2005 Legacy GT wagon: a bone stock rebuild

[birkhoff]

Keeping things rolling here.

 

First startup and run-in

 

In prep for the startup I did the following:

 

The previous day (April 11?) I finished up by filling in coolant (I used plain water since this was only going to be in for a few minutes) and oil. The next day I checked for leaks. None. I don't have any way to pressure test the coolant circuit otherwise I would have done that.

 

I'd installed a mechanical gage on the oil pressure port (nb: 1/8 inch BSP thread); you can see that in the previous pictures.

 

As for the run-in oil, up here we have a big box store called Canadian Tire. Amongst other things, they sell oil, cheap. Motormaster brand. It meets all the specs and is probably perfectly good oil, but I chose that since it is easy and cheap to get in 5 litre jugs. Moderate temps up here mean 10W-30 for the most part. Canadian friends here will appreciate the pic.

 

Cracked open the banjo fitting at the turbo to confirm oil pressure. Turned the key and nothing happened. Battery that I had pulled and stored since January and left on the charger overnight was dead. Rats!

 

Got the big boy out of my 300 SD (60 lb battery) and continued under jumpers.

 

With a helper watching the banjo filter, I cranked. It spins fast with the plugs out. Within a few seconds oil was puking out the banjo fitting! I had pre-primed the turbo some weeks before and evidently, oil pressure to the AVCS circuit builds fast. It was a bigger mess to clean up than I planned. Oil pressure at crank speed with cold oil was 70-80 PSI. So far so good.

 

Installed plugs, coils, airbox and cleaned the MAF.

 

Blew off and cleaned up the oil at the banjo, checked again for leaks (none) and we're ready to go.

 

Fired up without a moment's hesitation. Fast idled on its own, but threw an immediate CEL. I didn't even pull the code. I noticed I had forgotten to plug in the MAF line! Cleared the CEL and restarted. CEL gone, idle dropping to 750 but won't hold. Obviously. ECU needs to get acquainted with the new setup. After 5 minutes of nursing the idle and continuous monitoring for leaks and oil pressure the ECU had learned enough to hold stock idle speed except for a few stumbles when tweaking the accelerator. All completely normal in my experience with this car.

 

Coolant temperature rose and stabilized where it should. Rad fans came on a few minutes later. Still no leaks so everything holding pressure. Idle oil pressure now a little above 20 PSI.

 

At the 10 minute mark I shut things down and dumped oil and coolant. Pics of the 10 minute run-in oil show a nice clean build. A few threads of black sealant. Absolutely no metal. Maybe a little moisture in there, which is understandable, but nothing indicative of a coolant leak. I let the oil settle overnight and filtered through mesh and paper towel before taking some pictures.

 

So ends April 13, about 18 weeks after the engine first came out

 

Unfortunately I was so busy with this startup that I failed to get any kind of video or even a picture (of a running engine?).

[seabass07]

Nice. Hope this works out well.

[StkmltS]

Awesome!

[seanyb505]

congrats!

 

I chose to torque everything together before the install (turbo, plugs, coils packs etc). I really didn't want to have to fiddle with the spark plugs, and I've removed the trans for a clutch job previously without fiddling with the turbo, so I knew it could be done.

 

The weed sprayer oil primer worked well for me.

[birkhoff]

500 Mile Break In and Chasing Down a Few Gremlins

 

After the 10 minute oil change, I settled into the first 500 miles. As I said before, I don't drive very much, so this took a little time. My wife helped by doing a highway trip up island that put on about 200 km.

 

At the end of 500 miles did the obligatory oil change and moved to what I will be using for a while: 10W-30 Pennzoil. Drain oil looked pretty much the same as the first change. Pics below. Some small bugs in there from sitting overnight. Sorry bugs. Still a few flecks of black sealer showing up. Good thing I have all my banjo filters intact . I didn't use anything black in this engine, so these are leftovers from the factory build. Very hard to get them all out of the threads and whatnot in the block.

