WRX constant misfire on #4, can't find anything wrong

Hello, I am having an issue with my 2013 WRX. I put a rebuilt motor in it last fall. It has forged pistons and rods, rebuilt heads, factory cams, and gates race timing belt. Here are the upgrades that I installed with the motor: AEM fuel pump, IAG fuel rails with parallel feed lines and Aeromotive regulator, Injection Projects coils, Perrin intake pipe, Tomei headers and uppipe, and Grimmspeed BB500 turbo. I already had a PLM downpipe, Grimmspeed top mount intercooler, and 3 inch cat back with the previous motor that were retained with the new motor.

I also have a set of ID1050 injectors, but since things were changed with the intake track, I put the factory injectors in for the break in so I could focus on the airflow for the initial closed loop tuning. The initial tuning went well, and the break-in also went well. Since my other car is not well suited to winter and I wanted to make sure everything was fine with the motor, I kept the car out of boost and in CL until spring. I had no issues driving the car like that for ~ 13000 miles.

Unfortunately, before I could even start the closed loop tuning, it started to develop a mis-fire on cylinder 4. Since my last Subie died due to ring-land failure, also on cylinder 4 coincidentally, I immediately checked the compression. It hadn't changed since my last check during the break-in. The spark plugs were showing more wear than I expected to see, the ground side was eroded causing the gap to increase, probably the worst on #4. So I put in some new plugs and it seemed fine for about 50 miles. The things went downhill really quickly. It now has a very noticeable rough idle and almost constant misfires on #4.

I also had one instance of fuel injection circuit malfunction on #4. So, I decided to put the 1050's in. The o-rings on the factory injectors weren't in the best shape, but did not appear to be leaking. I updated the tables for the 1050's, and it is hitting 14.7 now at idle, but still with roughness and constant misfires on #4.

Here is what I have looked at, I'm out of ideas at this point.

Re-did compression, still fine.

Leak down, both #2 and #4 were good ~ 20-25%.

Swapped out factory injectors for new 1050s, and updated tables correspondingly.

Checked the ground at the intake bolt.

Swapped the coils between 2 and 4.

Checked power at fuel injector connectors,

Checked continuity of the lines between the ECU and the injectors.

Checked for obvious vacuum leaks, all hoses are connected and appear to be in good shape.

I'm new to tuning Subarus, so there may be something obvious that I am not getting. But I don't understand how there could be such a serious problem specific to #4, with the others happy. I can't find anything wrong with #4. Since the compression and leak down test didn't reveal any problems, I presume that means I don't have to be worried about valves or head gaskets.

Sorry it ended up being a kind of long post, thanks for reading it all if you got this far. Any help would be appreciated.

Jason

It seems like there may be a wiring problem on #4 injector - if you, or a friend, have reasonable circuit building skills, you can fabricate a simple circuit testing tool with male and female injector fitting and an LED in parallel to the injector.

Failing that, try wiggling the #4 injector wiring, or even the whole engine loom where the wiring runs, with the engine idling and see if that affects the misfire. Also, while you're at it, check the pins/sockets at the injector and ECU haven't been pushed out of place - it's not uncommon and can cause all sorts of problems.

Looks like there is a damage of rubber cover for spark plug causing sparks firing to the ground instead of spark plug. I had similar issue on Mitsubishi Pajero and had to replace entire set of wires as spark plug number 4 was getting damagef and out of timing very quickly despite being changed for new one every time.

Jason,

First off, cylinder leakdown of 20-25% is extremely poor for that engine so I would review that and see if the result is valid or not. I rebuild Subaru engines when they get up over 10%.

For reference that engine has 0-2% leakdown when stock up to about 60k miles when cared for, well tuned. I've tested dozens of stock engines over 150k miles that are around 5-8%. At 20-25%, combustion is absolutely suffering and for a stock engine to hit 20%, you generally have a cracked piston, but with a built engine, it could be a lot more things.

How confident are you in the leakdown test results? Any chance the test wasn't performed correctly or the test gauge has an issue and leakdown is actually much better than it appears? That test gets performed with a regulated 100 psi input, so perhaps only 70-80 psi was used?

If you hadn't mentioned leakdown was that severe, I would have first said to put a good set of OEM coils in because the ones you have, have not worked out for Subarus I've seen with them.

