Ok so a quick update so I don’t forget the details:
Turbo blown, replaced.
Late 2021 iirc I service abused the original vf46 to death by topping off the oil for an unknown amount of Km/months.
Failure mode was soot and varnish in the banjo bolt filters, filters removed from the stock locations.
Turbo was replaced with a used vf46 and the inlet pipe upgraded to Perrin with the stock location pcv system returns.
Misfires fiasco:
Around this time the engine started loosing power briefly, luckily I was datalogging the issue, so at least I knew the engine was mostly running properly. No DTC’s!
The miss would come and go, but at last on a 1000k plus weekend work mission the engine started missing a few times per minute, on the freeway with 500km to the next gig!
I put some octane booster in the fuel and kept going.
Then on the final leg, the ecu finally threw multiple DTC’s and went into limp mode.
All cylinder misfires…
So whilst I’d been peeping the logs, I had very mild smoothing set on the AVCS VVT angle traces (stock setting in MLVHD).
The smoothing masked momentary dropouts in the VVT traces.
I added per cylinder misfire parameters around this time iirc.
Then the penny dropped: spark plugs 20kkm overdue!
And that’s by the very optimistic factory interval spec which is around 100kkm.
Some say 60kkm is better even for platinum or iridium plugs.
Oil leaks:
Around this time the throw out bearing got really noisy, and stayed noisy until…
Oil leak from turbo area, pouring, did I mess up the oil drain gasket???
Pull the turbo again and redo the drain fitting, linished the mating face, oem gasket…
Oil leak continues but seems to lubricate/quieten the throw out bearing: win lol.
Checks upper rear right of block looking for oil leak: it’s the sump breather hose, one of the two primary *block* pcv hoses.
Replaced the cracked pcv pipe with a diy suction hose made from heater hose with a stretched compression spring inside.
Intriguingly this pipe does not appear on partsouq, opposed forces diagrams.
I did find it in the Baja diagrams, in the FSM, and in dealership diagrams that use Subaru FAST EPC. Weird.
Oil leak continues and gets worse, still negative fuel trims, blowing smoke after startup, on boost, high rpm, still driving the car huge km every week for work.
Can’t use fresh air because of the stank.
Still inlet pressure testing, click click click from centre rear of block.
Take a peek with the tmic off, oil on top of rear block.
Find the vacuum leak and source of the oil leak: it’s the crankcase vent, this joins the heads balance pipes.
Instead of using the oem hose, I made my own again from heater hose and right angle joiners.
This allows the pipe to flex eg when lifting the intake manifold slightly to float in gaskets…
Bingo!
Finally fuel trims started trending around 0% instead of -5-10%.
The engine stopped idling rich.
The oil leaks stopped, however the affected area around the right and left bellhousing and rack subframe area required multiple cleaning sessions with degreaser and brake cleaner.
There’s still a slight drip coming from the bell housing area, so I suspect a new rear main seal and re-sealing the rear AOS plate is in order when the stuffed clutch is replaced.
AVCS fail again ad nauseum.
Then the engine went rough again, AVCS sync fail!
Hmmm, what’s the fault mode, since the filters were recently pulled.
Vvt angle oscillation, as we’ve seen from Jezza’s research, the solenoids do get falling on the pistons, making sync impossible and the affected bank goes from zero to max angle and back repeatedly. Easy to spot when logging.
I went through my collection of used solenoids and got one that worked on the second try.
However sync was a bit iffy, and then I remembered advice from the legend Simon at my local dealership parts counter: an experienced tech mentioned the filters are also called baffles in the literature, and may be considered to work as dampers to smooth out the AVCS oil pressure.
I had a spare set of banjo filters in my spare reco block, so I cleaned these up and installed them in the ocv manifolds, not at the heads supply ends as per stock.
Works great, I now have really good sync once the oil is up to temperature.
The car ran beautifully for a month, it was amazing to finally look at more “normal” logs.
FLKC aka fine knock learning started to clear up, and fully went away after a UEC treatment.
I added the UEC at the BPV reference hose for the idle and soak stages, but also added it via the map sensor hose to get to cylinders 1 & 2.
I’d been messing about with the knock sum parameter in MLVHD, and worked out how to turn it into a useful parameter:
The raw data increments & resets at a certain count eg 60 hits.
The count might go up in one increment or more per data hit.
So initially I used difference over time filter to turn a knock sum event into a data point, not an increment.
But I wanted the severity of each increment to be revealed as well.
After much experimentation, *totally heuristic* since I’m no maths genius, I got to a working knock sum per second positive parameter and it’s quite interesting, being potentially the data the ecu uses to determine FBKC, feedback knock count which might be referred to as mechanical knock learning.
You can find more info on Subaru knock control strategy on the Cobb and RomRaider sites...
This is the formula math, if you're interested in the schema. Knock Sum parameter is "*Knocks" in the formula
round((([*Knocks]-[*Knocks-3])/([Time]-[Time-9]))> 0 ? (([*Knocks]-[*Knocks-3])/([Time]-[Time-9])) : 0)
A filter can be applied to show data corresponding to these knock events: [Knock sum per sec positive] =0
Or show these knock events over a severity threshold: [Knock sum per sec positive] <3
If you get to use it, please share your results!
Inlet pressure test smoke generator upgrade V3:
TBC… almost there
So the LV’s looked great for around a month, then the rich idle returned, and fuel learning went -5% but in the two upper airflow ranges.
Time to inlet pressure test, this time using just my breath to pressurise the inlet tract, from the bpv reference hose to have silence during the test, and to be closer to the intake manifold.
Tang tang tang noise as I pushed over 1psi in the inlet tract…
Hmmm what could it be? Adding soapy water spray got it making a weeee squirrely noise decreasing as the pressure very slowly dropped.
I had found some resistive wire, a 240V replacement radiant heater element at a garage sale, so why not ditch the hard to seal 12V soldering iron, and use some resistive wire crimped to heavier gauge wire to a 12v battery?
And it works great with the element in baby oil, just gets hot enough to heat the oil but not burn it.
So with the smoke in the inlet, bicycle floor pump for silence, I found the leak below the passenger tgv: lower intake manifold gasket.
These have really low torque spec due to the need for thermal expansion & contraction and I guess at some point I messed this up?
They’re kind of a 100kkm service item and mine weren’t far off that so I put in a new oem gasket with the air pump removed and the inlet lifted slightly with a crane.
However the old gasket was difficult to remove and when I inspected the pieces…
One section was missing! About an inch of one of the thinner sections in the middle.
Concerning surely, so I broke out the inspection mirrors, inspection camera and very bright led torch and had a thorough look in the ports and mating faces… nothing.
Hmm I really should have bit the bullet and fully pulled the intake manifold, but I’m always looking for a hard excuse to retire my not quite dead 340kkm long block, so I buttoned it up, tested inlet pressure again (all good) and fired it up.
Tap tap tap, clank clank clank and the offending piece was munched up!
And that’s where it’s at, ready for a test drive after another idle relearn following another UEC treatment yesterday.
Amazingly the gasket piece munch did not register at all in the logs I made at the time.
If I get another intake manifold gasket failure, I might try an aftermarket gasket, the type that appears in photos to have a raised seal around the ports as opposed to the oem flat graphite/fibre on a thin ally core type.
New “eyes”!
Todays mission: time to do the 2k clear headlight refurb on some spare headlights in the stash.
My rhs stock HID low beam was flickering and then died.
IME this might be the bulb, the igniter (bulb socket) or the ballast, luckily I have multiple spares to get up & running with fully refurbed “eyes” much needed!