 

Gremlins

 

0. Slight coolant leak at the turbo inlet pipe. Since this is a super easy to replace hose, it didn't make my 'better do it while you can' list. Of course it leaks! Double clamped it for now and added to my final Heuberger shopping list.

 

1. Oil leak at the PS pump inlet tube. This is very common, but can make a big mess, fast. Here is a useful link if it happens to you. Synopsis: Don't take it personally; eventually happens to everyone who pokes and prods at this hose; think: engine pull. The only proper fix seems to be the factory (metric) silicone o-ring. Just get it.

 

2. Loose cat in up-pipe. Prior to pulling the engine, in addition to the awful metallic sound that caused me to finally bail and rebuild, I had noticed a lighter metallic clanking sound at certain RPM. Difficult to isolate and had been around for about a year. However, almost immediately after being back on the road I heard it again.

 

No panic. Since I only re-used a few things like turbo, manifolds and pipes, and the turbo had already been swapped in a failed attempt to get rid of the sound earlier, it was easy to trace to the up-pipe. Here is the sound of a failing cat. . It does its magic best while dropping rpm just above idle. Putting a stethoscope on the up was definitive.

 

Luckily I had a spare up-pipe around and just had to swap that in (1/2 day of work, all said and done). I decided to pull the turbo to do this, rather than disturb the manifold to head joints. Not really a small job either way.

 

Once the old up was out, I could see the centre section of the cat moving axially up and down a couple of mm just by shaking.

 

3. Low boost pressure. I don't have any way to log the ECU on this car (waiting for a cable) but testing the seat of the pants way, I seemed to be getting low boost. Realized pretty quick that in my zeal to replace all rubber under the manifold I had pulled and tossed the restrictor pill. Kind of ironic because my practice is to save every little bolt and piece of rubber or gasket on a rebuild until the break in is complete. I do remember, as I was pulling out little pieces of hose, thinking -- madness. Keeping used vacuum lines? Toss it. So I have pretty much every hose replaced EXCEPT the turbo control lines!

 

Made up a restrictor pill from a mig tip, fabricated a mechanical gauge at the BOV and confirmed I am getting 10psi and then it tops out. I made my pill a little large (#55 drill) and I may have a problem with the BCS, but for a break in, 10 psi is probably a good idea. I'll get back to this later

 

SPACE RESERVED FOR LINK TO BOOST THREAD, IF REQUIRED.

 

 

4. Rough idle and hesitation on partial throttle. This has me somewhat concerned. It was present before the rebuild but I assumed it was due to cracked pistons. Obviously not, in retrospect.

 

It is subtle, and most people riding in the car might not even notice. Possible sources are intake, fuel, exhaust or ignition system. I realize that is a lot of ground, but I can (at least in the short term) eliminate a mechanical problem in the long block which is always the big unknown. Without any way to log, my hands are tied for the moment. No CEL, ever, which is my primitive way of logging up to now.

 

Basic tests for vacuum or boost leaks are negative, but I haven't gone carefully enough with that to be sure. Once I get a cable, if it seems isolated to one cylinder, I"ll move coils and plugs around to try to find the source. I have one ignition wire plug at the coil with a broken lock. I also have an annoying RF interference on AM radio at some frequencies. I don't remember that being around before. Both chassis grounds are in place and secure. Until I have an idea that it is one cylinder, that is about as far as I want to go.

 

TO BE CONTINUED IN A LINKED POST IF HESITATION CONTINUES.

 

Edit: Turned out to be a bad injector on #2. Swapping in a known good used one solved the problem.

 

5. Fuel consumption At this point, with a 60/40 mix of city/hwy miles I am getting about 17mpg average since rebuild. I don't know if this is good or bad. It is not awful.

[StkmltS]

Which cable are you waiting on and how are you going to log? I highly recommend BtSsm if you have an android device. It can monitor just as well as an AP or laptop.

