the dreaded throttle control error codes

[FFRWRX]

This is going to be a bit of a long one. Do others have a bunch of codes that are always there and you just ignore?

This is a 2006 WRX drivetrain (drive by wire).

I've got these error codes, and the car starts/runs/drive fine (most of the time, which I'll get to):

P1518 – starter switch circuit low input
P0852 - Park/neutral switch input circuit high
P1152 - O2 sensor - but I did replace it
P1153 - O2 sensor - but I did replace it
P0113 - intake air temperature sensor 1 circuit high
P0502 - vehicle speed sensor A circuit low
P0503 - vehicle speed sensor A intermittent

BUT...I bought this 818C partially assembled by a previous owner, and I could see the gas pedal was replaced, so an indication that there might have been an issue. While working on the car, now and then when starting it the gas pedal would not respond. I would turn it off and back on and that generally cured it. Now the car is done and I'm driving it, it will go into a limp home mode now and then; won't go above 3,000 RPM. With the throttle floored it will go up to 3,000 RPM and then bounce there as if hitting a rev limiter. I would turn it off and back on, sometimes a number of times, and then it would be OK for the rest of the drive. The additional error codes were then:

P2127 - throttle/pedal position sensor/switch E circuit low
P2138 - throttle/pedal position sensor/switch D/E voltage correlation

I replaced the throttle pedal with a new one, cleaned all the connectors (throttle pedal, throttle body, throttle control relay, ECU) but it still occurred pretty much every drive. Yesterday it did it again on a drive, but wouldn't recover this time. Multiple times shutting it off and restarting and still would not go above 3,000 RPM. Checked the codes and the P2127 and P2138 were gone, but this new one was there:

P2122 - throttle/pedal position sensor/switch D circuit low

Cleared the codes and tried it again....no luck. So in a way it is good that it has gone from an intermittent issue to a full-time one. Intermittent issues are always a pain since you never know if you've solved it.

I removed the throttle body, cleaned it internally and the connectors again, but that didn't help.

I checked the real time data on my code reader and it showed the throttle position as: 14.1% at idle, 19.6% when floored

The factory service manual gives a lot of trouble-shooting but it is all to do with the harness. Is there a way to check the throttle body itself? Haven't found anything on-line. I have no issue with replacing it if that will solve this issue. But I really don't want to buy another one only to find that isn't the issue. New ones are of course very expensive, but there are "remanufactured/refurbished" ones on eBay for around $200.

I measured around 5 volts at three of the throttle body connector pins with the ignition on. The manual says it should be 10V, but I've also read that 5V is correct.

Any suggestions or help?

Rick

[driveslikejehu]

Not sure if you mentioned this... Do you have a throttle stop? My first hard drive I got some of those codes, I think, and it went into limp home mode.
The DBW throttle needs a stop or it will error. Bob-in-Cincy has a post about it

[Ajzride]

The other thing to run to ground before you spend any money is.... grounding. This car is super sensitive to good grounding. I built a grounding circuit on mine instead of relying on the frame for good grounds.

[RPGs818SNA]

I second ajzride's post regarding good grounding and the symptoms you are seeing. The welded frame is fine for grounding as long as you have good connections to it. I found quarter inch fine threaded bolts tapped into the frame with aluminum antisieze effective. And like ajzride, I also ran ground wires between all the ground points for redundancy. Here's what overkill looks like.

188869

RPG

[FFRWRX]

I appreciate the quick help.

On the throttle stop...not sure about that. I've seen 2 different designs of pedals. Mine is the one that is completely enclosed with Hall effect sensors, or something like that, no copper traces. The pedal has a hard stop between 2 of the parts and it is pretty thick, hard plastic. I don't see how it could go past that. Here is a picture of it:

188870

On the grounding...yes, that is a possibility. I have grounded every ground connection I've found, but yes, there are a ton of them. I'll have another good look around and clean and re-do the existing grounds. I think I read somewhere about an intake manifold ground, so I'll check for that.

Rick

[lpmagruder]

When I was altering my factory harness, I had found the dbw lines were frustratingly close to the right length, but was like an inch short with how I had components laid out. I was about to clip the wires and extend them, but I checked the wiring diagram and noticed one of the wires (on my car, it was a white wire, pin B23 on ECU) was a shielded wire. So I chose not to mess with that and re-arranged a bunch of stuff to make it work at its original length.

Check and see if the previous owner extended that wire. If they extended it with just a solid wire, or had some intermittent short between the shield and conductor, I could see it causing issues.

[FFRWRX]

Looking at the wiring diagram, B23 goes to the pedal and that wire has not been modified in my car/harness. I did do quite an extensive wire diet, but left ones like that alone.

I took the IC off this morning to check the throttle plate opening. Of course it was working fine now. I put a small mirror near it that I could see from the drivers seat. Ignition on and had full movement of the throttle plate; confirmed with the real time data on throttle movement %. Put things back together and went for a drive. Everything worked well......for about 5 minutes, then limp-home mode again. Clipped a large ground from the TB to the frame but of course that didn't help. Don't know what else to try. The grounds look good. I'll check the wiring diagram again for grounds to do with the throttle system.

Funny thing is that when I got home and did some testing, the limp-home went from its usual 3,000 RPM to almost 4,000 RPM. This is just flooring it in neutral. I'll see about measuring the voltage at the TB wires that control the throttle plate motor. If that voltage is low when the limp-home thing is happening then I'll know it isn't the throttle body....maybe? Then again, if it is getting the proper voltage but the throttle plate movement isn't what it should be, maybe there is some feedback system to lower the voltage to limp-home. So then it could be the throttle body. The conclusion.....I don't have a clue!

[FFRWRX]

...OK, I borrowed your line.

Just to confirm. This is the engine grounding info from the wiring diagram:

188891

If I'm understanding this, terminals C15 and A5 at the ECM both go through a number of connectors and then to a large grounding tab. Would there be any harm in me adding an additional wire from those 2 connections, right near the ECM and grounding them? Likewise I could do that with D3 and D7; but why don't they ground those internally with the others? Why run an additional wire to ground? What if I connected all 4 together at the ECM and grounded them? Will I be creating some feedback between them that shouldn't be there? But they are all essentially connected together now, no?

[Ajzride]

Yes, you can connect all four of them to a giant ground lug right at the ECM. I used a grounding bus bar with the large left lug to the frame and the large right lug to the battery:

https://www.amazon.com/gp/product/B0091VHLW4/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&th=1

I have one of these grounding bars in the front bay, under the dash, behind the seats where the ECM is, and one in the engine bay. All of them go one side to the frame, and the other side daisy chains through each other back to the battery.

[RPGs818SNA]

It's ok to add a ground to an existing ground, Rick, but bad form to combine grounds on a single wire that weren't designed that way. A ground wire has some resistance, especially if it goes through several connectors. So whatever current goes through that wire to ground raises the voltage at the source. For instance if the wire has a 0.01 ohm resistance, a 10 amp current will raise the non-ground end to 0.1 volts. Normally that doesn't matter, but when the current is changing rapidly, it generates a varying voltage which can confuse some sensitive circuits.

It might be worth your time to check the voltage on all the ECM ground pins when the engine is running to make sure they are all adequately grounded. Compare them to battery ground to see if there is an outlier.

Good luck with all this. And remember that sometimes signal wires are flaky too.

RPG

[FFRWRX]

Been chipping away at things. I put some dedicated ground leads from the ECM (C15 & A5) to the frame right near the ECM. Went for a short drive, not expecting the issue to be solved, so I wasn't disappointed when it wasn't.

I started going through some of the other error codes I have. I disconnected the MAF plug and found that one of the wires had come out of the connector. It was the one at the end that goes to ground. So I repaired that connection and branched another lead off it to my dedicated ground. That eliminated the one code (P0113 - air intake temp) so that was good news.

I have an error code for the neutral switch, so took a look at that. Not too easy to get at (had to take some of the shifter mechanism off the tranny), but removed the sensor to test it. Seems like it is not working. With an ohm meter connected to the 2 leads, pressing the "button" that goes into the tranny doesn't do anything. So I ordered a new one; not too expensive and should have it tomorrow.

