Jump to content
Ultimate Subaru Message Board


  • Posts

  • Joined

  • Last visited

  • Days Won


Everything posted by GeneralDisorder

  1. This is why I said to buy a valve SET, replace/lap ALL of them, and send it. No need for a machine shop. Do it yourself so you know it's done. 🤷‍♂️ GD
  2. The 10-12 Outback's have problems with internal HG fire ring failure causing overheating. And they like to drop valve guides. If it's a CVT model those are the early one's - "countermeasure" torque converter is a must to prevent stalling when hard braking. Valve bodies and wiring harnesses and torque converters for days on those models. And yeah - CANBUS nightmare. I'm not a fan but I don't like smacking myself in the head with a hammer and some people like that I hear so I'm not going to judge. GD
  3. Probably a plastic separator plate. One of the black ones. Wasn't till about 97 they switched back to metal. GD
  4. Depends on where it's coming from. Camshaft plug (distributor block-off plate)? No. Separator plate or rear main seal? Yes. Honestly those engines are ultra simple. I can have the engine out with a cherry picker in about 45 minutes. Wouldn't look for a younger Subaru unless it's rusted out. Those were about the peak of everything that was good about the brand. I'll stretch that maybe up through about 2004 but the EJ22 in that '95 was a better engine. And the 4EAT auto is really a tough unit if at all cared for. GD
  5. No it's not a linear connection like that. It's completely parallel where everything on the network communicates simultaneously (kind of like the internet). Messages are being sent and received constantly and there's a multi-level system of priority, arbitration, error detection, and CRC checking that's all part of the message protocol to ensure messages don't collide with each other and that messages are received by the intended target(s). The problem arises in that the data inside the messages (which module the messages belong to and are intended for, and their payload) is proprietary intellectual property and without that information you can't decode the signaling. The network can be sniffed and all this reverse engineered but to reverse engineer just one module's message data can be hundreds of hours of labor. It can and is done in the performance aftermarket - for example we have a 2021 STI running on a LINK standalone and because it's an "enthusiast" vehicle there have been efforts by the manufactures of aftermarket ECU controllers, etc to decode the CANBUS data so their products can interface with the vehicle. But it's literally something you pay a CANBUS engineer ~$10,000 to decode for your product. GD
  6. No. No practical way and worse than that - no one at the dealership that is A: Skilled enough to do it, and B: Willing. You see every technician at the dealer level is paid "flat rate" - so they get paid book time. And when you can do a 2.8 hour spark plug job in 1 hour and get paid for 2.8 hours..... or you can troubleshoot some random CAN code that at best is going to take MANY hours and you will get paid probably for fewer hours than you actually spend.... which are you going to choose? They are going to scan it - write "unable to reproduce customer complaint" on the invoice, get paid their 0.5 for the scan tool diag and move the heck on to an actual paying job. They don't have the skills and the dealer isn't going to pay them for the research necessary to acquire those skills for one problem child car. It's not economically viable. Now you are beginning to understand..... Could a team of engineers with a laboratory equipped with a dyno that can angle and shake the vehicle while it is simultaneously driving under various ambient conditions and is hooked up to CAN sniffer equipment, and all 4 channels of half a dozen oscilloscopes find the problem? Given enough time..... YES! Is that likely to be funded by you or anyone else? No - as a society we will just throw that car away and build another. This is how it's basically always worked with technically complex manufactured products. Hell my LMTV (US Army truck) had an electrical fault that took ME eight months to find. Because it was intermittent and the troubleshooting documents and the engineers that designed that particular system (yes I spoke with them - multiple times) did not foresee this particular interaction of failures and so the symptoms could not be correlated with the actual problem. I eventually found it by pure force of will and by resorting to testing every single wire end-to-end for every component involved in the subsystem exhibiting the problem. I found a manufacturing defect in the truck's wiring harness by application of brute force, time, and a lot of luck. And that truck has no CANBUS in the sense that none of it's modules talk to each other. The US government spent $230,000 to purchase that truck in 2008. They sent it to Kuwait and then shipped it back to El-Paso. From 2008 to 2017 the