[birkhoff]

Which cable are you waiting on and how are you going to log? I highly recommend BtSsm if you have an android device. It can monitor just as well as an AP or laptop.

 

Cheap VAG-COM cable from China. RomRaider on a laptop. Log only.

[StkmltS]

Get the BtSsm app if you have an android phone or tablet, I think it works with most of the vag-com cables. All you need is a $5 OTG adapter and you're good to go. Then you only have to use your laptop when you want to reflash your ECU. No more carrying it around or hiding it somewhere in your car. Monitoring your engine with BtSsm is as easy as plopping your phone into the mount that you probably already have and plugging it in.

 

*Disclaimer: I don't earn money from BtSsm, I just really like it!

[xt2005bonbon]

Get the BtSsm app if you have an android phone or tablet, I think it works with most of the vag-com cables. All you need is a $5 OTG adapter and you're good to go. Then you only have to use your laptop when you want to reflash your ECU. No more carrying it around or hiding it somewhere in your car. Monitoring your engine with BtSsm is as easy as plopping your phone into the mount that you probably already have and plugging it in.

 

*Disclaimer: I don't earn money from BtSsm, I just really like it!

 

so much

[birkhoff]

Hey, I just found out that my mother-in-law has an android tablet, so maybe a copy of the highly regarded BtSsM is in my future after all. It's all Apples around my house.

 

Mother-in-law: Android is much better than Apple. More you can do with it. Cheaper hardware. A little harder to learn upfront, but once you get the hang of it . . .

 

What is going on with this world! My MiL is 85 years old! I just got my first cell phone (the usual daughter's-hand-me-down-for-cash deal) a few months ago. It appears I've got some serious catching up to do

 

Now, if and when my $10 Vag-Com comes from China . . .

[xt2005bonbon]

You will NOT regret the btssm app. You will wish you had this earlier.

[magnusonsubie]

you will not regret the btssm app. You will wish you had this earlier.

qft!!!

[birkhoff]

Ok, my cheap cable finally arrived and I got hold of an OTG enabled tablet.

 

My first run with BtsSm was a mixed success. It occasionally locks up and I have to disconnect the cable and restart the app. Otherwise I'm getting the hang of it. I don't trust myself to drive the car while this thing is running!

 

First step is to chase down the rough idle condition. I set up BtSsm and monitored misfires and AFR% in my driveway. No load condition.

 

Cylinder 1 is showing misfires. All others steady at 0.

It looks like the ECU counts misfires over some time interval and then resets. Is that correct? Or is it crank revolutions?

 

At idle speeds, I'm getting 1-3 misfires per window.

At 2000 RPM 10 -14 misfires. I don't really detect them at that speed under no load. Didn't really want to go much higher without load. Not sure my neighbours are pleased with this new toy!

 

Not all roughness/hesitation at idle registers at the misfire gauge, but some of it does. Sometimes I feel a hesitation and immediately get a +1 in the cylinder 1 misfire count.

 

The other gauge i've watched is AFR%. It sits pretty steady at 14.0-14.5 with occasional dips. That seems a little high, but I have no experience to judge, really.

 

My suspicion is that either the connector or coil pack on #1 is flakey. I remember that this coil has its connector tabs broken off. I think I'll start by pulling the connector, cleaning it up and maybe zip tying it to the coil. If that doesn't fix the misfire I'll move the coil.

 

Am I on the right track here?

[xt2005bonbon]

In my experience, if the app hangs, you are probably having issue with the OTG cable. A poor connection with the tablet/phone will do this to you. Assuming good cables are used, the app should work flawlessly (been using it for almost 2 years every time I drive..)

 

Regarding the misfire, this reported variable (per cylinder) was originally named 'roughness count'. I still don't know its actual unit (e.g. misfire per ???). The only thing I can tell you is that the ECU will throw a misfire CEL if that number will get higher than 30 or something. Also, this number will not go above 99. Once past 99, it'll reset to 0.

