LongLiveXJ

I have been chasing what appears to be an intermittent misfire on my 1997 Cherokee - 4.0, auto, 4x4. It has 148k miles on it. Drives great aside from this issue so I'd like to sort it out! But I'm having trouble nailing this down.
I've done a fair bit of forum surfing and found a few things to try but so far nothing has fixed it.

Symptoms:
- Severe engine/trans vibration around 1500-2500rpm, happens while stationary, under partial load when cruising, and across a wider range (500-2500rpm) when the vehicle is in reverse.
- Every once in a blue moon I will get a misfire code for cylinder 1. My scanner confirms that the misfire counter for cylinder 1 increases very slowly when the engine returns to idle after a rev.

What I've tried, loosely in order from memory:
- Can of lucas fuel system cleaner
- Reseated all fuel injector connectors / inspected for damage
- Tested fuel pressure - idle was oscillating 55-60spi, revving up a bit it smooths out and stays at 55psi
- Cleaned cap/rotor. Previous owner replaced both shortly before purchase (chasing down this issue? who's to say... did not tell me about this ) there was a little corrosion but nothing crazy
- Reseated all plug wires
- Replaced all plugs with NGK ZFR5N, gapped at .030" - plugs that came out looked ancient, maybe original to the vehicle (Champion RC12LYC), cylinder 1 and 6 were a little darker than the rest and most of the electrode on cylinder 5 was gone. Huge gap.
- Verified correct distributor indexing, afterward noticed that the mounting bolt has likely never turned and the distributor is original. There is some shaft play side to side but definitely less than either of my 1996s that both run perfectly.
- Replaced O2 sensors with NTK units - rear was definitely bad, scanner graph confirmed the voltage was jumping around more than it should have. Front was ok, replaced it anyway, both now look nice and smooth like they should.
- Verified harness around the back of the valve cover isn't touching the stud - it was when I looked, but hadn't worked its way through the sheathing yet and the wiring inside looked intact. I zip tied it up out of the way.

Next on the list:
- Replace fuel pump - check valve is shot and unit is 148k miles old, already have a delphi replacement on hand, will be installing sometime this week.
- Replace injectors - have 6 reman bosch 4 hole units ready to go, will swap them in after the new pump is in for a few days
- Timing light on all plug wires to check for shaky spark
- Compression/leakdown test

What I haven't ruled out:
- Plug wires
- A different spot in the wiring harness rubbing on something - though I've followed most of it through the engine bay and haven't found anything yet
- Throttle position sensor, although graph on scanner looks smooth and responds as expected
- Crank/cam position sensors - scanner graph jumps around all over the place, it looks like a count value that wraps around back to zero when it hits a certain value and I'm not sure what to make of it. I can attach photos of this if it comes to that.
- Flex plate bolts
- Torque converter
- Stuck/carboned valve - hopefully the compression/leakdown test comes back healthy

Anything I'm missing? What would be my best plan of attack? Any and all input would be appreciated, I'm really hoping to nail this down soon. Really hoping the fuel pump/injectors clear it up, but from what I've read on other forums with similar issues it could very well be something else.

Working on uploading a couple of videos from the engine bay and from inside the vehicle. For some reason it won't let me attach them to this post, they just refuse to attach with no error message.

Thanks in advance!

Dave51

Can you upload those photos?

cruiser54

FWIW, plugs, wires, cap , and rotor are 30,000 mile maintenance items......

LongLiveXJ

Idling:


Revved up to a max of around 3k rpm, starting at the 37-38 mark on the bottom scale, new data to the right:


The crank signal turns into that weird dip for a few seconds, and the cam signal increases in frequency.
I think the half-circle curve in the crank signal may be deceiving since the count appears to be decreasing. Since it's a count, that should be impossible, right? That value should only ever increase, until it hits its upper limit and wraps around? I'm thinking this might be a limitation of the sampling rate of the tool/ecu.

On a different note- a timing light confirmed that cylinder 1 is seeing uneven spark when the misfire condition occurs. Other cylinders appear stable. Couldn't get a good video of it in action.

