Lost with Megasquirt

Pete. (l33t FS) said:

Rich doesn't kill cats, rich makes the cats "go to sleep", which is a good chunk of why the OEMs do power enrichment.  The rest is keeping the exhaust valves happy.

 

This is how some modern turbocars run stoich even under high boost - they have better cooling in the exhaust area and good materials in the valves, and modern converters are a lot more tolerant of heat.  I am sure that the temperature drop across the turbine helps too.

IMHO, "go to sleep" is a little much in terms of what catalyst do when rich- there is still a shocking amount of NOx and HC conversion going on, but CO conversion drops off a little.  Little enough that it's an emissions problem most of the time when using enrichment to cool things down.  But OEM's use rich mixtures to cool the exhaust down (for whatever reason- valves, manifold, catalyst, sensors, etc).

But Roc (I'm sorry, lazy to copy your entire name, lol), if you want good emissions, target stoich on the sensors, and once you have a really solid basic tune for air, then really focus on the transient fuel compensation.  Keep it stoich all of the time, except for peak power for best emissions.  And I do mean stoich on the sensor- try to stay away from 14.7 or a specific a/f- since that's dependent on what fuel you run- virtually all pump fuel these days is E10, which si 14.3.  But the sensor don't care- stoich is stoich.

If you have problems idling at stoich, try adding spark first before bailing and running rich.  That may not work, but worth a try.  

Alfadriver, what do you mean by "stoich on the sensor"? 

Now I'm idling at 13.7 (when it wants to), 1050 RPM and 13 degrees of timing. I can't seem to go leaner, but I did not try adding timing like you mentioned. I definetly can't go below 950-ish RPM, otherwise it shakes the car and it just isn't smooth at all. Stock is supposed to be 700 to 800 RPM, I think 10* of timing. How did the guys in Stuttgart achieve that? I don't know the mixture they specify, but it certainly isn't 12:1!

While driving the car back from work today, it really misfired during idling, and even cruising. I parked it for about 10 minutes and came back to do some logs. I recorded 3 (I was just cycling the key between logs):

1- Bad one: It was idling at 14.5-14.3 and misfiring, but I don't see crazy AFR spikes on the log. 1050 RPM but not steady. Coolant 72C, IAT 47C, Idle duty 49%, injector pulsewidth around 3.1 ms. I tried playing with the cells it was at in the VE table to richen it, it didn't do anything.

2- Good one: Idling at a rock solid 13.7, steady 1050 RPM. Coolant 85C, IAT 51C, Idle duty 49%, injector pulsewidth 3.1 ms. I changed absolutely nothing and now it's good.

3- Bad one: Idling at anywhere from 13.7 to 14.7, with peaks at 16.2-16.4. 950 RPM. Coolant 85C, IAT 55C. Injector pulswidth went up to 3.2 ms. The engine was vibrating more than usual and misfiring. At some point, the electric fan kicked in and the AFRs went all over the place.

It's like a dice roll, I never know which Idle flavor I'll get when I turn the key!

Best would be to post those logs somewhere, but I can't do this here can't I?

Batch fire can be difficult to get to idle smoothly at stoich, or sometimes even at all depending on how short the runners are.  The fuel injected to one cylinder can be bounced back into the plenum and ingested by a different cylinder, so some are lean and some are rich.  You can band-aid that by idling more rich so the lean cylinders get a decent mixture.

If your original injection system was K-jet, those typically can idle really, really smoothly because they don't care about no steenking injector timing, they are always spraying.  This is probably why Mercedes in particular kept it for far longer than most other manufacturers.

Your idle variability may be a result of this, because your setup description smells like you are not running a cam sensor.  Sometimes it batches starting on 1 and sometimes it batches starting on 4.

I wonder if it would idle better if you doubled your squirts per cycle, assuming you are injecting one squirt per revolution.  This does carry the downside of cutting your injection times in half, and as Paul noted, small injector times mean your deadtime becomes that much more critical.  Certainly, if you could add a cam sensor and run sequential injection, it would be able to idle a lot leaner.  Heck, even if you did semi-sequential and batched 1 and 4, and 2 and 3, it would probably be a lot better.  (Toyota did this for a while in the late 80s/early 90s)  IIRC you can do this with your existing setup.

I don't have a cam sensor, but my injectors are paired 1-4 and 2-3, I don't only use one bank. Is that what you mean by semi-sequential? I'm at 2 squirts per cycle now, I could try 4. Is there a way to rig up a test bench in the garage and measure injector dead time precisely?

Wait, did I set this up right?

I regards to what Pete said, shouldn't I select semi-sequential instead of untimed?

In reply to Rocambolesque :

If rich is what it needs to even idle, so be it.  As long as the catalyst is warm, it will do a pretty good job.  Not perfect, but ok.  What you are seeing as 13.7 is 0.93 lambda.

In terms of "stoich at the sensor"- don't use air/fuel- that's assuming what fuel you are running.  All sensors will output stoichiometry regardless of the fuel, and every one I've used, they will output lambda- where 1.0 is stoich, lower is rich, higher is lean.  Again, I'm betting you are using normal pump fuel, which is actually 14.3 and not 14.7 (E10 has been the nominal fuel for quite a while now).

As for your data- if ALL of those tests were with it trying to run "13.7" (again, move to lambda- I know, broken record)- the fact that you are seeing some lean runs that are kind of variable, that *could* be hot fuel- so the system does need either cooled or run at a higher pressure.  Seems more likely than what I was just about to post, so I'll keep that out.

If on your data, if you are recording pulse width, and the idle pulse width is pretty close to the same on each test- then, yea- you have a fuel delivery problem.  So check the PW for the three starts, and compare them between each test.

In reply to Rocambolesque :

Yes, if I remember my docs right, "untimed" means that it is not particular about when it opens the injectors.  2 squirts/cycle with alternating, again IIRC and now I really need to get to the docs, means that it fires one injector driver on one revolution, then the other driver on the other revolution.

I think, anyway.

2/alt and the 1/4 and 2/3 pairing work great for an i4 with 1342 firing order. Should be no need for anything else. 
 

You can put the files in a google drive and link it or email me at kandpperformance at gmail dot com 

You say it sounds like misfires so I would be leaning to a more phyisical issue as well (hot fuel, air)

also not sure if this was mentioned but make 100% sure ignition timing is synced between ms and reality