Another very common issue specific to the 11-13 WRX is the ground wires in the main harness have poor crimps inside a section near the turbo, and the symptom is misfires. You have to cut a section of harness out near the turbo to resolve it. It's actually pretty simple, there's just a good bit of tar covering the wires that you have to pick off to get clean sections. I generally end up cutting the problem area out then splicing new sections in. Some have success soldering new sections in, but proper splices are the preferred method. The most common indication of this is the misfiring doesn't happen each start until the engine bay gets warmed. The issue progressively gets worse and then the car misfires all the time. This thread has good pictures of where the connections I'm talking about are and how to get at them, but the repair could certainly be done more cleanly than what's shown: https://forums.nasioc.com/forums/showthread.php?t=2550077

I'd also add a ground strap from the main intake manifold engine harness bolt point, to the negative battery terminal. I usually use 8 gauge for that.

Then I'd check both battery terminals are clean and secure on the battery.

The main engine harness connector up by the passenger strut tower isn't the greatest. Confirm the latch is all the way locked down and no wires appear to be pushed out the back side a little. The connector usually works fine til you have to use it...like when doing an engine job...and then sometimes they have issues when plugged back together.

It could be a combination of things. It could also be something I haven't mentioned, but I've listed all the most common causes on that particular car so they're worth checking before getting off into the weeds with less likely causes.

I hope something we've mentioned helps!

Hi Mike,

Dang, well it looks like I was misinformed as to what acceptable leak down is. My tester instructions actually say to run at regulated psi anywhere from 7 to 100, then it gets dialed in to zero before attaching the hose from the cylinder. I ran it at 60 for this test. I was also a little unclear as to what should be disconnected for this motor to get valid results, normally it is just radiator cap, oil dipstick, and PCV? I have an AOS on this motor, I disconnected the hose from that to the passenger valve cover, along with pulling the dipstick and radiator cap. Given that both cylinders 2 and 4 were getting the same readings, and 2 is not having misfires while 4 is, is that a strong indication that it was just my test procedure? I hope so.

You have given me a lot to go on now though, I'll revisit the leak down and then take a look at the rest of the list.

Thanks!

Jason

Jason,

I'd redo the leakdown at 100 psi and see how it goes. Also, it's possible perhaps you aren't quite on TDC during the test. Getting that just a bit off will make results suffer. I know that part is tricky, so if you get somewhat close with a ratchet, then perhaps switch to a breaker bar and move the crank the last little bit with tiny bumps to see if you find a spot where the results are significantly better, just be careful as the engine will try to spin as you get past TDC.

Disconnecting items on a leakdown test isn't required to get a proper result, but if the engine tests poorly, then you can open up the crankcase by unscrewing the oil cap to see how much of the leakage seems to be past the rings, cap the exhaust off, pull an o2 sensor out and see if there's significant flow past exhaust valves, feel at the throttle body for intake valve leaks, to get some idea if the leakage is mostly in the heads or block. You'll almost always feel some crankcase flow during a leakdown test, but if it's a lot, that's not a good sign. On a great engine you won't feel anything in the intake or exhaust.

I missed the leak-down test - Mike's spot on that that's WAY higher than an acceptable level, especially on a "performance" engine.

While sorting out the re-testing, I'd suggest picking up one of the cheap boroscopes that work with smart 'phones to have a good look at the piston crowns (all of them) and what condition the bores are in.

Well, either this leak down tester is a cheap piece of garbage, or there is something really wrong with the motor. It doesn't seem to matter what psi, what engine temperature, or how close to TDC, it ends up at ~ 20%. But, it is quite consistent at that reading, and both #4 and #2 read the same. I don't understand how both cylinders can be at the same level, but #4 is constantly misfiring and #2 is totally fine.

I followed your suggestion about TDC by setting the leak down at 0, then connecting to the cylinder at some random rotation, typically it would read very high > 80%. Then, I turned the engine over slowly, and I could hear the intake valve opening as air in the intake track, and eventually that sound would go away. Shortly afterward, the leak down % would minimize and stay at the same level for a fair amount of rotation. I think that all makes sense to me. Once at the minimal point, I would disconnect the hose and re-zero if necessary (but usually not), and then re-connect and pretty much all the time the best was 20%.

When at 20%, the hissing is audible in the oil fill tube, but not anywhere else.

tester example

I took some data during a cold start while running the engine for a warm leak down measurement, including that for reference.

cold start log

I have the RPM set at 1000, not sure why it ran down to 700 before recovering, but it hadn't been through a throttle re-learn. I don't typically idle the engine to warmed up without ever touching the throttle.

Is it worth going down those other avenues, or do I need to get a better leak down tester, or is this motor just done? That would obviously really suck.

Thanks,

Jason

Jason,

Depending on the particular engine issue, some are far more prone to causing misfiring. For example on a Subaru that's cracked a ringland, you often get leakdown in that range while the engine continues running happily, just making a bit less power, until the piston totally comes apart. On the other hand if a valve is significantly leaking, or oil or water is getting into a cylinder, it's far more likely to misfire because combustion gets more disrupted.

