OB99W

Here are a couple of links that might help with diagnosis: http://endwrench.com/images/pdfs/KnockingInfo.pdf http://endwrench.com/images/pdfs/Valve07MayEW.pdf

See this article for some additional ideas: http://www.catalyticconverter.org/news/news_page.cfm?Key=catalytic_converter-&News=120

Littlecars, I didn't see any response from you to what Log1call said about the injector wiring. You did say that you only lifted the intake, so perhaps you never disconnected the injectors. It's easy to accidentally reverse the connections to #2 and #4, but it's also easy to check. White tape on the wires at the connector usually marks #2, while #4 is black. Otherwise, I agree with those who have suggested that vacuum and compression testing can help pinpoint the problem without tearing everything down. What mdjdc mentioned about the intake gasket also merits checking.

Yes, although if you replaced the radiator, you'll have to transfer the speed-nuts from the old one for some of the positions.

Nipper makes a good point about the standard, especially if you need to be current. Another consideration is whether it makes sense to buy an inexpensive code reader as opposed to a more costly scan tool. I assume that anyone buying an OBD-II reader is doing so because they want info to help with a diagnosis. Well, while the trouble code is useful info, it's only part of the story. Being able to retrieve real-time data can be very helpful, and often even more important is ''freeze frame'' data -- it's a record of certain parameters that's stored in the ECU at the instant the trouble code is set. Having that additional data can make some trouble codes a lot less ambiguous. Consider spending somewhat more if those features are important to you.

Mike, are you suggesting that anyone here would be so cruel as to rub your nose in it? Although pulley failure can sometimes be relatively sudden, often there's some warning. My suggestion would be to not ignore any ''new'' engine noises or symptoms in the hope that they'd go away on their own.

By the way, if the engine was run enough while excessively rich, the oil might be fuel-contaminated. If it smells that way, or if the oil level seems higher than it should be, it could be prudent to change it.

Wonderful, a happy ending! Thanks for the recognition. Rob/rvac99, you also deserve credit for sticking with the problem until the resolution. It's sometimes hard for us who aren't able to use our own eyes, hands, etc., to be precise in diagnosing a problem, so it's gratifying when the person who's actually doing the job can give us enough feedback to work with. Congrats!

Progress, apparently! I'm glad to read that you're ''firing on all fours'' now. The unwanted conduction may have weakened sensor pulses, cross-coupled them, or some combination. From my experience with salt getting into the connectors, the best results I've gotten is from first cleaning with lots of water, then using alcohol as a water displacing and drying agent. Finally, drying thoroughly. Since B21 needed cleaning, B22 might also. You had mentioned adding some ground wiring. Depending on where it's located, sometimes that can cause ground loops that actually make sensor signals noisy. I'd remove that wiring and see what happens -- if things don't get worse, leave it off. Be sure that the body harness sensor shield (''sheath'') is back in place. It's there to prevent stray signals from getting to the ECU. When the ECU has been reset, it's usually a good idea to start the engine and let it idle for about 10 minutes (with the hood closed so things warm and dry) without touching the gas pedal. If none of the above helps, try that and let us know if the cam codes return.

The voltage output from the rear O2 sensor can easily be monitored by the dealer with a Select Monitor, or someone else can look at the real-time data with an OBD-II scan tool. That voltage, and whether it's steady or varying (and to what degree) can make determining where to next look a lot more obvious.

The weather here has been variable as well, and gambling with lightning's megavolts can have serious health consequences. If you can be patient, tenacious and hopeful, so can I.

That's sounding promising. Conduction due to an electrolytic solution will change with the voltage applied, so a modern ohmmeter (which typically applies a rather low voltage) might not reveal the full impact of the contamination. I sometimes use a hair dryer to accelerate the alcohol evaporation. (Heat guns are faster, but if used too ''ambitiously'' can result in melted insulation. )

It could be what I mentioned in the last paragraph of post #38. (Gee, this thread is getting ''a bit'' long. ) It will often appear to be greenish in color due to reacting with the metal pins of the connector.

I suspect that you're confusing milliohms (thousandths of ohms) with Megohms (millions of ohms). Even lab grade equipment doesn't readily measure resistance in the milliohm range, and for the purpose of automotive circuit testing it's rare that a measurement below 0.1 ohm (100 milliohms) would be necessary. When dealing with low-voltage/low-impedance circuits, leakage resistance readings over 100 Kohms (0.1 Megohms) usually isn't problematic. Some electrical connectors intentionally have a small amount of conductive material added to the plastic in order to help dissipate static charges, and in others contamination will sometimes cause slight conduction. The low readings on B21 #5 and #10 are explained by the fact that they directly connect to the oil pressure lamp and temp gauge respectively. If we're concerned with the cam sensor wiring, concentrating on B21 #18 and #20 is probably going to be more productive. If you want to check for possible shorts within the shield, B21 #17 and #19 (for the crank sensor) are within that shield along with the cam sensor wiring. Therefore, check the resistance from B21 #17, #18, #19 & #20 to each other (and #19 to ground).

While trying to find a short in the body harness, it should be disconnected at both the ECU (B135) end and the engine (B21) end. If those readings were obtained with the harness disconnected in that manner, then something very strange is going on. On the other hand, if the readings were obtained with the harness attached, that will cloud the issue. The shield (''braided sheath'') covers much, but not all of the length of the cam sensor wiring, and has the crank sensor wires within it as well. However, there should not be a ground reference once the body harness is disconnected at the ECU and engine even if those wires shorted to each other within the shield. Okay, but if you haven't been checking for shorts with the body harness disconnected as above, please do so. It does, however, seem that you might eventually have to peel back the body harness covering and do a physical inspection.

