[FIXED] 2006 Subaru Forester 2.5L hesitation/misfire

Could use your assistance on another vehicle that's getting ready to eat much lunch. I already asked the SD Meetup crew and got some great idea, but figured I'd put this up here anyway. Helps me keep my data/thoughts straight.

Another tech got this Suby with a severe knocking noise. Calls a rod knock, suggests a long block. We get one and put it in. Boomerangs the same day with a complaint of low power, misfiring and a flashing CEL. Other tech changes the plugs, the coil, finds a TSB that matches our symptoms and follows it. No help, now it's mine. Here's the TSB for reference:

 

Speaking of symptoms, they're repeatable and specific. From a stop under heavy throttle or WOT, there's a distinct hesitation to start accelerating. Once accelerating, it accelerates like you'd expect. During a WOT 1-2 or 2-3 shift, it also sets P0303 and P0304 simultaneously, along with a flashing CEL. It does not *feel* like a two cylinder miss during the WOT run. I don't know if the two symptoms are related or not.

Here's two scan data captures during test drives with the symptom present. Don't forget to hit 'Open PID List' to make stuff happen.

www.scanshare.io/share/dIDAXZ7HhEe8ktFnwj4THw

www.scanshare.io/share/xmVKOgicqEmf_EIOncd7wQ

I ran the MAF values through the ATG VE calculator and got 95% or better. Fuel enrichment appears to be there. Before you ask, there are no misfire counters or data PIDs to watch, so I can't graph the misfires and the engine data together. I only have the codes and the flashing CEL to let me know it's pissed off.

Relative compression looks good enough. Intake pulses show two cylinders slightly worse than the others, but nothing huge. Sorry for the upload - Autel makes it very difficult to share waveforms.

 

I didn't save the cam/crank waveform I took. :blush: But the tooth count and the cam/crank alignment looks correct. I compared the crank waveform to the crank sprocket tone wheel off the old engine, and they match. I visually checked the timing belt and found it in time.

I did notice the 'Knocking Correction' PID going as high as 12 degrees at times. Is the PCM pulling timing like the TSB suggested? I decided to either unplug the knock sensor, or salvage the old knock sensor and plug it in/insulate it as needed to make the PCM think a working knock sensor is installed. See if the hesitation goes away. That'll be Tuesday or Wednesday.

So what say you? Is the timing a problem, or a red herring? What else am I missing?

Sounds like a cluster you know what…. That’s a good one

Tyler,

not offering any help but a side note on another maxiscope issue you mentioned at the meetup regarding the time measurement; there is a work around for it:

If you choose a shorter time base this will increase the measuring resolution, you´ll just will have to work with the zoom function to get enough repetition on the scrren if wanted.

wrote:

Sounds like a cluster you know what…. That’s a good one

That's all I work on! :silly: If you or anyone else I haven't pissed off yet run across another similar YMME, I'd love a known good WOT run to compare to.

Juergen, your comments were not lost on me. I did try downloading the Autel MaxiScope software onto my laptop, then dug the capture files out of the Ultra and SD Card-ed them over. The MaxiScope software won't read them. It'll only read .MSData files, whereas the Ultra saves files as .DAT. Annoying. Let it be known that I never undervalued ShopStream Connect.

You're right about the time base, I think? I'll experiment with that next week. I just get lazy and choose 5sec/div or 10sec/div because I want a whole capture on one screen. 500mS/div would probably be better. This is why I value you guys - calling me out on my bullshit. :silly:

Just try to change the suffix from .dat to .ms, maybe it will work?

12 degrees seems normal to me, hows ignition look? cyl 3 and 4 share the same coil. an amp clamp wil help identify a weak coil pack (1 coil using less amperage than the other) (it requires significantly more kv to fire at WOT). I got tired of my maxiscope real quick and bought a 4425A can't wait to get my hands on some of the powered probes! I have 1 scheduled for 2 weeks from now I can check but hopefully this 1 will be long gone by then

Sent from my SM-G781U using Tapatalk

Quick update on this new problem child.