I also have 2 error codes for the vehicle speed sensor. The harness is supplying 12V and one of the 3 leads indirectly connects to ground. I ran an additional branched lead from that wire to a good frame ground. Still have the 2 codes, but just with clearing them and scanning again, not driving the car. The shifter stuff is still apart until I get the new neutral switch so can't drive the car. The speedo works fine, so does that mean the speed sensor is OK? Then why the error codes? I may replace that one too, but can't get one quickly, has to be ordered.

So no conclusion yet, but just checking out what I can and fixing what I find needs fixing.

As for combining grounds, not sure if that could cause an issue or not. Most of the factory trouble-shooting indicates that grounds should have a resistance of 1 ohm or less (from source to ground point). I tested a piece of wire that I had from my wire diet. This was about 5' of wire, a connector, and another 5' of wire. I measured .3-.4 ohms. I assume a short dedicated ground lead would have almost 0. If the factory manual allows for 1 ohm or less, it must be OK with zero. So the system must be designed to handle 0-1 ohm on each ground lead. As for combining them, they are already all combined by attaching to the car frame. There are also a number of ground connections that go to a common connector that is then bolted to the frame. I'm not saying you are wrong by any means, I suppose I'm just convincing myself that what I've done is OK.

Rick

Edit: Neutral switch replaced and one more error code gone. Limp-home mode still there. I should have the vehicle speed sensor on Wednesday so hopefully another couple codes will be gone.

[FFRWRX]

Initially I had these codes:

P1518 – starter switch circuit low input
P0852 - Park/neutral switch input circuit high
P1152 - O2 sensor
P1153 - O2 sensor
P0113 - intake air temperature sensor 1 circuit high
P0502 - vehicle speed sensor A circuit low
P0503 - vehicle speed sensor A intermittent

Plus the 2 connected with the throttle when the car went into limp-home mode; P2127 and P2138.

Fixed the broken wire for the intake sensor. Replaced the Neutral switch. 2 codes gone.

Without the car going into limp-home mode, I had 5 error codes left. I replaced the speed sensor today, ran the codes again, and.........no codes. No codes! They were all cleared. I went for a drive and for the first time ever the check engine light was not on. When I was almost home it went into limp-home mode and the check engine light came on.

Got home and checked the codes again. All 7 codes were back. The intake air temp one was gone since I did take care of that one. But all the others, including the 2 throttle ones were back (including the neutral switch one).

I'm pretty sure that the new neutral switch and the speed sensor did not suddenly crap out at the same time. So there is some common factor that triggered all these error codes. I know, you are going to say a bad ground somewhere. I'll hunt through the wiring diagram again, but don't remember seeing that all these circuits have something in common. And something in common with the starter wiring and the O2 sensors? Can't see it, but it must be something along those lines.

When I cleared the codes after getting home and scanning again, they all came back, so something that came connected/disconnected when driving was still connected/disconnected. And why did changing the speed sensor temporarily clear all of them? Very strange.

Rick

edit: just read on a Subi forum of someone leaving the green connector connected while trying to start/drive the car. It gave most of the codes I'm getting. I always do disconnect that green connector before driving, but it is like it shorts out and connects somewhere?

[RPGs818SNA]

Great to hear that you got everything working, even if just for a while. That suggests your sensors are OK, but you have, hopefully, just one wiring problem left to find. Maybe it’s physically near your new speed sensor and unintentionally wiggling it restored it’s connectivity for a while. If it’s not that green connector shorting to ground somewhere, I hate to say it, but it seems likely that there’s still a grounding problem.

The ECM has 6 ground pins, all of which go through 1 or 2 connectors before reaching ground:

B137 D3 BW through B317 to ground
B137 D7 BL through B317 to ground
B137 D2 BW through B100 and F60 to ground
B137 D1 BY through B100 and F60 to ground
B136 C15 BR through B100 and F60 to ground
B134 A5 BR through B100 and F60 to ground

If you haven’t already checked those intermediate connectors, that might be a place to start. You might also disconnect the battery and measure the resistance of each ground pin to frame ground. Also check engine to frame and battery ground to frame. If you find one connection higher than the others, it’s ok to tap that wire and run a wire to frame ground.

Good luck with this. Sounds like you are almost there.

RPG

[FFRWRX]

Thanks RPG. Heading out to the garage again with wiring diagrams. And you are right, this is sort of good news....in a way. There must be 1 thing that is causing all this.

I will check those ECM ground connections first thing.

As a last resort, I could try cutting the ECM wires that only have to do with the diagnostic connector and the green connector and nothing else (looks like just 2 wires, B27 and C16). If the limp-home still occurs then it is not something in that circuitry shorting.

[FFRWRX]

I disconnected the battery and pulled all the connectors from the ECM. Checked each lead that was supposed to be grounded and they all had a perfect ground. I snipped the 2 wires to the ECM that were connected to the green connector and the diagnostic connector, to make sure they weren't shorting somewhere. Went for a long drive and everything was good.

Went for a drive this morning and back to the limp-home mode. There was a really bad flat spot around 2500 RPM yesterday (but no check engine light) so I cleaned the MAS when I got home. It looked fine but gave it a good spray with contact cleaner and the flat spot was gone today. If I can get a throttle body for cheap I'll try that. Don't really think that is it but out of ideas at this point.

Rick

[Ajzride]

I think Bob has a spare TB that he loaned me so I could test if mine was an issue. Not sure if shipping to Canada would induce extra headaches.

[FFRWRX]

After wallowing in self pity for a while, I'm thinking of what to do next.

I have a line on a used throttle body that I could get cheap; like almost free. Even though I don't think that is it, it may eliminate one possible factor.

I wondered if the O2 sensors are an issue. A friend had a car that needed spacers to get the sensors somewhat out of the main exhaust stream. A fairly easy thing to try, but again, it drives fine until the check engine light comes on and it goes into limp-home mode, then there are multiple codes (always the same ones). So again, I don't think that is it, but could try the spacers.

Not much help searching Subaru sites for solutions. I've read of some people having this exact issue. They try all sorts of things and eventually stop posting. So no obvious solution there.

The car runs and drives great for 5-15 minutes usually. Then suddenly goes into limp-home mode. Won't go over 3,000 RPM and the throttle does almost nothing. Flat roads aren't an issue, but trying to go up a hill is very slow. I have to pull over many times if anyone comes up behind me. When it happens, it just happens. No bumps in the road or temperature factor or any other common issue I can think of. Just normal cruising light throttle and I can feel it suddenly lose power. So I don't go far from home.

I appreciate you listening to me whine.

Rick

edit: Forgot to ask this. I've read about all sorts of "learning" procedures for the ECM after the battery has been disconnected. I've never done any of them. Just start the car and drive. Here is one procedure (probably the longest one I've found). Is it important to follow a learning procedure or will just driving the car eventually sort things out?

1) Turn off the lights, aircon, stereo or any system in the car that draws extra current on top of the engine.
2) Disconnect the battery for 30 mins.
3) Reconnect the battery.
4) Before you start the car for the first time, turn the key to the ON position but do NOT turn the engine over. Wait 10-15 seconds so the electronic throttle body or IACV has time to go to the factory programmed home position.
5) After waiting, start the car and let it idle without any load, lights, A/C etc.
6) Every 20 seconds or so the idle will be adjusted up and down as the ECU tries to adjust it towards a stoichiometric fuel / air mix.
7) Leave the engine running for a full 10 mins but DO NOT TOUCH THE ACCELERATOR during this time or turn on anything that will cause extra electrical current draw.
Cool Turn off the engine, and leave the key in the OFF position for at least 20 sec.
9) As per step (4) turn the key back to the ON position for 10 - 15 sec without actually starting the engine.
10) Start the engine and leave to idle for a further 5 minutes without touching the accelerator and without turning on other systems in the car.
11) Turn off the engine again and wait at least 20 sec before restarting.
12) Take the car for a test drive as the ECU should now be fully retrained.

[Ajzride]

Your issue is not needing to relearn. I drive mine all the time after long battery disconnects and it's not ideal, but it never hits limp mode. There is something more nefarious going on here. I'm racking brain and coming up with nothing but a wiring gremlin. Do you have an immobilizer on your system? It's not the issue, but that would limit some of the trouble isolating things we could try.

[lpmagruder]

If you don't already have one, might be worth getting one of the Tactrix OpenPorts and logging a few long drives and see what goes crazy right before limp mode. You can get a surprising amount of data from the stock ECU with that thing.

