Rough Idle Still 2003 OBW

[Vinnyvango]

I've had this car for less than 6 months, but there's still a lot of history. I'll try to be brief (otherwise nobody will read all of it:-).

 

2003OBW 76K miles bought from a used dealer. It had a very slight miss at idle and some intake noise. I found the air cleaner housing was not installed right and had non-stock plug wires. So I bought it. It seemed like the mysterious rough idle, slide it into neutral problem. ........Then it overheated. Shop replaced the radiator and thermostat. Ran fine but I kept chasing the rough idle. New NGK plugs, stock plug wires, seafoam treatment, cleaning IAC did nothing to help. I even reset the valves. No improvement.......Then came the bubbles in the coolant overflow tank........Well, at least I've found a problem I can fix.

 

With the help of BlutoE I pulled the engine to replace the head gaskets. I used Subaru gaskets and timing belt and an Ebay idler kit with water pump. Took the heads to a shop for milling. ALL EXHAUST and half Intake valves leaking. 3-angle valve job and $400 the heads are done. I realized while installing the first head that the head bolts were mixed up (some hack has been into this thing before and mixed the bolts:mad:). There was some confusion over the tooth count on the TB. FSM says 44 and 40.5 (I think). Mine wasn't even close. I ended up at 47 and 43.5. The mark on the TB didn't quite line up on the driver's side. But, after pulling the pin on the tensioner all timing marks were straight up. I set the valves myself. Again, FSM was wrong. The sticker under the hood indicated 0.10" intake and 0.15" exhaust.

 

Engine in and fires right up. But it still has the rough idle. But this time, putting the trans in neutral helps, but it's even worse than before the HG. I hooked up the vacuum gauge at the brake booster and I get a rapid oscillation between 19-20 inHG at idle. When revved it jumps up like it should but the 1 inHG oscillation is still there. No CEL or pending codes.

 

So, here I am. It seems to run fine above idle. Gets decent gas mileage. No oil use. I'm stumped. I'm having a hard time believing I screwed up the TB or setting the valves. The wife doesn't seem to mind, but after this much work, it should run like a sewing machine.

 

Thank for all the help!

[BluetoE]

I guess you stumped 'em!

[hankosolder2]

Are the cylinder compressions equal? Is the frequency of the vacuum oscillation what it would be if one cylinder was low, or is it a slower drift?

 

I wonder if you could have a leaking injector? Remember the 02 sensor is working off the average of all four cylinders, so if one cylinder is running rich at idle due to a leaky injector, the computer would lean the other cylinders out too much trying to shoot for what it thinks is an appropriate mixture.

 

 

Nathan

[Fairtax4me]

Do a compression check and leak down test.

[jeffdpia]

Leaking HGs will usually wreak havoc on the O2 sensors. Mine does the same thing after a HG, TB job, just an older 96 obw. After replacing the upstream O2, idle and drive-ability really improved. Idle now somewhat smoother, but still low with vibration @500, have to replace downstream next.

[Vinnyvango]

Leak down test done. I did the passenger side first and it was as tight as a drum. I got a little leakage on the drivers side cylinders, but still below 5%. I'm guessing this was because the engine had cooled by the time I got to them.

 

But the timing mark on the driver's side cam sprocket was tilted inboard. So I pulled the timing belt to be sure. The timing marks on the Subaru belt were already gone. So I marked the position of the crank timing mark on the belt with a sharpie and pulled the belt. I did the tooth count thing umpteen times. I even compared my sharpie marks with the non-stock belt that came with my idler kit.

 

Installed the belt, pulled the tensioner pin and guess what. The timing mark on the sprocket shifted inboard again. But the tooth count is the same.

 

SO I've lost all confidence at this point. Do I trust the tooth count? AND tooth count is actually valley count on the belt, correct? AND the number of valleys on the belt is 47 and 43.5 correct?

 

Thanks. I'm hoping I don't look like George Costanza by the time this is over.

[Fairtax4me]

Don't use the marks on the belt. Never use marks on the belt, ever. If the mark on the sprocket looks a tooth in, set the belt a tooth further out and then pull the pin.

 

It's doing that because there is slack in the belt on the other side of the sprocket. If you turn the sprocket clockwise a hair with a wrench, set the belt on it, then turn it back to the mark, that should pull most of the slack out.

[Vinnyvango]

So the timing marks should all point up regardless of the tooth count?

 

Thanks

[Fairtax4me]

Not necessarily. But if the engine doesn't run right with it the way it is, try adjusting the position of the belt and see how if that affects it. It'll either be cured, or will get significantly worse. One tooth is not far enough off to make the valves start interfering, so no harm done if it doesn't work, just set the belt back to the old position and try something else.

[Vinnyvango]

O.K. I tried it. Seat of the pants, maybe a little better. But not fixed:(. It seems to have more compression. After a few minutes of running the timing mark on the driver's side cam was a little closer to vertical, but tilted outboard. (I guess it's either inboard or outboard a little) Tooth count on that side is now 44.5.

 

According to SWMBO there is no change. I feel like it's a little beter.

 

I plugged the vacuum gauge into the brake booster again. A little higher at 20-21 but oscillating still. I then tried the PCV hose. The same range but oscillating very rapidly. What could that mean?

 

My buddy has a 02 OBW so I'm gonna give his O2 sensor a try later in the week.

 

I welcome any input.

 

Thanks!

[Fairtax4me]

That means the PCV valve is working. The PCV hose is not a good source for vacuum readings because the PCV valve is in the way. It has a check ball in it that rattles around during normal operation. It's there to keep liquid oil out of the intake manifold if it ever backs up into the pcv hose. Protects the engine from hydrolocking in the bizarre case the cylinders suck in a gulp of oil instead of air. This would only happen if you had other major problems, but it would mean an end to an otherwise repairable engine.

 

The problem with the "marks" that the cam sprockets have to line up with is that they are in a peice of molded plastic, so they aren't very exact IMO. The "right spot" could be either side of the mark by a tooth or more. This is one reason I always make my own marks when doing timing belts.

 

Good news is you got a good result and that is a higher vacuum reading.

Did you try adjusting the position of the belt on the opposite cam as well or did you do just one and leave it at that?

[Vinnyvango]

The PCV valve was not in the vacuum path. On my 03 the valve itself is on the crankcase, not the plenum. So the oscillation still bothers me. Where should I connect the vacuum gauge?

 

And no, I did not adjust the other timing belt. The Passenger's side timing mark is directly in line with the seam in the head (no rear timing cover notch).

 

Thanks!