Even More Data 3.46

[grippo]

i gave the cam sync sensor a 180 fliparooski, and it fired right up and sequenced. the firmware definitely changed on this. see my bench testing comments from Jan-09 on previous firmware (holy cow - it's been that long?), now it's flipped:

Jan-09:
"i did not know if the sequencer would set my #1 TDC at 70° after cam sync, or 430° after cam sync. turns out that with cylinder offset set to 0, after wasted spark, and the cam syncs.... computer sets my number one cylinder at 430° after cam sync (or 290° before cam sync) - for some reason i would have thought it would be on the other cycle. but, it's consistent, so that's what is important."

http://www.microsquirt.com/viewtopic.php?f=90&t=23163

or... perhaps what has changed is that now i am programming a negative offset - that changed from previous. ***edited to add: i just tried it with old positive offset setting (all original settings), and it fired up okay... so it's definitely the firmware that changed***

also, i remember from my previous install, i set spark-fuel-offset to 2, and trimmed the fuel to hit the spray right where i wanted it in the cycle.

Cam sync is not a function of fuel-spark offset. Cam synch must arrive anywhere after Cam_Tooth and before tooth 1 (1st tooth after missing). So you want to put cam synch on the rev such that cylinder one tach cycle follows the missing tooth, as opposed to the 5th cylinder in the firing order.

[TheMonkey]

Cam sync is not a function of fuel-spark offset.

y, i get that now.

i was just commenting on which rotation the cam sync should be. i used exact same settings i had previously when it was sequencing, and never even touched the location of cam sensor. i had to flip it 180° this time to get it to work.

[TheMonkey]

Wow- motor is REALLY happy.

One thing that is irritating is the original beta power supply still does an occasional reset the moment the starter pulls the voltage too low on a cold start. It makes the IAC cycle right at startup.

Also- the cranking time seems to be longer than my bmw. Maybe just because crank rpm is slower. I think the datalog I posted has a start. Let me know if time looks normal.

Haven't fooled around too much with injection timing yet, but as it is now, it finishes the injection just prior to opening intake valve.

The new firmware forced me to change to tuner studio, and it is excellent. Great tools. Especially accel enrichments. The rolling graph made that painless.

[grippo]

We are really working to get the new sequencer hardware with a really awesome P/S out. It is next on the list.

As far as the cam sync, what was the last code you were running before you had to change the cam timing to get it to work with the latest code. I can compare the 2 code versions and see if I made any changes in that area.

[TheMonkey]

As far as the cam sync, what was the last code you were running before you had to change the cam timing to get it to work with the latest code. I can compare the 2 code versions and see if I made any changes in that area.

Al-

Previous code was 3.11.

I just brought 3.46 in and put it on bench with stim to compare previous scope shots from 3.11 on stim. It's different by 180.

3.11 cam sync signal is 430* before #1 TDC, and 3.46 cam signal is 70* before #1 TDC.

3.46 behaves like you described it above: Cam synch must arrive anywhere after Cam_Tooth and before tooth 1 (1st tooth after missing). So you want to put cam synch on the rev such that cylinder one tach cycle follows the missing tooth, as opposed to the 5th cylinder in the firing order.

[grippo]

I compared the 2 code versions and found that while the cam synch logic was the same, there was a change in when I check for cam synch. I vaguely recall making this change because I found a problem with cam sync timing. So it appears this fixed things, because the behavior of 3.460 in your engine is what I would expect.

[TheMonkey]

seq is running well on this car, although I found that it is having tach drop outs. so far, it's only happening when there is a sudden rpm changes. this motor winds up super fast while just idling and little stabs at the pedal. i think i'm close with the accel enrichments, but when tach drops out, it spikes lean. see pic below.

i'm suspecting that it is the wheel decoding algorithm that i may have to go to some advanced options. any suggestions on where to start?

my decoding is set to last interval, 0 time mask, 50% mask, next pulse tolerance for normal running 25%.

is there any tools to determine which variables to start playing with? isn't there a trigger logger in TS? i'm not seeing it.

[TheMonkey]

i turned on ABG filter to default settings, and bumped next pulse tolerance to 30%.

it is not having any tach dropouts anymore, but trigger +/- climbs when rpms pick up. even when just sitting steady at 2300 revs, tach +/- steadily climbs. no more tach -, it only climbs. AFR stays okay while it's climbing.

suggestions?

[grippo]

From what you say your problem is in the VR sensor which is producing extra teeth, since it is climbing. Forget the ABG filter - you have a 36 tooth wheel and don't need it, plus it only affects timing accuracy, not dropouts. Also, the masks have no effect when a wheel is used. However, the pulse tolerance does have an effect,and changing it to 30% or even 35% is fine - you shouldn't need more than that. I doubt it is due to your revving too fast for the algorithm - it works on small motorcycle engines that can rev way faster than a big block motor. Plus you say it also occurs while idling.

I though this engine was running fine last year, what has changed ? I know the code changed, but do you have the same code in both cars ? If one idles fine and the other misses teeth, it is likely in the hardware end.

[TheMonkey]

I though this engine was running fine last year, what has changed ? I know the code changed, but do you have the same code in both cars ? If one idles fine and the other misses teeth, it is likely in the hardware end.

last year, got it running on sequencer with original beta code, but never got out of the weeds with the phase shifts and dwell probs, so not clear whether this problem existed before.

couple years ago, it was running on ms2 with v3.0 board and needed to mod the vr circuit with an additional transistor to bring the square wave up to 5v. that was 2.883j code.

the wiring has been changed quite a bit too. mostly to keep any wires with current away from ms related sensors and wiring.

the dropouts kind of look like the dropouts i was getting on the bmw with sequencer, which an inline 10k resistor fixed, but i'm not sure why it fixed it. perhaps i should put one in on this setup too?