Hank Roberts

Good advice. I'd been thinking I ought to get the shop that worked on it to at least look again in case anything they'd put in was blown or done wrong -- they wouldn't be responsible if someone else had gotten into it first, of course. One thing might be confusing this, I'm checking -- The car has had an engine swap, a used Japanese engine that has about 60k miles on it, put in a couple of years ago -- so I'm asking the guy I got it from if he's sure the _engine_ is still a 1988! The car's a 1988 GL SPFI turbo. Definitely still single point fuel injection. The test leads are under the hood, I've found both the green pair and the white pair. Odd, they were pushed way down under a wire bundle, doesn't even look like they've been plugged in to each other. I haven't tried connecting either of them -- apparently don't need to, which puzzles me. I can read the ECU by looking up under the steering wheel -- I don't even need to drop the panel, though I did the first time to make sure what I was seeing -- just move a bit of foam to the side with my finger in the gap above the panel and I can see the red light on the ECU, it's blinking 3 slow, 4 fast, repeating -- the code is clearly a 34 -- it's showing as soon as I put the key in and turn it to the run position without starting the car. About the under-hood connectors -- they have right next to them a tan-colored multi-connector plug and a black multi-connector plug (which don't mate up to each other) -- which puzzles me and makes me wonder if this could be a different year engine. Too rainy and cold to do more today in the yard; tomorrow I'll go back to trying to make sure I know what the EG solenoid actually is! and see if I can measure resistance across it. Remember, I've now owned a Subaru for a total of about 24 hours. But this time tomorrow, I'll have owned a Subaru for twice as long!

I've also got to mount my ham radio antenna (I'm N6VSB) -- looking into this, I've found you can have a good DC ground electrically that is still not a good RF (radio frequency) ground for ham radio operation. Here's a discussion (follow links therein for much more detail): http://www.k2bj.com/Ground.htm/ He's using flat copper braid, tinned, for grounding all the parts of the vehicle together -- hood, trunk lid, several places along the exhaust, and so on. Remember this is grounding radio frequencies (the vehicle is serving as the 'ground plane' for the ham radio antenna). He has a good discussion of electrical ground and ground loops as well. -- And a note on terminology -- "dielectric" is confusing, it's actually a term that needs a number with it to describe how much electric potential is needed to cross it; there are highly insulating dielectric greases sold apparently for spark plug connections to 'keep the electricity inside' -- they allow the current to go where it should but not find other paths. There are specifically conductive greases. There's the high end six ounces for forty dollars stuff (!) http://www.2spi.com/catalog/vac/silver-filled-grease.shtml (silver when it is an oxide's still a conductor; copper oxide is an insulator) Whew!

>trouble tree Now we're talking. Can you tell me for the 1988 GL turbo, what reference book has this? I see people selling them (a four volume set, sometimes). I may have to provide it to the mechanic that worked on the car. And the way I approach things, that's what I'd like to have for myself anyhow.

Thanks, I'll hunt around in Berkeley area for some place that might do it. The old ECU removed from this 1988 GL is the size of a hardback book -- I'm guessing the PROM chip was not separate at that point, and maybe it's got some power relays inside this big metal box. Here's why I'm not just diving in and part-swapping til the problem goes away -- I've had bad carpal tunnel (failed surgery, got it too late to fix the nerve damage done by decades of keyboard/mouse work), I can't do much fine work myself --- so I try to understand things even when it costs more, rather than just pay other people to swap parts and tell me it's fixed.

Where is a place I can send an ECU for testing? I understand this is a debatable thread, I don't want to debate it here, I'm looking for information in this thread.

Is there a place I can send an ECU (pulled out of a 1988 GL) to have it checked? I know computers and computerized controls -- but never had a car new enough to have a computer in it til, um, yesterday. And from what I know of computers and cars -- who was the genius who thought of combining them? -- to properly test a piece of computer gear you plug it into a test board that measures all the outputs and provides all the inputs, step by step, a few hundred times , and eventually you isolate whatever's out of spec. Anyone do that? Or is this all throwaway stuff now?

No, it ran fine most of the way home -- and when I looked under the hood, the green connectors are not connected. But the light stayed on once it came on while driving, and comes on as soon as I turn the key now. I will be able to look at codes tomorrow or this weekend - I just had time to look under the hood to locate the connectors and try starting it a few times this morning, and call the mechanic who'd worked on it to tell them the problem hadn't really gone away after all, or had come back. Unfortunately the mechanic didn't write down the code numbers on the paperwork. The problem they claimed was that the computer ECU wasn't sending a proper signal to the EGR valve, replaced the valve, then replaced he ECU saying they'd found it had an internal short. Since it wasn't my car at the time, I figured, better they fix it before I take title -- which I thought had happened. Now, it's mine -- and I'm hoping they warranty their expensive work done four days ago!

