Sometimes runs way too rich - Computer or oxygen sensor?

Vehicle info: '92 Subaru Legacy L wagon 4WD non-turbo

Sometimes when I start my car, the exhaust is far too rich and detonates back in the catalytic converter. When it happens seems sporadic. The weather could be cold or hot, humid or dry, and the car may have been sitting for several days or only for 4 to 8 hours since the last time it ran. I can compensate by shifting manually to keep the RPMs way up. When the problem arises, I have to keep the RPM up to around 2500 when at a stoplight (so it doesn't peter out). When I take off, I'll put it in first gear and shift manually to keep the RPMs up high; otherwise, it will hesitate on acceleration and I hear the detonation in the exhaust system. Even after the temperature guage shows normal level, it is only after it has been at normal for about 15 minutes before the over-rich problem fades away. It only runs overly rich occasionally.

The other problem is that the "Check Engine" light comes on sometimes. I've had it in the shop 4 times to have them read the codes. The first 3 times, they got no error code. On the 4th time, they got no error code but when I started my car to leave then the "check engine" light came on and they hooked up while the car was running. I don't remember what was the code but it didn't make sense at the time and the mechanic said it was wrong. When the "check engine" light comes on is *not* when I have the first-startup overly rich mixture problem. They seem to be independent symptoms but maybe not. I suppose the "Check Engine" light might not be coming on but the computer is still screwing up the mixture when the car is started.

So my choices (that I can figure out) is to replace the engine computer and/or the oxygen sensor. I replaced my exhaust system about 4 to 5 years ago but they reused the original oxygen sensor (they were going to replace it but couldn't get one at that time). As I recall, this sensor check how much oxygen is in the exhaust and recirculates a portion of the exhaust if the oxygen is too low. Well, if it is running overly rich then wouldn't the oxygen level would be low?

Any opinions if it is more likely the computer or the oxygen sensor screwing up the mixture (so it is sometimes too rich on startup)? Might it be something else? Besides the 4 times to take it into the shop to see if they can nail the cause of the "check engine" light coming on (but never finding anything), I've even left my car at their shop over the weekend to let them start it cold to see if the over-rich problem crops up, but it never failed when they had my car.

Reply to
Vanguard
Loading thread data ...

Hi,

Seems odd the error codes weren't stored so you could do an easier job of troubleshooting!

I don't know about the '92, but my '90 doesn't throw an error code on a bad O2 sensor. Since it sounds like you're still on the original (did I read that right?), I'd certainly investigate installing a new one. Aftermarket units aren't that expensive (around $35 for mine vs $130 at the dealer) and seem to work well as long as you follow the instructions to keep overall wire length the same, etc. I don't know what the factory recommendation for changing that sensor is, but aftermarket units suggest changing as often as 15k miles. That's perhaps a bit often--maybe 30k would be reasonable? For reference, my last factory unit went south at around 75-80k miles.

If the O2 sensor's letting things run way too rich, you also run the risk of trashing your cat: they're a bit sensitive to continued over-rich conditions. The sensor's WAY cheaper than a new cat!

Other things to check: disconnect each sensor you can find, squirt the connections clean with some electronic component cleaner (better auto parts stores stock it), then put a little dielectric grease (same source) inside the connector before snapping it back together. Sometimes just a little dirt and/or corrosion can cause some screwy voltage signals to the ECU and it gets confused. There was a Subaru TSB out some years back concerning this.

WARNING: DON'T use any grease, sealer, etc. on any connections involving the O2 sensor. I've read where some of them rely on "hollow" insulation on the wire as a "vent path" for the wire, and clogging that will warm things up, also sending bad signals to the ECU. Don't know if Subie uses that style or not.

Is there any other "clue" that might help you narrow things down, like recently changing the air filter to a new brand or anything? Plugs, wires, cap, rotor are all good? I'm assuming you've already been down that path, but sometimes an otherwise "obvious" answer is overlooked.

Good luck!

