Jump to content
Ultimate Subaru Message Board

Recommended Posts

98 ej22 in a Legacy L. ~141k  I'm pretty stumped at this point and I appologize for a long post - bare with me here:


After an oil change I start the engine with the knock sensor disconnected.  I
then reset the ECU or clear the code.  Don't ask me why - it prevents
the bogging issue - it works for me.


After the latest oil change at 140k (Castrol 10w-40, FRAM filter), however, I have
been getting the typical knock sensor bogging issue even after the
tribal oil change dance.  I moved the knock sensor to the location of a factory ground point (on bracket above starter), reset the ECU and drove for about 100 miles.  Knocking? Yes.  Hesitation, bogging BS? No.


Corrosion was apparent on the mating surfaces so I sanded them back to clean metal.  I then tried the old knock sensor in the factory location, reset ECU and got no joy. 
I do observe the ~15mile learning cycle by the ECU before it begins to
bog.  It is at this point I was confident a new knock sensor would fix
my woes.  I purchased a new Knock Sensor from Autozone and installed in
correct location with about 15 ft-lbs making sure to orient the wire
lead so that it was clear of block protrusions.  Reset the ECU and got
no joy.  Now I started to suspect a bad connection so I cut the wire and
hardwired the old knock sensor in sans connector.  That resulted in no
change of symptoms (see note 1 below).


Now I was becoming befuddled and a bit ticked off. 
I can change the performance of the car by relocating the knock sensor
but a new one won't fix the problem.  I figured I didn't need a new
knock sensor and, at this point, I started to march down the path of "Gee, maybe there's too much knock!" so I returned the new KS to Autozone.

I drained/refilled and burped (to death) cooling system and installed a new thermostat (the bigger OEM type) and a new rad cap.  Operates right around 180-190F according to OBDII data so it is not running hot.


Then I went after the ignition system a bit: I installed $40 NGK wires and NGK Irridium plugs gapped to a tight 40 thou.  I also installed 500 pounds of copper all over the place to enhance grounding.  I cleaned
the mating surfaces of the igniter and it's mounting plate, although
the igniter is currently sitting on two ring terminals - one as a spacer
only - the other with a lead directly to a ground point on same sheet
metal platform igniter is mounted to.  The coils are about 25k miles old and I have an older set which I have swapped out with no change in symptoms.  The coil contacts are clean and I use dielectric grease on the plug wire boots.


Then I filled up with high octane gas and reset the ECU. 
With the original KS installed in factory location I still got the
bogging issue.  When I move the KS to my off-the-block location the
bogging goes away and there's no audible knock with high octane gas.  So, at least I know that high octane gas really has a higher octane.


all this I went to the junkyard to source an OEM connector and soldered
it in.  There is a short section where the sheilding is soldered
together but the rest is completely sheilded - see note 1 below.  I then
went back to Autozone to buy a new knock sensor.  I did this with the
hope that I would get another knock sensor (not the one I returned) but I
got the same one.  I'll ask and see if I can try another next time I'm
by there.


I purchased a BAFX Bluetooth OBDII
adapter (cool tool) and tried to read engine parameters with Torque
Lite.  All I got was engine temp and load so an actual analysis of
engine parameters was a dead end there. (If I buy Torque PRO will it
give more engine data or is my car just too old-school?)


this point I'm thoroughly stumped.  I will check the main harness
connections but I don't expect anything to come of it.  I saw some other
threads on USMB with good info but it seems the threads just died
without any real resolution - here, here, and here.
I can't imagine it is anything other than knock sensor related.  My
next steps are to check the car's recall history and see if I can get a
hold of a second new knock sensor to test the infant mortality theory. 
If a third knock sensor makes no change perhaps the ECU is to blame.


All pointers/tips/ideas are welcome and appreciated!



  1. I have a degree in Electrical Engineering and can surmise
    that the real need for the sheilding on the knock sensor wire is when
    ignition parts begin to leak their current i.e. a leaky spark plug cable
    that is jumping to metal on the engine somewhere rather than fully to
    the plug.  I can't imagine that there is any other source of electrical
    interference in the engine bay that could induce electric current in the
    wire - mine was a short, unsheilded piece which, I believe, means there has to be a
    LOT of unshielded current flowing somewhere else before any induction
    will take place in the short unsheilded wire to the knock sensor.
Link to comment
Share on other sites

I have a friend who feels your KS pain, we've been working with his problem for a couple weeks to no avail. So I have no real insight to that part of the issue.


