Just another misfire/rough idle thread...

[xt2005bonbon]

Also, when my long term fuel trims begin to deviate from 0, I avoid going WOT or pushing the car hard until I figure out what is going on.

[StkmltS]

So, what are you next troubleshooting steps?

Poke at things in the engine bay and see what happens?

Tonight I'll flash the ECU back to the exact ROM I was using before things got wonky to see if I did something wrong inside of the ROM. I have a propane torch (tank) I'll use to look for a vacuum leak, and I can bring my smoke tester out of retirement if needed.

 

Also, when my long term fuel trims begin to deviate from 0, I avoid going WOT or pushing the car hard until I figure out what is going on.

Post-rebuild I haven't come anywhere close to WOT. I probably haven't even gone 50% yet. I've hit boost a couple of times on accident (1-2psi), and a couple more times on purpose (up to 5psi). I'm proud of myself for how long I've managed to keep my right foot under control.

[xt2005bonbon]

Post-rebuild I haven't come anywhere close to WOT. I probably haven't even gone 50% yet. I've hit boost a couple of times on accident (1-2psi), and a couple more times on purpose (up to 5psi). I'm proud of myself for how long I've managed to keep my right foot under control.

 

Good job. Yeah, that can be quite hard to avoid boost all the times.

[StkmltS]

Try to wrap your mind around this beast. I've never seen Fuel Learn (%) jump up and down this much.

 

The chart is pretty busy around y=0, so I moved MAF(V) up 40 to get it out of the way. For MAF(V) pretend y=40 is y=0.

 

I've uploaded the raw datalog for you guys to play with if you feel so inclined. It was a little too large so I had to split it into two files. I'd be really interested to know if any of you gurus can pull anything useful out of the data.

BtSsm_20160927_064035_p1.csv

BtSsm_20160927_064035_p2.csv

Edited September 27, 2016 by StkmltS

[xt2005bonbon]

no time to take a serious look (not that I am a guru either ), but I can already see that when your fuel learn is pegged at a high value (e.g. +15%), the fuel correct stays positive and quite high, meaning it is trying to add even more fuel. So definitely running lean for some reason.

[heiche]

Fuel learn % (your purple plot) may seem to jump around a lot because it's switching between the 4 cells in your LV, depending on load at that point in time. Now if you keep the load fairly consistent, then you can see it start to "slowly" increase or decrease it as it deems fit.

 

Some of the spikes you see might be the ECU transitioning between one of the 4 cells to a different one. If the load moves to another cell, it doesn't just instantaneously change the fuel correction to that new cell's value, but rather it fairly quickly increases or decreases it until it reaches the new value. This makes the driving experience more smooth, and less twitchy.

[StkmltS]

What's the clear white plastic bulb-looking thing on the fuel rail between the injectors?

 

Sent from inner space.

[MTBwrench]

That's a fuel pulse damper. It soaks up pressure spikes caused by the injectors opening/closing.

[birkhoff]

From those logs, cylinder 1 is actually the outlier, with no misfires recorded in the log. Misfire counts over the combined dataset are (cylinders #1 - #4 respectively)

 

0, 4235, 1105, 114

 

So what is right about #1 that is not right with the others?? Try moving things between #3 and #1, for example. Try to move the zero. It is tempting to swap with #2 but there may be other problems with that cylinder.

 

There could be something global going on, like an MAF problem or airflow leak, but why then is #1 perfect and apparently not affected by any of this???

[xt2005bonbon]

What's the clear white plastic bulb-looking thing on the fuel rail between the injectors?

 

Sent from inner space.

 

That was I what asked you about a while ago.

[StkmltS]

I flashed back to my initial break-in tune and everything is back to the way it was before (misfires included).

 

Obviously the problem was in the ROM I tweaked and flashed on Sunday. I'm not sure (yet) what the deal was, but I suspect the "minor" changes I made to the MAF scaling weren't correct. Maybe I multiplied something by 15 instead of 1.5. Fortunately nothing terrible happened. If I have time tomorrow I'll compare the two ROMs and find out where my mistake was.

 

Thanks to all who chimed in with ideas.

 

Sent from inner space.

[xt2005bonbon]

ok. Back to the drawing board then.

[birkhoff]

Can you check the log you made with this ROM, or record a new one and confirm that you are misfiring in all cylinders except #1?

[StkmltS]

Can you check the log you made with this ROM, or record a new one and confirm that you are misfiring in all cylinders except #1?

 

I meant that it went back to the way it was before... before my fuel trims went bonkers yesterday.

Edited September 28, 2016 by StkmltS

[xt2005bonbon]

So, still seeing misfire counts as high as 30+ or it is settling to lower values now when you idle?

[birkhoff]

I meant that it went back to the way it was before... before my fuel trims went bonkers yesterday.

 

Yes, I understand, but before, I assumed from your description you only had misfires in #2. Is that still the case? I would not have noticed the misfires in other cylinders in your recent posted log just by scanning the spreadsheet. But if you do a column sum they show up. If indeed you have misfires in three of four (or four of four) cylinders, that is something to take into account.

 

Also, I thought this was a completely stock setup, but I see now you had a lightweight flywheel reported in post 1. How come? And how light? Rough idle and misfires are sometimes reported when LWFWs are installed, reducing the inertial mass compared to the stock flywheel. For example, see this NASIOC sticky

 

I have no experience with lightweight flywheels. But I recall momentum scales linearly with inertial moment, so a LWFW could be messing with the roughness detection algorithm, especially at low RPM.

 

Did you re-use the LWFW in the rebuild? If so, it should be put into the list of things to look at. As I said, include all things that didn't change during the rebuild.