Rick

Reply to
Rick Courtright

Only 15K (or even 30K) on an oxygen sensor? Geez, that's low. I was looking around and the price I got for a replacement O2 sensor was $82 (plus shipping) for a Bosch unit. Not real bad (but more than you paid).

Yep, the O2 sensor that is in there is the same one that has been in there since 1992 (i.e., it is the original one). I had the exhaust system replaced about 5 years ago. They were going to replace the O2 sensor but couldn't get one on order at the time (i.e., it would have to be backordered). So they put back in the old one and I've never went back to get it replaced. Geez, their maintenance schedule

formatting link
says I don't need to replace spark plugs until 60K intervals, and the O2 sensor is half of that? My recollection when I read the error codes (using the flashing lights method) was that once in 6 times tried that the code pointed at the O2 sensor. As far as other parts, I haven't looked or changed out the spark plugs, spark cables, rotor, or any of that type of stuff. When the problem arises, which is only occassionaly, and once the car gets hot (and has been hot for about 15 to 20 minutes), then the over-rich problem goes away. It won't reappear if I stop the car, like at the store, and then start again. Only if the car has sat for many hours might the problem reappear. I doubt spark plugs & cables, rotors, etc., are temperature sensitive. If, for example, the spark cable had degenerated over time, its resistance would be too high regardless of temperature fluctuation. While the oil inside the coil would heat up, I don't see how the coil's performance would alter depending on the temperature of its oil use for cooling. I'll probably be replacing the plugs, cables, rotor, and normal stuff for tuning and maintenance but I'd like to solve this over-rich problem first (big stuff first, little stuff last).

I can try the dry-and-grease method you mention regarding the electrical connectors for the sensors. In fact, I should still have some dielectric grease around from last time (I use it to avoid corrosion of the contacts within a connector that I take apart and put back together if it is outside the passenger compartment). I'll have to get some more contact cleaner spray, though.

I suspect that getting rid of the problem will end up being a piecemeal guess method: keep replacing possible components involved in the fuel injection system until I happen to replace the culprit.

Reply to
Vanguard

Yeah, seems low, but we've gotta remember the vendors want to sell sensors! Since US emissions requirements usually spec a minimum 50k miles for emissions related equipment warranties, perhaps 60k w/ new plugs and such sounds better? As I said, I don't recall seeing a replacement interval in the books I've checked...

Combining age and the fact your error codes tended to point toward that sensor, I'd probably put it in No. 1 position as a potential culprit!

The other stuff I mentioned (plugs, cap, etc.) shouldn't be too temperature sensitive, as you suggested. The backfiring you mentioned was what triggered that thought: spent a day once being totally baffled by a VW engine that would backfire like your Subie. It was rather intermittent... a better mechanic than I am found a carbon track in the distributor cap. It was a black cap (as opposed to the stock "orange" Bosch units) which made the track almost invisible.

Also, you might want to try resetting your ECU by disconnecting your battery for 10-30 minutes. That will clear any error codes, put the ECU into its default settings, and MAY help find other problems easier?

Best of luck!

Rick

Reply to
Rick Courtright

Did you check the engine temperature sensor? If this gives a bad signal, the mixture will be lean or rich, depending on the reading it's giving to the engine computer. If it is not off its specs, it won't cause an error code, but could still be giving the computer a cold (or cool) engine temp when it is actually hot and not needing the rich mix. Then the O2 sensor would read a rich mix, and when the computer can't correct the mix, it gives the error code for bad O2 sensor. If you have the manual, you can check the resistance the temp sensor is supposed to have at various engine temperatures to see if they are within spec. Ed B

Reply to
ed

The info above suggests yrou shop mechanic is either seriously inept at diagnosing engine malfunction, or milking you shamelessly.

