I got Volvo Amazon with B20.

----

36-1 toothwheel

I mounted a 36-1 toothwheel to crank from Ford (it's just like [this]) and Honeywell GT1 hall sensor. It did work when rotating crank manually, but engine didn't start. I get back to coil-type trigger, started the engine and looked hall sensor signal with scope. Scope showed that signal was lost when sensor was near the trigger wheel and it did work somehow when distance was out-of-GT1-specs, 4-5mm, but there was still few errors. I tried hall as primary trigger when distance was 4-5mm, but LCD showed Wxx growing all the time.

Jörgen told that the trigger wheel was not suitable for GT sensor; "But I have used the GT sensor a lot and have used it far outside it's specifications with good and bad results. What I do know is that the sensor is self adjusting, you'll notice this if you put a piece of steel in front of it at 1mm distance and then move it to 7mm distance. It will go high for a while and then selfadjust to see the steel again"

"I think this is what happens with the trigger wheel you have, it sees the solid disk for some reason, probably because there is not enough material in the trigger teeth. Moving the sensor further from the solid disk in the axial direction (of the crank) could help."

"If you choose to make a new triggerwheel you should know that the GT1 sensor does not work well with trigger made up of large raised areas with smaller lower areas as a trigger wheel made by drilling holes. This is because of the self adjusting feature."

Because I did weld ford toothwheel to crank pulley I did try another hall-sensor, made by Hamlin. New hall sensor did work and now my GenBoard/VerTwo is trigged from 36-1 toothwheel and it's controlling both, fuel and a single-coil (distributor) spark with IGBT.

<code>

primary_trigger=01 # multitooth (rising edge at the processor)

secondary_trigger=02 # Cam sync disabled

trigger_tooth=01 # 01*10 degrees after when the missing tooth is detected

tooth_wheel=23 # 36-1, 35 tooth

another_trigger_tooth=12 # also fire 18 teeth (180 degrees) later than trigger_tooth (for 36-1 wheel)

crank_minper=40 # anything from 01 to 50 is fine. If measured event-period (that is every 180degrees for a 4-cyl) is shorter than this, it's considered and error.

tooth_wheel_twidth1=0A # Tooth is 10 engphase units

tooth_wheel_twidth2=14 # Missing tooth is 20 engphase units

reset_engphase_after=D8 # 216 engphase units

</code>

----

Bosch wasted spark coil

What do you think about this [Bosch wasted-spark coil]? It's pretty cheap (~70 euros here!), but hard (?) to find from a junkyard. Seems like a good transformer.

I cannot understand the dwell(VBatt) function. Maybe I should calc with the i(t) = VBatt/R * ( 1 - e**(-t*L/R)) to see if they match Bosch values (R=0.5 Ohm, L=3.7mH for coil primary).

----

Trigger problems when cold (updated)

My GenBoard/VerTwo has 36-1 hall type trigger, it has run 1.0.12 all summer and worked just fine! Now when engine hasn't been warmed up with electric and it's cold outside (below 0'c) the car starts worse or it doesn't start at all. Cold starts has been problem for a while, I just haven't check Wxx errors before. On very cold weather (like below -10'c) while cranking only Wxx grows very fast and theres no RPM reading on lcd! When more or less below 0'c outside car starts but there is still (too many) trigger errors at crank. After engine starts, there will be no Wxx errors anymore. Also hot starts and 0'c are just fine, maybe 1-2 errors _may_ come (if any at all!) at crank but the engine starts very fine. So only very cold cranking is the problem!

I did check hall signal with Fluke and it showed on very cold engine (when only Wxx growed, no RPM on lcd) 62..56Hz, so 106..96 rpm on 36-1 toothwheel. One of my datalogs shows crank rpm 150 when engine is bit warm, sounds right to me. I did try to move hall sensor closer to the toothwheel, seems it didn't help. I have datalogged crank voltages with Fluke, hall voltage drops only about 13mV when cranking and battery voltage (-15'c or so) drops near 9v. Vems doesn't reset while cranking. I'm unable to confirm hall signal on very cold weather with scope, but Fluke still gave me believable (62..56Hz) frequency.

Now it's checked that firmware upgrade just few days before these real cold weather has nothing to do with the problem. I have 1.0.12 all summer, now I'm using 1.0.25 and it works just fine after engine starts.

