MAR 06/2006
Tried "try a 40..100k pullup (to 5V) on the VR signal (100k between EC36-pin27 and EC36pin29)." used a 100k resister between pin27 and 29, no diferent. Tried the 40k, a little better but nothing worth getting overly excited about. Just to double check, I DID NOT do one (resistor) to gnd at all, just jumped pin 27 and 29 with the resister, is that correct?
Also got frustrated and built a new 12-1 wheel using a ford sensor. Carved the missing tooth area to get the amplitude perfect (.5 v lower than the rest at 8v), made a slight difference at higher rpm (1200) but again, still has the same problem.
Hacked ofF the diodes by the input jumpers on the bottom of the board.
Nothing seems to be making any notable difference with the vr sensor.
Set distributer up with a drill on test bench. No gnd or gnd 5 loads other than fuel pump relay grounded by an injfet and the vr sensor to gnd. Still have 4 gnd5 wires and 1 gnd wire in test harness. Spin drill between 100 rpm all the way up to 1500 rpm, trigger drop out no different than when ecu is in the truck. Changed from multitooth to coil type trigger, 24 teeth and used a good, straight stock honda distributer on the end of the drill. Problem still the same. Used the 4 tooth part, set the tooth count to 4 and try again, thinking if the teeth were too close together on 24 it could cause a problem, it is ALOT worse when set up as a 4 cyl. Problem definately has to be either on the board or in my config. Messed around alot with the config settings, unable to make the problem any better.
Going to try and de-solder the diode from the bottom of the board that I soldered on (d27???), said it was optional, not sure what it does. If it protects something I dont care, I am so beyond caring about frying anything anymore.
TODO
Take pic of bottom of board.
Thought my broken gnd wire was going to be it. Just as bad as ever after wiring and ecu repaired, back to the drawing board.
Now that I have the RPM constant set right, the thing idles excellent (maybe ign idle control is working) but the trigger drops out way more than before (I can barely get it to start).
Replaced all my ground cables, added some new ones. The best I can manage is about 1100rpm safely, much lower than that and it will for sure start acting up. Here is a [pic of my board].
![[pic of my board]](http://www.vems.hu/files/EricN/lm1815area.jpg){kind=link}
Another issue that I think is related, if I hit any part of the engine with a tool I can make the ecu trigger. The closer I get to the distributer, the less force I need to do it. I need just to lightly touch the distributer with a tool and it fires away.
Also drove the truck today just for the heck of it. The problem actually happens at almost any rpm, it just isnt as noticeable when you are free revving the motor, add a load though, all hell breaks loose.\n
primary_trigger=01 secondary_trigger=02 tooth_wheel=17 trigger_tooth=01 another_trigger_tooth=04 crank_minper=0D tooth_wheel_twidth1=05 tooth_wheel_twidth2=19 cam_sync_r_edge_phase=8D cam_sync_f_edge_phase=21 reset_engphase_after=D8 ign_tdcdelay=6A ign_dwell14=2F ign_dwell6=B9 ign_crank_advance=14 ign_out=70 ignchmax=05
The full config is MembersPage/EricN/Config
I GIVE UP
The VR trigger is killing me, I have tried many things, even shorted pin 5 and 8 on the LM1815. It definitely worked a little better, but I still get the same dropouts.
- Stuck the distributer in the lathe, cut out all the runout. I now go from a highest peak of 7.4v to a lowest peak of 6.9v and it is not around the missing tooth. Even now, it still cuts out.
- Board is a v3.2. I tried the mod by cutting the trace off of pin 7 and putting a 100k resistor in its place. It didnt help. I also shorted pin 5 to pin 8. Would that not make the pin 7 mod completely unimportant?
BELOW THIS IS ALL GARBAGE, JUST WANT TO SAVE SOME FOR REFERENCE NEXT TIME I GET THE IDEA TO TRY IT
Started over with input trigger ideas. Took a VR sensor and 24 tooth trigger wheel out of a Honda Civic distributer. Ground off 1 tooth, pressed it on to a gutted Unimog distributer (took the whole advance mechanism out). Welded a plate inside, mounted the VR sensor on the plate. Now I have a 24-1 trigger wheel driven by the cam.
If I see right, this wheel is very prone to output bad signal at the missing tooth. If you can, record it with a soundcard. To be safe, divide by 50 (or so) using a 100k and a 2k (or so around) resistor. It might output just higher amplitude signal at the missing tooth, or maybe it has a ghost pulse too.
When cranking, I notice that every now and then the rpm dies off to zero for the same half second then comes back. Likely trigger signal shape problem.
If I touch the body of the distributer with a wrench (vr sensor in distributer is my primary input trigger) while the key is on but car is not running, it triggers the ecu and injectors and coils fire a quick shot.
- try a 40..100k pullup (to 5V) on the VR signal (100k between EC36-pin27 and EC36pin29). This raises the noise threshold (good), and unfortunately the minimum necessary signal needed to trigger (bad). If you measure the smallest signal amplitude (eg. with notebook soundcard, and 2 resistors to divide, see ElectronicDesign/TriggerSignalGenerator) that still triggers, you won't be bitten by "cannot start" problems as the weather gets cold and battery weak.
- If you have a v3.2 or older, a 150k (or 100..220k) resistor in series with C38 might help. See InputTrigger/RunOut . This should be able to pushe down adaptive hysteresis threshold under 65%. TRIED IT, DIDNT WORK, HELPED A LITTLE
- draw your GND and GND5 connections, with realistic line-lengths and take note of wire-thickness.
- Dont have a nice program to draw it in but I measured it out. The gnd5, I have 4 wires, 20" long, 18guage. It goes to the engine block. The gnd is 2 wires, 18 guage, 20" long, bolts to the engine block. The sensor gnds meet about 6" away from the ecu. They are 18 guage and the longest one is 10 inches. The vr sensor is the longest at 38". The VR ground wire goes right inside the ecu and is soldered to the ground on the board above the input trigger jumpers. The only things loading up gnd5 are 2 injectors, the fuel pump relay and the wb heater.
- nice, GND as good as can be!
You might also want to capture (in TerminalProgram) the output of Manmdkffmdfff command, especially around the strange behaviour. You can switch off the continuous output by mdf00 or reboot. Captured it, it is all in hex, I see I have to do some perl stuff to make it readable but cant make it work
Here it is\n
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
With the engine running, mda28 will force 10 crankdegrees advance, that is decimal 40 * 0.25 degrees; mda01 forces 0.25 degrees advance, mda00 restores normal advance calculation; see refcard on GenBoard/MenuSystem) and pointed my timing light at the crank, adjusted the distributer till #1 was firing dead on tdc.