What happened here? Jeep XJ funny business

My ignition is a bit worn and you can take the key out in ACC and kill the battery. This is not a new problem. So I had to jump it. 

It wouldn't start. I also ran the tank really close to empty and parked on a side hill. I walked up to the gas station and bought a gallon.

It still won't start and I'm getting worried. I checked for the sound of the fuel pump and I hear it running with the key in ON. I figure it needs a bunch of priming if I ran it fully out of fuel.

Still no dice. So for some reason I gave it half throttle and it fires right up. But it wouldn't idle for about the first 10 seconds, and I had to pedal it. Then it idled and ran fine. Started back up later with no issue.

What happened there? I'm kinda thinking IAC?



 

Did it ever kick when trying to start?  I've seen a couple times where vehicles would flood if you cranked them long enough with insufficient fuel pressure.  Not enough fuel to start but enough to wet down the chambers.

I haven't seen it on a Jeep but tons and tons of GMs from the late 80s/early 90s where carbon in the throttle body would cause flooding when trying to start.  Not enough air was going past the throttle plate.  Usually when they would get bad enough to cause this, the throttle would be sticky when closed.  Being low on fuel may have been a red herring.

Oh!  How low was the voltage?  I have seen some of those 80s era electronics still function even down at like 4v (modern stuff gives up the ghost below 9v) BUT they go really bizarre when that far outside their normal zone.  They often would run super mega soot-blowing pig rich.  So cranking with a weak battery may have caused the injectors to pour ALL the fuel in, compounded by weak spark.

I've had the 97+ Jeep XJ's act like that after the battery went dead and had to be jump started. They won't hold idle initially, have to keep it running with the throttle for a few minutes. Then it will be back to normal after that.

I can totally see being "out" (low) of gas being a red herring.

It didn't kick at all until I touched the throttle and then fired instantly. It cranked over like it was missing fuel or spark. It is a very low voltage vehicle too. Normal running is 13.2-13.7v depending on load. On the jumper cables I didn't check the voltage but it was cranking over just as fast, if not faster, than straight off the battery.

It's a Jeep thing.  You were told you wouldn't understand.

The ECU probably lost all the "learned" values it had accumulated over the time driving it before the battery died. Once it started and ran for a bit, the voltage was high enough to let it start to re-learn everything - throttle position voltage, intake vacuum pressure, rpm, ignition timing advance, etc. Drive it around for a while - highway, stop and go, different conditions so it will have a larger data set to work from.

Wouldn't hurt to verify the alternator is producing the amount of voltage and current needed to run everything and charge the battery fully. Old, flaky alternators and/or batteries can and will drive you nuts with weird, random glitches.  

Yep, you just lost the learned positions for everything. The transmission may shift a little oddly for a bit as well. Every engine is a little different, so the ECM is set up to compensate. For idle specifically, it has to be able to account for a range of ambient air pressures, temperatures, filter restrictions, engine compression as things wear, varying internal friction losses over the life of the vehicle, etc. When you shut the engine off, it stores an idle position offset (probably counts of the IAC stepper motor since it is an XJ) that is added to the base table commands. When you restart the engine, there's usually an additional restart adder that further bumps this up to ensure enough air for starting.

After the battery died, the offset was zeroed, and there wasn't enough airflow for the engine to fire. Some ECMs have logic to progressively open the IAC during extended crank, but no idea if the XJ does or not. Tapping the throttle got you the airflow, the engine started, and then it was able to start to relearn, which it can't do until the engine is running. It still takes some time to relearn, and each time it dropped below idle when you said you had to pedal it, it would bump up the IAC position, but it's a PID controller set up to be very stable, so it will take some time to fully adapt.

Check all of the grounds. I ended up running new ones on my TJ. It helped immensely with weird electrical stuff. 
 

Throttle open starting sounds like it flooded. 

In reply to gearheadE30 :

Interesting stuff. The only other "modern" car that I've ever killed a battery in is my E36 and it never did anything like that. The Jeep's computer I'm sure is much more archaic than the BMW's. Good info to help me understand these silly modern car systems!