Reply to
CompUser

If this happens right at startup it isn't the 02 sensor. When the engine is first started cold it runs in "open loop" mode and the 02 sensor is not used until the engine is at operating temperature, then it goes into "closed loop" where the 02 is used for regulating the A/F mixture. A more likely culprit, and a rather common one, is a bad coolant temperature sensor. These tend to fail over time as their resistance value shifts so the ECU will receive faulty temperature information, this can result in an improper mixture and symptoms like you have reported. A bad coolant temp sensor may or may not set a code, depending on whether it still produces readings that the ECU considers "valid" even though they are actually out of range. This is an inexpensive part and you can replace it yourself.

Reply to
mulder

JUst to be clear, is this the sensor with 2 wires that reports engine temp to the ECU? Not the one with 1 wire for the coolant temp gauge right?

Carl

snipped-for-privacy@x.files wrote:

Reply to
Carl 1 Lucky Texan

Reply to
mulder

Hi, Carl

On my '90 it's a two wire sensor with a green connector. It's mounted horizontally below the t-stat housing pointing to the right side (looking forward) of the engine. ISTR one of the books showing a slightly different location, but can't swear to that.

I don't know if I was lucky or not: when mine failed it just died, threw a code and triggered the CEL. It MAY have already been failing slowly as Mulder describes, but I'd just gotten the car and hadn't learned its little quirks. Failure mode is "full rich" so I noticed an immediate improvement in fuel economy (2 mpg or so?) when I installed the new one.

Rick

Reply to
Rick Courtright

Failure mode can go either way, depending on whether the sensor shorts or opens, or in less extreme cases whether the resistance shifts up or down.

Reply to
mulder

Hi,

I'm assuming we're defining things the same way: shorted is zero resistance across the terminals, open is infinity (or at least WAY high?) As I mentioned, it's been a few years but IIRC mine shorted--I know it failed "full rich."

Rick

Reply to
Rick Courtright

The manual I have doesn't detail specs, like resistance. I forget the publisher but it's like one of those Chilton manuals. I tried hitting the local library because they have the AllDataDIY automative database that is supposed to be very thorough, but all the kids had the computers tied up at noon, so I'll have to try again this week (and ask the librarian to bump any kid that didn't actually sign-in to use a computer). Hopefully it will show the location for the temperature sensor and what resistances I should see at various temperatures.

Reply to
Vanguard

Actually neither my favorite car shop or the dealer charged me when they couldn't see any error codes. The only cost incurred so far is the wasted time in having them try to get codes out of the ECU. I guess depends on your location and who you deal with as to whether or not they want to nickel and dime you to death. My recollection is that I had the car shop test the ECU three times but each time the "check engine" light did not come on for them and they got no error codes. I then decided to try the dealer figuring that maybe the car shop wasn't doing it correctly, but the dealer didn't get any codes, either. Between each attempt, I had cleared the ECU and then waited until the "check engine" light came on again. Each time, I was prepared to pay something for their trouble but they never charged me.

Because the ECU is turning on the "check engine" light but reports no codes is why it also might be screwing up the mixture. It already appears something is defective with the ECU. That doesn't preclude another component also involved in the problem but the ECU is already acting weird. If it is screwing up one thing, it could also be screwing up something else.

Reply to
Vanguard

Yep, it only occurs when the engine is cold, like when I start it up in the morning to go to work or it has been sitting over the weekend unused. It has, but less often, also occurred after driving the car to work but then sat in the parking lot for 8 hours. At one time, I thought it was more prevalent when it was cool and damp but that must've been when I first started noticing the problem.

When I first start the car, it is way too rich and want to die at idle, so I have to rev it up to around 2500 RPM. Although it is an automatic, I have to start out in first gear and manually shift up as I takeoff from a stop in order to keep the revs high. It takes around 15 minutes before the problem goes away. The temperature guage has reached the normal mid-point position but it won't be until about 10 minutes after that until the problem goes away (i.e., temperature must be normal hot but be normal hot for awhile).

I didn't get to recheck the ECU for codes this weekend. Got too busy. This week I'll be checking the AllDataDIY automative database at the library to see if it shows where is the temperature sensors (so I can differentiate between the guage sensor and the engine sensor) and what resistances it should have.

