OB99W

Good, that small fraction of an ohm is the resistance I'd expect if the connections are decent. There's no way to be sure, but unless the problem returns, I'd leave the starter alone.

You're welcome -- I hope any info I can offer will help. I had been concentrating on rereading about the electrical system problems in your previous threads, and had ignored this one: http://www.ultimatesubaru.org/forum/showthread.php?t=91349 . However, it might be a clue. Is there any possibility that this car had been in a flood? If the reason the carpet and seats smelled bad was due to mold/mildew, etc., it probably was wet at some time. Flooded cars usually have a lot of electrical problems; while they can be fixed, it's often very time consuming (and can be expensive) to do that. See: http://trucks.about.com/od/autobuying/tp/flooded_cars.htm Free CARFAX flood check by VIN: http://www.carfax.com/cfm/general_check.cfm?partner=PRL_A

The fact that the readings were changing like that indicates really poor connections. Voltage drop from a battery terminal to the end of the cable connected to it (or in the case of the negative, to the starter body) should only be a few tenths of a volt or so, with the starter cranking. I just looked back at your recent posts, and given all the problems related to the electrical system, I suspect there's a common "connection". If you haven't already done so, remove both battery cables at the battery terminals (posts) and clean the mating surfaces with a wire brush. Make sure when you put them back that the connection is tight. Do the same at the body (chassis) and engine ends of the ground cabling. Then check for voltage drops along the cables again. You could try moving the cables while measuring (which might require three hands ) to see if the readings change, which would indicate a problem. By the way, based on the readings you've posted, it seems you're using a digital meter. They're usually good for getting accurate readings, but make it difficult to see trends. If you can borrow an analog meter, it might be easier to determine what's going on. Tracking down intermittent problems can be a pain, but it should be possible to pin this one down.

Close, but just take those voltage readings with the key in the "start" position (engine cranking, or at least would be if things were working correctly) -- no need for "key off" measurements. The readings should be low, by the way, if all's okay. Of course you'll need another person to help, unless you have a way to safely attach the meter leads while you turn the ignition key. (I know some of you are thinking "use a remote starter", but I'd rather leave that out of this.)

Hey, it's okay -- I don't think anyone here can legitimately claim to be all-knowing.Besides, banging your head against a wall usually either gives you a headache and/or dents the wall.

It can be done, but it's a lot easier to keep the contacts in position if a press is used. I've used blocks of wood and a large C-clamp, although that's not the easiest or best method.

The "Main power terminal" reading is somewhat low -- a fully charged battery would usually read closer to 12.6v if all the connections are good. That reading is way too high, unless you left out a decimal point, etc. The resistance should be a fraction of an ohm. First, just short the two meter leads together (in the "Ohms" position, naturally), and make sure the reading is close to zero. If not, either zero the ohmmeter if it has that function, or just make note of the reading and subtract that from any resistance measurement you make. By the way, small voltage drops in cables that are carrying current can cause false resistance readings. (Resistance measurements are supposed to be made with all power disconnected.) If the positive battery terminal was connected while you checked the ground circuit resistance, the reading might have been misleading. If the ground circuit really measures 4 ohms, you have a ground cabling problem. Check the connections for corrosion, and if that's not the problem, the cable terminals might not be making solid connection to the cable itself (corrosion sometimes develops inside, where it's not visible). If you get a reading over a fraction of an ohm between the terminals of a battery/ground cable (end to end), it's bad. Sorry, but no, it doesn't have to be the contacts. Please check what I mentioned about the cabling before going back to the starter. If you can get someone to work with you, measure voltage drops while the key is held in the "start" postion -- from positive battery terminal to starter main one, and from battery negative terminal to starter body. Let us know what those readings are.

I didn't say you'd find the pics that way. However, you should have been able to find "the URL from which the pics normally would be available", as I put it (if they hadn't been taken down). For example, you'd know that they weren't being hosted by Flickr, but rather here at USMB.

With most Windows browsers at least, right-click on the red "X" or whatever replaces the pic, then click on "Properties", and among other info you should be able to see the URL from which the pics normally would be available. In this case it's "ultimatesubaru.org", so of course they'd usually be viewable. Unfortunately, I suspect they've been removed and replaced with the ones in http://www.ultimatesubaru.org/forum/showthread.php?t=91728 .