[FFRWRX]

Your issue is not needing to relearn. There is something more nefarious going on here. I'm racking brain and coming up with nothing but a wiring gremlin. Do you have an immobilizer on your system? It's not the issue, but that would limit some of the trouble isolating things we could try.

I really appreciate you trying to help me figure this out. I agree that there is something here, probably a simple thing, causing this. But being intermittent makes it harder to find. I only have the stock immobilizer system; whatever the 2006's had. I know aftermarket ones can cause a lot of headaches when they act up.


If you don't already have one, might be worth getting one of the Tactrix OpenPorts and logging a few long drives and see what goes crazy right before limp mode. You can get a surprising amount of data from the stock ECU with that thing.

Now that is what these forums are for! I never thought of that. I have the Tactrix cable and a laptop I bought a while ago. I used it to change the ECM program to eliminate the secondary air system and other stuff that wasn't in the car any more. I never thought of using it for data logging. I'll read up on how to do that. Should be very interesting to see what it shows.

Rick

[FFRWRX]

And here I thought data logging would be the easy part. Can't get it to work.

I've tried Romraider Logger and it won't read from the ECM. It gives me this error:

189400

Then I tried ecuEdit and get this error:

189401\

And all that was after watching a dozen videos on how to data log and reading dozens more articles. It seems that everyone expects you have the computer all set up and just want to know how to connect it to the car. The setup procedure (what to download, where to put the files, what various things to select) are skipped over.

Different frustration now.

Rick

[Ajzride]

If you can install TeamViewer I'll be happy to watch the screen and help you along. You don't have anything else talking to the ECM do you?

[FFRWRX]

Don't think anything else is "talking" to the ECM. No aftermarket stuff installed. I can scan for codes fine, so I would think that means the ECM can be accessed. But maybe logging is looking for something else?

I'll play around with it a bit more and if I can't resolve it I'll take you up on your offer.

[FFRWRX]

I realize this is not the correct forum for this, but since I started here...

Things have gone from bad to worse. It's almost funny now. Couldn't get Romraider to read the ECM, so I decided to delete it all from my computer and reload it. Found a very good video with step by step instructions. Loaded it all, put things in the correct place and ran it. It would pop this up for about 2 seconds then it would disappear and nothing would happen:

189457

So, deleted everything and downloaded it again....and again....and again. Now it won't do anything different, but when I go to the list of Windows programs, it shows this:

189458

It was previously showing all the various Romraider programs, now it only shows the Uninstall one. It is still there under Program Files (x86) with all the various components. I've tried various versions of Java since I've read that it has to be a 32 bit version and I'm pretty sure that is what I have now. But why is it not even loading the program properly in Windows now? Like I said, it is almost funny how I'm making progress in the wrong direction. Almost forgot what the original purpose of all this is for.

Rick

[Ajzride]

What version of windows are you running? How much memory do you have on your laptop?

[FFRWRX]

I'm running it on my laptop, not the Mac I'm using here. So I'm taking pictures of the screen. This is what I have:

189460

Romraider was running a few days ago when it wouldn't read the ECM.

[Ajzride]

I've always had the best luck running Romraider on a VM with Windows 7 32bit, but your don't have enough ram to run a VM. is your Mac a laptop? It might be able to run a VM.

[FFRWRX]

No, Mac is not a laptop. Could it have something to do with Java? I've read a lot of articles on how to load Java and all the steps to go through to make sure it is set up properly. Just went through a bunch of "environment" set ups but really don't understand all of this. The Romraider general set up page has a long article on setting up Java and all sorts of complicated steps. Does it really take that to get it to work, or is that for people developing stuff?

https://www.romraider.com/index.php/RomRaider/GettingStarted


I'll take a break for now. Again, I appreciate the help.

[Ajzride]

Unfortunately RomRaider hasn't been recompiled on the newer versions of Java, so all of the complicated steps are related to getting it to run on an old version. If it was compiled for the newest version this would be far simpler. Most likely we need to get all of the java and romraider stuff fully removed and start with a clean slate related to those two components.

[Ajzride]

My Romraider virtual box only uses 2GB of RAM, so you should be able to run it on your laptop. Once it cools down some tonight I'll go out and make sure it is logging correctly, then I'll zip it up and put it on dropbox for you.

[FFRWRX]

Some progress. I think I have the correct version of Java now. At least the program shows up here now:

189479

Still just flashes the Romraider title screen and then does nothing. There is no way everyone can be having issues like this with it. Anyone with any sense would have given up long ago. :confused:

[Ajzride]

Well rather than helping you, I've joined you in your misery. 4 hours sweating in the driveway and I still can't get romraider logger to read my ECU. ECUFlash will read it so I know my cable is good, but romraider won't. I even got out my2011 laptop I did my original tuning with. It's dead, but I pulled the hard drive and copied the exact revisions of installers I used for that machine from the downloads folder, and I still can't get it to talk. There is something jacked with the connection between romraider and tractrix, romraider won't see any available COM ports (edit: it doesn't need to for an open port 2.0). I'm off to the romraider forum for support. Once I get it all worked out I'll circle back with you.

[lpmagruder]

Sorry to have brought another frustration with my suggestion!

I ran out to check, maybe there is something specific to these cars we all deleted that causes issues, but mine reads and logs fine. 07 WRX donor. Don't connect the green thing for logging, just in case you're doing that. It's only for flashing.

I have an older laptop from 2010 or so, at one point it got updated from Windows 7 to 10 and I remember it being a big pain to get Romraider to read from my 03 WRX at the time. Pretty sure I had to do a full "clean" java uninstall, restart, and install the correct one, then change some settings to make sure it wouldn't auto update. Also, I don't think it will do anything until you get the separate definitions files installed.

You get logger definitions here: https://www.romraider.com/forum/viewtopic.php?p=66788#p66788
You get ECU definitions here: https://www.romraider.com/forum/topic360.html

[FFRWRX]

Well, isn't this getting interesting. At least we know who started the trouble.....and I thought it was me! :)

I'm sorry AJ that in helping me you are sweating things out. I'm sure you'd rather be sweating working on your car.

The suggesting for me to data log is still a good one, and maybe the only way to solve my issue. I have removed all traces of Romraider and Java several times and loaded both. I think I finally have a 32 bit version of Java loaded and working. And Romraider loaded, with the proper ECU and logger files. It will pop up the initial Romraider logo, then quit. I've done the steps I've found for tweaking the Java location and stuff like that, but there must still be one small thing missing.

Not giving up yet. I really don't have a choice if I want to get the car running.

[lpmagruder]

Have you tried right clicking and running as admin? Sometimes that can help. Also whitelisting it in any sort of antivirus software. Wish I wasn't 1,000 miles away and in another country otherwise I'd offer to drive up with my laptop!

I think most of those Bluetooth OBD2 readers these days have some sort of logging now too, I've got one for my phone . Definitely not as nice as what the Tactrix can do but if you've got a friend who already has one of those it may be another thing to check.

[Ajzride]

The OBD2 readers won't work on a 2006 because it has no can-bus. The form factor of the plug is the same, but the Subaru ECM is talking SSM/K-line instead of CAN.

[FFRWRX]

Strange stuff.

I removed all the Romraider files once again and reloaded it (with the various definition files). Again it didn't show up under programs in the side menu, but was there under "Program Files (x86"). I loaded it a 2nd time from the download folder, not removing anything first, and now it is there under the program drop-down menu. Still won't run, but strange. And yes, I've tried running it as admin. I don't have any antivirus other than what Windows has. I'll check and see if turning it all off helps.

I was just out in the garage with my code reader. This one:

189547

It does have live data, which I've used. It also has capture, which I've never used. I was just playing around with it and I think it can data log to a certain extent. Not sure how much but certainly worth a try if I can't get Romraider working.

[lpmagruder]

For what it's worth I was able to do some rudimentary logging with this El cheapo eBay Bluetooth OBD deal and the Android app called "Torque"

2007 wrx donor, not sure if that's the same as 2006 as far as the ECU is concerned

189554189555

Definitely not as nice as what you get with romraider but it's something.

[FFRWRX]

Worked with Aj through TeamViewer (isn't technology wonderful....when it works). Couldn't get Romraider to run. He set me up with VirtualBox, but that didn't work either. The settings in my BIOS won't let it run a virtual mode; this is beyond my knowledge of what is going on. I dropped my laptop off at the place I got it and explained the issue with virtual mode. He said he knew what I was talking about. I told him that was good since I didn't. He said to give him a day or so to see what he could do.

