Please help me fix my engine knock problem

[Bob_n_Cincy]

First a little history,
Last weekend we were running a Putnam park in Indiana. Everything was fine on Saturday for 5 sessions. Near the end of our first session Sunday I started hearing a noise that sounded like a air hose blower only during boost.
We packed up and came home to prevent possible further damage.

Some people that watched the video suggested that it was an exhaust leak going to the turbo.
They were exactly correct. (THANK YOU) The 2 bolts that hold the up-pipe to the manifold came loose and there was an 1/8 inch gap.

While reassembling it we also found a broken vacuum hose going from the intake tube to the waste gate solenoid. We repaired it.

Now when driving the car we are getting a lot of knocking at light loading.
Happens when you let up right before you shift.
We suspected maybe some bad gas. To verify, we drained the tank and filled with half 93 and half 110 race gas.
This reduced the knocking but it is still there.

We data logged a 15 minute easy drive. Below is a 5 second snap shot that had 4 knock events.
Any help would be appreciated.
Bob

PM me with an email address and I can send you the raw data.

http://thefactoryfiveforum.com/attachment.php?attachmentid=55755&d=1467673625

[JeromeS13]

These engines will always show knock events. There's no way to eliminate all of them. What's important to watch is Feedback Knock Correction. As long as that is -1.4 or less, you're fine.

[Bob_n_Cincy]

Jerome
I didn't record FKC. I didn't know what it was until now.

Here is a short 16 second video where you can hear the knocking.
Let me know what you think.
Bob


http://youtu.be/HBfSm4VhyqU

[STiPWRD]

You may want to check the spark plugs for color. It was a bit tough to follow the graph with all the different colors so I couldn't tell what your AFR was under boost but in general, to reduce knock, you could add a bit more fuel and pull back some of the timing. These two changes would however reduce power.

[Bob_n_Cincy]

You may want to check the spark plugs for color. It was a bit tough to follow the graph with all the different colors so I couldn't tell what your AFR was under boost but in general, to reduce knock, you could add a bit more fuel and pull back some of the timing. These two changes would however reduce power.

Thanks STI
My problem is happening mostly at light loads. Today I'm going try wiring the fuel pump to stay at high speed. I hope that make a difference.

I can email the raw datalog file if that would help.
Bob

[STiPWRD]

Thanks STI
My problem is happen mostly at light loads. Today I'm going try wiring the fuel pump to stay at high speed. I hope that make a difference.

I can email the raw datalog file if that would help.
Bob
Once you hit boost, your AFR should be within a fairly constant range of values. I've seen many tune for something in the mid 11's as a compromise between power and safety. The closer you get to an AFR of 12 under boost, the more power you can make but the more prone to knock you will be and the more important fuel octane becomes. A richer AFR in the low 11's is safer but comes at the expense of power. If I'm not mistaken, your 818 was tuned at Turn In Concepts and you are running a wideband O2, right? It may be worth a shot sending your data logs to TIC and getting their input.

[Bob_n_Cincy]

Once you hit boost, your AFR should be within a fairly constant range of values. I've seen many tune for something in the mid 11's as a compromise between power and safety. The closer you get to an AFR of 12 under boost, the more power you can make but the more prone to knock you will be and the more important fuel octane becomes. A richer AFR in the low 11's is safer but comes at the expense of power. If I'm not mistaken, your 818 was tuned at Turn In Concepts and you are running a wideband O2, right? It may be worth a shot sending your data logs to TIC and getting their input.

Dom Acia at TIC just put it on the dyno 1.5 years ago to make sure everything was right. I didn't want to increase power, so he didn't change much.
I do NOT have a wide band. I'm scheduled to stop over there tomorrow afternoon.
Bob

[Loring]

From the log, it looks like you're getting knock on tip in. I know you said it was on DFCO for shifting, but if you aren't knocking when you ease into the throttle and ramp up into some boost, you're probably not looking at a fuel quality or KS issue.

I'd check the FKC as well as the spark plugs. I'd also do a fuel pressure leakdown test to be sure an injector isn't leaking and dropping some fuel in during DFCO.