Good luck with the experiment, but just in case I didn't make my point, conducting heat into the sensor probably isn't advantageous. It is critical that the mounting area be clean and the bolt properly torqued, and the rotational angle of the wire lead can be important. I have some thoughts on knock sensor trouble. Of course, sometimes the problem can be due to wiring/connection issues (anywhere from the sensor to the ECU), and not failure of the sensor itself. If the sensor is actually bad, you can often see cracks in the encapsulating material -- it may be that besides heat, ozone might play a part in that deterioration. Ozone is generated whenever there's a sufficiently high voltage discharge in air. While we'd all like to think our ignition systems have perfect connections and insulation, that's not very likely, so ozone is typically generated in at least small amounts. The actual knock sensor is a piezoelectric device; cracking of the covering shouldn't directly cause the sensor to fail. However, cracks would expose the piezo element and internal connections to it to the potentially damaging kind of atmosphere found under a car's hood. So, make sure it's not a connection or mounting problem before condemning the sensor. If you're getting frequent failures due to cracking, perhaps checking for ignition voltage leakage (tight connections at coil and plugs, no spark "leakage", etc.) might be worthwhile.

Since I've never used the compound on a knock sensor, I don't have any hard data. However, I don't think your experience with the ignition modules can be applied to the sensors. The ignition modules generate heat, and dissipating it more readily should help extend their lives. However, the knock sensors are passive devices, and heat is likely their enemy. I suspect that using the compound wouldn't lengthen knock sensor life, although it probably wouldn't significantly shorten it either. I'm going to give that a bit of thought, and post later.

How could I argue with that logic? (Sorry, screwbaru2, I thought you were agreeing with the "exhaust bubbling" theory. )

I guess it wouldn't hurt to look... ... but knock sensors are designed to "hear" the "ringing" of the block caused by the sharp rap of knock. I'm not sure that exhaust gas bubbles in the coolant would have a similar effect.

It seems that disconnecting power to the ECU for too short a time can "confuse" it. Although it doesn't address the same model year as yours, see http://endwrench.com/images/pdfs/RoughIdle.pdf -- the problem appears to afflict more than just those mentioned in the article. I suspect that the "learned" parameter values get corrupted, and a longer disconnect time allows the ECU to truly reset to default values.

While you're at it, you should probably lubricate the caliper slides. That's especially true if you notice any significant difference in wear between the inner and outer pads.

The knock sensor "hearing the exhaust bubbling into the coolant" theoretically shouldn't be sufficient to set P0325, based on Subaru's apparent usage of that code. Anything that could be interpreted by the ECU as knock should just cause it to retard engine timing. Loss of coolant at a cylinder due to HG leak might have lead to localized elevated temperature, resulting in "real" knock that could be detected. However, that still should just have caused the ECU to retard timing, not set P0325. I wonder if heat distorted the piezoelectric element and changed its characteristics enough to cause the code. OBDII P0xxx code definitions are supposed to be generic, but that doesn't mean a manufacturer won't "bend" things a bit. Perhaps Subaru isn't revealing the entire circumstances under which P0325 is set. For example, does the ECU "think" the knock sensor or wiring is defective if it already has retarded the timing the most it can, but is still getting indication from the sensor of what appears to be knock? Generic knock sensor codes (for a single sensor) are: P0325 Knock Sensor 1 Circuit Malfunction (Bank 1 or Single Sensor) P0326 Knock Sensor 1 Circuit Range/Performance (Bank 1 or Single Sensor) P0327 Knock Sensor 1 Circuit Low Input (Bank 1 or Single Sensor) P0328 Knock Sensor 1 Circuit High Input (Bank 1 or Single Sensor) It might be that conditions that more appropriately should set P0326 were programmed in the ECU for P0325. Conjecture aside, since the knock sensors have a history of failing, I'll continue to look there before assuming other causes of P0325 ("when you hear hoofbeats, think horses, not zebras").

Since most Denso reduction drive starters are fairly similar, the following link should help anyone replacing contacts in the starter solenoid: http://www.expeditionportal.com/forum/showthread.php?t=12746 .Note the use of a press to hold the contacts flat while the nuts are tightened.

