Running good only in D-Check mode - What's going on!?!

n2x4 · Post by **n2x4** » Sat Jun 27, 2009 10:02 pm

My EJ22T swap in my Impreza developed a cutting out issue under WOT. I can run 3/4 throttle fine, but when I go WOT, it bogs out. I recorded videos here:

http://www.youtube.com/watch?v=pWUkJNTFumg

In the first video, you can tell when I go WOT with the "thunk" of the pedal hitting the floor. The RPM's drop. At the end of the video, I hold it to the floor, and never let off, thus killing the engine.

http://www.youtube.com/watch?v=l0OV-w2GEtA

In the second video, you can see my scantool voltage reading on the TPS. Everything looks ok there.

I replaced the MAF and TPS sensors, just to see if they were to blame. Not the case, still had the exact issue.

I had some stored codes that I tried to clear, so I connected the D-Check connector (green one). With it hooked up, the car ran great!

What does the D-Check mode do differently that could cause the issue to go away? I'm very confused.

Basically, my testing shows that with d check connected, the car runs fine and the cel blinks constant (all clear signal).
Disconnected, I have my throttle issue as shown in the video. Any ideas?

n2x4 · Post by **n2x4** » Sat Jun 27, 2009 11:54 pm

Drove the car in d-check mode. This time I got 2 codes-

35 for Canister Purge (not surprised, I think this solenoid goes bad on almost all Legacy's)

24 for IAC. This one seems strange to me. This code has come and gone several times during my swap. It started with a super high 3K rpm idle and now it's died back down to normal, I don't understand what the problem is. I guess I'm experiencing the symptoms of "Failsafe Operation" prevents abnormal engine speed using "fuel cut" in relation to engine speed, vehicle speed, and throttle sensor position

Driving in D-Check mode is interesting. When I'm at WOT, the CEL light stays illuminated. When I drive normally, it flashes the codes. It still drives fantastic. Why is that?

Post by **Legacy777** » Mon Jun 29, 2009 12:26 am

I'm assuming you're using the stock ECU?

Have you double checked all your wiring? Do you get those codes to come up with just the black connectors?

Have you tried resetting the ECU?

log1call · Post by **log1call** » Mon Jun 29, 2009 1:55 am

Hmm, It's pretty hard to say what is wrong.... given that it has a had a swap done.

The motor shouldn't run any different when you put the green connectors together. There is something wrong there. I'd suspect wiring to and from the ecu.

The canister purge could be a pipe off, the canister is full of fuel, has a crack, or the solenoid isn't working.

The IAC code could be a pipe off.

The cutting out could be caused by a pipe being off.

I'd check all your vacuum pipes are routed and connected correctly.

The strange behaviour with the green plugs I'd check seperatly.

n2x4 · Post by **n2x4** » Mon Jun 29, 2009 3:52 am

Legacy777 wrote:I'm assuming you're using the stock ECU?

Have you double checked all your wiring? Do you get those codes to come up with just the black connectors?

Have you tried resetting the ECU?

Stock ECU. Black connectors do show these codes, though I've yet to try and drive the car with black connectors connected so I'm not sure how the car behaves in normal code reading mode. I haven't tried to reset the ECU. So far my attempts were unsuccessful. I disconnected the battery for a few hours, and that didn't work. I'll try it again and then see if I can get any luck using the connector method.

log1call wrote: Hmm, It's pretty hard to say what is wrong.... given that it has a had a swap done.

The motor shouldn't run any different when you put the green connectors together. There is something wrong there. I'd suspect wiring to and from the ecu.

The canister purge could be a pipe off, the canister is full of fuel, has a crack, or the solenoid isn't working.

The IAC code could be a pipe off.

The cutting out could be caused by a pipe being off.

I'd check all your vacuum pipes are routed and connected correctly.

The strange behaviour with the green plugs I'd check seperatly.

I think the issue is either a bad IAC and the ecu runs in some "failsafe" mode when in d-check, regardless of the sensor value or it's a wiring issue.

I'm positive there are no leaks, and no lack of fuel pressure. I've tested for those already several different ways.

I'm ready to accept that the issue may be electrical. I know that if I did the swap, I have to accept the consequences

I just needed to find out if the ECU behaved any differently when in D-check. If not, then I will try and track down electrical issues - I know I can't expect you guys to begin to guess where to start on the wiring - lol. Just know that all connections were soldiered and heat shrink was used on every wire.