truck accrued 2,045 miles of use, and was then deemed "surplus to needs" and (probably) because of the electrical fault and some minor damage (not related to the electrical and easily repaired with only bolt-on replacements) it was not selected to be upgraded to full armor so was sent to auction. This is the DOD - with a budget of $860 billion. It was deemed not economically viable due to a single wire not having been soldered correctly at the factory that "in practice" no one could actually isolate. So the whole truck was sent to the corn field. Sadly - unless you want to become an expert in CANBUS networks and sniff out the problem yourself - at the cost of great expenditure of time and effort - Subaru could care less if they lose you forever as a customer. It's a statistics thing. A convert will come along that had a bad experience with a Ford to replace you. It really isn't to anyone's financial benefit except you to spend the effort to find the problem because in the end it will require so many hours to do so that it will far eclipse the value of the car and you won't pay for that so no technician is going to spend the time. It's a bit of a societal problem in general but the ever-increasing complexity, the drive for ever-increasing profit by creating the artificial need to replace products frequently, and the ability to release products while they are still in a software beta state and "online" update them later if and when there's enough complaints is really reaching an intolerable level. At least for me. Basically everything we buy is now total garbage and we have forgotten the two most important of the three R's. Everyone thinks it's all fine as long as it's "Recycled". No one bothers one bit with Reduce or Reuse. Just throw it "away" (where the hell is that exactly?) and get another one. Someone will recycle it.... probably into a landfill or into your lungs via a convenient incinerator. GD
  7. Woof. That's rough since either the ECM or the TCM could be at fault and the car won't operate without them so you can't just disconnect them and see if that stops the codes. At that point I would start with disconnecting and cleaning both of their connectors (Deoxit is a good choice for connector cleaning) and see if the issue returns. Followed by probably replacing the TCM since that doesn't require programming (I don't think anyway), and if that doesn't do it..... might have to replace the ECM but that requires programming the immobilizer, keys, and VIN, etc. Could still be grounding issues also - clean every connection you can find. Also these low voltage coms can be disrupted by outside influences - bad alterntors, aftermarket stereo and accessories, etc. Remove all aftermarket wiring and return the vehicle to 110% stock if it's not already. The ECM's seem to be getting less reliable for Subaru's. We had one that seemingly got smoked when the customer smashed his COBB Access Port cable in the door - we ordered a replacement and after programming it worked but continuously threw a code for one of the intake runner control valve actuators - after troubleshooting it down to the ECM itself we bought a second brand new unit and that solved the problem - so #1 (maybe) killed by a short on the OBD-II port, #2 bad out the box from the dealer (returned as faulty). GD
  8. Don't need a machine shop. Toss in some new valves, lap them, and set the lash. Resurface with glass and 120 grit. Send it. GD
  9. Knock sensors are simple devices and the ECU is only looking for a resistance value generally. Unless you plan on running poor quality fuel I would just stick a resistor in the circuit and send it. One thing you learn from years of tuning and owning a dyno - knock sensors are one of those "last 1%" items. Generally the engineering and preparation that goes into their use makes them worthless and inconvenient for motorsports, engine swaps, etc. They don't like noisy environments like custom exhaust headers, custom engine mounting situations.... they like very quiet grandma friendly environments. They will usually tell the ECU to pull a bunch of timing from erroneous "knock" detection in the types of environments we stick them in with regards to engine swaps, high performance modifications, etc. You won't find them on race cars. They are not applicable to that environment. Thus why I suggest to just bypass it and forget about it. Alternatively you could find/measure/guess the frequency response of the stock knock sensor (it's typically chosen based on cylinder diameter), and source something in the same frequency range and change the connector, etc. Knock sensors are just piezo-electric microphones that are tuned to hear a specific resonance frequency of the cylinder bore size. Figure out the frequency and find one similar from any other model. Problem solved. GD
  10. None. All USDM models have the ECU under the dash or under the passenger floor. We did not get the EJ20 except the EJ205 turbo. EJ18, EJ22 and EJ25 only for USDM. GD
