More Data

[grippo]

I believe the flanking is due to how the delay teeth/ trigger offset was set up. You have tach periods of 60 deg and the ignition putputs alternate, so you have a max of 2 tach periods (120 deg) to dwell + spark advance. At 2400 rpm that is 25 ms / rev and about 8.3 ms for 120 deg of rev. So that is way more than enough to get in a 5 ms dwell and spark. PROVIDED you have delay tooth and trigger offset set up so that you can start dwelling right after the tach pulse and keep going for close to 120 deg before you spark. If you don't spark, the code will do it for you. Dwell cannot start until a tach pulse arrives and you must spark before 120 deg has elapsed (= 2 tach cycles). That is probably what the flanking is about - you want to spark after the 120 deg elapses, but the code won't let you - it cuts the dwell short (sparks) at the tach pulse. So if you play with delay teeth and trigger offset I believe you can get rid of the flanking. At some point, but way past 2300 rpm, you will have to reduce the dwell but you should still be able to fire a decent spark. I guess what I am saying is that you want to adjust delay and offset (but keep offset negative) so that at max rpm you are sparking say 5-10 deg before tach arrival. That allows you to tune advance a few deg either way plus includes a little prediction slop, but with a 60 tooth wheel that should be less than a deg. And, to get still more, go sequential and that will make 2 tach periods = 240 deg. I estimate that will get you up to about 8000 with close to a 5 ms dwell.

[TheMonkey]

Al-

I had 12 delay teeth & 12° offset originally, and then this last MSQ I tried 15 delay teeth & -6° offset after you had suggested a negative offset.

I don't entirely follow what you are saying, but I read this part which seems like the important part:

... I guess what I am saying is that you want to adjust delay and offset (but keep offset negative) so that at max rpm you are sparking say 5-10 deg before tach arrival....

Without ignition setup, I still haven't been able to look at the offset wizard with a timing light. But, I'm pretty certain my TDC of #1 cylinder occurs exactly at tooth #14 after missing teeth. this is using the convention of starting first tooth with #0.

My 60-2 wheel has teeth every 6°.

If I expect my max advance at WOT max RPM is 36°, then, that would be at tooth #8, #28, #48. You suggest having a negative offset with a spark event happening 5°-10° before tach arrival. Using an even tooth of 6°, that would be a tach arrival at tooth #9, #29, and #49.

My options of delay teeth / offset would be:

15 / -6°
14 / 0°
13 / 6°

etc..... up & down

so, the lowest delay teeth option (with the options of 9,29,49) with negative offset would be: 29 delay teeth, -90° offset. And be cautious about leaving at least 5-10° fudge room in spark before tach signal?

is this what you are suggesting?

i'll try it tomorrow and report back.

if this is the case, and non-optimal inputs results in bad ignition signals, couldn't this routine be written into the code such that an input (like my original 14/12°) could be filtered into an optimum combo (like 29 teeth /-90° offset) in the code?

[grippo]

Optimizing the inputs is best done in a pc side program where you have a wheel and TDC marks and sensor pos and you play with it and see the outputs. The end result would be the trigger offset, delay teeth, skip teeth, etc. The logic is horrendously complicated because there are so many options, and then you double them for 2 vs 4-stroke. Every number of cylinders has it's own logic, wasted vs sequential has its own logic, etc. We just don't have the time to write this program right now.

Meanwhile we are just muddling along. I can't be sure what I am telling you is 100% correct, so just test it on the stim. If you use a 12-1 tooth wheel you can better see what is going on on the scope and it's easier to count the teeth, and then you can just scale it up to 60 teeth.

I tested your msq again last night, exactly as it was, at rpms of 1600, 3200 and 6000 with pulse tolerances down to 1 % and everything was rock solid on the scope . But then I tried a wheel simulation which purposely dropped pulses every so often, so the missing tooth counter would count down. When this happened, the code evidently compensated because there was no loss of synch, but I would see 2 ign outputs per 360 deg and the whole thing jittered. This went on for a sec or two, then back to normal pulse train, then kept repeating the whole sequence. This kept happening even with 35% Pulse Tolerance. So I have something to investigate: If there is no loss of synch, then why is the missing pulse causing the ignition timing to go out of whack ? I will aslo look at whether this happens with 8 cyl. I also have a whell simulation which has jitter in the wheel timing and I will try that and decrease the tolerance and see what happens. What I am interested in is whether I can get misses which are + then - so the net count is 0 like you are seeing, yet it causes the timing to go out of whack like yours does.

