CEL has been on about 50% of the time since install [4000 mi, 6 mos] of a used motor following seize-up of original [probably a bad timing belt change]. The sensor is said to have been deliberately transferred from the dead motor to this one, 'to match my ECM'. I am sure I can change it [no thanks to Haynes, but thanks to USMB!!] but before I do that, I could

 

[a] try a tank or two of Premium; if it stays OFF, I really may have an engine knock and replacing the sensor won't do any good; then $90 goes to better gas. Light coming ON tells me nothing.

 

Measure the resistance while the light is off, and put in a resistor instead of the sensor [or just use 500K]. If the CEL comes ON, there is a defect elsewhere, like in the wiring and replacement is pointless. If the CEL stays OFF, that would point to a bad sensor, right?

 

thanks for the elegant recent discussion on TB; this is an amazing Board!!

If you know it's a knock sensor code change the sensor. They go bad with age. Mostly cracks. The code reports a problem with the sensor not an engine knock.

Got alm0st similar question. Do you think that a bad sensor make the motor rattle as if a rod was bad? Someone told me that the sensor can't cause a knocking sound, but I'm looking forward for confirmation. Thanks.

 

_____________________

Autopartswarehouse.com Car Solutions.

I'm chasing this condition on a 98 Legacy 2.2. Replaced one that looked really good with a used one from a 97 2.2 that is unknown but looked good.

 

Neither had any cracks, corrosion, connector looked good.

 

I hate to buy a new one with all the used ones that I have but it's looking like that may be a smart move. Dealer says VIN# is mandatory - I think he said there were 8 different ones. I think he said they start at 80 bucks.

 

Any concensus here on whether this is a recommended dealer item like plug wires on a 2.5, WP gaskets, cam seals, etc?

 

Thanks,

 

Dave

Any consensus here on whether this is a recommended dealer item like plug wires on a 2.5, WP gaskets, cam seals, etc?

Yes! Dealer only. Otherwise you'll just throw good money at it and still have the problem in a short while.

 

Yes, the VIN is necessary. Seems there were several in use for the same model year, (well, that's what I was told :-\ ).

 

If you do it on your own, pay attention to the angle of the sensor when you install it (as per the pix on this board). If it's the wrong angle your CEL will come back on.

 

Look at it this way: it'll be done for the next 100,000 miles.....

 

When I hit 260k I'm just going to pull it, replace it and the O2 sensor again, and be done with it, I've had enough of the CEL thing.

Got alm0st similar question. Do you think that a bad sensor make the motor rattle as if a rod was bad? Someone told me that the sensor can't cause a knocking sound, but I'm looking forward for confirmation. Thanks.

 

_____________________

Autopartswarehouse.com Car Solutions.

The sensors job is to tell the ecm when the engine is pinging/knocking. The ecm retards the timing until the sensor stops sending the signal. If the sensor is bad (open) then it doesn't signal the ecm and the engine pings/knocks. So yes to your question except it doesn't make the engine knock it lets it knocks. Also I used a bosh knock sensor on the 99 imp seems to be working fine. Make sure the angle is correct on installation it does make a difference.

checked the code again at AZ, 0325, knock sensor.

disconnected sensor, inserted 500K resistor to ground: CEL still on

disconnected battery, CEL still on

 

looks like wiring fault,not sensor? anyone tell me which pin in ECM as if open circuit, i won't find it by looking for continuity!

i'm not sure that the test completely rules out the sensor.

 

have you read the knock sensor installation procedures? there are two key steps...just "installing a sensor" is not appropriate. the sensor is supposed to be installed at a particular angle and at a particular torque. i do not know why, but that's how it's intended.

 

someone else just had a knock sensor code last week and reinstalled the same sensor (properly) and the code went away.

 

i'd pull the sensor and have a look. the sensors don't fail internally, it's just the rubber housing that cracks and deteriorates. it's very easy to remove. pull yours, check for cracks on the bottom and reinstall it properly - get the angle right and the torque on the bolt right.

checked the code again at AZ, 0325, knock sensor.

disconnected sensor, inserted 500K resistor to ground: CEL still on

disconnected battery, CEL still on

 

looks like wiring fault,not sensor? anyone tell me which pin in ECM as if open circuit, i won't find it by looking for continuity!

 

Knock sensors don't operate by resistance values, their signal is audible and is measured in hertz, Khz.

but several people have described installing a resistor as a temporary correction of the CEL on this particular sensor. the ECM is said to see this, in the self check, as acceptable.

 

grossgary, i agree the existing sensor is probably bad but i would like to be as sure as possible, before i drop 90 at AZ or even more at Subaru!

but several people have described installing a resistor as a temporary correction of the CEL on this particular sensor. the ECM is said to see this, in the self check, as acceptable.

 

grossgary, i agree the existing sensor is probably bad but i would like to be as sure as possible, before i drop 90 at AZ or even more at Subaru!

Hmmm... normally I would have to see it to believe it, but I'll take your word for it. Learn something new everyday.

 

Rockauto.com has the OE knock sensor for the '99 2.5 Legacy Outback for $43.00, but not for the Impreza Outback 2.2 motor. Not sure which you have.

checked the code again at AZ, 0325, knock sensor.

disconnected sensor, inserted 500K resistor to ground: CEL still on

disconnected battery, CEL still on

 

looks like wiring fault,not sensor? anyone tell me which pin in ECM as if open circuit, i won't find it by looking for continuity!

I may be wrong about this too but I don't think you can clear codes on an OBD II car by disconnecting the battery. I believe it has to be done with a scanner, or it should clear itself after three drive cycles after the problem is fixed.

[...]looks like wiring fault,not sensor? anyone tell me which pin in ECM as if open circuit, i won't find it by looking for continuity!

ECM/ECU pin #3 (white wire) is the knock sensor terminal, according to FSM. It's not a direct connection, however -- that white wire goes to pin #6 of connector B21 (12-pin gray), which connects with E2 (where the wire color changes to Blk/W), continuing to the knock sensor. Shield continuity is via pin #7 of B21/E2. The shields are connected to ground via a Blu/G wire from pin #56 of the ECU, again not directly, but via "Shield Joint Connector" B83 (4-pin jumper).

 

 

but several people have described installing a resistor as a temporary correction of the CEL on this particular sensor. the ECM is said to see this, in the self check, as acceptable.[...]

One test for the knock sensor is to unplug the ECU connector and check at pin #3 for between 400k-ohms and 700k-ohms to engine ground. If that's the range of resistance the ECU finds "acceptable", then a resistor of value somewhere in the middle of that range should fool the ECU that an intact knock sensor is in place. That should prevent generation of the P0325 code, but of course offers no knock detection.

One test for the knock sensor is to unplug the ECU connector and check at pin #3 for between 400k-ohms and 700k-ohms to engine ground. If that's the range of resistance the ECU finds "acceptable", then a resistor of value somewhere in the middle of that range should fool the ECU that an intact knock sensor is in place. That should prevent generation of the P0325 code, but of course offers no knock detection.

 

thanks for the detail!!

 

i am happy to say TWO HOURS battery disconnect was the trick [or possibly three start/stop cycles, as that happened as well, in testing the battery disconnects]. CEL is now off and sensor itself is apparent problem. next i will check it out physically.

 

is any draining of coolant necessary? posts vary on this point.