<code>

primary_trigger=01 (normally its not filtered 01, now trying 09 too)

secondary_trigger=02

tooth_wheel=23

trigger_tooth=01

another_trigger_tooth=12

crank_minper=32

tooth_wheel_twidth1=0A

tooth_wheel_twidth2=14

cam_sync_r_edge_phase=8D

cam_sync_f_edge_phase=21

reset_engphase_after=D8

ign_tdcdelay=A5

ign_crank_advance=14 (while maximum advace on n[] table is 46 degrees)

toothrel_normal=8A

toothrel_missing=00

</code>

Above is my normally working configuration. On Marcells advice I did try later trigger tooth:

<code>

ign_tdcdelay=69 (thats 52.5 degrees, originally 82.5 degrees)

trigger_tooth=04 (thats 30 degrees later)

another_trigger_tooth=15 (also 30 degrees later)

</code>

Tried this on 0'c day, with these settings engine sounds very differend while running! Now I check'd with a timing light and when using later trigtooth (above or any similar with trigtooth=2..3) I need to add more tdcdelay than calculated to get the ignition right (mdaxx). But still sound of the engine isn't right! Actually later trigtooth I tried, more weird my engine sounded.

The 11msec dwell explains it all. When trigger (pulse) to spark (according to actual ignadv) time (time depends on RPM, always the case at very high RPM) is less than the dwell (at actually VBatt), the dwellstart is started timed from previous spark. You really don't want this to happen under 1800 RPM. Not seeing any doc about your ign cannot be sure, but even possible that your ign module autofires if dwell gets too long because of this (if engine slows down in the meantime, this can be significant at low-RPM). Since this will not change this month (even if it's planned to be improved for multitooth wheel), I'd just go to the yard and get an ignition transformer / module that works with max 4 .. 4.5 msec dwell.

* Sorry, I really should make doc of my system. My ignition is a single-coil (distributor) spark with IGBT (as mentioned before on this page casually). Dwell is just calculated value from measurements, 4*L/R => (4*0.0091 H)/3.1ohm=11,74mS exactly. Worked before so I haven't even try to lower it. 3.1 Ohm resistance sounds like an old type ignition transformer, so it likely needs min 6msec dwell, this justifies to get a newer type. If it was v3.x, you likely would go to wasted spark at the same time.

* Still, this is just explanation for problem of using later trigger tooth. But dwell6 increases dwell on cold day as VBatt is lowered, so... similar effect as you see with later trigger tooth.

* Even though on a very cold day I got trigger errors even on flood clear mode. It shouldn't make trigger errors just because dwell is too long. Indeed, wheel error shouldn't increase (except max 1..3 at every cranking), it's independent of dwell.

** But I have to see again on a colder day that if moving hall closer to trigwheel really helped or not.

When I rotate my engine to TDC, I see that it's 11th tooth just passed by the hall sensor after the missing tooth (=wider gap).

GenBoard/VerTwo hall signal goes thru some resistor and protection zener (but it's not inverted like with optocoupler).

So what next? I could not suspect hall sensor or toothwheel anymore, because everything works just fine after engine is hot (cranking rev is higher). Maybe try add capacitor near hall sensor, but Hamlin hall has internal cap [see its pdf].

---

Trigger problem continues 19.2.06

Now I did check the trigger signal with the scope. Signal was what it should be. Engine was cold and battery energy wasn't that high as when hot. I did look pulsewidth (pulse-/pulse+) from scope and on one crank rotation it was 4+7mS and 8+12mS (I think first one was pulse-) so engine speed varies from about 156 to 85rpm!

It look like Wxx errors grows on every crank rot. Using later triggertooth does not help. Now I also have good short dwell coil. This was also tested on flood-clear mode.

I have narrowed problem to:

* maybe reverse polarity VR ? I had exactly the same problem earlier. Check (take and publish) InputTrigger/TriggerLog (or scope also helps a LOT if available.)

* maybe missing tooth amplitude too high because of wheel toothshape ? => try C38 mod (unless it's v3.3). Even a soundcard reveals is the amplitude varies a lot.

* Hall sensor, it has self-adjusting feature. Maybe on slow speed it adjusts and pulse get wider. I doubt this one a bit.

* Vems does not handle that wide speed/pulse variation. If missing is detected from somekind of mean value of trigger pulsewidth it would go wrong in this case? Should I try to bend 1 tooth so missing gap gets wider?

* My question is how does vems handle multitooth sensing? Should vems handle this wide variation? I just want to make sure this isn't vems related! Maybe I should table test it with really clean 36-1 signal variating big time.

** first focus on the wheel-error increment, you should eliminate that.

* Also where else than junk yard I could find VR sensors? One with M12x1 thread in it would be best but not required.

** VR sensors are everywhere (in most cars today) but finding one that can be mounted with less than 4 hours of work can be a challenge