Edited September 28, 2016 by birkhoff

[StkmltS]

So, still seeing misfire counts as high as 30+ or it is settling to lower values now when you idle?

 

Still high counts, but it doesn't really have a particular value that it "usually" goes up to. Sometimes it'll top out in the 40's, other times like this morning it only got up to around 20. The only thing consistent is that it misses when speed = zero.

[StkmltS]

Yes, I understand, but before, I assumed from your description you only had misfires in #2. Is that still the case? I would not have noticed the misfires in other cylinders in your recent posted log just by scanning the spreadsheet. But if you do a column sum they show up. If indeed you have misfires in three of four (or four of four) cylinders, that is something to take into account.

 

Also, I thought this was a completely stock setup, but I see now you had a lightweight flywheel reported in post 1. How come? And how light? Rough idle and misfires are sometimes reported when LWFWs are installed, reducing the inertial mass compared to the stock flywheel. For example, see this NASIOC sticky

 

I have no experience with lightweight flywheels. But I recall momentum scales linearly with inertial moment, so a LWFW could be messing with the roughness detection algorithm, especially at low RPM.

 

Did you re-use the LWFW in the rebuild? If so, it should be put into the list of things to look at. As I said, include all things that didn't change during the rebuild.

 

I've seen roughness in cyl#3 three or four times recently, but each time was very brief and it stopped missing as quickly as it started. If I remember correctly each time it only went up to about 8. I just brushed it off as a new rebuilt motor thing (and it's occasionally happened since the problem started). Well, this morning cylinder 3 decided to join the party and take over the show. Notice how the misses don't change with RPM, but they DO change with speed.

 

If I get car time this weekend the first thing I should do is dig out the wiring for the injectors and coil packs and hope I find a partially broken wire. There are like a million things I should do first so who knows what I'll actually end up getting accomplished.

 

I installed the LWFW and new clutch after the problem started. I chose my new clutch setup based on price and reviews. A new dual-mass FW is stupid expensive, and none of the local auto machine shops I talked to wanted to resurface my used one.

BtSsm_20160928_060549_idling-and-slow-driving.csv

Edited September 28, 2016 by StkmltS

[heiche]

That graph is crazy! When you said it misfires at idle, but not while moving, i figured it was sort of a trend. But that graph illustrates it's absolutely that way.

 

How about this test: next time it's misfiring, let out your new clutch a little with the e-brake on, like you're starting off up a hill. Does that stop the misfiring even though you're at zero mph?

[xt2005bonbon]

Mine was super similar btw. I don't recall ever really misfiring when I was idling and moving. I had to be at a full stop for it to begin misfiring.

[birkhoff]

I've seen roughness in cyl#3 three or four times recently, but each time was very brief and it stopped missing as quickly as it started. If I remember correctly each time it only went up to about 8. I just brushed it off as a new rebuilt motor thing (and it's occasionally happened since the problem started). Well, this morning cylinder 3 decided to join the party and take over the show. Notice how the misses don't change with RPM, but they DO change with speed.

 

If I get car time this weekend the first thing I should do is dig out the wiring for the injectors and coil packs and hope I find a partially broken wire. There are like a million things I should do first so who knows what I'll actually end up getting accomplished.

 

I installed the LWFW and new clutch after the problem started. I chose my new clutch setup based on price and reviews. A new dual-mass FW is stupid expensive, and none of the local auto machine shops I talked to wanted to resurface my used one.

 

Misfires disappear with the car moving. While moving, you also add in the extra inertial moment of the drivetrain. Could that be a key? I vaguely recall you tried an experiment where you coasted with the clutch engaged. Did the misfires come back while coasting?

 

Just to be sure, when you say LWFW you mean light weight, aftermarket, whereas, single mass FW is what we often install, in place of the stock dual mass flywheel by using the 2008+ WRX setup. Light weight is what you have, correct?

[StkmltS]

How about this test: next time it's misfiring, let out your new clutch a little with the e-brake on, like you're starting off up a hill. Does that stop the misfiring even though you're at zero mph?

I'll give it a shot tonight. I'm pretty sure I've tried it before, but I can't remember what happened.

 

Misfires disappear with the car moving. While moving, you also add in the extra inertial moment of the drivetrain. Could that be a key? I vaguely recall you tried an experiment where you coasted with the clutch engaged. Did the misfires come back while coasting?

 

Just to be sure, when you say LWFW you mean light weight, aftermarket, whereas, single mass FW is what we often install, in place of the stock dual mass flywheel by using the 2008+ WRX setup. Light weight is what you have, correct?

 

SO FAR any movement causes the misses to stop. Here's a nice chart showing me coasting to a stop on a highway off-ramp yesterday. It makes no sense at all, but it's very repeatable and reproducible (under the conditions I've tried so far). I haven't yet tried starting on top of a hill and going from 0 to >0 in neutral.

 

My flywheel is this 12lb Competition Clutch WRX replacement. I'm not sure where the threshold is between "lighter" and "lightweight", but 12lb vs 26lb is lightweight in my book.

BtSsm_20160928_173526_coasting-on-a-highway-off-ramp.csv

Edited September 29, 2016 by StkmltS

[xt2005bonbon]

I was experiencing just that too. I think you can rule out the flywheel for causing that.

[birkhoff]

Compelling log.

 

I'm thoroughly stumped on this one. With trans in neutral you have very little added inertia in the system and yet, nothing happens until the wheel speed hits zero.

[heiche]

What if, just for giggles, you temporarily disconnected the vehicle speed sensor for a quick test? Now it won't know whether you're moving or not. Will that affect the misfires, for better or worse?

 

Also have heard of some odd behavior associated with the neutral position switch.