[Bob_n_Cincy]

I made a better chart and a link to the 25 seconds of raw data.
Bob


Raw Data
https://drive.google.com/file/d/0Bw0RlpAc9ArhaEtNM0J2eXV6MUk/view?usp=sharing

http://thefactoryfiveforum.com/attachment.php?attachmentid=55822&d=1467781165

I expanded the bottom of the chart below.

http://thefactoryfiveforum.com/attachment.php?attachmentid=55828&d=1467820002

[STiPWRD]

Bob, I'd reeeeeaaaallllly suggest getting a wideband O2 sensor, it can be great for diagnosing issues and crucial if you ever decide to do any tuning. The stock narrowband sensor is designed to operate between 14-15 AFR and is not accurate outside that range. There are some great options out there, I went with the AEM UEGO because you get a wideband sensor and gauge for $165, and it can be data logged along with engine data via RS232 and Rom Raider.

With that said, given that the knock events were under low engine loads I would think it's an issue with one of your spark plugs or injectors. Your IDC is low so the fuel pump isn't working very hard, I doubt that's the issue. Another idea is to verify that the knock sensor is working properly, I know you can hear some noise in the video but it may be worth checking as sometimes these sensors go bad.

[ssssly]

Wow, those are crazy busy logs.

Generally when posting logs for this sort of thing all you need is RPM, Manifold pressure, AFR, IDC, Timing, Knock.

Your timing is bouncing around on stable throttle input prior to knock, leading me to believe your knock counter is probably pretty high (which will cause problems diagnosing things via logs). The ECU is already doing weird things to try to keep from knocking.

Since you don't have a wideband you can't really plot the AFR direct from the ECU log. You need to log the actual voltage to get close (you really need a wide band to tune or diagnose).

Based on the above it is also very possible that as a result of the large air leak, your AF learning is way out of whack, which can lead to low load knock.

Reset your ECU and then do some steady RPM logs at 1000, 1500, 2000 RPM and then gentle acceleration from stop to 3-3500.

At that low RPM without change in load, if you still knock, you most likely have an injector problem.

If no knock without a change in load, but knock at low load under gentle acceleration you most likely have a timing or scaling problem.

[Bob_n_Cincy]

Changed our plugs today. This is on that came out of our 100k engine. No clue how old. Maybe original.
Didn't fix our problem.
Bob55847

[Bob_n_Cincy]

Wow, those are crazy busy logs.

Generally when posting logs for this sort of thing all you need is RPM, Manifold pressure, AFR, IDC, Timing, Knock.

I can repost with only this info.

Your timing is bouncing around on stable throttle input prior to knock, leading me to believe your knock counter is probably pretty high (which will cause problems diagnosing things via logs). The ECU is already doing weird things to try to keep from knocking.

Since you don't have a wideband you can't really plot the AFR direct from the ECU log. You need to log the actual voltage to get close (you really need a wide band to tune or diagnose).

I am looking into getting one

Based on the above it is also very possible that as a result of the large air leak, your AF learning is way out of whack, which can lead to low load knock.

That was my thought also. I checked the pipes after the MAF the best I could without taking things apart. Idle is fine.

Reset your ECU and then do some steady RPM logs at 1000, 1500, 2000 RPM and then gentle acceleration from stop to 3-3500. I'll do it this weekend when I get back.

At that low RPM without change in load, if you still knock, you most likely have an injector problem.
I am also going to force my fuel pump to always be in high speed.


If no knock without a change in load, but knock at low load under gentle acceleration you most likely have a timing or scaling problem.

Michael just called me with an interesting observation.
We are running a 2 quart accusump. On the track (high rpms and high oil pressure) a quart or 2 is in the accusump tank.
On the street, at hot idle and only 15psi of oil pressure. The 2 quarts is pushed into the engine. This will overflow the pan by an inch or more. If this oil level makes it to the bottom side of the piston, could it be causing piston slap?

We will turn off the accusump and test tomorrow.
Bob

[Bob_n_Cincy]

Thanks ssssly for the advice on what info is needed. Here is the same chart with only the needed info.
Bob

http://thefactoryfiveforum.com/attachment.php?attachmentid=55848&d=1467869885


55868

[Bob_n_Cincy]

Tried reducing oil level, did not help.
Still trying to figure out what is causing knocking.

[walt555]

Didn't you have low psi when on the last tune around 12 psi. Maybe the exhaust leak was preexisting from the last tune. you fixed the leak and just need a retune. +1 on the reset ECU

[ssssly]

No problem.

