OB99W

The EJ engines are fussy. Be sure to use original spark plugs and OEM Subaru wires. You won't be needing those. EDIT: Also, check/clean or replace the PCV valve.

While it's not impossible for there to be a transmission problem that excessively loads the engine, it's not common. Unlike years ago with uncontrolled carburetors, the engine management system on modern cars compensates for loading, so the drop in idle speed from 700 rpm (I assume that's when in neutral) to 500 rpm in drive isn't normal. When idling in drive, if you take your foot off the brake, does the car creep forward, does the idle smooth out, and what rpm do you see?

No, '99 was the "lucky" model year for the problematic MAFs. Previously, they were hot wire, and in 2000 the SOHC engines got a MAP-based system.

There are two basic types of MAFs, "hot wire" and "hot film". The wire type runs hotter than the film, and tends to not contaminate as readily. Many hot wire ones have the ability to self-clean by elevating the wire temperature way above that of usual operation. Not so with the film type. The '99 used a film type MAF. They've been somewhat troublesome, and there was a bulletin issued (WWG-90). If you have a defective one, it should be replaced. Check with Subaru of America at 1-800-782-2783. Please let us know the outcome.

Thanks, 2000 Legacy, for recognizing the many people who make USMB what it is. Okay, the belt doesn't have as much mileage on it as I had presumed (based on previous info). That's good. Yes, manually turning the engine while in gear could be "difficult". I guess I could have mentioned it being a bit easier when in neutral. Not to nitpick (but so that anyone who isn't aware of this has a better understanding) -- under the conditions described, the "arrows" would be positioned at 1:30 o'clock. That's because there's 45 degrees between them and the "lines" on the cam sprockets. By the way, at the crank sprocket the relationship between arrows and lines is 90 degrees. Unfortunately, the notch isn't very deep, and rust takes its toll. Based on your having determined only a couple of degrees discrepancy between the cam and crank positions, it's fine. (The test isn't precise, it's "ballpark".) If it had existed, a single tooth jump at the crank would have shown up as about 15 degrees off. The full correct part name is actually "neutral position switch". On some vehicles, that sort of part is used as a starter lockout, but on your Subaru with manual trans the "safety" lockout is a clutch-operated switch. (Auto trans cars have an "inhibitor switch" which allows starting only in Neutral or Park.) Some folks refer to all such trans-related switches as "safety", even though it's not alway accurate. For 2003 Legacy, should apply (hosting site Russian, info in English ): http://www.subaru-car.ru/leg3/leg3_trans_4-1.pdf http://www.subaru-car.ru/leg3/leg3_trans_4-2.pdf http://www.subaru-car.ru/leg3/leg3_trans_5.pdf Personally, I'm willing to stick with anyone who's as dedicated as you've been. I do have further thoughts on your problem, but I'll post them separately.

So it was due to a "critter" and a hose -- but the "critter" wasn't of the 4-legged variety. Agreed, it's a common problem, might as well replace it You're welcome.

In and of itself, a hole in the filler neck shouldn't have that much impact on running. How hot did it get where the car was stored for the last several months? The more volatile portions of gas can evaporate (especially if there's a hole in the filler neck), and what's left behind doesn't ignite as easily. If there was a significant amount of fuel already in the tank, then adding five gallons of fresh gas to it probably wouldn't make enough of a difference if indeed the rest of it was in poor condition. Getting the code(s) read could be helpful, as might draining the tank and using completely fresh gas. By the way, if the car was stored where "critters" could get to it, look around for chewed wires and hoses.

How do you know I'm not in the process of gathering material for one? Yes, but it's better to not use the arrows during a belt change, since besides the difficulty of holding things in place, there's more chance of interference at TDC. That's why the "lines" are there. The arrows are normally used to obtain correct positioning for valve clearance adjustment on solid adjuster (non-HLA) engines. You set the arrows at 12 o'clock for cyl #1, then rotate at 90 degree intervals, checking/adjusting the valves in firing order.