What I can tell you is that you need a blue tooth scanner with SSM capabilities to read our ecu completely. Elm327 protocol scanner is not equipped with the coding needed for pre 00 subaru ecus. There are options out there for this but they are not cheap like the bafx unit you have, or the no name pos I have lol.

Link to comment
Share on other sites

What CEL code are you getting?


Cutting wires is never a good idea!


Most engine sensors are very low voltage and any corrosion or poor connections can set a code.


When the engine detects Knocks, it retards the timing to prevent knocks and thus the loss of performance.


5-30 oil is recommended for most Subaru's.


Since this started after the oil change, you might want to start there.  New Oil and a Subaru filter are way less than you've spent in parts and time.

Link to comment
Share on other sites

welcome, i'm from westminster.

good troubleshooting so far, really good!


knock sensors are like $15 on ebay if you're paying high prices at AZ and want to try a different one.


i'd test resistance:

1. between knock sensor connector and corresponding pinout on the main engine harness connector (passengers side rear of engine bay)


2. between the body-side engine harness connector (same passengers side rear of engine again, just body side connector instead of the engine side one it plugs into) and the ECU in the passengers front seat foot well (it's under the carpet, very easy to get to).


Swap out another ECU.  PM me and I can ship you one to test if you cover costs.

Link to comment
Share on other sites

The sensor is being "number" by moving it farther away. This could indicate there is very slight knocking occurring which is not audible but still detectable by the sensor.


Knocking is caused by low octane fuel, too high compression, or a lean AF ratio, not by ignition components. It can also be caused by oil getting into the combustion chamber.

Another factor is the EGR(if equipped). If the EGR is clogged or inoperable knocking can occur under low/medium load conditions such as during normal cruising at part throttle. The a/f mixture is lean under these conditions which leads to high combustion chamber temps. High temp plus lean mixture equals knocking. EGR helps to combat this by lowering combustion chamber temps.


Try a strong fuel system cleaner (Lucas works well), new fuel filter if it hasn't been changed recently.

You may also want to check fuel pressure with the engine under load.


A bad PCV valve can lead to excessive oil consumption, even if the valve still rattles it may be damaged. Get a new one from SUBARU, not autozone.


Make sure you have no air leaks. Check all of the breather and PCV hoses for cracks/ splits etc. Check vacuum hoses also.


Front O2 sensor. Is it old? If its reading a false rich condition it will lean out the mixture.

Link to comment
Share on other sites

First of all, thank you for all the responses!


I found out about needing something that can read SSM the hard way even though it is an OBDII car - oh well.

No CEL.  Cutting wires is probably not normally a good idea especially if you don't know how to put them back together but it was worth a shot.  I always use a 10w-30 or 40 when I change the oil and it has been fine.  The idea has crossed my mind to try changing the oil but I'm incredulous to believe it would have an effect.  Stranger coincidences have occurred.  I will give it a try, though.

Grossgary, thanks for the warm welcome! I think testing impedance further along the circuit is a good idea.  That could have the effect of moving the min/max signal/voltage range to out of spec levels.  I may take you up on that ECU swap.  I'll try all other suggestions first.

Fairtax4me: I understand what the knock sensor is doing.  The thing that is not clear to me is to what extent it should retard the timing when knocking is detected.  I have to imagine that it is overreacting to a normal amount of knock when sensor is in factory location.  I think it was you who, in another post, pointed out that the ECU is constantly searching for optimum operating conditions and pulling back timing when knock is detected.  Can knock really be that bad in my car that it is just working fine and engine performance is suffering due to some other issue?  As far as I'm familiar with knock it is not severe and, as I said earlier, the audible knocking is not present with high octane fuel but no change is observable in poor performance between 1500-2500 rpms.


I will pull the EGR and clean it out.  I have never touched it so its a good place to check.


I have replaced the PCV a while back - will try a genuine Subi one, though.  BTW, I don't think she's burning much, if any oil - I never have to top it off and I always check level at fillups.


I have checked for leaks using the tried and true can of carb cleaner and have found none.


Front O2 has been replaced prior but I will check its connections.  Agewise?  I'm guessing I put that on there maybe ummm, 45k miles ago.


I did do seafoam about three tanks of gas ago.