That last graph isn't showing up.

[Bob_n_Cincy]

No problem.

That last graph isn't showing up.

Try now

I reset my ECU every time we get out of the car. we use the battery switch as a key.
Bob

[R.Spec]

Can you get logs of feedback knock? Also you are logging too much data and losing accuracy. Take out a lot of parameters.

As well. Have you pulled knock sensor and retorqued and angled properly on the block? I have seen this situation a lot with transmission noise, ground issues. and even seen it as an early sign of a motor failure. Happened to my girlfriends car just recently, only knock under light throttle, heavy load. Full throttle, no problems. Then motor was toast.

[Bob_n_Cincy]

Can you get logs of feedback knock? Also you are logging too much data and losing accuracy. Take out a lot of parameters.

As well. Have you pulled knock sensor and retorqued and angled properly on the block? I have seen this situation a lot with transmission noise, ground issues. and even seen it as an early sign of a motor failure. Happened to my girlfriends car just recently, only knock under light throttle, heavy load. Full throttle, no problems. Then motor was toast.

I can get some new logs this weekend with less parameters.

We can easily hear our knocking. I have heard it many years ago with cheap gas or fuel starvation.

Our knocking at light throttle and heavy load, it goes away at heavier throttle.

What was broke in GF motor?
Bob

[ssssly]

Something is going on with the injectors. (or you are logging so much stuff your times are synced up right).

The knock is being caused by lack of fuel. If you look at the plots, the IDC drops and then before the sluggish factory AFR can adjust, the engine knocks, then yanks timing.

Your timing is also a bit high in the low load range.

After you check to make sure the knock sensor is torqued in properly and check the bolts around it to make sure they aren't loose/rattling and setting it off, if it is still knocking try the following.

Check your AF learning at idle and then through the low RPM range. If it is off by more than 4, you need to rescale your injectors.

I would also pull ~10 degrees of timing out of you bottom 15% of the load table.

[Bob_n_Cincy]

Thanks ssssly
We use our battery switch as the key. Is it possibly that we a not giving the AF learning a chance to adjust to our 12 year old stock injectors?
We are basically resetting our ecu every time we get out of the car.


Found out that I can get a snapshot when a knock event occurs.
Attach is 2 knock events we got today.

55922

55923

55924

[ssssly]

Well. If you reset the ECU every time you start the car, you have no AF learning when you turn the car on.

So from a cold start, unless your injector scaling is perfect, your AFR is going to be off for at least the first heat cycle (~15 minutes).

If you have not manually scaled the injectors, I would suggest doing that regardless. With injectors that old, you should definitely adjust the scale to the observed output. Unless you cleaned and flowed them, they will be off a bit.

[ssssly]

The car should not be knocking at a 14.5 AFR at low load though. And the car should have been fine at 43 deg of timing there (41 deg after it pulled 2 for knock).

You have no learned knock and the while the ecu logged a knock event profile, it didn't actually register a knock.

So I would also look for something physically hitting something else on the engine. IE a loose engine mount bolt, the wastegate arm hitting the frame, loose bolts around the turbo and intake manifold).

If it was actually repeatedly knocking, you would expect to see recorded misfires and something on the knock counter.

With how close the wastegate arm is to the frame in these, I would start there. Pressurize the WG line and make sure you have clearance at full extension. And put a wrench to the engine mounts to eliminate the simplest mechanical culprits.

[Bob_n_Cincy]

Thanks ssssly,
I will double check to see if there is anything loose mechanically.

Today we were diving without helmets and the engine cove off. So we could hear better.
Now we can create the sound at will.
In third gear, 3000 to 3500 RPMs, 10% throttle we get a constant sound of waterfall of marbles into a metal tray. That is the best way I can explain the sound.

Bob

[C.Plavan]

I remember the knock sensor being bad in a few cars, or loose.

[Bob_n_Cincy]

Today we stop by Turn In Concepts (TiC). TiC is a small company that specialize in performance Subaru's.
https://turninconcepts.com/

Our 818s looks right at home in their parking lot.

56085

Tony Barber of TIC took our car out for a test drive so he could determine what was causing our noise.

