Is My O2 Sensor Working?

[Guest]

My 91 Legacy GT doesn't quite run properly. It is pretty standard, only things are de-snork, de-resonator, simota panel filter, de-cat, runs 10 psi (standard), and thats about it. No cracks in coils, new platinum plugs, new oil, drives great on boost, but runs rough on over-run and a little rough at idle. It seems to fluctuate about 50-100rpm at idle or is that normal for this car? I wasn't sure so I thought I would try out Evoscan to capture some data and its clear from the graph that the idle is not constant. The other thing I noticed it the randomness of the o2 sensor output, or again is that normal. I can't see any relationship between the o2 data and the rpm and/or vehicle speed. I am not sure if all the readings are all correct but I am sure of voltage, rpm, speed, coolant, TPS, and possibly engine load.

www.paulhaywood.com/subaru/evoscan.jpg

Anyone got any ideas.

[Guest boostin]

Yeah, clean your idle speed controller.

[Guest]

Is that the Idle Air Controller? I took that off and cleaned it with carby cleaner. No different.

[Guest boostin]

Hmmm. My 1990 BF5 GT ran a shitload better after I threw 5 or 6 earth leads around the engine bay. The lights were brighter too. You could try that...

FYI - the o2 sensor is ignored at idle. Its only really used while cruising.

[log1call]

Try logging to a spreadsheet and the information is easier to interperate, well in my opinion anyway!

The O2 sensor is read,(and acted upon), at idle and moderate throttle. At full throttle(or near), the signal from the O2 is ignored.

The O2 reading should range about the 450 millivolt reading and it should go from one end of it's travel to the other and back again several times in a minute... perhaps ten times. The distance it ranges from 450 millivolts in either direction varies quite a bit. If everything is in good order the ecu will change the mixture quickly and so the range stays close to centre, if things are wrong then it takes longer for the ecu to get control of the mixture so the voltage ranges further.

The exact readings you get when logging are not very important, it's the behaviour you want to watch. If you pull a spark-plug lead off the revs should change a bit and the IAC valve should change it's duty-cycle to compensate, at the same time the O2 should jump/stick at one end for a second then resume it's regular ranging back and forwards.

To get good readings it's best to turn off the sensors you don't need. To give the readings some relevence it's best to have revs and throttle, then you can see where you put the foot down etc. The other sensors should be chosen to suit the job at hand, tuning or diagnosis.

As always, the results of these readings need interperating just as the trouble codes do. Just because the O2 code is set or it's readings are out... It does not mean the O2 sensor is defective... It could have a misfire. If the IAC valve has set a code or it's duty-cycle seems out... It does not mean it needs puling off and cleaning... It coud be an air leak somewhere.

[log1call]

"Just because the O2 code is set or it's readings are out... It does not mean the O2 sensor is defective... It could have a misfire. If the IAC valve has set a code or it's duty-cycle seems out... It does not mean it needs puling off and cleaning... It coud be an air leak somewhere.".

I just thought to mention...

The O2 sensor reading being out could also be caused by an air leak, but then it's reading would be towards lean. The IAC reading could be out because of a misfire, but then it's reading would be in the other direction.

By comparing all the readings we are interested in at the same time, and by comparing the possible causes of one "incorrect" reading, to the possible causes for some other incorrect reading... we can deduce the real cause of the problem.

[log1call]

Having deduced the cause of a fault it is always desirable to verify our conclusion.

There are a couple if ways of doing this....

First, and most important, by making sure the fault/symptom/effect can be reproduced reliably. To test the symptoms it is desirable to devise a test that will make it happen when we want it to. This not only is a help in deciding what factors are involved, but it also means we can test the repair after we have done whatever we end up doing.

Secondly, by substituting the suspect part onto another car or fitting a KNOWN good part to the car with the fault. This is a standard workshop practice these days. At the end of every diagnosis flow-chart I have read on EFI it has said... As a last test/resort swap a know good part into the car.