I started working down the list of ideas we came up with during the last meetup. Unplugged the knock sensor - instacode + runs like ass. :lol: Reconnected the knock sensor, unbolted it from the block, bundled it up in a rag and zip tied it out of the way. Both the Knocking Correction and Ignition Timing PIDs show more advance, but still runs the same way.

Checked coil current ramps. Noted the 3-4 coil had a slightly lower peak, especially during a WOT snap in Park. Tried a new coil, no improvement. The secondary waveforms said I was wasting my time, but I felt the need to try it anyway.

Moved injectors front to back, or 1 <-> 3 and 2 <-> 4. Easy enough to do, and faster than a balance test. No change. Service advisor thought to check if the injector connectors had been reversed, and one bank was! Fixed it. Zero improvement.

Went in-cylinder on #3. Cranking and running compression looks good. No backpressure during a WOT snap. I added a spark tester and ignition probe on the #3 wire to check advance. Saw very little at idle or during a snap. Ran out of time last week, but I want to revisit this. I also didn't think to compare the actual advance to the indicated advance on the scan tool. :blush:

While we're on the subject, has anyone tried FreeSSM? Freeware Subaru software. Supposed to emulate the OEM tool.

sourceforge.net/projects/freessm.berlios/

I only ask because I'd still like to be able to watch misfire counters during a 1-2 or 2-3 run. Service info says the counters are there, Snap-On says they aren't. I'll try the Launch and Autel next week.

3 and 4 are a TDC pair. And the ECM probably syncs on 1 and 2. Maybe try scoping CKP, CMPs and coil current with the Pico. Then throw 720* cursors on the waveform to see if the 3 and 4 ignition events are off.

wrote:

3 and 4 are a TDC pair. And the ECM probably syncs on 1 and 2. Maybe try scoping CKP, CMPs and coil current with the Pico. Then throw 720* cursors on the waveform to see if the 3 and 4 ignition events are off.
 

Added to the list of tests to do next week.

Paging Chad. Paging Dr. Chad. :lol:

Not overly clear on what the problem is here ?

You said someone said the engine was knocking, so they changed the bottom end ? So was this other problem present before this ?
Was only the bottom end changed ? Any noises thereafter ?

is it off cylinders ? just down on power ? other ?

For more Subaru specific diagnostics and logging, this works fine, usually with the Tactrix adapter

www.evoscan.com/evoscan-gps-obdii-cables...penport-20-usb-cable

The specific problem now is a flashing CEL with P0303 and P0304, as well as a noticeable hesitation during acceleration from a stop.

The long block was replaced with a reman. There are no noises now. The customer maintains that this problem was not present before the long block swap.

The codes suggest it's misfiring, but my butt-o-meter doesn't agree. I also don't feel a lack of power. Just a hesitation as I noted above.

The Evoscan link doesn't work.  

weird, it worked when I sent it ! isn't now. whole webpage seems to be down

Certainly on the 2001+ turbo cars in the UK, valve/seat erosion is common and can cause misfires and poor running, but strangely often won't even show on a compression test. It just reduces valve clearances ( ie cam/lifter ) to nothing so the valves do not seat properly.
It's very rare I ever work at the n/a cars, although I think even those are quad cam motors at that age ? Maybe something there ?

Speaking of symptoms, they're repeatable and specific. From a stop under heavy throttle or WOT, there's a distinct hesitation to start accelerating. Once accelerating, it accelerates like you'd expect. During a WOT 1-2 or 2-3 shift, it also sets P0303 and P0304 simultaneously, along with a flashing CEL. It does not *feel* like a two cylinder miss during the WOT run. I don't know if the two symptoms are related or not.

This would be a good place to graph Crank frequency, with a sync.

there are no misfire counters or data PIDs to watch, so I can't graph the misfires and the engine data together. I only have the codes and the flashing CEL to let me know it's pissed off.