I thought our noise was related to a fueling or timing issue. Tony wasn't 100% sure, but he was suspicious of a rod knock.
Tony advised us to inspect our oil. The oil was changed 2 track days ago, so maybe 500 miles on it.

The oil out of the oil pan looked good.

This is oil dumped out of the filter. You can see all the small specks of metal.

56086


This is the inside of the oil filter. Again, you can see all the small specks of metal.
56087

I was expecting silver colored specks. These specks look more brass colored.
Is this Normal color for rod bearings or could this be from some where else?
Bob

[RM1SepEx]

How many miles on the donor? The brass flecs are babbit bearing material plain bearings are gone, either rod of crank journal. Rods are most common based on feedback that I've gotten on Subaru motors. Older tired motor raced as much as yours, no surprise

[Blwalker105]

Bob, that's more than likely bearing material and quite a bit at that for those few miles. Coupled with the fact that the noise is getting easier to reproduce means the situation is degrading rapidly. If that much material is in the leftover filter housing, I'm betting the actual filter media was chock full. Did you cut the ends off and spread it out? There are usually only two sources of anything that color in most engines: bearing material and valve spring shims.

[C.Plavan]

That does not look good. :(

[Bob_n_Cincy]

How many miles on the donor? The brass flecs are babbit bearing material plain bearings are gone, either rod of crank journal. Rods are most common based on feedback that I've gotten on Subaru motors. Older tired motor raced as much as yours, no surprise


Bob, that's more than likely bearing material and quite a bit at that for those few miles. Coupled with the fact that the noise is getting easier to reproduce means the situation is degrading rapidly. If that much material is in the leftover filter housing, I'm betting the actual filter media was chock full. Did you cut the ends off and spread it out? There are usually only two sources of anything that color in most engines: bearing material and valve spring shims.


That does not look good. :(

Thanks Guys,
The 04 FXT motor has about 100k on it.

I think the root cause: I was NAIVE. As we were tracking the 818 with base power setting (195whp) that I could get by without an external oil cooler and not have proper gauges for racing.

Now it is decision time. We have a track days schedule on the 19th, 30th, and 31st this month.
Rebuild this motor quickly or steal the donor motor from the unfinished 818.
Bob

[STiPWRD]

Swapping a shortblock in 6 days is doable but going back out to race without proper data logging (oil temp, oil pres, AFR, IAT) is asking for trouble. Sorry to hear about the engine though, on the bright side now you can upgrade.

[RetroRacing]

Please get one of these, you will be much happier when tracking a car. We check ours after every weekend.
http://kandpengineering.com/

[ssssly]

The guys at TIC are top tier Subaru guys.

Gold specs are almost always bearings.

Unfortunately that only adds another question.

Has the knock been being caused all along by the bad bearing?

Or is the engine knocking from something else, and the knock caused it to throw a bearing?

The thing that worries me is that in your original logs, the fuel cuts befor the knock event.

And thrown bearings don't generally cause true knock or sound like a metal, marble waterfall.

That being said you can rebuild a boxer in an afternoon if you know how to work on motors. I would be very suspect of those injectors and double check all the ignition wiring.

But whatever you do, listen to the guys at TIC if they give you advice.

[Bob_n_Cincy]

The guys at TIC are top tier Subaru guys.
Gold specs are almost always bearings.
Unfortunately that only adds another question.
Has the knock been being caused all along by the bad bearing?
Or is the engine knocking from something else, and the knock caused it to throw a bearing?
The thing that worries me is that in your original logs, the fuel cuts befor the knock event.
And thrown bearings don't generally cause true knock or sound like a metal, marble waterfall.
That being said you can rebuild a boxer in an afternoon if you know how to work on motors. I would be very suspect of those injectors and double check all the ignition wiring.
But whatever you do, listen to the guys at TIC if they give you advice.

Hi ssssly,
Totally agree with your opinion on TiC.

The sound has not really changed.
I'm sure Tony has heard 100s of rod knocking engines. But he wasn't 100% sure. He probable listened to it for 1/2 an hour.

Most people would not have noticed our knock in a normally sound insulated car. We also have a relatively quiet exhaust.
From what I'm hearing, it might be multiple cylinders.

The fuel cuts in our logs were a result of leaving off gas pedal during a shift. Then it would knock just as we started to get back into the gas.