I did manage to data log with my code reader. It can log about 10 minutes of driving info before it is full. Of course the first 10 minute drive went well with no limp-home mode...why should things be easy? So I cleared the reader memory and set out again. It went into limp-home mode while data logging this time. I've only had a quick look at it and can't seem to get it downloaded into the computer to study it. I can look at it on the screen, one parameter at a time; I will spend more time with it. I think Romraider would capture much more data so I still want to get that working.

Rick

[FFRWRX]

I got a used throttle body for almost nothing. I swapped it for the one that was on the car. I didn't go for a test drive or run codes or anything since I'm pretty sure this is not going to make a difference. Why?.....

I removed the original one and left it connected and positioned so I could see it through the rear window. With the ignition on and slowly flooring the pedal, the throttle would open fully some of the time. Sometime it would then go to about 1/8" open and only go from closed to that position even when flooring the throttle. Cycling the ignition on and off would sometimes return it to normal operation (opening fully) and sometimes it would only open that 1/8". When I swapped throttle bodies the "new" one did the same thing. Sometimes it would fully open with the pedal floored and sometimes not.

So with the engine not running, what would be causing this? It isn't like the car is moving or bouncing to move loose wires or sensors. The ECM isn't looking for running parameters (O2 sensor data, airflow data, etc). There is the pedal, the ECM, and the throttle body controlling this with the engine not running. So something in the ECM is triggering the throttle body not to open? But not all the time. What else is it looking for?

Haven't got my laptop back yet, so can't data log. But now I wonder if that will show me anything useful.

Rick

edit: throttle control relay maybe? Won't hurt to get a replacement.

[RPGs818SNA]

I agree with your analysis, Rick. With the engine off, it should just be the pedal, throttle control, ECM, and relay. There are a few mid-wire connectors that need to be good too. I’d put a voltmeter on the pins of each of those components to see which signal is jumping around making the throttle jump around. I was sure hoping it would be a loose ground, but now I’m hoping it’s a bad connection somewhere that’s easy to find and fix. I’m hoping it’s not the ECM. Here are the relevant schematic pages so all can follow along.

189777189778189779

Wish you the best of luck finding this gremlin.
RPG

[Ajzride]

I believe there is a primary and secondary signal from the pedal to the throttle body, check that they are in sync.

[Bob_n_Cincy]

I removed the original one and left it connected and positioned so I could see it through the rear window. With the ignition on and slowly flooring the pedal, the throttle would open fully some of the time. Sometime it would then go to about 1/8" open and only go from closed to that position even when flooring the throttle. Cycling the ignition on and off would sometimes return it to normal operation (opening fully) and sometimes it would only open that 1/8". When I swapped throttle bodies the "new" one did the same thing. Sometimes it would fully open with the pedal floored and sometimes not.


The action you describe here is close to what I would expect from a pedal that doesn't have a travel stop.
After clearing codes or cycling power to the ECU the throttle body works fine until the first time the pedal goes into over-travel.
Over-travel causes limp mode. Try you test again but only go to 1/2 pedal each time.
Bob

[Bob_n_Cincy]

I believe there is a primary and secondary signal from the pedal to the throttle body, check that they are in sync.

Here are the voltages you should be looking for. 189780

[FFRWRX]

The action you describe here is close to what I would expect from a pedal that doesn't have a travel stop.
Over-travel causes limp mode. Try you test again but only go to 1/2 pedal each time.
Bob

I'll try only 1/2 pedal, but....are you sure? Whenever it goes into limp mode while driving I am just cruising along not even altering the pedal much.

This came up earlier and I can't see how the pedal could over-travel. This is the pedal with the Hall effect sensors (or something like that inside, not the copper traces/wipers). There is a very solid stop built into the pedal assembly. Is it this pedal design that you've seen over-travel?

[FFRWRX]

Found an excellent video series on hunting down the limp home mode issue. This is part 3 where he finally solves it. He has an amazing amount of diagnostic equipment and really know his stuff.

https://www.youtube.com/watch?v=Mm0QJM8lbz0

What he shows is that the main relay powers up and then that powers up the throttle control relay. That allows the 12V to go to the ECM. Apparently that has to happen within 400ms or the system goes into limp mode. Spoiler alert on the video. He finds a loose connector on the back of a fuse box that prevents the 12V from consistently going to the ECM.

189788

So the key is turned on and that sends 12V to the main relay (orange lines). That closes the contacts that allow the 12V to energize the throttle control relay (red lines). That pulls in the contact that flows the 12V to the ECM (green lines). This has to happen within 400ms. If that doesn't happen, or there is an interruption to the 12V to C1 on the ECM, it goes into limp mode.

So what makes sense (right now anyways) is for me to measure the voltage on pin C1 on the ECM and go through a number of key on/off cycles and do I get no initial voltage on that pin when it goes into limp mode.

[Bob_n_Cincy]

I'll try only 1/2 pedal, but....are you sure? Whenever it goes into limp mode while driving I am just cruising along not even altering the pedal much.

This came up earlier and I can't see how the pedal could over-travel. This is the pedal with the Hall effect sensors (or something like that inside, not the copper traces/wipers). There is a very solid stop built into the pedal assembly. Is it this pedal design that you've seen over-travel?


See the caption under this picture.
189791
I experienced limp mode in my first 10 minutes of driving back in 2015.

[FFRWRX]

See the caption under this picture.
189791
I experienced limp mode in my first 10 minutes of driving back in 2015.

I saw the caption under the picture. I also saw that it says that particular pedal design does not have a built in stopper. You have not shown the complete webpage where it shows the other pedal designs and it does not say the other designs don't have a built in stopper. Please see what I've written in post #45 and much earlier in post #5 where I show a picture of my pedal. The one you referenced states it is an STi pedal. Mine is a WRX, as shown in the 3rd picture on that webpage, which definitely does have a built in stopper.

So I'll ask again, are you aware of pedal over-travel with the WRX pedal design with the built-in stopper?

[lpmagruder]

For what its worth, I have a 06/07 pedal which is the next picture on that page. I've been driving around for a while with no additional stop and have been to WOT a few times at least and never seen it throw a code or go into limp mode. Fairly certain it's just that one pedal design that has an issue.

I do want to eventually build a stop because the sheet metal adapter bracket FF provides is pretty flimsy, and I don't like the idea of the force getting reacted that far up on the pedal long term, but that's just a reliability thing I think.

[Bob_n_Cincy]

So I'll ask again, are you aware of pedal over-travel with the WRX pedal design with the built-in stopper?

Rick, I have had both pedal designs in my car back in 2014. I don't remember exactly when I had the over-travel issues.
My trouble shooting method always revolved around attacking one DTC code at a time. Have all your codes been resolved?
Bob

The stopper for the plastic pedal was mounted to the floor/carpet. 189841

[FFRWRX]

My trouble shooting method always revolved around attacking one DTC code at a time. Have all your codes been resolved?


Normally I would agree with you Bob. But in this strange case I replaced the neutral switch which eliminated that code. Then replaced the speed sensor which eliminated that code. When it goes into limp mode and the codes related to the throttle pedal appear, so do those previously eliminated codes. So I think something is triggering a number of codes to appear, even though the main issue is not one of those components.

Going for a drive later to test out another theory.

Rick

[Bob_n_Cincy]

Rick, do you have have a actual wire from battery negative to engine block? Or are you going through the chassis?
Bob

[lance corsi]

I’ve been watching with interest. I’m glad I chose a cable operated tb.

[FFRWRX]

Rick, do you have have a actual wire from battery negative to engine block? Or are you going through the chassis?
Bob

My ground lead uses the frame. My reasoning is:

1) Millions of production cars do it this way and it works; including Subaru I imagine (though my car came with the donor stuff already stripped out so I don't know for sure).

2) The car starts fine which means the ground conducts a couple hundred amps without issue.

3) The ground lead at the battery is short and goes onto the frame; clean connection. There is a robust lead from the engine block to the frame at the rear.

4) If it was a ground momentarily losing contact, would it cause this exact symptom every time? Honestly, I have no idea.

I could run a booster cable clamped on the battery and the other end onto the engine, if I can figure a way to do that while leaving the car drivable; hood and rear deck lid partially open.