 

It is believed to be OK to get an ocassional count of 1 per cylinder. But in your case, something is definetely up. I would swap coil with another cylinder and see if the misfire follows.

 

Also, you did not specify if this issue occurs with warm or cold or any temp engine.

 

Finally, what are your AF learn values? Pull an 'LV' to see these values and report.

[birkhoff]

With my 12 yr old son as co-pilot and in charge of the tech, I got some data.

 

 

First, the image below is a snapshot of LV. I have logs with fairly high load so I assume my fuel corrections are running well into the 40g/s range, where there seems to be some sort of alert (orange colour).

 

As for misfires, by messing around I determined that the worst misfire count is generated between 2000 and 2500 rpm under no load. I was able to run the Cyl 1 misfire up to 40 per window. That, of course, generated a CEL event. P0000 and P0301. I assume the 0000 code is a hardware issue with the log cable.

 

Swapping #1 and #3 coils made absolutely no difference.

 

Swapping #1 and #2 injectors completely cured the problem for cylinder#1.

#2 however, is only showing a max of 3-4 misfires per window, more than before but nothing like what it was previously doing to #1. Does that make sense?

 

I guess that is enough evidence that the misfire problem was primarily caused by the injector. All other cylinders (except #2 now) are showing occasional misfires, but never more than a count of 1. At almost any given time, the three of them are sitting at zero.

 

MAF still seems high to me. It pegs at 1.38 (lambda) under closed throttle, no load, which is probably normal, but on a hard pull we were getting numbers around 0.98 and almost never much below that. I'm thinking this is awfully lean.

 

All my logs are short (tech problems) but I'd be happy to upload one or two if someone wants to look. Or generate some new ones if there are suggestions.

[xt2005bonbon]

Did you check your valve clearance before you put the valve covers back after the rebuild?

 

Good to hear it may just be an injector problem but it is odd that there is no similar high count in #2.

 

Your fuel trims are a bit too negative than I'd like. I'd suspect you have a boost leak.

[birkhoff]

Did you check your valve clearance before you put the valve covers back after the rebuild?

I did not.

 

Should've. Heads were reconditioned and set up by the only shop in town that does this subaru work. Does 5+ sets per week, same guy every time.

 

But should've.

 

Good to hear it may just be an injector problem but it is odd that there is no similar high count in #2.

 

 

Well, the old #1 injector spiked a 17 count in the first few seconds of life in its new home, then settled down. Still showing 3-4 quite frequently, and cyl #2 was a flat 0 before, as is cyl #1 now. Moving the injector definitely fixed things for cyl #1, it just didn't mess things up as much as I expected for the recipient #2.

 

 

Your fuel trims are a bit too negative than I'd like. I'd suspect you have a boost leak.

 

That is possible, although I did the 'blow into the BOV pressure line' and it holds good pressure for quite a long time. 10 seconds or so? I may have to set up for a proper pressure test from the inlet side of the turbo intake.

 

I've attached a CSV log that starts with some variable throttle stuff with no boost, then ends with a 3rd gear foot to the floor pull with boost rising to 8 PSI at the end. Software cut out at that point, but I was reaching the end of my test hill anyway.

BtSsm_20160714_185521.csv

[birkhoff]

Injector in #2 has settled into its new home and will happily throw 15-20 misfires per cycle at around 1750 RPM once everything is warmed up.

 

All others remain stable at 0 with very occasional 1 (maybe once per 10 min in one cylinder, chosen at random).

 

I'll try and dig up a replacement for #2 injector and then get back to logging fuel trims, AFR and whatnot without a bunch of raw gas confusing the A/F sensor.

[xt2005bonbon]

You could always send the injector to deatshwerks or witchhunters for servicing. I think it is like 25-30$ or something.

[StkmltS]

Hopefully it's just something simple like that.

[birkhoff]

You could always send the injector to deatshwerks or witchhunters for servicing. I think it is like 25-30$ or something.