If you graph Crank frequency (and a sync) with a scope, you can use a third scope channel to sync with a scan tool data pid. Maybe, a shift solenoid? Or, some other pid that you can use to sync the scope with the scan data? In the past, I have use a Power Probe, connected to a scope channel, to manually insert a marker into the waveform.  It might be worth trying something like this. Or, it might not be.

In my experience, a flashing CEL has always been Primary/Secondary Ignition, or an Injector circuit.  I would graph the Crank frequency. If it shows misfires, I would take a very close look at Primary Voltage and Current, along with Injector Voltage and Current.

For the hesitation, on your next drive, I'd be interested to see Accelerator and Throttle positions, IAC, FT, and Injector pulse data, in addition to the previously posted pids.



 

wrote:

If you graph Crank frequency (and a sync) with a scope, you can use a third scope channel to sync with a scan tool data pid. Maybe, a shift solenoid? Or, some other pid that you can use to sync the scope with the scan data? In the past, I have use a Power Probe, connected to a scope channel, to manually insert a marker into the waveform.  It might be worth trying something like this. Or, it might not be.

 

Might be able to use CKP Hz and RPM PID to sync. Or just use CKP Hz to identify the upshifts if the misfires are repeatable. Could also scope the MIL control if it's hardwired at the ECM.

Might also be good to use a channel for injector current for this test. The ECM might disable injectors when the MIL starts flashing............

Great ideas, guys. I'll get some waveforms and post them up next week.

This is why I bring the fun ones here. Even if it doesn't directly lead to a fix, I still learn new techniques.

Evoscan link is working again.

Is the n/a version a quad cam ? does, it have VVT ?

wrote:

Evoscan link is working again.

Is the n/a version a quad cam ? does, it have VVT ?
 

N/A, SOHC. No VVT, just VVL.

Never worked on one, and very rarely single cam.

Makes it more simple anyway lol

Update with more waveforms. Thanks to everyone that's replied so far!

Before going down the scope path, I still wanted to get a scan tool with better data. Thought about Evoscan? :huh: But I'd have to get a special cable. Tried the Autel Ultra - wouldn't even attempt to talk to it. Got desperate and tried the Launch Pro Mini... Success! I couldn't transfer the data or the screenshot, but this is a shot of the data during a WOT run with the MIL flashing:

 

So the codes say #3 and #4, but the Roughness Monitor favors #3. :huh: While free revving in the bay, I can get both #3 and #4 to count up:

 

So I start thinking about the VVL system, since it's programmed to engage above 3500 RPM. I went in-cylinder on each cylinder to see if I could identify an issue. Here'a shot of #2 (theoretically known good). The red trace is the oil pressure switch for the VVL solenoid on the same bank. Short version, the oil pressure switch changes state when the solenoid is activated.

 

I generally take the slight dip in compression after engagement to be a good indicator of VVL operation. I also see a slight difference in the intake pocket. Here's #3:

 

It looks similar enough to #2? I'd appreciate your thoughts. As always, e-mail me at wrenchturnsyou@gmail.com for the original captures.

I also had a look at the cam, crank and an ignition trigger signal. One, to recheck the cam/crank sync, two, to check up on ignition advance, and three, to see if I could ID the true misfiring cylinder. To be honest, I couldn't make heads or tails of it. :silly: Again, I'd welcome anyone to have a second look.

Finding known good waveforms for this is a little challenging. I found two that looked similar to mine, but they had different relationships.

My cam/crank at idle:

 

Known good from Harvey Chan at iATN:

 

Known good from some guy at Rotkee:

 
 

Your crank/cam look the same, albeit your scope with the black screen just seems a pretty low resolution or capture speed ?

On the second red trace, why is there a dip at the bottom before it goes high ?

On the Quad cam motors, it is not uncommon for the mesh screen for the VVT solenoids to gather dirt causing issues, and/or the dirt getting into the solenoids. But as said, naver worked on any n/a single OHC motors, only turbo cars.