[FFRWRX]

My latest theory was this.

The throttle control relay directs power to the ECM. So if there is an issue with that, I think (from the video I referenced above) the car goes into limp mode. I accessed that relay and tested it; it tested fine. But I did notice it gets pretty warm. Not hot, but warm. Of course that might be normal since it is a metal encased one, not the usual plastic. My theory was that even though the relay tested fine, maybe after driving a bit it heated up and stopped working. So...

I soldered some leads onto the wires on the contact side of the relay; one going in and one coming out. I put a volt meter on the lead with power coming out (going to the ECM) and left that on the passenger seat and went for a drive. Of course the first drive the damn car wouldn't go into limp mode, but the second drive it did. But the voltage coming out of that relay did not change. So, another theory gone.

Nothing left to test that I can think of. When I get my computer back and can data log maybe I'll find something there.

Rick

[RPGs818SNA]

Well, it’s not the relay, so you’re one step closer to finding the culprit. That leaves the pedal, throttle control, ECM, and wiring. I just noticed a common ground from the ECM pin B30 RG to connector B83 which sends a ground to the pedal, MAF, both oxygen sensors, fuel temp, and fuel tank pressure. It’s called the Shield and Sensor Ground Joint Connector and is shown on page E/G(TB)-06. That’s probably worth measuring.

Incidentally, the 1.5” frame tubes have a conductance similar to 0 gauge (quarter inch) copper wire, and they form an interconnected network. So if you have good connections to the frame, you can’t do much better.

If the ECM pin B30 and connector B83 are good grounds, I’d try to monitor each pedal wire for smooth voltage change as you depress the pedal. Somehow the ECM seems to be getting a bad signal.

[FFRWRX]

I just noticed a common ground from the ECM pin B30 RG to connector B83 which sends a ground to the pedal, MAF, both oxygen sensors, fuel temp, and fuel tank pressure. It’s called the Shield and Sensor Ground Joint Connector and is shown on page E/G(TB)-06.

That is the one part of the wiring diagrams I find confusing. On the ground details, these are clearly grounding tabs:

189896

On other parts of the wiring diagram, this is clearly a ground as well:

189897

But what you are indicating I find confusing:

189898

I know it states "ground joint connector", but it doesn't actually show it grounding anywhere. Following the 6 wires that come from the B83 connector, none of them go to a ground. The connector just looks like it connects them all together, but where do they ground? Or is B30 on the ECM a ground?

[RPGs818SNA]

Since the sensors produce an output relative to their ground, and the ECM measures that output relative to its ground, it's best to use a ground from within the ECM to ground the sensors. So ECM B30 is a ground from the ECM and is distributed to the sensors via B83 interconnections. B30 should measure low resistance to the other ECM grounds that go to the engine or frame.

[More detail than necessary] Each ground, whether engine, frame, ECM, or battery will have different sources and levels of noise on them. The engine has spark plug noise, the battery alternator noise, the frame lighting switch noise, etc. The ECM ground will have have noise based on what it's grounded to. But if it distributes that noisy ground to the sensors, they will add their signal to that noise, and the ECM will subtract its ground noise from the signal plus noise to get a clean signal from the sensors.

RPG

[Bob_n_Cincy]

189896


The picture above are your intake manifold grounds.


189907
This is mine on the passenger side


189908
This is mine on the drivers side


189910
The (Subaru OEM) black wire with yellow stripe goes to may battery negative
The Other wire on the starter bolt is my ground strap to chassis frame.
bob

[FFRWRX]

The picture above are your intake manifold grounds.


I do have one on each side of the manifold. I'll take them off and clean them just to be sure they are good.

I'll see about checking the voltages at/from the pedal to see what I am getting.


I just noticed a common ground from the ECM pin B30 RG to connector B83 which sends a ground to the pedal, MAF, both oxygen sensors, fuel temp, and fuel tank pressure. It’s called the Shield and Sensor Ground Joint Connector and is shown on page E/G(TB)-06. That’s probably worth measuring.


Checked the resistance from pin B30 on the ECM connector to ground. No resistance, so things area good.

This would almost be fun if it wasn't so frustrating at times.

[FFRWRX]

189910
The (Subaru OEM) black wire with yellow stripe goes to may battery negative
The Other wire on the starter bolt is my ground strap to chassis frame.
bob

I assume neither of these 2 wires are stock Subaru? I don't have anything on that bracket or starter bolt.

[Bob_n_Cincy]

I assume neither of these 2 wires are stock Subaru? I don't have anything on that bracket or starter bolt.

The black with Yellow stripe was out of my donor.

https://www.youtube.com/watch?v=Ahm8RUt_yqY

The other wire I made myself for the ground strap to the chassis.
Bob

[FFRWRX]

Got the computer upgraded and Romraider works now. I data logged a drive this morning and fortunately (?) it went into limp mode after driving for about 10 minutes. It's a huge CSV file, almost 12,000 lines.

Just had a very quick look at it and the first thing I noticed was that when it goes into limp mode the "accelerator pedal angle (%)" does not register properly. Before the limp it is accurate, cruising along it is around 10-15%, floored it is of course 100% with the "throttle opening angle (%)" being the same. After it kicks into limp mode, the pedal angle does not go above 9.02% (with a corresponding throttle opening of 14.08%) even when I am flooring it.

So, is it a normal thing whenever it is kicked into limp mode for any reason that the actual throttle pedal angle is not being measured? What is being shown is obviously a throttle pedal angle after some software manipulation.

This is pretty cool to be able to capture all this data. At least I have something to study now that will hopefully help.

I appreciate all the help to get me to this point.

Edit: "Main Accelerator Sensor (V)" goes to 0 when it goes into limp mode. It is above 0 every line before that.

Rick

[Ajzride]

The accelerator pedal has primary and secondary power feeds (B21 and B22),primary and secondary signals back to the ECU (B23 and B31), and redundant grounds (B29 and B30). It would take a lot of bad connections to result in 0V signal. I wonder if you have a bad pedal. Perhaps sit in the garage with the engine off key on while monitoring/logging all the signals while you work the pedal up and down for a few minutes and see if anything jumps out about the signal dropping to zero.

[RPGs818SNA]

Some have had success cleaning the pedal contacts with contact cleaner sprayed around the pedal pivot. It looks like yours may come apart for a more thorough cleaning. If it's a wire, it's likely the red one from pedal pin 6 to ECM B23.

[Bob_n_Cincy]

Just a note to help you troubleshoot.
The power to the pedal should be about 0.0v and 5.0 volts.
At zero pedal push the return signal should be around 1.0 volts.
At WOT the return signal sh0ould be around 4.0 volts.
These voltages should be referenced to the intake manifold as this is where all the sensor and ECU grounds terminate.
Bob

[lance corsi]

All this to eliminate a cable? I’m not a fan.

[lpmagruder]

If you want I can take a look at your file this evening. Not sure that I'll pull anything you haven't already though.

It's gotta be getting "something" otherwise you'd have no throttle control in limp mode, so it's weird that it would just show a flatline at 0v after that.

[Ajzride]

If you want I can take a look at your file this evening. Not sure that I'll pull anything you haven't already though.

It's gotta be getting "something" otherwise you'd have no throttle control in limp mode, so it's weird that it would just show a flatline at 0v after that.

Perhaps primary goes to zero and secondary is alive, causing limp mode due to only one signal? Would be nice to know what all will induce limp mode.

[FFRWRX]

The accelerator pedal has primary and secondary power feeds ... I wonder if you have a bad pedal. Perhaps sit in the garage with the engine off key on while monitoring/logging all the signals while you work the pedal up and down for a few minutes and see if anything jumps out about the signal dropping to zero.

A bad pedal would probably be my conclusion as well, except that I've replaced it with a new one (original Subaru part) and I suspect the person that did the original build replaced it before that. It seems to take about 10 minutes of driving before the limp mode kicks in. So it may or may not happen sitting in the garage trying it. I will be checking the voltage and grounds that you suggest, that are also in the Subaru diagnostic manual.


Some have had success cleaning the pedal contacts with contact cleaner sprayed around the pedal pivot. It looks like yours may come apart for a more thorough cleaning. If it's a wire, it's likely the red one from pedal pin 6 to ECM B23.

