Subaru Outback Forums banner

21 - 33 of 33 Posts

·
Registered
2010 Outback 3.6R 2014 Legacy 2.5i 2003 Legacy L special edition (retired to backup)
Joined
·
1,257 Posts
knock sesors should set a code for knock sensor failure (open or short to ground, high voltage, etc)

you can just unplug it and see if the car runs.
 

·
Registered
Joined
·
17,809 Posts
I am thinking the knock sensor is sending bad readings to the ecu and its pulling timing.
That's not a valid test.

If the original knock sensor is, in fact, causing ignition timing to be delayed due to actual engine knock, then new, zip-tied knock sensor would not be detecting that same engine knock.

The true test would be to install the replacement on the engine and then see if the engine runs properly. If it does, then fine. But if the result is the same as with the original sensor (timing delayed, loss of power), then there is knock in the engine (or something that is being interpreted as knock) and the sensor and ECM are doing what they are supposed to do, i.e., delay ignition in order to save the engine from catastrophic damage.

The knock sensors are "tuned" to detect vibration events that are typical of engine knock. They have to be mounted on the engine and oriented in a particular direction to be effective. But, they can't distinguish between an actual engine knock and something else that has the same characteristics. There are reports that something as simple as loose heat shields on the exhaust can result in a knock sensor sending a "knock" signal to the ECM.
 

Attachments

·
Registered
06 BAJA EJ251
Joined
·
85 Posts
Discussion Starter #23
That's not a valid test.

If the original knock sensor is, in fact, causing ignition timing to be delayed due to actual engine knock, then new, zip-tied knock sensor would not be detecting that same engine knock.

The true test would be to install the replacement on the engine and then see if the engine runs properly. If it does, then fine. But if the result is the same as with the original sensor (timing delayed, loss of power), then there is knock in the engine (or something that is being interpreted as knock) and the sensor and ECM are doing what they are supposed to do, i.e., delay ignition in order to save the engine from catastrophic damage.

The knock sensors are "tuned" to detect vibration events that are typical of engine knock. They have to be mounted on the engine and oriented in a particular direction to be effective. But, they can't distinguish between an actual engine knock and something else that has the same characteristics. There are reports that something as simple as loose heat shields on the exhaust can result in a knock sensor sending a "knock" signal to the ECM.
The sensor that is installed is a new sensor, its installed as recommended. I will grab my stethoscope and see if i can find any knocking.
 

·
Registered
Joined
·
17,809 Posts
The sensor that is installed is a new sensor, its installed as recommended. I will grab my stethoscope and see if i can find any knocking.
Okay, so that's two knock sensors (original, and this new, installed one) that seem to causing the ECM to pull back on the ignition timing. Barring the second one having the same fault as the original, it's more likely that both are indeed picking up knock type vibrations, rather than either being faulty.

But as @cardoc noted, knock could be due to other factors. One I don't see mentioned in this thread is a restricted (i.e., partially blocked) cat converter. Has this been considered?
 

·
Registered
06 BAJA EJ251
Joined
·
85 Posts
Discussion Starter #25
Okay, so that's two knock sensors (original, and this new, installed one) that seem to causing the ECM to pull back on the ignition timing. Barring the second one having the same fault as the original, it's more likely that both are indeed picking up knock type vibrations, rather than either being faulty.

But as @cardoc noted, knock could be due to other factors. One I don't see mentioned in this thread is a restricted (i.e., partially blocked) cat converter. Has this been considered?
I deleted the second cat and it looked perfect, the front one looked clean as well. I tested the back pressure in front of the first cat and had 0 psi. I check the engine with a stethoscope and i couldn't hear and knocks.Nothing seems loose. The odd part is it only pulls negative timing when under a lot of load (pulling out on a very steep hill). lol I am really not sure where to go from here. Is it again forum rules offer a cash reward to who ever figures it out lol?
 

·
Registered
Joined
·
400 Posts
Is it again forum rules offer a cash reward to who ever figures it out lol?
OK, I know how to FIX this problem. Sell this 06' and get another one.

PROBLEM SOLVED! (this one, but what problems in the next? ....can't answer that!)

Check's in the mail, right? ;)
 

·
Registered
Joined
·
17,809 Posts
A little levity does help at these times . . .

I deleted the second cat and it looked perfect, the front one looked clean as well. I tested the back pressure in front of the first cat and had 0 psi.
. . .
The odd part is it only pulls negative timing when under a lot of load (pulling out on a very steep hill).
When you did that back pressure test, was it under the same conditions, i.e., pulling out on a very steep hill and trying to accelerate in 1st and 2nd?

I'm looking at the logs for patterns and as ideas arise, want to clarify, just in case . . .
 

·
Registered
06 BAJA EJ251
Joined
·
85 Posts
Discussion Starter #29
A little levity does help at these times . . .



When you did that back pressure test, was it under the same conditions, i.e., pulling out on a very steep hill and trying to accelerate in 1st and 2nd?

I'm looking at the logs for patterns and as ideas arise, want to clarify, just in case . . .
I did the pressure test with the car not moving, i did it cold and warmed up. it was 0 psi all the way to red line
 

·
Premium Member
2001 VDC/SC One of a Kind
Joined
·
13,988 Posts
The ECM is putting the car at 100% engine load as soon as the throttle opens, as low as 30% throttle. At 100% load the ECM should be in OL fueling and running the fuel rich, like down to 13:1 or so, not keeping it close to Stoichiometric, 14.5-14:9, as it's doing in the log. The ECM is pulling fuel like it's in CL fueling, because it is and it shouldn't be.