You're welcome. I enjoy the challenge, especially if the problem is a little unusual. If it wasn't enjoyable, I wouldn't participate in the forum. (Although there are times when it gets frustrating enough so that I'd like to reach through the computer and put my own hands on the car.) There are numerous other forum members who are quite knowledgeable and freely offer advice, so please don't single me out. Some of those members are certainly more experienced than I am in certain areas (it's hard to be expert at everything), but they might not have as much free time available to post as I do. Even though I reread things before posting, some errors still slip through. In my previous post I wrote "If the belt has jumped a notch at the crank, ...". I meant to say "tooth" where I wrote "notch".

It could be trans slippage, if the engine revs are going up but the car isn't correspondingly accelerating.

Welcome to the forum. How many miles are on the odometer? Other than the knock sensor, has any other maintenance (preventive or otherwise) been done by you? Do you have any information as to what work was done on the car by the previous owner? Does the problem present itself any differently if the engine is cold versus warm? Can we assume that the CEL/MIL light isn't on? What happens if, while the hesitation is occuring, you give it substantially more throttle? I suppose that's enough for now.

This post is long, but I think it's worth reading. (But I'm biased. ) It's not necessary to clear the codes, etc. Until the problem is fixed, the misfire (p030x) codes will keep coming back. We know that the other two were set by disconnecting things, so they can be ignored. Thanks for the readings, they're useful. Let me go into a little operational theory first. The MAP reading with key-on/engine-off is the barometric pressure. The ECU gets that at start-up, and uses MAP values with the engine running relative to that base. If a MAP reading with the engine running is subtracted from the base barometric reading, it should be equal to the intake vacuum, and should agree with the gauge reading (within reason). I created the table below, using the measurements you provided (and my calculator ). It's in the form {MAP Base} minus {Engine-running MAP} = {Calculated vacuum} versus {Vacuum gauge}. 29.2 - 15.6 = 13.6 v 12 29.2 - 18.8 = 10.4 v 10-11 29.2 - 19.7 = 9.5 v 10.5 29.2 - 18.6 = 10.6 v 11 29.2 - 17.7 = 11.5 v 11.5 29.2 - 16.5 = 12.7 v 11.5-12 29.2 - 17.4 = 11.8 v 12.5 29.2 - 14.7 = 14.5 v 14.5 29.2 - 13.2 = 16.0 v 16 29.2 - 12.9 = 16.3 v 17 29.2 - 12.4 = 16.8 v 17 Considering the accuracy of a typical vacuum gauge, and the difficulty in reading it precisely, it appears that the MAP sensor is working well -- the calculated intake vacuum is in reasonable agreement with the gauge. So, it looks like the low vacuum readings at cold start are real, and the MAP is just reflecting that situation. The rich mixture may be a result of the low intake vacuum. Engine systems see low vacuum as an indicator of higher load, and respond by providing more fuel -- the FPR receives lowered vacuum and ups the pressure to the injectors. The fuel trim values in the -12 neighborhood indicate that the ECU is trying to correct what it sees as an overly rich mixture. However, it has no control of the fuel pressure, just the injector pulse width (duration), which is what the fuel trims are for. The ignition timing ("Spark Adv") readings ran from 12 to as low as 4 degrees of advance. More typical would be 10 to 15 degrees with low engine load, especially at idle. The retardation from the usual amount of advance will have an impact. It may be that the knock sensor is "hearing" things related to the problem. Let's look at the possible causes of low intake vacuum. We've ruled out leaks, because you couldn't find any, and because when you disconnected the brake booster vacuum line the engine ran better, rather than worse or stalling. Poor general engine condition, such as badly sealing rings or leaky valves, can be a problem, but the high compression readings you previously got indicate otherwise. That leaves mostly "breathing" problems. Tight (insufficiently opening) valves can cause poor breathing, as can excessive carbon buildup on the back side of intake valves. (This is likely why the dealer wanted to pull the heads.) Restricted exhaust can also contribute to low vacuum, but the fact that running the engine at 2000 and 2500 rpm didn't lead to lowered vacuum, but rather an increase, tends to rule that out. However, there's