Edited 5/2 changing the title -- current best guess, it was the passageways/ports that carry the exhaust gas from the exhaust side to the valve, and from the valve back to the combustion chamber, that were packed solid with crud. They're now being cleaned out. Hoping ... ---- Well, I _almost_ got it home before the Check Engine light came on again. Maybe this is a clue -- I'd been at the highway standard 65-75 speed til everyone came 'round a curve and saw stopped traffic, bridge work going on. Sudden stop, then creeping along at 2 to 4 miles an hour for a while, then a stretch of a mile or so at 25, then everything cleared out, I sped up to 65, and the Check Engine light came on. And it's stayed on -- I let the car rest overnight, started and stopped the engine five times to see if that'd change anything -- when the key is turned right to the running position but the engine hasn't turned over, I can hear the fuel pump and some relay under the steering wheel clicking and the Check Engine light flashes. As soon as the engine starts up, no more flashing or clicking. Suggest anything to anyone? I know where the leads are (under the hood in front of the driver, for the 1988 SPFI) and how to take down the codes, so I'll do that this weekend. I know grounding may help, and I know how to test resistance and will do that across whatever sensors I find. And I know the reports that the EGR relay coil was fabricated with no solder connecting the leads, apparently trusting the plastic coating to maintain contact and that these do corrode over time and the resistance goes up -- so assuming I can find that relay coil I can check it. And I know a lot of questions to ask the mechanic over at the other end of the Bay who thought this problem had been fixed before I bought the car yesterday. I may have to take it back to him before doing anything since he does warranty his work - but I'd sure like to know if he is doing it right. Wish I had the ability to really work on this myself, I don't, so beyond trying to understand and check things, I may be looking for a mechanic in Berkeley a lot sooner than I thought.

Where do you get these? I have a new-to-me 1988 GL 4wd wagon, and sure have been envying the brush bars on the front of similar Subarus. Hadn't come across them available anywhere.

So, did you get yours running? I solved my '69 Dodge van problem -- it was sitting on about a 4 degree slope, gas tank end downhill. The fuel pump of course is on the engine end -- uphill. Way uphill. After I added the sixth gallon of gas to the tank, the fuel level was high enough in the fuel line that the fuel pump was able to suck. Runs fine now. D'oh. And no, the gauge hadn't said empty ....

I found different information -- including in several places the warning NOT to connect the green connectors first, because this erases all the stored code information you want to find out. The original of what I'm posting below: use a Usenet newsreader (if nothing better, use "Google Groups") and search for the newsgroup (I think Google calls it a community or something but it's a newsgroup) alt.autos.subaru search for: "Curtiss Priest" +"Technical Note" +Subaru This is quoted from his note on ECU code reading: QUOTE: The following description refers to a 1988 DL. +++ the 1988 SPFI version did have the wires to clear codes and display codes inside the hood (next to the diagnostic connector block). 1. The Green single contact plugs -- this, when connected with the ignition key in the On position, does clear all codes and extinguishes the "Check Engine" light 2. The White single contact plug -- this, when connected, causes the ECU to emit "memorized" codes of faults Further. If there is no current fault, the ECU displays 5 short flashes, a pause, and 5 short flashes, etc. (with neither white nor green connected) When the white block is connected, the ECU then finishes its stream of 5 short flashes and changes over to displaying one or more fault codes... END QUOTE -- see original for full info

I love the Kurt Lewin quote. I haven't seen anyone post about grounding the ECU itself, til your post. And, since I'm just about to pick up a friend's 1988, which had the ECU's internal ground signal blown up -- ECU had to be replaced, and the new one properly grounded -- this interests me. I do worry about ground loops -- too many different contact points to the frame would be as bad as too many different ground rods hammered in around a house, I'm sure current flow between them would be a big problem.

>silver Yes, in California I've been using fairly low temperature silver rosin-core solder for all my electrical/electronic work, it's readily available. Much, much safer now and down the road. I've also been using "hard" silver "solder" -- it's actually silver brazing, the bond interpenetrates the metal surfaces rather than just stucking to them -- much higher melting points -- for jewelry/sculpture soldering of brass and stainless steel for many years. But for those silver mixes, you need to get the metal red hot -- and then drop the piece into boiling water to blast the much nastier flux off, so it won't corrode in a few years' time. I suspect that the boiling water would be good for cleaning any soldered electrical connection -- it makes the flux remaining on the joint literally pop off, leaving a much cleaner surface than I can get if I just let it cool even slightly and then try to sandpaper it clean. The water has to be actually boiling for this to work on sculptural soldering, however. I want to get myself a meter and start checking all the resistances between various points under the hood and the negative battery terminal, having seen people report these numbers start increasing after a few years as the ground starts to corrode. Maybe some day we'll have sensors on our sensors, to detect failing ground connections on our sensors .... "and so ad infinitum."