The code definition you have for P1121 may be misleading you. Codes P1xxx are brand-specific -- it appears that "TPS A RAN" isn't the Subaru code definition. For Subaru, P1121 is "Neutral Position Switch Circuit [High Input, MT Vehicles] [Low Input, AT Vehicles]". It seems your transmission neutral/inhibitor switch isn't making good contact (or connectors/wiring to it are the cause), and therefore the ECU doesn't know whether the car is in gear or not. Check that switch, etc., and you'll likely find the solution.

The definition for code P0106 is "Pressure sensor circuit range/performance problem". It's one of the many codes that doesn't point to a specific defective part. The barometric pressure sensor itself might be bad, but there could be problems with electrical connections to it or the pressure sources switching solenoid valve (however, given the conditions under which the code comes up, it's probably not electrically-related). Also, there could be clogged or cracked vacuum hoses connecting the same parts, or the inline filter might be clogged.

The definition for code P0325 is "Knock Sensor Circuit Malfunction". It's set when the ECU checks the knock sensor circuit and determines there's an electrical problem with the wiring, connection, or the sensor itself. It doesn't mean that knocking was detected. Engine knock of the type that the sensor is designed to detect doesn't tend to occur at idle, so what you're hearing is likely totally unrelated. Assuming the sensor is connected correctly and its mounting bolt is properly tightened, it's likely that the trouble code is being triggered by a defective knock sensor (they typically develop cracks).

Since "ignition key warmer" isn't part of the "winter package" that I'm aware of, I wonder if excessive contact resistance or current draw through the ignition switch might be the cause of the heating. Or is the lock just hot from the car sitting in the sun, and transfers heat to the key? (The steering wheel is often too hot to comfortably hold after being in the sun.) Does the key get hot during nighttime driving?

Measuring the circumference of each tire isn't a bad idea -- but if there's binding and it's bad enough, even a significant difference of inflation pressure in otherwise similar tires can cause problems. If you have alloy wheels, especially ones that have seen much road salt, they tend to get corroded and not hold air for very long. By the way, 16 flashes of the ATF temperature warning lamp only indicates an electrically-related transmission problem, but not specifically what it is. Have you read the trans code(s) yet, or are you assuming that duty solenoid C is the problem?

As long as in addition to dealing with the loose screws the mechanic properly resealed the oil pump, that should be okay. Having positive crankcase pressure means that either your engine has excessive blowby (combustion gases getting past the piston rings), or the PCV isn't functioning correctly yet (perhaps it's still somewhat "gunked up").

I'll give you some general info that might lead to a solution of the problem. Obviously, a single stroke of the master cylinder should move the caliper pistons (or wheel cylinders, if drum brakes) so that the friction material makes contact with the rotor (or drum). The master cylinder has to match the requirements at the wheels, in terms of fluid volume delivery (and pressure, but let's forget that for this discussion). If you change the calipers, you have to make sure that the area of the two pistons doesn't materially exceed that of the single one in the calipers being replaced. Otherwise, it's possible that not enough fluid will be delivered by a single depression of the brake pedal to sufficiently move pistons at the wheels, requiring pumping to compensate. If you find that the pedal is hard enough after pumping, the master cylinder is probably not defective, especially if you can hold pressure on the pedal for a while and it doesn't sink. However, if the "new" brakes have greater fluid volume requirement, the master you're using might not be adequate. Since I don't know the specifics of your swap, Gary, I don't know whether that applies to your circumstances or not.

That should be the location, and it's typically square and brown in color. You should probably still try tapping on it to see if running conditions change, but I'm less inclined to consider it given the additional info you've since provided. Often if the engine runs better without the MAF connected than with it, the MAF itself has problems. As McDave suggested, if another MAF is available, it could be worth trying. Then again, both MAFs may be good and the problem due to poor connections or otherwise. You had previously mentioned that the car ran "fine" for about 10 miles with the used MAF, but until things are warmed and the ECU isn't running open-loop, that period isn't very telling. As has been already mentioned, if the front O2 sensor is disconnected, that will also cause the ECU to go to default settings, since it will look like the sensor hasn't warmed up. Mike, please check your Private Mail.