Looking at the FSM, I see that connecting the Orange connectors is just grounding a pin at the ECU. I'm not sure how grounding a pin on the ECU would help the issue at this point.

I appreciate all the input so far. I think I need to get another IAC installed just to rule out the possibility of a faulty valve. Then with that either fixing the problem or ruling it out, I can try to trace wires. Before this - the swap's been running great for a few months!

vrg3 · Post by **vrg3** » Mon Jun 29, 2009 3:50 pm

I am pretty sure this problem will go away when you fix the problems you've already identified. As you've noted, the ECU does have a failsafe mode in which it cuts fuel. When active, it shuts the injectors off if the throttle is open above a certain percentage. I know it activates this mode when it detects a problem with boost control or pressure sensing, and I'm guessing it's doing it also because of the IAC valve.

The ECU does do some things differently when in D-check mode. My guess is that, in order to allow the other systems to be checked, it doesn't cut fuel then. Maybe that's why it lights the CEL steadily when it would otherwise be cutting fuel? An interesting thing I've noticed is that at least sometimes my ECU does a similar throttle-based fuel cut in D-check mode when there are no codes. Silly ECU.

Anyway. You've already done the detective work you need. Go ahead and replace the purge solenoid since that's easy. You don't need another OEM valve; there are scores of other solenoids that'll work. You just need it to be a 12-volt-operated normally closed vacuum solenoid that doesn't ground through its case.

The IAC valve problem I think is unlikely to be related to the plumbing. You could try swapping out the valve like you say, but you might consider first quickly going over the wiring. If it's an intermittent problem it could be moisture getting into the connection or something, and that can be fixed with a little dielectric grease.

You probably won't have to replace the entire IAC valve in any case. The little portion on top, held in with two Philips-head screws, is the solenoid. If you mark its position you can then remove it and swap one over from the IAC valve of any first-generation Legacy except a 90-91 manual non-turbo. That should make things easier since you won't have to locate the turbo-specific valve or separate the valve from the coolant lines or intake manifold.

n2x4 · Post by **n2x4** » Mon Jun 29, 2009 6:46 pm

vrg3 wrote:I am pretty sure this problem will go away when you fix the problems you've already identified. As you've noted, the ECU does have a failsafe mode in which it cuts fuel. When active, it shuts the injectors off if the throttle is open above a certain percentage. I know it activates this mode when it detects a problem with boost control or pressure sensing, and I'm guessing it's doing it also because of the IAC valve.

The ECU does do some things differently when in D-check mode. My guess is that, in order to allow the other systems to be checked, it doesn't cut fuel then. Maybe that's why it lights the CEL steadily when it would otherwise be cutting fuel? An interesting thing I've noticed is that at least sometimes my ECU does a similar throttle-based fuel cut in D-check mode when there are no codes. Silly ECU.

Anyway. You've already done the detective work you need. Go ahead and replace the purge solenoid since that's easy. You don't need another OEM valve; there are scores of other solenoids that'll work. You just need it to be a 12-volt-operated normally closed vacuum solenoid that doesn't ground through its case.

The IAC valve problem I think is unlikely to be related to the plumbing. You could try swapping out the valve like you say, but you might consider first quickly going over the wiring. If it's an intermittent problem it could be moisture getting into the connection or something, and that can be fixed with a little dielectric grease.

You probably won't have to replace the entire IAC valve in any case. The little portion on top, held in with two Philips-head screws, is the solenoid. If you mark its position you can then remove it and swap one over from the IAC valve of any first-generation Legacy except a 90-91 manual non-turbo. That should make things easier since you won't have to locate the turbo-specific valve or separate the valve from the coolant lines or intake manifold.

I do have a spare IAC solenoid from a 92 automatic car - I think I'll give that a shot. I'm pretty confident in my wiring, though I could recheck my grounds once I've ruled out the solenoid.

I was doing some searching on the rslibertyclub site as well, and I see some guys have experienced similar issues with the Neutral switch in the transmissions. If there's a bad signal, the ecu can freak out when it goes back an forth from registering in gear and in neutral. I did notice the signal flutter once when using the scan tool. Again, another easy thing to check (transmission only has 50K miles on it so I doubt it's the sensor - more apt to believe a bad wiring connection in this instance).