It's strange that I read that in so many places. I must be the only one that has pedal with no internal contacts. This is one of them taken apart. Some sort of Hall effect with a square magnet that rotates with the pedal. So nothing to clean inside it.

190128


Just a note to help you troubleshoot...The power to the pedal should be about 0.0v and 5.0 volts...
Bob

Yes, I never did get around to measuring these voltages so I'll do that as the next step.


All this to eliminate a cable? I’m not a fan.

Not helping, Lance. The car came with that donor equipment already installed. If I had a choice I would have gone to a cable system long ago. But that is not a feasible option at this point.

--------------------------------------------------------------

The strange thing is that this happens after about 10 minutes of driving, and pretty much every drive. Something reaching a temperature that causes it? The coolant temp was pretty stable for a few minutes before this happened. If it was a loose connection or something like that, I would think it would happen at different times. And it is just cruising along when it happens; no bumps, no crazy throttle inputs. And to clarify a little as to what happens in the limp mode. The car will idle, but maybe a little slower than normal. When the pedal is pressed (a little or floored, makes no difference) the revs will go up to around 3,000 RPM max. So the car can be driven, but it is like you can only accelerate at just above idle power.

I put the file in Dropbox so if anyone wants to take a look, please do. There is probably a lot more data there than I needed, but since I wasn't sure what I needed I captured quite a bit.

Line 4290 is where I floored it and let it rev to 6,000 RPM and then backed off, shifted, and hit it again.
Line 10148 is the interesting one. That's where the limp mode starts and the accelerator voltage goes to 0.

File link in Dropbox: https://www.dropbox.com/scl/fi/yv563ishopz0skp2segqb/romraiderlog_20230915_121326.csv?rlkey=5jv62g0l7po wr4itqusl3rjcp&dl=0

Rick

[lpmagruder]

So I just pulled up your logs. My initial takeaway is that as soon as your main throttle position sensor loses voltage, the sub sensor voltage spikes. Maybe this was just you actually mashing the throttle since it went into limp mode, so maybe this is nothing. The vertical line I marked is when it happens.

190129

Here's a real zoomed in view of right where it happened. I guess if it were me I'd probably frustratedly mash the throttle so maybe that's what happened :-)

190130

Since the units here are V I would lean towards thinking that the main and sub sensor voltages are "raw". So it seems like something is killing that signal after 10 mins.

One thing to maybe check next, is to measure that conductor as close to the ECU as possible. Measure it when you know the car is NOT in limp mode, while datalogging, and see that your volt meter measurements match what the ECO sees. Then go drive it, let it go into limp, then take it back into your garage but don't turn it off. Presumably, the ECU will still be claiming to see 0V there. Check and see if it's actually 0V, or if you have the same behavior as you did when you checked it before limp mode on the volt meter. That'd at least help it isolate it to either something between the throttle and ECU, or something in the ECU itself (or another signal it's getting)

Or, maybe I just have no understanding of how the ECU works, and it's just pretending the sensor doesn't exist after it's seen something funky. That could make sense, if it's determined it can't trust data from that sensor it could just be shutting it off. In that case, try doing something to intentionally disconnect that line from the ECU, then reconnect it, all while it's running and datalogging. See if you see it go to 0V in the logs and stay there, or if it goes to 0V then comes back to life.

If you had an intermittent short/break I would expect to see that signal bounce between something "real" and 0V. Seeing it randomly jump to 0V and just stay there for the rest of the drive is strange.

[lpmagruder]

EDIT: Nevermind, looks like you did look into this earlier, sorry!


Here's another thought - and I'm sorry if someone mentioned this already (did a quick scan of the thread and didn't see)

At the beginning of the thread you mentioned O2 sensor codes along with intake air (this is measured at the air flow sensor), among others. These have in common a junction at the "joint connector" and go to ECU pin B30:

190135

I'm not sure what that junction actually looks like, but it's a place where a bunch of wires are presumably crimped together so there's some likelihood for something going wrong there. If something there shorts or does something it shouldn't, then all those sensors would see an issue.

[Ajzride]

Log the Open/Closed loop fueling status and see if limp mode corresponds to the switch from open to closed.

[lpmagruder]

Sorry for all the posts - I should just be collecting my thoughts into one post.

Last thing I'm seeing, is I keep running into the test mode connector when I trace from the secondary sensors you mentioned. I think test mode status is a field you can select when logging. May be worth throwing into your log parameter list and see if it's kicking on.

From this page; most of your original codes are there (minus the throttle stuff): https://iwireusa.com/blogs/iwire-university/subaru-test-mode

[lpmagruder]

Log the Open/Closed loop fueling status and see if limp mode corresponds to the switch from open to closed.

It "should" be in closed loop off boost, low engine load, low throttle angle, at least that's how it was on my 03 when I was tuning. When his car went into limp mode his manifold was in vacuum, 2300 rpm, ~10% throttle angle.

Though, that got me looking at the O2 sensor he logged. I didn't notice initially but he logged the rear, not the front which is used for closed loop fueling. Is anyone here actually running the rear sensor? I'm just running the first one and was under the impression that the rear was only used for cat efficiency checking. That signal seems to be jumping from 0V to ~1V without much correlation to what the car's actually doing.

And when it's at 0, it's pegged at 0, not bouncing around a small voltage. Again, I've never logged or looked at this sensor to know what it should be doing, so there may be nothing here too.

190136

It's a little late to fire mine up tonight, trying to stay in the neighbor's good graces. But I can try logging tomorrow and see what my (unconnected) rear sensor is logging like if you think that'd be useful.

[FFRWRX]

This is all really good stuff!


So I just pulled up your logs. My initial takeaway is that as soon as your main throttle position sensor loses voltage, the sub sensor voltage spikes. Maybe this was just you actually mashing the throttle since it went into limp mode, so maybe this is nothing.

One thing to maybe check next, is to measure that conductor as close to the ECU as possible...

In that case, try doing something to intentionally disconnect that line from the ECU, then reconnect it, all while it's running and datalogging. See if you see it go to 0V in the logs and stay there, or if it goes to 0V then comes back to life.


I didn't notice the sub sensor voltage going crazy when in limp mode, but you are right. I don't remember flooring it as soon as it happened; I could have, but don't think so. I'll see about measuring the actual voltage at the ECU since I did wonder if what is showing (0 volts) is real.


Log the Open/Closed loop fueling status and see if limp mode corresponds to the switch from open to closed.

Can't find that in the data logging parameters. Maybe I just don't know what it is called.



Last thing I'm seeing, is I keep running into the test mode connector when I trace from the secondary sensors you mentioned. I think test mode status is a field you can select when logging. May be worth throwing into your log parameter list and see if it's kicking on.

From this page; most of your original codes are there (minus the throttle stuff): https://iwireusa.com/blogs/iwire-university/subaru-test-mode

I saw a post somewhere that showed most of the codes I get when in limp mode. I thought at one point that maybe the test connector wires were connecting somewhere. I did disconnect them from the ECU and went for a drive and it still went into limp mode. But maybe something related to that.

I don't see that the test mode connector is a parameter to log...would be great if it was.




Though, that got me looking at the O2 sensor he logged. I didn't notice initially but he logged the rear, not the front which is used for closed loop fueling. Is anyone here actually running the rear sensor? I'm just running the first one and was under the impression that the rear was only used for cat efficiency checking. That signal seems to be jumping from 0V to ~1V without much correlation to what the car's actually doing.


Yes, pretty sure the rear sensor is just for checking the cat, so maybe no point in logging it. But I did log the front one:

190142

But you are right, it didn't actually log those. Next time I'll check the screen to see if they are being recorded before I take off. I did read somewhere that some people had an issue where the front O2 sensor wasn't working, but didn't show an error code, but the car did go into limp mode. So I unplugged it the other day and went for a drive to see if it made any difference. It didn't. I did plug it back in but I'll give the connector a shot of Dioxit to make sure the connection is good.

This is all good stuff. I'll report back after doing more checking/driving/logging. May be a couple days. There is a British car show I want to go to tomorrow (the largest one in North America). No, not taking the 818 even though it has the reliability of a British car right now. I've got a Tiger I want to make sure is ready for the trip; no ECU, no power anything, nothing to data log, almost boring. :)

Rick

[lpmagruder]

For what it's worth I just drove around a bit monitoring (forgot to actually hit the log button) my (deleted) rear o2 sensor, and it's flat lined at 0V. If you like most are not running the rear sensor and it's seeing that voltage fluctuations, may mean something.