 

The ones in there went through Deatschwerks about 30K ago. I bought a backup set used off the forum and had them sent directly to DW and then to me. So I could use DW again, but it is a one month turnaround given two border crossings.

 

I'm trying to find someone up here that does cleaning and calibration.

[birkhoff]

Your fuel trims are a bit too negative than I'd like. I'd suspect you have a boost leak.

 

Can I ask a dumb question about this again? Remember, I'm completely new to this tuning thing, so be patient if you can.

 

If fuel trims are negative, doesn't that means that the ECU is 'learning' from closed loop operation that the stock tables are probably injecting too much fuel for a given MAF value during closed loop. That would mean a too-rich condition which is kind of the opposite of what I would expect for a boost/vacuum leak.

 

Very little run time has been under any kind of boost (which I agree would turn things around the other way). However lots of run time has been at loads above the 40g/s in the top cell of the open loop learning table. In other words closed loop, but under vacuum.

 

Or do I have this backwards somehow?

[xt2005bonbon]

First of all, I am no tuner . But essentially, we suspect a boost leak when the trims become noticeably negative.

Let's first assume that your MAF and o2 sensors are in good working order (important assumption!).

Let's now assume that the MAF sensor measured X amount of air entering the air intake.

Now, let's assume that Y amount of air (with Y<X) actually entered the intake manifold.

As a result, the ecu will pull fuel making the trims negative.

 

So that means that between the MAF and the intake manifold, air had to escape somehow. And the logic is that it would escape if you had a boost leak, which would be located post turbo (e.g. between the turbo and IC, between IC and throttle body hose, or between throttle body hose and throttle body; also if the IC is stock, that sucker can leak).

 

From personal experience and what I've seen on the forum, a common location where this type of leak occurs is between the TB hose and TB itself especially if you run higher boost. The stock TB hose has a tendency to come apart.

 

If you have not been able to locate a leak with the 'blowing' test, you may want to do a smoke test. If there are no leak, then, you need to confirm that the MAF and o2 sensors are in good working conditions. If they are, confirm that the MAF wires are good. If they are, then time to adjust the tune..

[birkhoff]

Front O2 is brand new. MAF sensor is original, so an unknown as you say, but recently cleaned.

 

Maybe it comes down to this point: When the trim cells report negative, is that instructing the ECU to deliver more fuel or less, in open loop?

 

I was assuming less, in other words, add the trim amount to the base fuel quantity in the fuel/MAF table to determine the learned fuel delivery level in open loop.

 

If X g/s is going past the MAF sensor, and Y is sneaking in through a leak (under vacuum conditions) then X+Y is the delivered air charge. If fuel calculations only see X, then AFR should drift lean. In closed loop, feedback from the O2 sensor should tell the ECU to add more fuel, which then gets translated to the open loop fuel trim tables as . . . a negative?

 

To me, boost leak is just vacuum leak spelled backwards, but maybe that is my mistake.

 

Sorry, thinking out loud here. Feel free to ignore! I'll clear my ECU and runs some more logs.

[xt2005bonbon]

Maybe it comes down to this point: When the trim cells report negative, is that instructing the ECU to deliver more fuel or less, in open loop?

 

^It instructs it to deliver less.

 

 

 

We usually call it a boost leak when metered air has escaped between the MAF and intake manifold.

 

Conversely, we call it a vacuum leak when unmetered air has entered the system between the MAF and the intake manifold.

 

So, in the case of a boost leak, MAF measured X. But then the o2 sensor noticed less than X entered the combustion chamber. Then, the ecu is instructed to decrease fuel delivery. This will show up as negative values on your learning view.

 

In the case of a vacuum leak, say MAF also measured X. But then the o2 sensor noticed that more than X got into the combustion chamber. Therefore, the ECU is instructed to add fuel, thereby rendering positive fuel trims.

 

 

Also, I would not reset the ecu until this gets sorted out.