Here's another aspect -- from the aircraft builder side: "Automotive wiring is covered with PVC insulation which gasses-off cyanide gas when it melts or burns,...a major contributor to deaths in aircraft fires where otherwise folks were uninjured. Automotive and Radio-Shack switches are not intended for high-vibration and can internally disintegrate due to the abusive aircraft environment. When they do they can spark, short-out and overheat and sometimes be the source of the previously mentioned fire. " Link: https://www.globalair.com/discussions/georges_hangar/article.asp?msgID=475#replies

Here's a somewhat different take I just happened on, from the boating side. The thoughts about corrosion and vibration may apply for cars too. <a href="http://www.stripersonline.com/cgi-bin/ubb_547C/ultimatebb.cgi?ubb=get_topic;f=5;t=001468;p=0">How to rewire boat?? - StripersOnline - SurfTalk</a>: " Here are a few pointers: 1) Always use tinned multi-strand marine grade wire (the tin prevents the copper from oxidizing). Anchor products are arguably the best, and not cheap if you buy from West Marine or Boaters World. 2) Use only marine grade crimp connectors on everything. 3) Use a good 'double' ratcheted crimper for all connections (Anchor makes this too) - NEVER SOLDER connections because the vibrations in a boat may cause them to break where the wire flexes by the 'stiff' solder. 4) For general lighting and accessories use 14 gauge wire for total loads less than 15 amps and a total run of ~50 ft. If the load is more or you have a long run, increase the qauge accordingly. Check out Blue Seas website for exact load/run values. 5) Use marine grade heat shrink that has adhesive on the inside to make water tight connections. Do this even on terminal ends where the end of the wire is exposed. To seal the ends, just make sure its cover the ends and after you heat it, press is closed with your fingers. 6) Get a good heat gun - don't use a hair dryer, you'll waist a lot of time. 7) Use electro magnetic grease on all fuse and bulb contacts. 8) While your at it, replace all running and navigation lights with LEDs. They are brighter and won't drain the battery. That about covers the basics."

I've been looking for more on this engine control stuff; here are three good posts you can find with a Usenet newsgroup reader (like Google 'Groups'). I'll add others. These three stood out. Search for the subject, in the newsgroup alt.autos.subaru written by W. Curtiss Priest Reading Subaru Engine Control Unit (ECU) Codes Technical Note July 5, 2000 1980-1992 Subaru Automobile Reading Engine Control Unit (ECU) Codes 1988 Model DL ... (same codes found elsewhere but some additional thoughts) How-to repair SPFI cold idle problem Technical Note #2 November, 1998 1980-1988 Subaru Automobile (there's a shielded wire that frequently breaks internally, up close to firewall) Repairing EGR Solenoid Valve on Subaru 1988 Model DL ... (not the valve, the coil goes bad; how to test and how to replace)

I don't have an answer, but this sounds familiar -- you said you'd killed the battery and were going to recharge it. I'm facing _exactly_ this problem now with my 1969 Dodge. Gone through all the same steps, recharging repeatedly. Currently, unless I've missed a block in the fuel line (tomorrow's task) -- the best current guess is simply that I have a weak battery, that when "fully" charged turns the starter over but can't simultaneously make reliable ignition sparks. A "Duh!" if true. If not that (sigh) you'll be where I was last week with my friend's '88 GL -- You'll want to read up on all the ECM (Engine Control Module) postings. (This seems to be the dark secret of all post-1983 or so automobiles, they have computers in them and no absolutely certain way to test them, semiconductors being quantum mechanical things that get vaguely off spec so odd things happen). Seems to come down to (my amateur newbie ECM summary): ECMs usually don't go bad, and can be obtained cheap(!) from junkyards -- and swapping one in can resolve problems if yours has gone bad, but(!) they will help only after you've made all the input sensors work right. Otherwise, they'll tell you the same problems. Older ones (pre 88??) erase themselves; later ones hold old codes in unerased memory forever until cleared so you have to swap the ECM, clear it of old codes, then (uh, I dunno, run the engine a while, off and on, which won't help you much!) get them to take readings on your current car. MAYBE, if it's an engine control problem, "garbage in garbage out" then a replacement ECM would tell you engine to wake up and run right. They're computers. Check all the grounds and connectors for sensors, ground wire connections. Find out what your particular model's ECM does -- whether it's just making idiot lights to warn you of things or it's actually changing how your engine operates (which they do, more and more, for later years). I haven't found a good source for that year by year but am assured that by my 1988, having an unreliable ECM means having an engine that's sometimes confused. Little stuff -- you have a fire extinguisher handy working with fuel, eh? -- and eye protection around the battery, and someone within hollering distance? Forgive me saying the obvious but a spark around a battery can splash acid everywhere. -- starter spray is very volatile ether, you can spray it into the air intake opening, leaving the air filter on (avoids having the engine open, less risk of fire where you don't want it) -- it has to mix with air on the way in, but you don't want evaporating out all around your face when you crank the engine.