  11. Nope. That's a CANBUS module also. Most cars think you are stealing the stereo and won't even attempt to start. GD
  12. In other news it won't be long before I retire out of the Subaru (and consumer vehicle) repair world. I'm going to move to retired military hardware. Both the CUCV (square body chevy truck military platform), and the FMTV's like my M1079. Stuff you can actually repair and that was built to be durable and to keep your privacy intact. Maybe head up to the northeast and do some snow plowing with M1083's with sanders and plows. I've had a lifetime's worth of plastic consumer trash and I'm going back to my roots and my first love - OD green, desert sand tan, and 3-color NATO. GD
  13. The early DBW (drive by wire) isn't so bad in and of itself - the throttle body's have actually proven to be quite reliable (sadly not the pedal assembly however). It wasn't till about 2008 that most models went completely sideways with CANBUS networks. Do see plenty of the 05-07's with bad computers though. For that matter I see older Subaru's with bad computers though - just getting past 2004 you have to worry about programming them with VIN numbers if they require inspection and that's not always possible. Generally everything I see on the road now I just absolutely cringe at. In the last 10 years it's not even so much the complexity and the repairability - now they are a MAJOR privacy concern. The car industry is just absolutely a wreckage when it comes to privacy concerns. It's literally the wild west and your data is the stagecoach full of railroad payroll. Read: https://foundation.mozilla.org/en/privacynotincluded/articles/its-official-cars-are-the-worst-product-category-we-have-ever-reviewed-for-privacy/
  14. Flashing cruise is a result of the check engine light. Cruise control is automatically disabled as a driver alert for all system faults. It's to annoy you into getting it fixed. Battery seems fine although Optima went straight into the toilet when they sent production to Mexico so I wouldn't recommend buying any of their products. Not that any of the battery manufacturers make really excellent consumer batteries anymore. Most are crap now. AAA stats show batteries are only lasting 2-3 years now due to quality and heavier electrical load on modern cars.
  15. You would have to consult the network diagram but virtually all "modules" are on the network. Some things that are modules are not obvious - like the entire gauge cluster - that's a module. Everything down to the electronic parking brake has a module. The more advanced diagnostic tools have moved to an interface that shows network "topology" so you can see all the modules laid out sort of in their network diagram. The Thinkcar tools are popular for this: As for "A" - no it likely won't point to anything unless it's a very advanced vehicle that has more than one network. Not likely in this case.
  16. U codes are network errors. This almost always comes down to a wiring or module fault. All you can do is hook up an advanced scan tool and follow the clues. If there are no clues the only choice is to clean every ground, check every connection, and if it still comes back you start disconnecting modules while monitoring the network (in cases where a module is actually crashing, or corrupting it) or replacing modules (at great expense for parts and programming since used one's cant be VIN coded) in the order of most to least suspect based on the opinions and experiences of the dealer technicians that have seen these problems in the past. But at the end of the day it's a LOT of guessing. And if you aren't educated guessing it will take 10x as long. So factor that into your cost analysis of if you want to tackle it yourself or not. I have plenty of vehicles with computers but I stick with vehicles that don't have networks of them. Discreet modules only that only share data by analog signaling or raw sensor data. The automotive CANBUS networks are very poorly engineered in terms of redundancy and error detection - in fact they basically have neither and you are left to swap parts randomly and guess at every turn because not only is there no error detection to point you to a cause, but all the network data is considered proprietary intellectual property. GD
  17. I've had a couple customers encounter this and every time we have tried to lube them up and get them working they inevitably come back with the wheels locked up again so we just decided that the customer's choices are replace it or disable it. They can also leak so we have deleted them a time or two as well. Not that hard to do. GD
  18. CVT is junk, CANBUS is junk. 10 years old - time to trade up or trade back to something that's not a rolling plastic computer. Is your phone 10 years old? Is your TV 10 years old? Exactly..... GD
