1k-2k oscillating idle. WTF did I miss?
 

I'm at the end of my rope here. 90 Miata with M45, Megasquirt ECU, crank overlay, and AEM water-methanol injection pre-supercharger.

On Sunday, my car's idle started oscillating. It gets to 2,000 at start up, then goes down to 1,000 or so, then back and forth. I did observe that the idle occasionally sits at 1,400-1,500, and actually did so when I put the A/C on, but it seemed only when the compressor was working hard.

I've tried to check all of my vacuum lines by pinching the lines, and I tried running a new vacuum line to the ECU just in case there was a problem there. The only one that gave an effect was pinching the line to the ISCV, but that may have just been because I was basically stalling the car. I swapped out the ISCV, but it did not help.

My cam angle sensor is still lined up with the marks I made, so I am not sure if it could be timing.

My only other ideas at this time would be doing a test with propane to keep looking for a vacuum leak, or disassembling the S/C and throttle body in search of the cause; the dashpot looked OK, but who knows, and I currently have the idle speed screw fully closed (previously open a couple turns). I also wondered about the S/C bypass valve, but it seems to be performing properly, and I tried to blow into it and didn't find a leak.

I have a boost gauge on the car, and it still makes 10PSI at redline. I've had the supercharger for over a year, but the most recent changes are the ECU, crank overlay, and water-methanol.

If you have any thoughts, please let me know. I've tried to check for the ISCV, vacuum leak, cam angle sensor, and I put in a GTX PCV maybe a month ago. I am wondering if the water-methanol has somehow screwed something up since it's the most recent change, but I don't know.

Cap the throttle body input and the IAC valve input with the engine running. Does the car stall immediately?

Thanks for the tip!

I tried this, and the engine did NOT stall immediately. It stumbled hard as if to stall, but did not. I thought I had a pretty good seal on the inlet, but maybe I did not, and my water methanol nozzle probably isn't perfectly sealed (it's before the throttle body). If Seafoam can stall an engine via the brake booster line, then this should have worked... right?

I picked up a propane torch and used it to try a propane test. I tried spraying unburnt propane along various connections and pipes on the inlet, but I did not get any noticeable change in the RPM oscillation nor shift in RPM's.

I used a couple cans of brake cleaner tonight to check it out.

It would appear that I have a vacuum leak on Cylinder 4's injector. When I saturate it with brake fluid, my idle stumbles. I tried this a few times and also double-checked my FPR and brake booster line since they are nearby, but it appears that the injector is the problem. No stumbling from the other 3 injectors.

I tried grabbing the bottom seal from a spare injector and put it on there, but the leak persisted. I probably need to put in all new seals to verify the problem.

Does this sound like a viable smoking gun? Or should I keep hunting? I suppose I could have a few leaks...

I got my mp62 and 9lb flywheel to stop oscillating through tuning. I don't have my laptop in front of me to recall exactly what i changed but it was in terms of recovery times. The other thing I played with is the idle screw itself. I opened it up and let the MS trim back the idle valve. You want to let more air through the idle screw and let the idle valve do the fine trimming. If you have the idle screw closed too much, you are relying on the idle valve to trim a larger range than it needs to...paired with with idling the higher load from the supercharger, the valve oscillates and cannot recover if it's target is too narrow. Finally, I bumped the closed loop idle to start around 12-1400rpm and slowly converge on the hot idle rpm.

I will have to double-check my idle to see if it has room for improvement. Typical operation had an oscillation of maybe 50RPM at any given time. That was with the idle screw open a couple turns. I actually had the MS idle control turned off for awhile initially.

This problem is more serious. For example, my AFR starts at something low, like 12 (it varies) at 1000RPM and then leans out to max on the UEGO as it goes to 2000RPM, at which point I think it cuts fuel and starts over again at 1000RPM.

Fingers crossed that the injector leak was the problem...

I replaced the injector seals. Checked it again with brake cleaner, seems to be leak free.

BUT the idle is still oscillating. :vash:

This is what is still throwing me off. It does not stall immediately. I removed everything up to the TB itself, and capped both the ISCV and PCV line (so it doesn't suck air thru the crankcase), and it still did not stall.

If I idle it with the ISCV capped off and ISC all the way in, the idle is stable around 1200RPM. If I open up the ISC a few turns, though, it becomes unstable again.

I checked timing - 13 degrees indicated, 12 degrees with the light.

I tried adjusting the dashpot idle stop screw a full turn either direction, but no effect.

I unplugged and capped off the S/C bypass valve - no effect.

I unplugged the TPS - no effect.

PLEASE let me know if you have anything I should check!

Attached are the idle settings that I currently use, for comparison.

IIRC, this is the process that I used to establish a steady idle:
Open up Tunerstudio with the car idling. While slowly opening the idle bypass screw, observe the Idle DC until it reaches a minimum value. This is theoretically the point that the idle bypass is flowing sufficient air to maintain your target rpm without additional air from the idle air control valve (I don't remember if the actual idle dc will match the closed idle dc at this point or not...so it is closed or near closed). Give the idle bypass screw 1/4-1/2 turn back closed to ensure the iacv is still contributing and can modulate a small percentage of the airflow without flowing too much air that you can never acheive your target rpm.