I don't know this from personal experience, I'm just learning as I buy a friend's '88 GL in California. The computer apparently doesn't just monitor sensors and turn on an idiot light, it also has outputs that depend on the inputs. It's not just idiot lights, it's idiot engine management. "It never does just what I want, but only what [the sensors] tell it" or inother words "Garbage In, Garbage Out" -- if all the sensors are not good, even if the computer itself is good. Now as to exactly WHAT the computer will be sending bad instructions to .... um, got me now. Anyone KNOW what functions are affected by the computer on these older Subarus? I'd sure like to know. For sure. Word I've rec'd by email is -- the ECMs are usually good from junkyards and may be as cheap as $20. I've found third party new ones on the web for about $150; remanufactured Subaru for around $350 (bet that's what your mechanic was looking at) and new Subaru for (gulp) over $700. Everyone's response at first seems to be ignore the light (except the California Smog Board, which, um, disagrees). But -- from what I can find out -- ignoring bad output from an engine computer is rather like ignoring bad output from, well, any computer. You get troubles. I dunno. If you can find a cheap computer and swap it in -- after making sure all the sensors are good and the electrical grounds are good and the wires are all good .... write if you get it to work. I've never owned a car newer than 1983, til now. I dread computers.

Me too, if I can put this on a 1988 GL turbo wagon (that I don't quite own yet but will next week ...). Email also sent.

I've just been facing the same question in California with a 1988 -- last possible explanation came 'round that it's the actual computer itself that needs replacing. Most people have said go to a junkyard -- try searching for "ECM" or "Engine Control Module" -- and apparently these things often are good out of scrapyards and are dirt cheap. I've come across third party manufacturers in the $150 range and non-dealer (maybe remanufactured) in the $350 range; that's half dealer new price for the model I'm looking at replacing. If you've already fixed and checked all the inputs -- all the sensors and relays are good, the wires all are intact and insulated except where they should be making good electrical contact -- and the outputs from the ECM are likewise electrically good, including the ground -- then do what I didn't, try to find one cheap you can swap in. I can tell you, from the experience of the guy whose car I"m buying, that ignoring an ECM light on can degrade performance, clog up the PCV valve and tubing and channels, and I dunno what else. They're not just idiot lights, they're idiot _drivers_ and what they do changes how the engine works, so should be correct. Note that I know nothing at all first hand, I'm passing on what I"ve learned in several days of extensive reading, while facing this dilemma.

Me too -- just in the process of buying an old friend's 1988 GL 4wd Turbo. Waiting on a replacement computer, since nothing else fixed the check engine light (and it'll have to be fixed in less than a year for its next smog check). The car has had both front axles replaced -- one last week, one last year -- and the mechanic told me the right front wears out faster because it's driven ...um, harder? more directly? -- and looking at the history, this car has had front axles replaced before, too. What else is there to look for? What about improvements -- adding more cooling for example? What are recommended oil grades (this is a used Japanese engine, supposedly under 60k at this point, that looked quite unworn going in). Also -- the previous owner used "Sea Foam" for a rattle on cold startup. I wonder if it gunked up his engine -- anyone know? His Check Engine light was on and being ignored, and he ended up with his PCV valve, the tube leading to it, and the channels in the engine block leading to that all gunked up completely by the time I got it checked. Pricey to take apart and clean out. How about the air conditioning -- any easy take-apart for that? I'd bet the condenser is full of mold by now.

OH, and a followup question -- I'm assuming a bad computer has to be replaced because it manages the engine -- right? These things don't run right without the computer, do they?

Pretty close to what I came to ask about -- I'm buying an old friend's 1988 GL Turbo 4wd wagon (he needs the money, I need something with low range gears for field botany work in the mountains of N. Ca.) It's got 245,000 miles on the body and maybe 60,000 on a replacement used Japanese engine. Garage did an inspection, fixed some little stuff, found that because the owner had been ignoring the Check Engine light for 10 months the PCV valve and channels leading to it were totally gunked up; cleaned those. Got the "check engine light" off, had it checked and it passed CA smog OK. Paid them, drive it away -- and then the light started coming back on after 10-15 minutes and eventually stays on no matter what the garage did. After quite a few free hours of work, they concluded everything else is OK, and that it's the actual onboard computer chip that's failed -- it's not signaling "ground" properly -- they've verified that the actual electrical ground connection is good. $900+ replacement cost, for a new one. Either now, or in 11 months when it has its biannual smog check done again. ---> So, same symptoms, I think, assuming you've also checked everything else?