Ignition page of MembersPage/PeepPaadam/Porsche
I need to upgrade to overlapping dwell firmware, because when redlining (7000 rpm) it loses teeht occasionally near 6500rpm like Marcell predicted.
So, what I'd really need is someone who could point me in the right direction with the new primary trigger settings.
Wheel is 100-1 .
- Firmware 1.0.38
- and config (mcd, mct). Config updated here: MembersPage/PeepPaadam/Config
Car runs now very good, bug was in the swapped coil signal wires (two of them were swapped due to same colours and my mistake). Which setup do you use now ? ign_dualout=off ? Are you sure that all 8 cyls are running even at low-RPM ?
Thank you guys for the input and sorry for all the fuzz.
HW verification
- all 8 cyls have power on the ignition ? Yes, verified.
- power on the injectors ? Yes.
- same fuel pressure on both rails ? Yes, measured.
Firing order : 1-3-7-2-6-5-4-8
The reasonable wiring for this engine is that ignch1 and 0 drive cylinders 360 crankdegree apart. ignch3 and 2 also, etc... That is:
- ignch0,1 cyl 1,6
- ignch2,3 cyl 3,5
- ignch4,5 cyl 7,4
- ignch6,7 cyl 2,8
so use this wiring and don't change it! only change h[2] in configuration ! :
| cyl | ignch | EC36pin.. |
| 1 | 0 | 35 |
| 6 | 1 | 33 |
| 3 | 2 | 34 |
| 5 | 3 | 36 |
| 7 | 4 | 11 |
| 4 | 5 | 12 |
| 2 | 6 | 24 |
| 8 | 7 | 10 |
- ignchmax=07 (in any case)
For COP and proper setup, camsync is necessary. Without camsync, we have a problem.. While other wiring makes ign_dualout impossible, this wiring allows 2 different configs:
- ign_dualout ON
- ign_out=72 means ign_dualout
- h[2]=07 05 03 01 07 05 03 01
- "quasy-wasted-spark" setup. A full energy spark every 360 crankdeg (higher sparkplug wear). With a normal wasted spark setup, sparkgaps of opposite cyl are connected in series: the exhaust gases drop lower voltage and dissipate only 10% of the energy: so 90% goes to the cyl we want to ignite the mixture in.
- fully sequential (ign_dualout OFF)
- ign_out=70
- h[2]=07 05 03 01 06 04 02 00
- without camsync (primary_trigger=02) this is not possible. Though engine runs, it runs in a limp-home mode. Car runs good in this "sequential" mode when RPM > 1500 ! All cylinders are firing and no misfires. Tons of power.
- but RPM < 1500 it tries to resync for every crank-rotation (multitooth.c syncronize_igncount() called when detecting missing tooth and camsync disabled. Acts only when ignstate==0, that's why it's only seen at low-RPM ), and only fires 4 cylinders (the last 4 counting down from ignchmax: entries 7..6..5..4 from the h[2]. And does NOT touch entries 3..2..1..0)
- it would be possible to remove the resync-condition; during cranking, engine would kickback 50% of the cases: and start 50%. Good solution is not possible without camsync. It's a good question if hack patches are feasible to cure what cannot be cured
OLD, BAD wiring
With the old, BAD wiring (cyl1=ignch0 ... cyl8=ignch7) the following was the h[2] ignition sequence (ign_dualout=off):
- h[2]=07 03 04 05 01 06 02 00
This is no longer used, since the h[2]=07 05 03 01 06 04 02 00 has same effect with the reasonable wiring.
Meta notes
It is possible to help, if data published in wiki is accurate
- config
- tables
- publish mcd and mct dump (obligatory. These are used to reproduce setup). MegaTune .msq is optional (not used).
- firmware
- wiring
- behaviour
- measurements / logs
When updating, remember to delete old, confusing info (or move to a history page, but don't expect anyone to review or consider that page).
Also remember that phone is no substitute for these written information. MembersPage/BmwVenesz (despite much experience with motec) was calling everyone from Dead-Sea to Mount Everest and still couldn't start the car. Than he put on car on trailer: we counted the teeth from missing_tooth to cyl1,4 TDC and made the configuration changes in 5 mins that started the car. Had he properly written down the ignition wiring and missing_tooth position (we requested several times), we could have made the simple config changes saving him 1 week of telephone and frustration, and 2x140 km trailer ride.
MegaTune and firmware
- burned all values to ECU in MegaTune. Dumping the config, it showed ign_out=72 (good, that is ign_dualout=ON)
- Do I need to update the firmware from 1.0.38 to something newer?
- you can try 1.0.53 if no better idea, but shouldn't be different than 1.0.38
- if you apply camsync, or start to lose teeth at 6500 RPM, or dwell-limit near 5000 RPM, it will make sense to upgrade to overlapping dwell firmware (1.0.61 or later), that eases the cam-constraint and shortens the longest-interrupt-path (but still experimental and requires more config-options; table-testing with given config), but definitely make it work smooth (at least idle .. 2500 RPM) with non-overlapping-dwell firmware (1.0.55 or older)
Marcell is thinking about possible hacks that would make this not supported setup (no camsync when camsync is needed) perform better (without negative effect on the innocent).
All outputs are logiclevel (0 IGBT inside ECU). Logiclevel outputs pull to GND (0V) when inactive, so when you measure DC voltage with DVM, from 5V as base, you read -5V
- Wiring is separate, i.e. every ign output is to one coil respectavely, also I wired separate +12v supply and grounds for them.
- Switched +12V to coils, they are fused (two banks, 15 amps each) and grounds from the engine. Signal wires come separately from VEMS to each COP.
Car runs ONLY on primary trigger which is stock 100-1 tooth wheel on the flywheel and I don't have cam sync (not supported setup).
I tried to fire COPs in pairs, as wasted spark, but it did not make any difference.
Note that I fired BMW m50 engine COPs also sequentially by using just stock 60-2 trigger wheel, same low-rpm problem. No surprise, the same there: camsync is needed for proper operation.
The engine runs, but high dwell (3 .. 3.5 msec ?) was needed.
- "little voltage" from VEMS is undestood now (it's cause by measuring DC-mean of a part-time-on signal)
- what outputs ?
- what conditions ?
- what's connected ?
- measured how ?
A: I measured all logic level ign outputs and nothing was connected at that time. Used DVM in usual DC mode and it read 1 volt on ign. event. As Marcell said, it's ok.