[FFRWRX]

I am running the rear sensor as well as the front. I could unplug the rear one and see what that does. It shouldn’t do anything, but who knows.

[FFRWRX]

I'm really at a loss now.

I measured the voltages for the drive-by-wire pedal. They were all good.

The pedal connector (not connected, but ignition on) gave .68V-3.3V, pedal not pressed to pedal floored. Normal is supposed to be less than 4.8V. This is from connector B315 pin #4 to ground and pin #1 to ground.

Measured at the ECU (pedal connected, ignition on) pretty much the same numbers, measuring B135 pin #23 to ground and pin #31 to ground.

The above voltages are lower than the max "normal" of 5.5V for the first one and 4.8V for the second one. Are mine too low?

I can't data log the front O2 sensor. Romraider has it listed when I choose what to data log, but as soon as I turn the ignition on, they are no longer available:

Ignition off:

190344

Ignition on:

190345

I have read on-line about this happening. There didn't seem to be a reason other than the data log definition doesn't allow it for some reason.

I decided to go for a drive and data log again. I did unplug the rear O2 sensor. The car went into limp mode faster this time. Maybe because I had it idling for longer before driving off. The last time the "Main Accelerator Sensor" voltage went to 0 when it went into limp mode. This time that did not happen. So there was really no change in the data logging when it happened. The only way I can see that it happened is that the Accelerator Pedal Angle maxes out at 9.02 when in limp mode, just like last time.

I put this latest log in Dropbox if anyone wants to see it.

Data log from today:

https://www.dropbox.com/scl/fi/httj9zjr8oa94tmxbfah6/818C-data-log-2.csv?rlkey=t2wha09w38x04faxz82j6dbts&dl=0

I'm really at a loss now as to what is going on. The car is basically not drivable as-is (not fun getting back home in limp mode). And everything I test checks out fine. The only thing is that the max pedal voltages are maybe a bit low; testing at 3.3V max compared with an allowable of well over 4 volts. Anyone measure theirs as a comparison? I don't know what the next step is since everything keeps checking out fine.

Rick

[Ajzride]

Have you tried just letting it idle in the driveway for an hour to see if it hits limp mode while not moving? And perhaps doing that twice to see if there is some correlation (engine temp, AFR, etc)?

[Bob_n_Cincy]

Rick, on line 2657 of your datalog the main accelerator signal goes to 0.04 volts. anything below 0.50 will put the car in limp mode.
At the same time your sup accelerator signal stay good around 1.3 volts.
This tell me the supply voltage to the pedal is good.

see this section of the chart. 190348

on line 2692 of your datalog the main accelerator signal goes back to up to 0.86, a normal value.
An the accelerator pedal is operating correctly. Once the car goes into limp mode it will mot recover until cycling ignition.

See this section of your dstalog 190350

I suggest carefully inspecting the red wire connection between the pedal assembly and the engine eCU. If this doesn't working, try a new pedal.
Bob

[Ajzride]

I suggest carefully inspecting the red wire connection between the pedal assembly and the engine eCU. If this doesn't working, try a new pedal.
Bob

He’s on his second pedal, I would focus on the wire.

[RPGs818SNA]

And if not that wire, it may be time to look at the ECM. The fairly consistent 10 minute time to limp sounds a bit like it could be an ECM internal heat issue. I've read that they almost never fail, but you've checked pretty much everything else.

It would be fairly easy to strap an ice pack to the ECM to see if you can get past that 10 minute mark. If so, you may have a simple cooling issue, or, worst case, a faulty ECM with a heat sensitive connection, like a broken trace that pulls apart when heated.

I'm impressed by your perseverance, Rick. That's bound to pay off shortly.

RPG

[Bob_n_Cincy]

He’s on his second pedal, I would focus on the wire.
I agree if he has already replaced the pedal.
The red wire is shielded by a ground wire. So it is possible that the wire got pinched somewhere during it's life. An easy test for this is to disconnect the ground.
190353

Rick, Did you splice the wires going to the accelerator pedal? If so, That is where i would inspect first.

This is a chart of the Accelerator voltage . You can see that voltage drops below 0.5 volts multiple times.

190354
Bob

[FFRWRX]

I'm impressed by your perseverance, Rick. That's bound to pay off shortly.
RPG

My perseverance is due largely to you guys helping with suggestions on what to do next. Otherwise, don't know what I would do.

Letting it idle on the driveway might work, but how would I know if/when it goes into limp mode? If I data log and check for low Accelerator voltage that would probably work. I'll keep this in mind, but other things I can check first. I have considered if it is engine compartment temp triggering it. The ECM (or ECU, seems to go by both names) is in probably the coolest spot in the engine compartment, but heat could still be an issue. I've never felt its temperature right after a drive, so I'll consider that.

I didn't notice the Accelerator voltage dropping so low just before it goes into limp mode. Good find there Bob. In the first log I was looking at when it went to 0 and figured that was the start of the limp. But it could have been the 1 line before that when the voltage drops to 0.02V that kicked it into limp mode.

And in this latest one, it is always above .68V until it drops to 0.04V at line 2657 and it goes into limp mode. Even though the voltage goes higher later, it can't recover until reset.

190357

Obviously that is significant and a common factor. I'll check the voltage on the red wire. It doesn't go into limp mode after hitting a bump or anything like that, so a bad connection doesn't seem like it. The grounding of the shield around the red wire (Q in the part of the wiring in the post above) goes to a joint connector that connects to another shield, and also with pin C6 on the ECU. Maybe electrical interference from somewhere that is triggering it due to a bad ground in that system? I'll check that. I'll also take off the grounds I find and clean them; I may have missed some of that previously.

I don't think I spliced any of the wires with shields on them. I'll take apart more of the centre console to see the wires. It is of course difficult on a completed car to see all of the wire runs.

Thank you once again for the help/suggestions. Back to the garage.

Rick

[Ajzride]

You could always run a temporary new wire for the questionable one to keep from tearing the console apart.

[FFRWRX]

Haven't made it to the garage yet. I could run a new wire, but according to the schematic it should be a wire with a shielded ground. It isn't a huge deal to take the console panels off. I designed it to come apart without any tools to get at the wiring and fuseboxes that are in there.

[Bob_n_Cincy]

Haven't made it to the garage yet. I could run a new wire, but according to the schematic it should be a wire with a shielded ground. It isn't a huge deal to take the console panels off. I designed it to come apart without any tools to get at the wiring and fuseboxes that are in there.

At this point, don't worry about not using shielded wire. Regular wire will at least get you running reliably.
I would pull the red wire's pin out of the connector at each end. Could be a bad crimp of a pin.
The red wire in the OEM harness is not long enough to reach the gas pedal. So i suspect it was spliced.
Bob

I do not suspect heat as the issue as all other sensors going into the ECU are working normally.

I did a chart with the sub accel pedal showing good signal compared to the bad main.
190379


I also did a chart on your battery voltage. It doesn't look very good.
190380

[FFRWRX]

At least I'm pretty sure it is. Maybe I shouldn't have said that.....I might jinx it.

I put another wire from the pedal to the EMU to parallel the red power wire. I branched it into the existing wires very close to the end connectors. I left the "stock" one in place. Went for a long drive yesterday evening and.....no limp home. Went for another one today, pretty long, and no limp home. My wife kept saying that now I have to disconnect the new wire and see if the limp mode comes back. So I went for a drive today and after 12 minutes I snipped the new wire and kept driving; no limp home.

Here is what I'm concluding. When asked in this thread if I extended the accelerator pedal wires I kept saying no, which is true. But I bought the car with the donor stuff already done, with the stock wiring harness in it. I didn't think that the pedal wires might have been extended at that point. I took the harness out and did a serious wire diet, but left the pedal wires alone. Looking at them now under the wire sheath, they have been extended and several times. Looks like the red wire from the pedal goes into a white wire for about a foot, then into another wire, then who knows as it disappears into a major bundle of wires.

I disconnected the wires at the pedal and the ECU and measured the resistance of the red wire. It was about 12 ohms, but was jumping all over the place when I touched the harness. Now with a digital meter it does tend to jump around when the reading is changing, but this seemed like more than that. So I attached my extra wire and went for the drives.