The one thing I haven talked about much yet was our oil. We use Rotella T6 5w-40. The oil was less than a month old with maybe 500 miles on it. It came out very dark compared to new oil.
Even though our logs show the coolant staying right at 190f, I believe our oil got much to hot at our track day.

Our plan is to fix or replace the engine and add an oil cooler and gauges.
Bob

[ssssly]

While yes, its knocking as you accelerate from a shift. In all but one case where the IDC didn't even move after the knock, the fuel cut is prior to the knock (assuming that the time slices aren't far enough apart to create a lag due to logging a ton of stuff).

Is easier to compare it to the timing line. The ECU pulls timing in the shift, and at knock. If it was pulling IDC in response to the knock, the timing and IDC would always pull at the exact same time.

Your plots do not show this. You have IDC rising through knock induced timing pull and IDC dropping prior to knock and timing pull in separate places.

[Blwalker105]

Most rod knocks can be duplicated with the vehicle at rest and the engine brought up to 2000-2500 rpms then revved up and down a couple hundred rpms. Was this the case in your situation?

[Bob_n_Cincy]

Most rod knocks can be duplicated with the vehicle at rest and the engine brought up to 2000-2500 rpms then revved up and down a couple hundred rpms. Was this the case in your situation?

We did not know this. So we were just revving from 700 to 5000. We didn't hear any knocking.

When Tony was listening with the car at rest, he was able to make it knock when accelerating in the 3000-3500 RPM range.
Thanks
Bob

[Mechie3]

All my blown motors sounded worse in the 3000rpm range. I've pulled a motor, swapped in a new factory shortblock, and put it back into a car in a weekend. I've also done it several times. Should be easier in an 818. Might be more work, but I would be inclined to swap in the other donor motor so you're not rushed rebuilding this one. Check the cams for wear. Don't mix up your cam caps, and if they're mildy scored you can hand buff them. If they're worse you need to find a shop that can deck and line hone. I remember only finding one shop that would do that when I was looking about 4 years ago.

[Bob_n_Cincy]

All my blown motors sounded worse in the 3000rpm range. I've pulled a motor, swapped in a new factory shortblock, and put it back into a car in a weekend. I've also done it several times. Should be easier in an 818. Might be more work, but I would be inclined to swap in the other donor motor so you're not rushed rebuilding this one. Check the cams for wear. Don't mix up your cam caps, and if they're mildy scored you can hand buff them. If they're worse you need to find a shop that can deck and line hone. I remember only finding one shop that would do that when I was looking about 4 years ago.

Thanks Craig,
We decide to swap the whole motor this weekend.
We are going to add a heat exchanger, and remote filter to our accusump oil circuit.

We will pull the block apart at a later date to see if rebuild or replace makes most sense.

Sitting at my barn is a collection of blown motors, 07 Honda 400cc ATV, 06 NA 2.5L impreza, 07 NA 2.5L impreza, and now 818 04 FXT 2.5L. Not able to keep up with the demand.

Going to cancel our track night Tuesday.

Bob

[Mechie3]

This was my garage back in 2009. Two blown blocks, one new block.

http://i139.photobucket.com/albums/q291/mechie3/1018092014a.jpg

[Bob_n_Cincy]

I opened up the oil filter.
Here is a picture. You can see metal particles in the folds.

56184

[C.Plavan]

Do you really think the Accusump worked? I was always worried about the oil pressure, and its small size- the Accusump runs out of oil pressure quickly with the pressures the EJ throws out.

Have you thought about a drysump or that new killerB valve setup for oil?

[Bob_n_Cincy]

Do you really think the Accusump worked? I was always worried about the oil pressure, and its small size- the Accusump runs out of oil pressure quickly with the pressures the EJ throws out.
Have you thought about a drysump or that new killerB valve setup for oil?

Hi Chad,
I feel confident that the accusump is working. Right now we are only running a manual valve on it. Canton has recommended that I add an electric valve so it only dumps at defined pressure. I will make that change with the replacement motor.

At 1.2G (no aero yet) I'm not sure the accusump is even necessary.

I'm not ready to jump to a dry sump yet.

I feel our oiling problem was heat related. Our almost new oil cam out pretty dark.

I feel we made a mistake in not having a oil cooler on for track days. even with stock hp.

Do you have any extra oil coolers system parts around you want to sell?

Thanks
Bob