I'm convinced that D-check mode does handle things a little bit differently. While I've yet to find and solid documentation on this "theory" in any FSM reference, it does make sense. Kind of nice for a backup - I don't do much with this car other than Auto-X and weekend drive, so still can take it out and run WOT once in a while:twisted:

I think I'll address the code 35 last, as mentioned it's an easy fix and one that I'm not too concerned about at this point. It's just a PITA to get to

Thanks again for all the feedback!

vrg3 · Post by **vrg3** » Tue Jun 30, 2009 3:33 am

If the neutral switch signal is unsteady it's probably the switch going bad. What you can do to get a little more life out of the switch is to remove it and replace the crush washer that seals it against the transmission housing with a thinner one. I don't think that's what's causing your problem, though, because it ought to only cause little hiccups unless the ECU was actually throwing a neutral switch trouble code.

If you have the other IAC valve handy, then, yeah, totally do that first. It's an easy swap.

I think it's unlikely to be a bad ground. The IAC valve is grounded through the ECU by the ECU's output circuit.

The purge solenoid can be hard to get to, but you don't have to remove it if you don't want to. Just install a new one somewhere else.

log1call · Post by **log1call** » Wed Jul 01, 2009 1:32 am

I don't think that the diagnostic mod does anything to the motor. It just makes it set trouble codes after one instance of a fault instead of several.
The limp mode is something different. The ecu sets limp mode if there is a fault likely to cause damage to the motor.
These characteristics, (and diagnostic mode for that matter), aren't exclusive to subaru, most modern cars have the same facilities.

I'm a mechanic and would like to know if there is some change in the running of the motor while I am trying to diagnose it, so.... could you tell me what it is you think the ecu does in diagnostic mode so I can check it out please?

vrg3 · Post by **vrg3** » Wed Jul 01, 2009 3:38 am

On what are you basing that supposition about diagnostic mode (by which you mean D-check mode I guess?) changing the sensitivity to number of instances? That is not how our ECUs work.

Also, EJ22T ECUs do not have a limp mode beyond the fuel injector cut we've described here.

I don't have a good understanding of what the ECU (not the motor) does differently in D-check mode, but it does at least what I've described above.

log1call · Post by **log1call** » Thu Jul 02, 2009 1:57 am

Hi,
I'm a mechanic and I have read lots of manufacturer's manuals and training manuals, for subaru, toyota, nissan, honda... if they have a diagnostic mode it makes them set codes with less instances of a fault, whether its fifty missfires instead of one-hundred or one instance of a MAF plug going open circuit instead of three instances in one drive cycle. They all work the same.

If they put the engine into limp mode it would be a disadvantage to the diagnostic process, not a help.

log1call · Post by **log1call** » Thu Jul 02, 2009 2:03 am

All ecus have a variety of limp modes. If they are a bit too hot they speed up the idle and advance the timing, if the knock sensor isn't working they don't allow all the advance factors to be added to base timing, things like that. The 2.2 T won't be any different. Depending on the fault, and the possible damage it could cause, the limp mode varies.

log1call · Post by **log1call** » Thu Jul 02, 2009 2:27 am

The codes that get set are not nessecarily indicating a faulty component. They are only an aid to diagnosis. If you read the workshop and training manuals and you will see it is so, if you get trained as a mechanic you are taught it is so.

The IAC code can be set because there is an air leak and the valve has gone right to the end of it's travel and still can't get the revs down. The O2 sensor code can be set because the mixture is too rich or too lean and the ecu can't adjust the IPW enough to get the mixture within an acceptable range.

The diagnostic part is where you realise that the two or more codes are related and that there is only one explanation that fits all the circumstances.

Sometimes there are two faults, then we have to recognise that and isolate them and deal with the one that could cause the other... first.

log1call · Post by **log1call** » Thu Jul 02, 2009 3:07 am

I have long been an admirer of your work connecting laptops to ecus Vrg3, and I mean no disrepect, but, what use do you think it would be to have a diagnostic mode change the tune of the motor? Why would they do that?

I am imagining that you are basing your statements on your interpertation of ecu programing. So what do you understand diagnostic mode does?

These are genuine quiestions, asked in good faith. I would like to know if there is some change of tune when in diagnostic mode.