another possible cause -- valve timing being wrong. That can be due to a jumped timing belt, as I mentioned many posts ago. I know that you and DaveT looked into it, but not completely, as per post #50. Apparently the cam alignment (relative to each other) was checked, but not at the crank. Unfortunately, due to the fewer teeth being engaged at the crank sprocket (and other factors) versus at the cams, if a jump occurs it's much more likely at the crank. I previously said "Timing belts wear, and 85,000 miles is getting near the end of life for one. Also, the tensioner and idler pulleys, especially if they weren't replaced the last time the belt was, can develop some play.". In addition, during t-belt replacement some people "forget" to reinstall the timing belt guide found on manual trans engines (it's not used on auto trans ones), which if in place minimizes the chance of jumping at the crank sprocket. I would suggest fully checking the timing belt alignment. (Hey, it's another "no-cost" diagnostic procedure. ) When you and DaveT last did this, you ran out of time to get to the crank mark. I assume that means the left (driver's) side cover was pulled and the other loosened enough to check the cams, but you thought that it would be necessary to remove the lower cover (and therefore accessory belts, crank pulley) to see the crank sprocket. Well, there's another way to check belt alignment that doesn't require all that additional work, but not many are aware of it. It's "quick and dirty", and mostly good for checking "ballpark" t-belt alignment, but that's probably all you need to do for now. There are two sets of alignment marks on the cam and crank sprockets. The ones that are used for belt replacement are the "lines". Using them makes the job easier because it orients things so that the sprockets don't tend to "spring" out of place during the job. The other set of marks, the "arrows", are there as indicators of cylinder #1 TDC. We usually tell people to ignore the arrows, as they confuse t-belt replacement. However, they can be put to good use in this diagnosis. (The following is dependent on the crank pulley not being in terrible shape. You'll need to be able to spot the timing notch at the rearmost of the crank pulley rim. To help visualize where this is, imagine that it would have to be near the timing scale on the engine in order to check ignition timing with a timing light.) 1) Remove the left belt cover. 2) Remove screws as needed to loosen the other cover so that the right (passenger) side cam sprocket can be seen. 3) Using a wrench on the crank pulley bolt, turn it clockwise until the arrow alignment marks on the cam sprockets are pointing straight up (12 o'clock position), facing the notches in the rear belt cover. 4) Determine where the timing notch in the crank pulley is pointing. If all is well, the crank pulley notch will be pretty close to pointing at the "0" (zero) degree line on the timing scale. If the belt has jumped a notch at the crank, the pulley will point about 15 degrees from the "0". Note that the above doesn't allow full inspection of the timing belt components. With a manual trans engine, if the t-belt guide is in place, even a loose belt will sometimes stay in place. The looseness can cause timing errors, and noises can be generated that the knock sensor sometimes picks up, causing spark retardation. Since you and DaveT had covers off/loose, I assume you might have noticed if the belt was loose. I'm primarily mentioning this so that anyone reading won't incorrectly assume that not completely pulling the covers still allows for sufficient inspection. Being able to check the idlers and tensioner is important. Soapbox: Checking the t-belt won't necessarily result in finding a problem. Remember, every test result is valuable, if not as verification of a fault, then as a rule-out. Since we're trying to keep expenses down, rather than throwing parts at the problem, I'm concentrating on diagnostics. Dealer and independent shops sometimes don't spend lots of time on diagnosis (often justifiably, because labor costs can outweigh the expense of trying parts). If the labor is yours, depending on how you value your time versus your money, time-consuming diagnosis can be the way to go. Whew! If you think it took a long time to read the above, just imagine how long it took to write.

In the factory configuration, the switch is used to control a relay coil. As long as you don't exceed the switch's contact ratings, you could use it however you like.