Still working on getting the misfire videos up somewhere.

cruiser54

I'd be putting new plug wires on it. They're likely older than the plugs were!!

Dave51

Yup.

The CKS/CMP signals are supposed to look like this:

Dave51

Check the resistance of the old injectors to see if one or more are off.

LongLiveXJ

Well, after a fresh set of wires from Napa I can confirm that the issue is still present. Didn’t make anything worse, though!

Now I’m suspecting the cap and rotor again? Maybe the distributor itself?

Injectors all measured 12.2-12.5 ohms on my meter.

cruiser54

Cap and rotor should be good for 30,000 miles.

Dave51

OK great.

As it could be physically clogged, (and if you're curious) you could try switching the injector to another cylinder and/or cleaning it with some carb cleaner (give it very brief 12V hits while backflushing).

Is this blip reproducible and coincide with the misfire?



While it doesn't accurately reflect what is happening, it could still be significant (a low sample rate alone shouldn't do that).

Dave51

That said, I would be aggressively looking here:

LongLiveXJ

I replaced the fuel pump this past weekend. Didn't hurt anything, made it a whole lot easier to start (more on this in a second), but did not solve the main issue.

Yesterday I replaced all 6 fuel injectors with the 4 hole units. Also didn't hurt anything, in fact I'd argue it runs a bunch smoother now, but again the issue is still present.
I was going to try swapping injectors between cylinders but the o-rings came out so mangled I opted to just replace the lot.

Interestingly enough, after this job the misfire counter no longer increases on cylinder 1, or any of them for that matter. Wondering if injector #1 was bad/failing after all.

Regarding the crank sensor signal, that dip appears at 2k rpm no matter if the issue is happening or not. If I rev it up quickly past the 2k mark it will not make that funky curve. In reverse, the same applies (dip around 2k) but the issue is present across a much wider band of RPM, and the dip doesn't widen proportionally.
The tie to reverse is still throwing me for a loop. It's very obvious - park idle is smooth, reverse idle is choppy, neutral idle is smooth, drive idle is smooth - and if you rev while in reverse it bucks and complains like no other.

The last two variables are the crank and cam sensors and I'm starting to suspect the former. Even after replacing the fuel system it will still start hard every other time (but significantly easier than before. 4-5 cranks vs 20), mostly when warm, so I'm working on getting an oscilloscope going to try and see what the sensor is doing.

I'm not looking forward to replacing the crank sensor if it's actually bad, but oh boy wouldn't that be nice if it was the root cause. Will hopefully update with scope screenshots soon.

Thanks for the advice so far, gentlemen!

LongLiveXJ

Crank sensor may be due for replacement, after all.
Cold started first crank today after sitting overnight.

Idle, no misfire:



Next two pictures are revving through the misfiring range, during a misfire event:



Dave - it seems as though my crank sensor matches your known good data when the engine runs well, and drops pulses when the engine runs poorly. Thank you SO MUCH for posting that reference waveform.

Do you know why it outputs 4 pulses? Is this sensor effectively 4 hall effect sensors stacked on top of each other? Is this redundancy or does it serve a larger purpose? To me this seems like a really weak point in the design of the sensor... lots of room for error, evident here.

I'm going to search out a replacement crank sensor tomorrow. Fingers crossed this solves it.

LongLiveXJ

Answered my own question... flex plate has sets of 4 windows in it, triggering the sensor 4 times. That makes way more sense!

Dave51

IMO you have enough evidence to change out the CkPS.

2 more thoughts:

When you noted

another possibility could be the flexplate is falling off, shifting the reluctor windows and causing the CkPS to miss pulses.

Second, depending on the oscilloscope, overloading it with data can cause aliasing (inaccurate waveforms). If I overload my Hantek I get a waveform like



even tho the CkPS is fine.

It sounds like you have a pretty strong cause and effect. Get a Mopar CkPS as at least anecdotally they have a better track record (although I've used AZ without a problem).