I think I have a course of action:

  • probe along harness to ECU for circuit faults (specifically knock sensor circuit)
  • pull EGR and give it the once over
  • replace PCV valve with Subi PCV valve
  • pull O2 sensor and test voltage output, check circuit for faults
  • change to lower viscosity oil (on both grades)


Thank you to all who posted.  I appreciate all the ideas.  I will reply with outcome when completed!

Link to comment
Share on other sites

why doesn't piston slap cause the knock sensor any issues? wrong frequency?

Different frequency. The frequency of spark knock is always the same whether its noticable or not. It's also practically the same for every engine. It can vary slightly, but not by much. A Ford knock sounds the same as a Chevy knock sounds the same as a Subaru knock. It's a distinctive sound and when you hear it coming from a car driving by you know what it is no matter what the car is.


No amount of knock is considered normal. Knock is abnormal spontaneous combustion of the AF mixture that is not initiated by the ignition process. Optimally there should be 0 knock occuring in a gasoline engine. But with emissions laws the way they are, engines have to run a very fine line where the AF ratio will be as lean as possible without causing knocking. But with all the variations in temperature, atmospheric pressure, fuel quality, engine condition, etc. the fuel injection system can't always maintain that exact perfect AF ratio. If a knock is very slight it won't hurt anything in a normally aspirated engine. There are plenty of cars that knock audibly for 100k miles or more, but audible knocking can and will eventually cause damage to the engine. It's also a factor in the formation of NOx emissions which are one of the major forms that the EPA and CARB have enforced so many regulations in order to prevent.


I had a similar issue last summer which I "cured" with a resistor in place of the knock sensor. I had yet to try a new sensor to see if the condition will return. This is on a 96 with 235k miles and it does burn a bit of oil, which I figured is probably the primary factor. I've only noticed occasional audible knocking without the sensor, and have probably put 8-10k miles on in the year.

Link to comment
Share on other sites

The sensors I had at the time measured around 580k ohms so I aimed for that. I think I put together a 100k and a 470k in series.

Soldered and heat shrinked with an eyelet on one end and a connector I cut off an old Knock sensor to plug into the harness on the other end.


I bolted it down on top of the sensor that way I could switch back and forth for testing purposes. Originally it was to be temporary, to be on for a few weeks while I tried different grades of fuel and such. The problem I was having would only pop up under very certain conditions so I wasn't sure if it was the KS or something else. If it actually was something else, the false sensor masked the problem. Haven't noticed any recurrence of that issue since then, but the problem was only happening when it was hot out.

Link to comment
Share on other sites

So out of my list of things to do and check....


...I did none.


Not long after starting this thread the car started making a noise exactly like

.  Knowing that I had a dodgy tensioner I proceeded to replace it and the two smooth and one toothed idler pulleys.  Stealership parts with double bearings were purchased (one old pulley was single bearing and had some play).  Also, PLEASE BE CAREFULL with those damn pulley bolts or else you're in for some thread tapping fun - don't ask!  I have since moved the knock sensor back to its original location and all seems well.  I will post a followup months from now to aid any who come accross this thread and wonder if this really fixed the knocking.  So far, it appears to have done the trick!


Initially, after completing this job the sound changed to a whine.  Since I have driven the car several miles, the whine has gone away and I'm left with a textbook failing water pump sound.  I took it to our family mechanic who instantly said, "That's the water pump."  To whom I said, "The water pump is 4k miles new."  Then he couldn't believe it was the water pump but came around to admitting the remote possibility that I have had a brush with "infant mortality" as they call it in various production businesses.


My question now is this: what, if anything besides a bad water pump could make the bad water pump noise?  This is really just a sanity check for me because the sound is coming from the water pump area and it sounds like a dying water pump but, hell, I just bought the darn thing!!!  My only other guess is that I botched the thermostat installation during the coolant change and the thermostat is rattling around, but I doubt it because IT SOUNDS LIKE A WATER PUMP DYING - sorry - I'm going crazy.  I just don't want to take the car apart one more time and have something ELSE pop up!


This noise is NOT accessory belt related as the noise exists without said accessory belts connected.  Does anyone else have "infant mortality" experiences with water pumps that they'd like to share?  BTW: it is a Duralast from Autozone.  I may spec out a new one from the stealership but I would like to be pretty darn confident I'm spending more $$$/time where it needs to be spent before I pull the trigger on a water pump.


Thanks for reading!