Welcome to the forum. As you noted, the P0136 code refers to an O2 sensor circuit problem. Therefore, besides the sensor, the wiring could be the culprit. Based on what you said, it's possible that the wiring was damaged or a connection was disturbed during the work done by the dealer.

Good, you're closing in on something. With those connectors pulled it obviously isolates the cam sensor wiring in both the engine and body harnesses. Based on what you said, the red wire which connects B135 #2 to B21 #18 is partially shorted to ground somewhere within the body harness. Since that wire is shielded for much of its length, and yet B135 #10 (the shield connection) isn't showing conduction to the other wires, in theory that shouldn't leave much of the length vulnerable. I'd proceed by making a reliable connection from your ohmmeter to the wire with the short and ground. Then try flexing the body harness from B21 to the firewall, and also from the ECU end -- I've found that sometimes it requires pushing, pulling or twisting the wiring to see a change in meter reading that could lead to finding the short. You could also try disconnecting B22 from E3, since several ground wires run through there, and see if the short is affected. By the way, I understand that the measurement is varying, but what range of resistance readings to ground are you seeing?

It seems that you're on a path to finding the problem. However, you have me a bit confused, both as to where you're seeing conduction, and the connector you're referring to. As I mentioned in post #9 of this thread, ''E2/B21(gray) are a 20-pin connector pair where pins #18 & #20 are used for the cam sensor''. E2(Engine) mates with B21(Body). B21#18 corresponds to B135#2 (at ECU), and B21#20 to B135#9. - Where you refer to B22, did you mean B21? - I'm assuming you're disconnecting B21 from E2 (as well as disconnecting the ECU). - Are you saying that there is conduction (non-infinite resistance) between B135#18 and B135 #20, or that the conduction exists from those pins to ground? It might be important to note that there are two separately connected shields (''sheaths'') for the cam sensor wiring. As you know, the wires for the cam sensor have a shield covering them in the body harness from the ECU to B21, and the shield's connection to ground is made via the ECU (so that when B135 is disconnected, the shield is no longer grounded). In addition, the cam sensor wires in the engine harness from the sensor to E2 (mates to B21) are also shielded, but that shield is directly connected to an engine ground (so pulling the connectors we've been discussing still leaves that shield grounded). Since I'm unsure exactly how to interpret your findings, the following might not apply. Sometimes salt from road deicing finds its way into connectors that are exposed, causing unintentional conduction between pins. That conduction can vary with humidity, etc. The engine connectors (E2/B21 in particular) could be experiencing the problem. I've dealt with such a problem by decoupling the connectors, spraying them thoroughly first with water and then with alcohol to displace the water, and allowing the alcohol to dry before reconnecting.

That sounds like you nailed it. Congrats, again.

This thing is apparently bugging me . Looking at both the cam and crank sensor pulses might be revealing. The reason I'm mentioning the crank is because it's the reference signal, and the cam trouble code(s) can be generated when there's a timing discrepancy between the crank and cam pulses. Since there's been a lull in this thread, I thought I'd post some further conjecture on what might be the cause of the problem. The crank sprocket has 6 reluctors in two groups of three -- each of the ones in a group has a corresponding one 180 degrees around from it. Each group of three corresponds to two cylinders, one group to #1 & #2 and the other to #3 & #4. If there was a sufficiently weak pulse from one of the groups of three, it could affect just #3 & #4. It might then cause the ECU to see a timing problem between the crank and cam pulses, triggering the P0340. A weak pulse can be due to a damaged reluctor tooth. It could also be that the crank sensor's output is slightly low in general, and normal variation in output from tooth to tooth might bring one below the triggering threshold. Just a thought, might not be worth 2 cents.

Besides the things that Cougar mentioned, the main relay powers a few other circuits. If you don't find a short at one of the O2 sensors (could be due to melted wire insulation or internal short), other connections are: - MAF, yellow wire - Fuel injectors, yellow/blue wires - The sneaky one -- under the dash, often hanging down where it can get stepped on and short to something, is the diagnostic connector for the Subaru Select Monitor. It's a 9-pin yellow one, and the yellow/red wire going to pin 5 is the supply lead.

Okay, I'll spell it out. Subaru 1996 ECUs reset their monitor status to ''not ready'' whenever the engine is turned off. The EPA and state inspection agencies know about this and exempt the '96 models (see the links in the post I linked to above). The ECU in an early-production '97 model could have the same programming as a '96, and therefore have the ''not ready'' problem. If you have an early '97 and can convince whomever is doing the inspection of this, then the car should be exempted. If not, then contact Subaru of America, give them the VIN, and they should be able to provide a letter of explanation that will do the trick. (Either that, or tell Subaru that they owe you a newer ECU .)

I wonder if you have an early production '97, or if someone replaced your ECU with a '96. See: http://www.ultimatesubaru.org/forum/showthread.php?t=99635

You're welcome, thanks for getting back to us, and congrats on resolving the problem.

Since a new/rebuilt ECU can be expensive, and because (as Cougar said) ECUs don't typically have this sort of failure, I've been trying to suggest alternative possibilities that fit the symptoms. Although apparently under the circumstances it isn't practical, if certain test gear was available, some further checking might reveal something that hasn't been obvious, allowing us to be more targeted. For example, even though it's been replaced, I'd certainly feel more comfortable if we could verify that the cam sensor was outputting proper pulses. Perhaps a used ECU can be found at a reasonable price.