OB99W

Skip, take a look at page WI-148 (not -138, that's my typo). It'll show both front O2 sensors. (And why they put the 4-cyl info in the H6 supplement is beyond me, but...)

Well, that's why step #3 is a test procedure for checking the heater current draw. If the current is low, it's likely that the ECU is making the ground, but that there are other poor connections in the heater circuit. If it's zero, either the heater or wiring are completely open, or the ECU isn't making the ground. I believe that when the heater is on, it normally should draw about one-two amps. The "0.2 A" referred to in step #3 would indicate a fairly high resistance in the circuit (60 ohms or so including the heater), and even with a poor connection I'd expect "more than" that current. Of course, step #3 requires either the Select Monitor, or an OBD-II scanner that has the feature. If you could get to the main relay and ECU terminals, you could measure resistance from the main relay to the appropriate pin(s) for the front/right heater at the ECU, but that seems about as annoying as getting under the car.

Sorry, there are only nine steps -- you saw my post before I noticed my error. I did edit and correct before you posted, however (11:39 vs 11:45) -- does that count?

I don't see a problem, other than possibly having a bad connection due to corrosion at the sensor connector go into "hiding". Since you're pulling the connector, why not just start with #9, and check the heater resistance. If it's open, your troubleshooting is completed. Measuring voltage as in step #8 is best helpful if there is current being drawn. Any problem that's affecting voltage only to the front/right heater (and not the other two heaters) probably won't cause a noticable voltage drop on the voltmeter (most have a high input resistance and consequently draw little current). Still, it's certainly worth measuring.

Good point. Well, it was supposed to be for the H6!

That particular diagram, since it specifically addresses P0031 (front/right low), doesn't show the wiring for the rear sensor, whose heater also connects to the same power source. Info concerning P0051 (front/left low), for the other front sensor, uses the same diagram. The diagram for troubleshooting code P0037 (rear low) is different, but shows the rear sensor's heater connection to the same relay terminal as the front ones. Therefore, all three heaters are powered via the same relay contacts (at least on the '01 ).

If you have specific info or links, especially for the 2003, I'd appreciate if you'd post them. The info I have available is for the '01, so I'm not certain it directly applies to the '03. The '01 diagram indicates that the contacts on the main relay switch power to all three HO2S -- the two fronts, and the rear one. If the relay contacts are bad enough, the voltage should be low to all three heaters; P0037 and P0051 should also be thrown as well as P0031. I suppose it's possible the main relay just beginning to go bad, in conjunction with a slightly poor contact in the front right O2 sensor's heater circuit, might trigger only P0031. The ECU controls the ground end of all the HO2S circuits, so voltage sensing and heater current are ultimately determined by the condition of all connections, wiring, the heaters, and the ECU itself.

Responses below based on info for 2001 H6. Should be RH/Front. Perhaps, or a bad connection, or a problem with the ECU (it controls the ground end of the heater). Probably not fuse or relay; they feed all the O2 sensor heaters, so code for just one doesn't seem to fit. I'll send something for the 2001; see your PM.

All the diagrams I've seen show the FWD "switch" connected to an input on the TCU; that, of course, would mean that without the TCU, the switch/fuse doesn't do anything.

This previous post of mine may help: http://www.ultimatesubaru.org/forum/showpost.php?p=390602 The key cycling has to be done carefully; if you're sloppy, the pulses aren't seen properly. Stop after the tenth cycle with the key in the LOCK position, but don't remove the key. That's when the horn should sound. If that isn't clear, when I get a chance I'll post verbatim what it says in the owner's manual (pages 1-13 to 1-16). By the way, there's no mention of the brake pedal in the procedure.

The locking lug nuts obviously aren't as rugged as the solid steel ones; they certainly wouldn't tolerate significant over-tightening. What happened may be a sign that you were correct about the nuts having been over-torqued when the wheels were mounted.

You may be hearing a leak that closes as parts expand with increased temperature. A sufficiently long piece of vacuum hose, one end held to your ear and the other end moved to suspected areas, can help pinpoint this type of problem. Of course, a stethescope with just an open tube at the end would probably be easier and more comfortable to use, if you have or can get one. You also might find this of interest: http://endwrench.com/images/pdfs/LegacyIdleSum04.pdf

Even with an impact wrench that's set too high, if a torque stick was used, they may have been tightened to something near spec. (Note that I said "near"; use of torque sticks certainly doesn't guarantee accurate/consistent tightening.) The weight of the car shouldn't affect your ability to loosen and then properly tighten the lug nuts. The only issue could be wheel centering, and as long as the wheels are correct for the car (in the sense that the centerbore fits the hub correctly), that shouldn't be a problem.