No, brown would be Br. BY = Black/Yellow(stripe) W = White WY = White/Yellow(stripe) YL = Yellow/Blue(stripe) The colors in your pics agree with those in the diagrams I posted. If you need more info, you'll have to explain either what's wrong or what you're trying to do.

Is this sufficient?

I haven't seen it here. Thanks for posting it, John.

As close to the center of the intake manifold as possible. On the EJ engines, that usually means a connection just to the front of the throttle body. Be sure to use a "T" so that whatever was connected there still gets vacuum while the gauge is connected.

You're welcome. As to the wire looking "stretched", it might have gotten snagged at some point, cutting through the insulation and breaking some of the wire strands. The ones that were left would likely have eventually corroded.

Thanks, John.

Hmmm -- although the break is where I expected it to be based on the measurements, I wonder how it got damaged there. After all, with the harness running between the fender and the inner fender liner, its really not exposed. At least it's unlikely to happen again, and the repair looks solid.

Got that additional shot? Although it's probably not a common problem, you never know -- it just might help someone else.

As you know, if the system is running OL, it's supposed to ignore the A/F sensor. Since nothing much changed with it unplugged (and certainly not to default values that improved things), it would seem that the basic problem isn't being affected by the sensor. This doesn't mean that it's behaving like a new one, but since we have other things to investigate, let's (again ) put off replacing it for now. Sounds good to me on changing the neutral (position) switch. Those links do show that certain prevalent problems can be caused and/or addressed by ECU programming. However, I don't think a reflash should be considered now, specifically because the problem doesn't seem to be common on the 2000 (and you already had a "newer" ECU installed once). Yes, disconnecting the A/F sensor would cause the P1133 code to be set. I'm sympathetic to your situation, and trying to be conservative. We'll see what results you get with some further testing. I'm hopeful that eventually something obvious will jump out. As usual, you're welcome.

In the spirit of not spending money, how about some more diagnosis with tools you already have? We didn't address the questionable MAP readings. I wrote up a procedure that should give us some insight. Please make notes of readings as follows: 1) Begin with engine cold 2) Connect vacuum gauge to intake, using "T" 3) Connect 3130, route cable so scanner can be read alongside vacuum gauge 4) With key on, engine off, read live data "MAP" 5) Start engine, idle, get "MAP", "Spark Adv", and vacuum gauge readings 6) As engine warms, in one minute intervals, repeat those three readings 7) Once engine is warm, take those three readings again, but at idle, 1500 rpm, and 2500 rpm (nine readings for this step) Let us know what you get.

Just to make sure I'm reading this correctly, are you saying that disconnecting the A/F sensor resulted in no change in the way the engine ran? Based on the new readings, the neutral switch (it's not a "safety") isn't making reliable contact. Whether it's sufficiently bad to be causing the problems you're experiencing is uncertain. I'd replace it, since it's likely to only get worse with further use. The engine not responding to throttle when the clutch is disengaged is definitely interesting. I wonder what would happen if the clutch wasn't disengaged when coming to a stop. (Yes, I know the engine would stall if you did that when still in gear. ) However, what if at the point you'd normally declutch and apply the brake, you instead left the clutch engaged but shifted into neutral and braked? I realize that's not the "normal" way to drive a stick, but if you can do it a few times and see if the engine behaves differently, it might be revealing. By the way, if you have the time, the following is also "interesting": http://www.arfc.org/complaints/2010/subaru/outback/?q=&page=1 Search for the word "clutch" on this page and the following ones. You will find an amazing number of complaints sounding just like yours, but about *2010* Subarus. http://www.arfc.org/complaints/2000/subaru/outback/?q=&page=1 A similar search on the 2000 reveals *zero* such complaints. http://action.publicbroadcasting.net/cartalk/posts/list/2143255.page The "fix" for the 2010 model is to reflash the ECU.

Congrats. The closeup photo is great for seeing the broken wire, but doesn't make it easy to tell where on the harness it happened. How about informing us?