I have been a mechanic since before computers were ever in cars at all... even the simplist of them controlling emissions out of breathers. I have followed all the developments in ecu use, I am familiar with all of their operating principles although not their specific combination of features for every individual model. In my experiance and as far as I know, no matter what make of ecu, car, protocol or standard, they all interact with cars, from a mechanics point of view, in the same way. I know from following your work Vrg3 that there are many different ways the ecu processes the data it receives, but, the basic behaviour/relationship between ecu and engine is always the same no matter what system is being used, whether it's the latest can system or the oldest manufacturers priority system. The obd2 systems make more of the logical conclusions(aided by more sensors), that mechanics used to have to arrive at by their own reasoning, but the old systems can still be used to derive nearly as accurate diagnosis's.

Any other mechanics out there agree?

vrg3 · Post by **vrg3** » Thu Jul 02, 2009 3:53 pm

Wow, calm down.

I don't think you're correct in your assumptions. Your choice of career and reading material notwithstanding, one device made by one manufacturer exhibiting a certain behavior doesn't imply that a similar device made by another one would exhibit the same behavior. If you have some information to offer that's based on either experience with this model or based on documentation about this model, that would be useful.

In my experience, the EJ22T ECU throws a trouble code the instant a fault is detected, regardless of whether it is in U-Check mode or D-Check mode. This is consistent with the wording in the Subaru factory service manual.

In my experience, the EJ22T ECU has only one limp mode, and that is the injector cutoff I've described here. This is not consistent with the factory service manual, but it seems that the Hitachi ECU is not quite capable of as much as the JECS ECU that most models covered by that manual have. It does have other failsafe modes (leaving the radiator fan on at all times, or sticking to open-loop fueling or base timing), but I don't believe it would be accurate to call them limp modes. It sounds like we don't exactly disagree on this except perhaps for the last point.

In my experience, the EJ22T ECU cannot be made to throw an IAC valve trouble code because of an air leak. It will do what it can to make the engine idle, and if it fails the engine will simply stall with no trouble code thrown. The IAC valve trouble code is thrown when the ECU detects an electrical problem with the IAC valve. The factory service manual is a bit vague about this one, but it seems consistent to me -- it says the code is thrown when the "monitor circuit" detects an abnormal signal.

In my experience, the EJ22T ECU does not throw an oxygen sensor trouble code when it can't get the fuel mixture right. It throws a different trouble code -- code 41, for fuel trims. This is backed up by the factory service manual.

As I have already explained in this thread, in my experience, the EJ22T ECU absolutely does change how it operates when it is in D-Check mode. It cuts fuel based on throttle position, throws no trouble codes, and stops doing it once I put it back in U-Check mode. n2x4 has also now observed that his ECU behaves differently in D-Check mode as well. I have also offered, in this very thread, a possible explanation for why it might be desirable for the ECU to behave differently in diagnostic mode.

Please stop trying to pull rank; who you are doesn't make what you say any more or less true. Please stop making assumptions; what is true for some or even most other cars may not be true for this car.

asc_up · Post by **asc_up** » Thu Jul 02, 2009 9:06 pm

My friend swapped an EJ20G into his GC8 awhile back. He had the exact same symptoms, but it would run fine in D-check mode.

We found that part of the engine harness had slipped down and was being cut by the tire. I believe the wire that was cut was the MAF signal wire (no MAF code was thrown). So I'd just go through and check all your harnesses to make sure there aren't any shorts occurring.

n2x4 · Post by **n2x4** » Thu Jul 02, 2009 9:42 pm

asc_up wrote:My friend swapped an EJ20G into his GC8 awhile back. He had the exact same symptoms, but it would run fine in D-check mode.

We found that part of the engine harness had slipped down and was being cut by the tire. I believe the wire that was cut was the MAF signal wire (no MAF code was thrown). So I'd just go through and check all your harnesses to make sure there aren't any shorts occurring.

Did you guys get the chance to run the scantool on the engine when this issue was occuring? I'd be curious to see if the ecu got a reading from the sensor with the problem. My MAF readings look fine. But with that thought, I'll check the wires and test to see if the car runs in D-check mode with the MAF completely disconnected...