As far as I know, the TCU is inactive in reverse; operation in that gear is determined by the manual valve position. Assuming the inhibiter switch is functional, the TCU should normally know what gear is selected; with the TCU disconnected, it should revert to whatever control the manual valve allows anyway.

Based on the usual smell, that might work best with oil that has already hit the exhaust system.

I though this problem sounded familiar, so I searched a little, and found this: http://www.ultimatesubaru.org/forum/showthread.php?t=81603 . Do we not think this is possibly related?

Not quite. Any time a ground short is there and power is applied, the lamp will be at or near full brilliance. However, when the short is removed, the test lamp and the normal load will then be in series; there will still be a path to ground through whatever is that load. Therefore, the test lamp's light output will be reduced, to whatever degree the voltage division between the test lamp and the normal lamp circuit determines. The test lamp would only go out completely if both the short were eliminated and the normal load were removed (in other words, no path to ground existed at all). The reason I've suggested a test lamp rather than just a meter is that the eye is sensitive enough to readily detect changes in brightness, but you have to pay closer attention to meter readings. Furthermore, lamps have non-linear resistance with respect to applied voltage, and that complicates interpreting the readings. If you find it useful, you could put a voltmeter in parallel with the test lamp; when the short is removed, both the lamp brilliance and the meter reading should drop. I don't have a wiring diagram for the '02, just the '01. Fuse #5 in the fuse/relay box is shown as 10 amp -- is that the value of the one you're referring to? There is a tail/illumination relay, at least on the '01. However, it appears to precede the parking switch in the circuit, so isn't a likely point of a short. There are a few ways to approach this. One is to start pulling connectors, and see if you can isolate the short. Another, if you suspect a shorted lamp, is to pull all of the ones on the circuit, one at a time; that includes the front clearance lamps, the rear finisher and tail lamps, and the license plate lamps. There's also the trailer connector to consider.

I didn't mean to suggest you were being argumentative, but rather that I didn't want you to take my own comments that way. Hopefully anyone with a cranking problem will be able to differentiate between when the solenoid doesn't pull in properly, and when it does but the starter still doesn't crank reliably, and complete an appropriate repair.

Skip, not to argue the point with you, but as I said there are two common failure modes. One is insufficient current to the solenoid coil; if due to a problem with the ignition switch, an added relay can resolve that. On the other hand, worn contacts in the solenoid of Nippon/Denso starters is a well-known and common problem. This is one of many Web pages devoted to replacement parts for the problem: http://www.nationsautoelectric.com/densoparts.html

Good point. Hopefully it would be easy to at least tell a DOHC from SOHC by just lifting the hood (err, bonnet ).

You're welcome, and good luck.

There are basically two separate issues. The first is whether enough current is available for the solenoid to pull in and hold. If there isn't, and it's due to a poor ignition switch, then adding a relay will help; even a marginal ignition switch contact will usually be adequate to energize the coil of a relay, and then it's the relay contacts that have to be good enough to deliver power to the solenoid coil. Once the solenoid is functioning correctly, the second concern is whether current for the starter motor itself can get from the battery to the motor. That's dependent on the copper washer at the end of the solenoid plunger bridging the "fixed" (non-moving) contacts with sufficiently low resistance, and the cable from the battery to the starter also having solid, low-resistance connections. Except for the rare possibility that a bit more solenoid current could temporarily overcome a marginal situation with the solenoid's starter contacts by slamming the plunger in a little harder, the two issues really are independent of each other.

Oh, and see if you can get any paperwork indicating when the timing belt was last changed. Otherwise, at 144k miles it could be a long time before it's needed, or overdue and a catastrophe waiting to happen.

I've never tried this, so I'm curious; what happens if you manually select first or second with the TCU unplugged?

Obviously, there's a short to ground of relatively low resistance. There's a "trick" I sometimes use when troubleshooting this kind of problem. Take one of the blown fuses, and wire a lamp of reasonable current draw (a clearance, stop, or turn-signal type, for example) across the fuse; use long enough leads so that you can place the lamp where you can see it anywhere around the car. Plug the fuse/lamp back into the holder, and turn on the lights. If a low-resistance short to ground exists, the lamp will light at very near full brilliance. You can now go around and try moving suspect wiring, remove suspect bulbs, etc., while watching the test lamp (out of the corner of your eye?) at the same time. If the test lamp gets dimmer (due to voltage division, if we need to get technical ), you've found the problem.