Because you're not confident in the leakdown tester, I feel buying or borrowing a known good tester seems a good next step before you think further down the path of the engine potentially actually being damaged. Hopefully you get a good tester, results are acceptable, and you can work your way through the other ideas to diagnose this.

Ok, do you have a recommendation for a good tester? You can probably tell leak down testing is new to me. Just browsing on summit, it's clear that I have a very cheap tester.

Thanks,

Jason

I have a SnapOn one but I remember one of my old techs got the Mac Tools one and it worked great at a lower cost maybe 200 vs 300 something I paid.

lonagacre makes one I’ve not used, but I have a few of their tire gauges and they’re pretty well respected, so it’s likely a decent tool at a lower price point. It looks to be around 140 bucks from Pegasus Auto Racing.

Check the fittings included with whichever you pick. You may need an adapter hose in which case make sure the adapter hose works well or it can be the reason the test isn’t accurate.

I actually have some old craftsman adapter hoses that work fine. I just replace the o rings periodically.

You keep mentioning #2 & 4 cylinders, but I can't see any mention of #1 & 3?

What were/are the compression test results and leak-down results for ALL the cylinders?

The impression I get is that either you didn't check them, or they were a LOT better than 2 & 4?

You need ALL test results and, while usually within 10% is cited as the tolerable range, with a freshly run in re-built engine they should be much closer than that,

WTF? Tried that last post twice and it's not posting most of what I typed?

If 1 &3 are consistently 'good', and 2 & 4 are 'bad', that's NOT a tester problem.

Leakage into the crankcase is a PISTON and/or BORE damage problem - just get a boroscope and check that.

Personally, I wouldn't bother - it's engine out time, with new pistons and a rebore to be expected. That still means double checking the fueling and ignition side as they may have been significant contributors to this... Ignition misfire means unburned fuel and bore wash, mis-firing/lean injector means possible pre-ignition and/or detonation (possibly confirmed by spark plug condition).

Ok, so the real reason I could not find anything wrong with #4 is ... I was looking at #3. Somehow this is not a SBC. I was a little suspicious when I could not find the other 4 cylinders, but in too much of a hurry to worry about that I guess. I'm sure I saw at some point that cylinder #1 is on the passenger side, but forgot about it.

Very sorry to have wasted your time, but I really appreciate the leak down testing help, and general Subie troubleshooting knowledge.

It did end up being a bad coil. After swapping and seeing the issue switch to #2, I put the old ones back on and it is fine now. Looks like you were right about these coils I bought, one died already. I will get some new OEM coils.

Here are the new leak down results, let me know if I may still be doing something wrong. I purchased a Longacer tester, LNG-73014 from Summit. Now, all four of the cylinders are reading 0%. I was very skeptical, so I measured each one a bunch of times, always with the same result. My procedure was, leave the gauge disconnected and turn the crank over until hearing the pressure from the compression stroke come out of the hose, then stop when that dissipated. The instructions said, just like Mike, that the pressure would be 100 psi, and needs to have >100 regulated psi supplied to get valid results. The instructions for this tester also said it could be either TDC, or BDC, but not in between because the pressure can cause the crank to turn suddenly. I tried to hit TDC as best I could and didn't have any problems. Then, after connecting the cylinder hose to the tester, use the regulator on the tester to get 100% at the left gauge and the right gauge gives leak down. It always went down to 0 at 100% for each cylinder. Just to see if I could get a different result, I turned the crank over for each cylinder over a few more twists of the bar and re-tested. I would get between 50-70% and could clearly hear hissing at the exhaust tips. So, I really hope these results are now valid. The engine is running the same as it was before having this problem so I am very happy about that. It's also nice to have the 1050's in, at least that part has gone smoothly so far.

The next thing I was going to try, was capture the injector pulses on a scope. I picked up a couple of injector connector extensions and exposed a part of the wire for the #2 pin on the injector connector. When I was verifying the pin on the injector, I realized my stupid mistake about the cylinder #'s. I was going to ask, even though I don't really need to do this right now, is it ok to connect a 1 MOhm probe to the injector wire like that? Or could it cause the pulse to get messed up or hurt the ECU?

Thanks again,

Jason

That's all great news Jason. Yup I totally believe 0% leakdown and your test method sounds fine except I would test at TDC for consistency, then if you still think something is wrong you can test at the bottom to see if the bores were done properly all the way down.

I've scoped injectors many times, never hurt something.

Thanks Mike, and thanks everyone for the suggestions and info.