  19. Valves are bent. Probably all of them. Sounds like you're on a budget with more time than money. Tools and free internet education is cheaper than a mechanic You can often return tools after use - the various auto parts stores do this on purpose to get you in the door to buy parts (not required that you purchase any parts though). Amazon and ebay sell complete valve sets CHEAP - never had an issue with any of them if setup properly. Get HG's from Subaru ($150 for both). Get an Aisin timing kit and some cheap fluids. Unless you keep all the tools you buy (or hopefully you already have some of what you need). this can be a $500 repair. Or if you go to a shop and get the heads rebuilt and a new short block from Subaru because it's old and burns oil, etc then it can easily be a $6,500 repair. GD
  20. It depends on the level of computer integration you are willing to deal with. Mechanically the EJ engine ran till 2012 in the Legacy and could be had with a 6 speed manual. Those do tend to have fire ring head gasket failures and sometimes dropped valve guides but if you correct those two issues they are mechanically pretty sound. The CANBUS and the electrical system.... as long as you don't keep the car past about 200k it will probably be alright. Don't get anything with a sunroof and if it's ever been in a flood just forget about it. As for non-CANBUS cars - 2005 and some 2006 is about the last years. They have some issues with drive by wire, airbag indicator/dome light circuit boards, bad HVAC/radio solder joints, head gaskets weeping oil, manual transmission center differential and rear input shaft bearing failures, and the last few years we are seeing a more engine computer failures either causing really bad misfires due to ignition coil driver failures or phantom misfire codes that aren't real misfires but require the computer to be replaced to stop them from throwing codes. The problem is starting to be that older Subaru's that are actually reliable and would otherwise be desirable are suffering from parts availability. Engine ECU's for the 2005+ models have the VIN number coded into them and can only be programmed one time so you can't easily use a used ECU from the junk yard - and Subaru is actively discontinuing the parts - there's already models that you can't buy a new ECU and a used one will not work due to registration issues, etc. At some point it becomes a real problem. And then there's the availability of drive axles, etc - can't get any axles for 2004 and older from Subaru now. And aftermarket are absolutely trash. GD
  21. And don't keep it after the warranty expires. The FB engines and CVT auto's are hot garbage. Just had a 2019 come through the shop with a bad CVT at 34,500 miles. Dealer parts guys just roll their eyes when you ask what the hell is up with these cars...... total crap. GD
  22. Lug pattern and offset is the same. From there it's a matter of tire choice. GD
  23. Take it apart. If it needs a gasket - make one. If it needs an O-ring - make one. Clutches are often pressed on. Sometimes splined on the drive plate and bolted or snap ringed. Often the clutch bearing is pressed onto the housing. What gives is that no one repairs them anymore. It's cheaper to have China make another one. So there's no market for repair parts. Thus no one makes them, no one stocks them, and they never will. Why make/catalog/stock something no one will buy? GD
  24. The slave cylinders from that year range have a Subaru service bulletin about weak internal piston return springs. It causes the fluid to be displaced in the system - the pedal will not return to it's upright position and the slave piston will not retain pressure against the fork which causes insufficient throw of the fork and failure of the clutch to fully release. Also the rubber hoses always leak. Get a new NABCO slave cylinder and a new hose. Look harder - it has a steel line AND a hose. GD
  25. Anything is possible with enough engineering knowledge and time and a CANBUS sniffer. You have this experience and knowledge or money to pay for it? It smells like no since if you did you literally wouldn't need to ask these questions. One thing you cannot get around in the state of CA is that they are now doing checksum calculations of the original software on the ECU so even if you were able to make changes with a Tactrix cable such as disabling the immobilizer system, etc - it wouldn't pass smog because changing even a single bit value of the original code would change the checksum and it would fail due to software tampering. iWire is located in CA and could advise you on the particulars. Don't be surprised if this is economically unfeasible - you essentially have to recreate the Tribeca in a software emulation just to get the engine computer to be happy enough to pass a government mandated 30 second inspection. You really want to do that? F*ck that state in my opinion. Register the car in Utah or one of the other states used by https://www.dirtlegal.com/ https://iwireusa.com/blogs/iwire-university/canbus-part-2-can-bus-and-your-subaru-swap GD
  • Create New...