It doesn't help to take the car to the dealer for an appointment because the engine is hot and has been hot long enough for the over-rich problem to disappear. I've left my car on their lot over the weekend and told them that they need to check it right on startup for the problem but it did not occur when they tested (it is sporadic as to when it will show up).

I would think that if a sensor were out-of-range, high or low but within range, or open, short, or broke, that the problem would be consistent (i.e., it would be over-rich everytime I started the engine when it was cold).

Reply to
Vanguard

Ya know - this does seem a little backwards doesn't it? I mean, if the sensor thought the engine were cold all the time - wouldn't it start easily but get poor mileage as if the 'choke' were still on? And if the sensor indicated to the ECU that the engine were aleays warm - you'd get 'lean' on start up(maybe a difficult start in cold weather) THEN near normal A/F ration for warmed-up driving.

I dunno, maybe one of the O2 sensors instead. Does it actually seem over rich when you start - smell gasoline, any black smoke?

Carl

Vanguard wrote:

Reply to
Carl 1 Lucky Texan

There is no black smoke. There is a light blue tinge on cold start (when the over-rich problem occurs) and you can smell gasoline. I have started when cold when the problem appeared, drove 3 miles to a restaurant, and restarted it about 40 minutes later and the problem was still there. Only after the car has gotten hot (normal range) AND been hot for 10 minutes, or more, does the over-rich problem go away (when it happens). It's so sporadic that I haven't found any consistent condition under which I can reliably recreate the problem. It's like shooting at a thin flock of birds flying around in haphazard paths in the air by using a bow and arrow. So far, I haven't manage to get a hit when anyone is trying to read the codes. Argh! Although I'll be testing for the codes again this week, I surely won't be surprised if it says there aren't any error codes. However, the check engine light does come on now more often then it did before, so the flock that I'm shooting that arrows at is getting denser.

That's why I suspect the only way to cure the problem is to start replacing parts in piecemeal fashion. However, I won't know if the problem got solved until after sometime later after seeing the check engine light doesn't come on anymore and the over-rich problem goes away. I'm just not sure which order I should follow in replacing parts, from most likely to least likely (with the most expensive part last but not necessarily in price order before that). PCV, oxygen sensor, engine temperature sensor, knock sensor, purge valve (EGR), and ECU (engine computer) seems about the right order. However, the ECU comes in LHD and RHD versions, and an in-state salvage yard lists a RHD unit for only $30 (plus shipping). There might be mounting problems (mine is a LHD unit) but a perforated strapping iron strip could be used to affix the box to the body. They want $100 for the LHD unit, like mine.

I read the AllDataDIY automotive database regarding the O2 sensor. After all the hype that I heard for AllDataDIY, I was disappointed that it didn't describe remove/install procedures for many parts and the pictures were difficult to read drawings instead of actual pictures. Of course, they refer to a Subaru tool part number for removing the O2 sensor. Does it really take a special tool, or will a deep-well socket or box wrench work? They really didn't describe the location very well and the picture is poor (very few references to the objects in the picture). It looks to be atop either the exhaust pipe before the front catalytic converter or maybe even atop the front catalytic converter itself. Hopefully I don't need a special tool since the cost of which would obviate any savings from me buying the part and replacing it myself compared to having the shop do it.

Reply to
Vanguard

Hi,

Problem w/ using regular wrenches/sockets on O2 sensors is the sensor's often buried in sheet metal so you can't get to the flats easily. And I think you may find it easier to disconnect the Y-pipe at the heads and downstream so you can drop it a bit to get at the location with your wrench. Many auto parts sell a socket to do the job--it's a deep set socket with a slot up the side so you can move the wire out of the way while installing/removing. If your parts shop handles K-D tools, they make one. Shouldn't be too expensive... seems it was in the $10-$14 range when I looked after fighting and finally getting the sensor in w/ regular wrenches.

Rick

Reply to
Rick Courtright

MotorsForum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.