I have off tomorrow so I can dedicate the day to troubleshooting

It's interesting to see the two schools of thought produced in this thread. One thing's for sure: D-check does some crazy stuff.

asc_up · Post by **asc_up** » Thu Jul 02, 2009 10:03 pm

Unfortunately we didn't. I do remember checking everything with a multimeter and everything SEEMED to be ok, though. The wire wasn't completely cut though, it just had the coating worn away, that way it would appear to be ok when checking with the multimeter but when moving I think it would vibrate and touch the chassis which caused it to ground out.

ericem · Post by **ericem** » Thu Jul 02, 2009 10:45 pm

log1call wrote:All ecus have a variety of limp modes. If they are a bit too hot they speed up the idle and advance the timing, if the knock sensor isn't working they don't allow all the advance factors to be added to base timing, things like that. The 2.2 T won't be any different. Depending on the fault, and the possible damage it could cause, the limp mode varies.

This is exactly what i wanted to know and definitly explains the relation between high idle and fans not shutting off at operating temperature.

log1call · Post by **log1call** » Fri Jul 03, 2009 3:28 am

Ok... I didn't realise that questioning would cause such offence.

I will defer to those with more knowledge on the subject and look on from the side-line in the hope that I might learn about these things.

For anyone trying to learn about these systems(or any of the systems in use on your car), I would recommend finding training manuals if they can. They have explanations of how the system works, as opposed to the workshop manuals which tend to gloss over the theory in favour of specific results to be expected from the system.

n2x4 · Post by **n2x4** » Fri Jul 03, 2009 9:19 pm

Ok, I've spent the afternoon troubleshooting the wagon-

New IAC solenoid, new knock sensor, rechecked the MAF sensor wires and repaired a cracked wire.

Also discovered that the coilpack had a crack on the side of it - replaced this for another good one.

Same results, same codes. I tried to clear the codes with both connectors connected, and it wouldn't clear them. Apparently if there's one code that's still active it won't clear the non existent one?

At this point, the only thing left to do is pull the harness from the car and check it again. I'm out of ideas. I think I'll race it in D-check mode and go from there.

vrg3 · Post by **vrg3** » Sat Jul 04, 2009 12:37 am

Hmm.

You can make the CPC code go away by going to Radio Shack, buying the 50-ohm 10-watt resistor, part 271-133 (http://www.radioshack.com/product/index ... Id=2062292), and wiring it between the ECU's purge control pin and +12v. Maybe that'll help convince the ECU to clear the codes.

log1call · Post by **log1call** » Sat Jul 04, 2009 12:49 am

Some trouble codes can be set when there is nothing wrong with the sensor/actuator.

The trouble codes are in a certain order so that we can avoid trying to fix non-existant problems.

The codes are arranged in an order such that you cure the problem that could cause some un-related code first. This is an industry wide policy that has been in place since they first started have self diagnosis.

You are meant to check/cure the lowest numbered code first, then clear the ecu, then go for a drive in diagnostic mode before moving on to the next code(assumng there is one).

Start with the lowest code, check the sensor as per the book, follow all the instructions in the manual and be thorough at each step of the way. Don't try to skip some step of the diagnostic procedure. When you are SURE that component is good, and that it's signal is getting into the ecu, clear the memory and go for a drive in diagnostic mode.

The arrangement of the codes in order is not fool-proof, there are some codes that are in a lower place in the scale because either they are more likely to be the cause or just that there are two or more likely condidates for the position depending on the specific complaint.

log1call · Post by **log1call** » Sat Jul 04, 2009 12:56 am

I know I was going to sit this out but I should also mention...
I pull ecus out and have a look inside them if the symptoms or codes don't make sense. They can get leaking capacitors, burnt resistors, exploded transistors, cracked circuit boards, bad solder joints or water damage.

Given that it is a simple job tp get the ecu out, I do it fairly early on in the diagnosis if the car is a few years old.

Once again, if you are going to have a look at the ecu... be thorough, use a magnifying glass, examine EVERY component and solder join.

log1call · Post by **log1call** » Sat Jul 04, 2009 1:08 am

I just remembered too, there is a theory that a faulty main relay can set codes 24 and 35. If you go to...

http://cid-4ca3c3459aaa7f7f.skydrive.li ... ublic?uc=1

and download the "basic diagnostic", file you will see on the last page there is a mention of it.

There are also a couple of diagnostic pdf files there,(04-dtc criteria and Diagnostics) if you have a look through them you will see what it takes to set trouble codes... the number of incidents of a intermittent/momentary fault.

legacycentral bbs

A forum for 89-94 BC-BF(BJ) Legacy Owners and Fans

Running good only in D-Check mode - What's going on!?!

Running good only in D-Check mode - What's going on!?!