To understand where I'm going, I'll try to explain it.

100% load is at WOT and higher rpms than what we see in the log. At high engine loads, generally over 80% in a Subaru, the ECM should be in OL and running rich to avoid detonation and provide the torque requested. When the ECM is in OL the STFT should read 0.0 (ZERO) because the ECM isn't correcting fuel trim for emissions sake. It is fueling the engine based on the OL map and relies solely on the AF sensor feedback to run the fueling based on the map and makes correction to timing and fuel depending on knock sensor feedback.

The injector pulse doesn't match either, but then again the ECM is running CL with 100% load calculated, which is wrong. The injector pulse never goes above 18.18% and at 100% load I would have expected at minimum 25% IPW with the 185cc injectors on the engine in order to provide fuel sufficient for a high load value.

With 100% load value, ECM in CL fueling, which should be verified with RR data whether it shows OL or CL when it does this, the ECM going to 100% load right at throttle opening with low TPS values, and the knock correction tells me the ECM is erroneously reducing timing based on load value calculated and this is causing the detonation and reduction in fueling which makes the engine run like shite.

The TPS voltage looks good.
The ECT looks good.
The AF sensor is apparently working.
The rear O2 is apparently working and the cat as well.
Injector pulse matches rpm/AF sensor feedback close enough.
The engine load value is all wrong.
The fueling is all wrong for what should be OL fueling based on engine load calculated and the ECM is running in a CL mode and map.

It's looking like the computer is crapped out.
 

·
Registered
06 BAJA EJ251
Joined
·
85 Posts
Discussion Starter #31
The ECM is putting the car at 100% engine load as soon as the throttle opens, as low as 30% throttle. At 100% load the ECM should be in OL fueling and running the fuel rich, like down to 13:1 or so, not keeping it close to Stoichiometric, 14.5-14:9, as it's doing in the log. The ECM is pulling fuel like it's in CL fueling, because it is and it shouldn't be.

To understand where I'm going, I'll try to explain it.

100% load is at WOT and higher rpms than what we see in the log. At high engine loads, generally over 80% in a Subaru, the ECM should be in OL and running rich to avoid detonation and provide the torque requested. When the ECM is in OL the STFT should read 0.0 (ZERO) because the ECM isn't correcting fuel trim for emissions sake. It is fueling the engine based on the OL map and relies solely on the AF sensor feedback to run the fueling based on the map and makes correction to timing and fuel depending on knock sensor feedback.

The injector pulse doesn't match either, but then again the ECM is running CL with 100% load calculated, which is wrong. The injector pulse never goes above 18.18% and at 100% load I would have expected at minimum 25% IPW with the 185cc injectors on the engine in order to provide fuel sufficient for a high load value.

With 100% load value, ECM in CL fueling, which should be verified with RR data whether it shows OL or CL when it does this, the ECM going to 100% load right at throttle opening with low TPS values, and the knock correction tells me the ECM is erroneously reducing timing based on load value calculated and this is causing the detonation and reduction in fueling which makes the engine run like shite.

The TPS voltage looks good.
The ECT looks good.
The AF sensor is apparently working.
The rear O2 is apparently working and the cat as well.
Injector pulse matches rpm/AF sensor feedback close enough.
The engine load value is all wrong.
The fueling is all wrong for what should be OL fueling based on engine load calculated and the ECM is running in a CL mode and map.

It's looking like the computer is crapped out.
Looks that way to me too... I found a used one online for around $100, we will see if it fixes the problem.
 

·
Registered
06 BAJA EJ251
Joined
·
85 Posts
Discussion Starter #32
@plain OM found a knock pretty sure its coming from my clutch, video below. I am guess this may be the cause of my issues, knock sensor hears knock from clutch and pulls timing thinking its ping? I have already ordered a ECU I guess we will see soon enough.

 

·
Registered
Joined
·
17,809 Posts
we will see if it fixes the problem.
That would be interesting. If it does fix it, then I would really like to see another log (same parameters) of the same hill pull as in your last log.

I am guess this may be the cause of my issues, knock sensor hears knock from clutch and pulls timing thinking its ping?
Can't discount that possibility.

That's going on the premise that knock signals are causing the timing to be pulled leading to the poor engine torque. (@cardoc suggested the ECM might be messed up and is misreading things like the engine load etc. That's something else.)

In Romraider Logger, under the "Switches" tab, there should be a PID "Knock Sensor Signal", "Knocking Signal #1" or something similar. This is a binary that is 0 when off, and 1 when on. I'm not certain, but do believe that every time the ECM receives a valid knock signal, the switch changes from 0 to 1 and back to zero. If I'm right, then if this parameter is included in a log with all the other parameters, it's possible to actually "see" the knock events. It would be interesting to use this to verify if, during the hard acceleration up the hill, there is, in fact, a lot of knocking.

Along the same lines, if the noise in the video is the source of knock signals, and is causing the ECM to pull timing, it should be possible to "see" the resulting knock events in a log made under those same, stationary, conditions.
 
21 - 33 of 33 Posts
Top