How do you test if knock sensor sending false readings to ECU?

[dcolemanca]

I have 97 Legacy Outback and think I may have the infamous knock sensor issue being too sensitive and retarding the ignition leading to some slight hesitation at lower revs.

 

I live in SF bay area and always use 87 octane gas.

 

It's not really obvious but it does feel like a slight hesitation when I accelerate.

 

I don't have a check engine light on so the sensor is probably working OK.

 

I was wondering if I can test this condition by disconnecting the knock sensor and see if the hesitation goes away?

 

I would just do this for a test run so minimal chance of any damage.

 

What is the latest thinking on the solution to this problem?

 

Has there been any improvement to the sensor so I should buy a new one?

 

What's the best way to reduce the sensitivity? I see the use of a teflon spacer being recommended, does that work?

[cookie]

and he is a bit of an electronic wizard. Do a search on his posts if he does not chime in here.

[outback_97]

http://www.geocities.com/hobiegary/hesfix.htm

 

Assuming you've read this?

 

Steve

[blitz]

and he is a bit of an electronic wizard. Do a search on his posts if he does not chime in here.

 

Wizzard! Thanks for the vote of confidence tho.

 

The ECU does a continuity check of the circuit by measuring the voltage drop across the sensor's resistance, so you'll need to substitute a resistor in place of the sensor rather than just dissconnecting it.

 

Just connect the resistor from the single lead to the engine ground, simple enough. I think most of the sensors measure out at around half a meg. Mine measured at 560k so that's the resistor I used. I'm guessing you could be off about 20% in either direction and the ECU would be none the wiser as it's really only a rudimentary check for a short or open condition in the sensor circuit.

 

If you just disconnect the sensor (ECU sees open circuit), the engine will go into safe mode and pull about 10 degrees of timing / throw a CEL (which can be reset with a quick batt. dissconnect).

 

I did this about 45k miles ago and never re-connected it. I never needed one on any car I've ever owned before, why start now? They're good in theory, but crude in practice. It'll be a better system when they implement two sensors through logic and compare phase angle and intensity, etc.

 

Hope that helped.

[blitz]

I should add that that you can do the resistor trick just as a temporary test to see if the sensor is giving you the hesitation. You can always re-connect it afterwards. A couple of occasional light pings isn't going to hurt anything. If for some reason it starts knocking like a diesel, don't continue to drive it that way.

[dcolemanca]

Thanks for all the responses - I'll try the resistor trick and post back with the results later.

[Commuter]

At what rpms are you getting the hesitation?

 

My understanding is that the knock sensor problem affects the mid range (eg 2000 to 3000 rpm sort of thing). If you are having hesitation when starting off with a light throttle application, then it is probably carbon build up.

 

Ask me... I know. 97 OB with 425,000 km. Mostly highway driving, Esso or Sunoco gas, and it still happened... twice. The second time, a sluggish O2 sensor was agravating the problem.

 

Also, the knock sensor issue should disappear briefly after an ECU reset. The carbon issue won't.

 

Commuter

[cookie]

425,000 Km! You might be the first one to get that year of Outback up that high.

[Commuter]

425,000 Km! You might be the first one to get that year of Outback up that high.

I doubt it, but yeah, it is a lot.

 

Headgaskets and a conrod bearing failed at 260k km though... Otherwise, its been a pretty good car. I expect to keep it for a few years yet.

 

Long ways to go to match my 85 Civic though... 712,000 km over 14.5 years. Engine was never touched. Clutch was even original.

 

Commuter

[cookie]

have gotten out of an early 2.2. Probably the distance to Mars.