When it didn't go into limp mode with the new wire cut, I'm reasoning that my disturbing the original red wire may have pulled a poor joint into a less poor joint and it didn't loose contact.

Here is what my temporary wire looks like:

190400

190401

There is no good way at this point to run it where the rest of them go, to the rear of the console and then through the firewall into the engine compartment. It would mean fairly major work to open up that path again. I'm thinking of running it forward in the console into the forward compartment, over to the right side rocker panel, through that to the rear, and then to the ECU which is on that side. Shielded wire does not seem to be easy to come by and I don't know if that is required now. If I run it where planned it will be behind aluminum panels most of the way that should shield it from any electrical interference. Thoughts?

If there is a way to pull the contacts from the plastic connectors I'll do that and add the new wire in properly. But I don't see how to do that.

A couple other things. The check engine light did go on part way through my first drive with the new wire. After I swore once or twice I realized the car was still running fine, and continued to do so. I checked the code when I got home and it is a new one for me:

190402

I looked it up and it seems to be something like the system expected the temperature to be higher after that running time. I'm not really concerned at all about that one.

And yes, my battery does seem to be a bit on the low voltage side. I haven't yet gotten around to cleaning all the connections, so I'll make sure and do that. And if I need a new battery, not a big deal at this point.

This has by far been the most difficult car issue I've ever had to solve. If it is (and right now I'm very sure it is) just a bad wire/splice issue, then it is a pretty simple one to correct. But finding it......what a PIA!!

Again, I thank all of you for keeping me going and not giving up.

Rick

[Ajzride]

Congrats man!! And all it took was a new laptop to track down a faulty wire.

If all the PO splices can be seen inside the cabin just cut them all out and put in a new splice you know is done right.

[RPGs818SNA]

Congratulations Rick! Well done. Glad the fix is a fairly simple one.

If the red wire was extended with a bad joint, the others were probably extended too. It might be a good time to replace all 4 of the pedal to ECM wires. Your Hall effect pedal is awesome in that it has no sliding connection to go bad, but the signal is fairly weak. I recommend shielding at least the red wire. Running the wire behind aluminum panels won’t help that much.

One option is to use shielded 4 conductor wire like this:
https://www.amazon.com/Shielded-Speaker-Electrical-Appliance-Transformer/dp/B0CBK2P7JD/ref=sr_1_2?crid=2IS7H13PRFR14&keywords=22AWG%2B22%2F4%2B4%2BCore%2BShielded%2BWi re%2BCNC%2BWire%2B(30FT)%2BUL2464&qid=1695646792&s=industrial&sprefix=22awg%2B22%2F4%2B4%2Bcore%2Bshielded%2Bwir e%2Bcnc%2Bwire%2B30ft%2Bul2464%2Cindustrial%2C124&sr=1-2&th=1

Connect the braided shield to the ECM pin C6 BY wire. The other splices can be near the pedal and ECM. Let’s hope that’s the end of your pedal worries.

RPG

[lpmagruder]

Congrats man, I've been following this and I feel like even sharing your stress from afar, I bet it feels great to have it nailed.

For the connector, you can depin/repin them if you have the right crimp contacts and seals, I've had mixed success re-using original connector housings. Maybe this one's ok since it's inside the cabin, any connector from the engine bay I've tried to depin, I have about a 70% success rate getting it all back together without breaking something, the plastic gets a little brittle with age/heat.

Good news is almost any Subaru connector you can get replacements thru iWire, the dbw pedal connector is one of the ones in here: https://iwireusa.com/collections/5-9-pin-replacement-plugs

You can typically get either a connector + pins and seals, or a pigtail that you'd crimp your own wires to. The "right" way to do it is probably to get the crimp kit and run a whole new group of wires back. Be careful with the shielded one that it gets the shield trimmed back and secured (heat shrink) before crimping, then the other end near the ECU tie the shield to ground making sure you keep it isolated from the internal conductor.

[Ajzride]

You can also use the existing wires to try and pull the new wire by attaching them with some solder temporarily and taping them up good.

[Bicyclops]

mil spec shielded (and un-shielded) wire and supplies at: bandc.biz

Ed

[lance corsi]

Yay! You got it! It was beginning to drive me crazy! Congrats!

[FFRWRX]

I cut the heat-shrink from the extensions the original builder did. They were all twisted together and the heat-shrink applied.........no solder!?!?!?! One on the red wire did not look very well twisted, so that was probably the one that was giving bad resistance measurements when I moved it. This is kind of surprising since all the other work was very well done.

I am replacing the 4 wires that go to the ECU and I've soldered the extension connections on the other 2 wires.

The path the bunch makes to get from the rear of the console into the engine compartment makes it pretty much impossible to pull new wire through. I'll run it forward and then through the rocker area.

Just waiting for the shielded wire to arrive, or as HJS said "now we play the waiting game".

Rick

[lpmagruder]

I've noticed a lot of otherwise good mechanics are total hacks when it comes to wiring. It's a whole different skill set.

I did most of my harness lengthening/shortening with this sort of splice, with solder then heat shrink over it: 190562

Then after I was done someone told me that a crimp connection was better than soldering for vehicles, since vibration over time can start to crack the solder, so I guess we'll see. Hope my harness isn't a ticking time bomb.


While you're in there you may want to look at the other lengthened wires since the previous builder probably did something similar. From memory, there's the brake and clutch pedal sensor wires, as well as one of the headlight bundles (forgetting if I had to extend the R or L one) I didn't end up needing to, but I've heard some people extend the starter solenoid wire as well. I remember earlier you had a code for starter voltage or something similar so that one may be intermittent too.

[FFRWRX]

I do basically the same thing when joining wires; twist, solder, heat shrink. I have also read that crimp connections are better. I've done some crimped ones but I only have a cheap crimping tool. Apparently the good ratcheting ones do very nice crimps.

I've done soldered ones for decades and never had an issue. Of course I never keep cars that long, so I don't really know if they cause an issue down the road (hey, a pun!).

I'll have a look for any other previously done lengthening of wires.

[FFRWRX]

I replaced the 4 extension wires that go from the pedal to the ECU. Grounded the shield to ECU pin C6 and everything is working well now. Kind of hard to believe that such a small thing (one shoddy connection) can cause such headache.

Gone for several long drives. Went to a cruise night and then another daytime "cruise" yesterday. Sort of a "where's Waldo" but there really isn't any hiding with this color:

190729

190730

So much interest in the car and many positive comments on the color; especially in the sun when the green takes on a gold effect.

I am getting some stumble now and then and it was dying when idling a few times. Got a P0301 error code (cylinder 1 misfire). I'll look into that but I don't think it will be as hard to track down as the previous issue.

Rick

[Bogusenterprises]

After almost three years I am finally starting the engine. Donor is a2006 impreza wrx . The engine fired right up but will not run above 2500-3000. Preliminary troubleshooting when the key is turned on the throttle relay powers up and then immediately shuts down. Why? I have done alot dieting to the wire harness. The wire harness to pedal has been shortened with soldered splices but I did not touch the shielded wire. I have checked those wires for continuity but not for resistance. The same goes for ECU to TB wires.

[FFRWRX]

Don't think I can help you with that, but hope others can. I sold my car last year.

[Bob_n_Cincy]

After almost three years I am finally starting the engine. Donor is a2006 impreza wrx . The engine fired right up but will not run above 2500-3000. Preliminary troubleshooting when the key is turned on the throttle relay powers up and then immediately shuts down. Why? I have done alot dieting to the wire harness. The wire harness to pedal has been shortened with soldered splices but I did not touch the shielded wire. I have checked those wires for continuity but not for resistance. The same goes for ECU to TB wires.

My car is Drive By Wire (DBW) and has Electric Throttle Control (ETC). I don't have a "throttle relay," so I'm unsure what you are discussing. The main fault that you get that limits RPM is a pedal that gives the ECU and out of range fault. On an 818 this is usually caused by not having a pedal stop to limit pedal travel. Here is a picture of mine. 212213

[Ajzride]

I guess I never floored mine (I've only been in go-kart stage for 3 years now), because I don't have a pedal stop and I've never hit limp mode.

[lpmagruder]

Not every pedal has that problem. There's 2 dbw throttle styles. Realistically both eventually need a stop for robustness reasons, but only the early styles have the over travel issue. This has been discussed a few times here, and iWire also has an article on it.