Idle Oscillations

[Jim Blackwood]

I have an oscillation cycle that is about 2 seconds long.
I've made minor and major changes to the idle PID settings to see if the IAC could control it but it may not be capable of it as it is too slow to react. It IS capable of making it much worse, but it rarely opens before the end of the downslope. It actually works best with the PID's zeroed out.

But I have come across something that could help. While MAP definitely lags the RPM drop that begins the cycle and Inj pw shows a slight decrease, the cylinder mass is definitely dipping before the RPM dip begins. It begins to dip Almost a quarter second before the RPM begins to drop. I'm also seeing a small (or maybe not so small considering it's about 4ms to start with) drop in the inj pw as much as 1/4 second in advance of the airmass. Meaning the cycle begins with a dropoff in the pulsewidth. Now with the MAP and TPS holding steady, what else could cause it to do this?

Any thoughts on where to look? There is no consistent correlation with the IAC at the beginning of the first dip. I know some of you have studied EFI theory and would love to see your thoughts.

Jim

[kur4o]

You have described very good the first part of engine surge[drop of rpm], not sure why the second part[surging] is missing in your case. When you see the drop of airflow is likely a missing cylinder, that decrease the engine velocity and less air is sucked for the other cylinders. Usually advance change is faster reacting for that case to stabilize rpms to target, iac is much slower to react.

I suggest start playing with spark advance at idle[migth not be too much or less and not the optimal] and return to stock or play with idle under,over speed advance tables.

[Jim Blackwood]

Good idea. That sounds like just the sort of thing that might do the trick. At present the advance is rock steady at idle. I'm not seeing any evidence of misfire and the surge is too consistent and steady for there to be an intermittent cause. With a 2 second period at 600 rpm that means 20 revolutions of the engine during the cycle, or 10 revolutions during the downswing so I don't think a misfire would have much impact.

Of course using timing to correct the problem is just covering it up but if it works it works. Still I'd love to get to the root of the problem. Something seems to be causing a dip in the injector pulsewidth. If I could find the cause for that I think maybe I could eliminate it at it's source.

I had one good trace (unfortunately not saved) where there were a few seconds of steady idle before the oscillation started. I think I may be able to reproduce it. It clearly showed the MAP lagging the rpm dip by about half a second meaning the MAP is not causing the dip in the PW, but the airmass and PW were leading it. Can you think of any other PID channels I might look at to track it back any further than the PW?

Jim

[kur4o]

The pcm uses timing to control idle rpm. Some advance added and the rpm rise, some retard added and rpm drops. It acts much more faster than the iac adjustments. I suspect that table is zeroed and timing is not controlling idle at all[The fact that advance stay rock solid, it should fluctuate 2-3 degrees around the target]. I remember similar condition as you have described when I zeroed that table.

If you maxed the timing at idle[reached the MBT point}, there will be no room for adding advance so you need to back it a little, so the timing adjustments can oscillate and fine adjust rpms.

[Jim Blackwood]

OK I'll try that tomorrow. Having some fun with the MAP and AFR traces and trying to sort those out which I really need before doing much more but at least maybe I can get the idle steady. The AFR PID channel decided it didn't need to show a voltage anymore, probably something I did. It comes in on EGR pin 55. The MAP is clipping at 105 kPa even though the voltage goes on up. Seems like that has to be something in the trace but that limit doesn't seem to be accessible. I even tried setting up a user defined parameter but so far no luck. I can see days passing before I get those sorted out. Also my MIL won't go out and I'm not showing any codes. Got no idea what's up with that.

OK I got the MAP to work by using the sensor voltage and multiplying it by 44.xxxxx whatever (101/2.27) so one down and a couple more to go. Just have to think creatively sometimes.

AFR works on my simulator (cloned ecm) so I'll have to check the input on pin 55 and see what I have there. That leaves the MIL. I remember a switch to turn it on and off but I can't seem to get back to it.

Jim