that with this trigger hardware setup.
- Could be possible to start up faster with a wasted spark trigger, especially with a smart MembersPage/YasecElise type trigger
- or with cam-HALL signal also connected to a 4th trigger input (the cam-HALL only used for masking away every 2nd crankhome pulse in this setup, but theoretically it could be connected directly into the exhaust-cam signal input but that would need firmware support for this marginal gain)
No wheel errors during running (wheel error counter stays at W01 after startup). Because of how they mounted the distributer shaft, the HALL trigger became nonstandard, as it coincides with (masks out) the 134 BTDC cyl5 crankhome pulse, not the 62BTDC cyl1 (see InputTrigger/AudiTrigger for more explanation on the auditrigger). Since the distributer has "only" appr 190 camdeg window to set position, it seems impossible to rotate the distributor to standard position (without removing the shaft). It would make little sense to rotate anyway, since this setup allows firing 72crankdeg earlier during initial startup, and the 360 crankdeg off-phase was very easily accomodated in configutation:
- the orig "cyl5 TDC follows tooth0, with ign_tdcdelay=130 crankdeg"
- rotated 360 degree, that is: "cyl1 TDC follows tooth0, with ign_tdcdelay=58 crankdeg"
- I have to strobe it to verify this 4 crankdeg offset compared to the actual pin location 62 btdc cyl1 (according to not precisely verified "internet information"):
h=00 48 36 24 12 .. .. .. h=66 30 20 30 20 .. .. ..
This assumes 1,2,4,5,3 firing order, and the non-standard setup as it is wired currently:
- cyl1=Stepper-D ( using orig logiclevel-input power-output stage and factory coil), entry "0x66"
- cyl2,5: power-input wasted spark coil, ignch2, entry "0x20"
- cyl3,4: power-input wasted spark coil, ignch3, entry "0x30"
The 2 VR sensors
- look almost identical (one has grey boot, the other has black)
- with same pinout,
- measure 1001 and 985 Ohm, respectively
- has 3-pin AMP-style male connectors.
- using a DVM in DC mode, and a wrench, we checked in 1 minute that the mid-pin must be GND for proper polarity (and the pin at the side must be the signal). The proper polarity (with LM1815 that senses falling edge) means signal is positive when wrench gets closer to sensor and negative when wrench comes away
- Of course, in the factory wiring the GND (which is shared between the sensors, and probably with other sensors) is the other, so I need to swap the 2 wires in the harness to reverse polarity . This confirms what Miska documented earlier.
- the holes at the very top of the clutch housing, see [pic15] and [pic14] look like those for VR sensor (maybe not: hard to tell with engine mounted). Not sure what this can be good for, though....
- See [pic17] for the 2 good candidates for crank135 and crankhome: both on the right side of the engine, close to each other.
- the front sensor is the crankhome. This sensor closer to the sky, it sits deeper in the housing, see [SJM pages]
I had to scope it anyway, to verify crankhome polarity and HALL window relative position. (this, if logged during running and different loads, would also show cambelt sloppiness).
Found some metric (M6?) screws, used a washer and an extra female to make it proper length, and mounted sensors in the 2 holes close to each other.
- grey one reads flywheel teeth
- black reads the pin behind the flywheel
That is, with normal mounting. I will have to check (and document) if I followed the standard (colors: no other difference between sensors), or got it with colors swapped.
Cam-HALL sensor - finally fixed and positioned
There are 2(-3) possible cam-HALL setups, with exactly the same function.
- timing belt sprocket, at the front-end of the camshaft
- inside the distributer
- dummy distributer
- real distributer (that's what I have)
I found that ...
- I see the window (I have a one window shutter-wheel mounted). A 5-window wheel would be easier to setup, because a random position is likely good there, while a random position is likely bad with the one-(small-)window wheel.
- after unclamping the M13 screw, I can move the HALL sensor (moves together with the "Bosch" housing)
- DVM measurement: the sensor gets 5.08V
- Note that 3B Hall needs 12V supply (MembersPage/GintsK) Sorry! Only 7A n/a 20V engine have 12V supply in accordance with BOSCH manual. 3B and AAN - 5V.
- That is weird. We did NOT change the 3B harness, and the HALL sensor receives +5V supply from the ECU (from same point as the TPS supply). So I would NOT dare to apply 12V there (and possibly damage the TPS or other sensors).
- But I can disconnect the HALL from the harness and provide 12V from other means. This will not hurt if the HALL is damaged anyway, and will reveal if our HALL sensor needs 12V. The almost factory engine which was in (in a broken state, before our built engine was installed) was hacked in unknown ways, so anything is possible
- (when sensor disconnected): I can pull the HALL signal to GND with a small wire (and measure EC36-pin13 go from 3.97V to 0V)
- but the sensor will not pull to GND
- examining a bit closer: the HALL sensor connector in the car looks suspicious at least, almost like if the male receptacles were broken off, maybe the sensor is good just no contact to the female connector ?
Conclusion: either the connector tricked us badly, or my HALL sensor is bad.
1.1.23 allowed running without HALL (with bad starting: every 2nd attempt on average). The factory motronic setup continues to run if HALL sensor is disconnected after startup. But 1.1.23 also allows to start up the engine (with 50% chance) and good running (without retard: retard makes little sense when only HALL is faulty, since timing is done from crank anyway). This does not effect normal operation, the same code runs perfectly when there is proper HALL signal, of course.
Spare HALL sensor - obsolete
I verified that my spare HALL sensor works (from the audi distributer with the slightly bent shaft)
- black: GND
- red: +5V supply
- green: pulled to +5V via 4k7
- surprisingly, the sensor pulls output to GND when nothing in the shutter.
- the sensor releases the output (=> 4.96V by pullup) when ferrous material is placed in the shutter
But I don't know if such a spare sensor makes things easier. Not a bolt-out, bolt-in thing.
Although I have the spare sensor in the distributer with bent shaft, a small HALL sensor, and 1GT101 sensor too, it seems difficult to mount it in a sane way (either on front side wheel or inside distributer). Had the front-side wheel have 5 (or 7 or any odd number would work) spokes, instead of 6, that would make it easier to watch those with a 1GT101 HALL sensor.