Link to comment
Share on other sites

I said the noise wasn't related to the accessory belt system.  Well, I have removed them so many times lately that I mistakenly thought I had heard it without them installed.  I was wrong.  I was pulling everything off again yesterday evening.  After pulling the accessory belts and getting down to the timing cover I started the car to try and pinpoint the noise but - no noise.  So I put the A/C belt on - no noise.  Then I put the PS/Alt belt on - NOISE.  A brand spanking new Duralast reman alternator DOA.  Not even a hundred miles on the POS.  I took it back for a refund.  I'll be purchasing another reman (damn!) from Subaru which is around the same price.  So I think I can put this whole fiasco to bed....




on my way out the door to get the Subi badged alternator before the parts dept. closes I backed into the car of the real estate lady who's selling the house across the street :banghead: .  Her car banged up a steel railing for some stairs that lead down to the basement level of the split level house that she is selling :banghead: .  She was inside and thought her clients had shown up - HA!  She was nice about it and joked that now I had to buy the house.


I will update in a month or two about how the car is running for those who stumble (no pun intended) accross this thread looking for solutions to their hesitation/power loss issues.


Thanks again everyone for ideas and moral support! :headbang:

Link to comment
Share on other sites

My house is oddly offset from my neighbor's across the street - due mostly to the floorplan being mirror-imaged I guess and the garage on the north instead of the south. I have had to dodge their trucks, campers and son's landscape trailer many times. BUT, the previous owners over there TWICE backed into a daughter's car. I have since builr a flower bed at that corner of the driveway to discourage folks from parking there.


If you still have the noise after the alt, look in the power steering tank for bubbles while idling. PS might be sucking in air - though it sounds like you have found the problem.

Link to comment
Share on other sites

The verdict is in.  Replacing the timing idler pulleys and tensioner fixed my hesitation issue.  I just got back from a 450 mile trip to Spruce Knob, WV and the car performed really well loaded down with camping gear and mountain bikes.


Thanks again.

  • Like 1
Link to comment
Share on other sites

  • 1 month later...

Everytime I read something like this it amazes me at how sensitive these cars are to such minute things as a t belt roller. In my 10 plus years as a pro mechanic I have do e maybe two honda belt rollers ever. Almost every suby I mess with needs them done though or didnt get them done in time. Glad you found the issue.

Link to comment
Share on other sites

  • 2 weeks later...



zombieforce, I have come to see how sensitive the Subis are, too!  It truly is becoming a PITA but a challenge I can't refuse :-P


I regret to say that it has been acting up again.  Clearly, the timing belt pulleys had a positive effect on changing the (lack of) performance issue in the 1500 to 2500 RPM range but it was still there.  There are obviously several things contributing to poor engine performance.


In discussing the issue with an experienced dirt bike/motorcycle (nut) friend of mine he suggested verifying that I had the correct spark plugs installed.  He mentioned he had run into trouble running too hot of a plug in his dirt bikes before which led to poor performance.  I went on Amazon, Autozone and Advanced Auto websites to see what temp range spark plug they recommended for the EJ22.  All three recommended a R5 temp rating for the NGKs (for all tip materials).  NGK recommends running a R6 if you check their website.  Furthermore, they have great technical info on spark plugs - interesting reading.  The thing that struck me most is that the difference in tip temperature for plugs that differ by one temp rating can be as much as 70 to 100C!!!  Therefore, if you're off by one rating you can really screw the combustion process up.


I got plain jane BKR6ES-11 plugs - standard copper plugs to replace the R5 iridiums I had in there.  That has improved my engine performance tremendously!  It was amazing.


Alas - there is still a hesitation problem which I haven't been able to divine yet.  This problem is more of a loping issue as opposed to the previous and most definitely identifiable engine timing retardation due to timing belt slop and wrong plugs.  I'm pretty sure that my next issue is not timing retardation related because I tried the old nylon washer under the knock sensor trick to no avail.  I'll probably move it off the block to make absolutely sure it isn't related to that but it doesn't really have the knock sensor symptom anymore anyways.


The wierd thing about this layer of the cake is that this week I noticed that the problem was evident when the tank was empty.  However, the next day with a full tank the car was working absolutely perfectly without any indication of an issue at all.  At this point I don't have enough evidence to tell if it is a coincidence or not but I will pay more attention to it as the tank empties and gets filled.  I have replaced the filter before but I honestly can't think of a reason why that would make a difference depending on how full the tank is.  I think I put in some drygas a while back but I can't remember.  It's worth a try for a quick fix but beyond that I'm still searching for the answer.


I'll keep updating as work progresses.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Create New...