[TheMonkey]

not a lot of new info from me, but while it's fresh in my mind, i'll share what happened today.

started by changing the delay teeth to 29, and offset to -90°. at first i was really encouraged because feathering down on the throttle to the trouble zone of 1500 rpm did not create the lean/rich puffs. i went for a drive, and twice the car stumbled really bad coming out of the hole, just on the other side of the intersection. so, i took it easy home and put the scope on.

i found that the new 'trouble zone' of phase shifts moved from 1500 rpm to 2000 rpm. the 'flanking' seemed to disappear altogether. i did another scope run, sloooowly increasing revs to find where it started to lose it, and right at 2000 rpm i tried holding it steady there, and it completely fell apart. the ignition dwells looked like a circus. also, at the end of the bunches of phase shifts.... brief flanking showed up again.

arbitrarily, i changed settings to 20 teeth, -36° offset, and had time for a 30 second scoping.... slow rev up to 3500 and back, revealed zero phase shifts and zero flanking. i'll test this setting more tomorrow, hopefully it's gone, but there really seems to be something in the code going a bit weird. behavior changes with settings. but... one thing that haunts me a bit is the fact that i needed to put a resistor inline with my VR sensor to make it work, so maybe it is hardware related. crazy.

[TheMonkey]

... When this happened, the code evidently compensated because there was no loss of synch, but I would see 2 ign outputs per 360 deg and the whole thing jittered..

what do you mean 'jittered' ?

......What I am interested in is whether I can get misses which are + then - so the net count is 0 like you are seeing, yet it causes the timing to go out of whack like yours does.

remember though... when my ignition goes out of whack, it moves to the next cylinder dwell spot... not just a move over by a tooth or 2. isn't a miss of + or - on the count, just by a tooth?

[Bruce Bowling]

.. one thing that haunts me a bit is the fact that i needed to put a resistor inline with my VR sensor to make it work, so maybe it is hardware related. crazy.

Actually, after talking to Al this mourning, its not crazy...

I will let Al chime in on specifics and solution but on the hardware side, what happens is that with some missing tooth wheels the "missing" gap actually will produce a small signal blip, sorta like there is a tiny tooth. Its OK because the amplitude of this blip is small and will not trigger the hardware. However, the voltage amplitude from the VR sensor is proportional to engine RPM, so as the RPMs increase so does the voltage. To a point where the little blip actually becomes a tooth - and the missing tooth is no longer missing...

Of course this screws up everything, but in the software, after sync, it will throw out any missing teeth. So even if the missing tooth becomes a tooth the s/w will throw it out... however Al found a condition where it may not be doing this correctly, when it correlates with the tach pulse area. I will let him speak on this but I think what you may need to try is to add one to your delay teeth number, and then subtract off the tooth spacing off of the trigger offset value. So if you have a 36 tooth wheel, this is 10 degrees... you will need to make the trigger offset number more negative by 10 degrees (trigger offset is always a negative number for this mode).

What the resistor does is basically reduce the input VR signal amplitude (really in actuality the current is reduced but this is converted to voltage with a resistor on the transistor collector), so the missing tooth amplitude stays below the hysteresis trigger threshold....

- Bruce

[grippo]

By jitter I mean the timing moved about 1 or 2 teeth back and forth.

I am trying to reproduce what you are seeing: trigger+/- = 0 and the ignition pulses phase shift by 120 deg. Last night I was able to find the cause of what I was seeing - turning 1 ignition pulse into 2 separate pulses when it detected a missed tooth. This was caused by the fact that the missed tooth, in 1 out of 3 times, occurred such that the missed tooth was a tach pulse tooth. When this happened I inserted an extra tooth and went ahead and processed it like a tach tooth. This was the cause of getting everything out of whack. IIf you miss a tach tooth it should have just reset/ resynched. When I put this logic in, that is what happened, so on 2of the 3 missed pulses, the missing pulse was inserted correctly, on the 3rd one you got a resynch. Then I moved Delay tooth by +1 and trigger offset by -6 deg more (to -12 deg) and this got rid of all problems, because now the missing tooth was never a tach tooth.

I will be spending the weekend looking at how to make this system more robust for missing/ extra teeth. Plus I may give you some code to try to verify whether this 120 deg ignition phase shift is really occurring when there are no missing/ extra teeth.

[TheMonkey]

using the settings of 20 delay teeth, -36* offset, i am repeatedly seeing successful driveway sessions through the entire RPM band.

i'm tempted to hook up the coils, although i'd like to hear back from Al first on what he untangles this weekend in the wheel decoding. and sure... if you have a code to create sort of a trail of what is happening, i'll try it.

this also makes me wonder whether my Pontiac sequencer problems were code related, and a shift in delay teeth & offset would have got me going. hmm... unfortunately, though, i wasn't able to test the ignition outputs on the car without having them hooked up and causing backfires, etc.

it's possible that this phase shifting was occurring in the MS2, but was never found with a single coil. i'm not sure how the coding works in terms of before/after routing... but it's possible that it's in code before routing (tach1 gets mixed up with tach2, so that router still thinks it's going to correct spot?). just thinking out loud, i guess.

[TheMonkey]

i put almost 100 miles on the car today without any puffs or stumbles using 20 delay teeth, -36° offset.

[Bruce Bowling]

i put almost 100 miles on the car today without any puffs or stumbles using 20 delay teeth, -36° offset.

Sweet Now go hook up 'them coils

- Bruce