MSPNPDIY extreme oscillating idle
 

Hello All,

I have a 1991 Miata with a Megasquirt PNPDIY unit. The car has a turbo and inter-cooler with the GM IAT and Variable TPS upgrade. The car ran great with the stock injectors and coil packs. Eventually getting it tuned better and better, my clutch started to slip so I decided I wanted to upgrade that along with bigger injectors and coil packs. I now have Flow Force 380cc injectors and the LS coil packs (still running in batch and waste fire modes). Ever since I installed these items the car will not idle once it is warm. I left all my settings the same with the exception of adjusting the dead times for the injectors and the dwell times for the coil packs. I know my fuel table would need a total re-tune, but I can't even get it to idle once its warm. I never was able to take a log before I changed the injectors and coils. I've checked my grounds since I noticed my TPS bouncing around a bit, but it did that with the stock injectors so I thought it might not be an issue. I reinstalled the stock injectors tonight and the car idles totally fine again. Please note I totally started over with my tune so some startup parameters might be funky, but seeing how it was a good starting point before with the stock parts, I was hoping the tune "reset" might have fixed the issue. Attached are the current tune and log. I am not using any throttle input at start just to show the oscillation then finally it stopping.

Thanks everyone for any input you might have.

Did you adjust required fuel?

Yep, it went from 11.2ms for the required fuel on the stock injectors to 7.1ms with the larger injectors. The odd thing is it looks like from the crazy AFR oscillation that the injectors just stop injecting once the duty cycle gets to low.

Yes, larger fuel injectors do usually have a larger deadtime. Did you adjust that as well, using the values flow force provides?

Otherwise being able to change injectors and fix the tune solely with reqfuel was a pipe dream from the moment it was conceived.

In my experience OEM injectors need a much lower VE value in the fuel table around idle and low loads. Newer injectors manage a flatter transition from 50+ kpa to idle, less of a gradient. If idle cells were around 20ve, they'll land around 40 when tuned with EV14's. If the deadtime is correct the extreme low load cells won't drop like a rock or spike upwards weirdly. Of course you need to tune these extreme low load cells by hand, o2 delay and the fact it's all momentary transitions makes tuning with any autotune impossible.

Without even looking at your tune or log I assume it was hitting untuned idle cells and running wicked lean, then it would spike load because it's stalling but the higher load cells are closer to correct so it catches and runs. Load then falls and it's too lean again, repeat until end of time...

So go back and tune the new injectors. Set deadtime and reqfuel and assume you need a whole new VE table because you do...

Yeah, I adjusted the dead time and battery correction for the new injectors as well. I tried looking at the circular "loop" the VE table cursor would follow and tried to adjust the higher load cells as you mentioned by getting the lean spike to drop. I will have to get more aggressive since I seemed to make it worse by fattening up those cells (perhaps the wrong ones). I know each situation is different, but should I expect a wide variance of values in the idle range? I appreciate the input on the behavior of the new style injectors pattern at lower idle. I'll try to play with it more and report back!

AFR will have a pretty significant delay at low RPM, even more so at low loads. You were misinterpreting which cells were lean and which were closer to OK.

If you have a wide range of values around idle it won't idle. RPM and MAP are going to jitter a bit when idling, if the cells vary too much it's not going to keep pulsewidth smooth and it will speed up or slow down making it all worse. It also averages 4 near cells depending on actual load and rpm, so messing with the binned values to keep idle to the same 4 cells on a MS2 is also helpful. Otherwise MS3 has idle VE which is even better.

Try to change deadtime from 0.4ms to at least 0.8ms

I finally figured it out with the correct balancing with the four or so cells when it is warmed up at idle. Thanks again for pointing me in the right direction!!

So I did this and WOW what a difference it made in getting the car to respond to changes and not have the oscillating idle. Its odd that FlowForce was suggesting the 0.409ms deadtime, but all I know is this greatly helped. Thanks!

So I just wanted to update this thread one last time, I was having grounding issues with my TPS, coolant and other sensors bouncing around when the motor was running. Low and behold I found that when the ECU was unplugged I could still get the TPS ground wire to ground out against the chassis. I know all the sensors should go the ECU then to the grounding point on the engine and so finally I found this post about it: NA ecu grounding 90-93

I just had to cut the black and green striped wire from one of two ring terminals that bolt to the back side of the intake manifold. Just to help anyone else with a 90-93 that has a ground loop issue like I did. Cheers!