polychromeugand

Sorry for the delay. Yes, with the TCU disconnected nothing electronic can change the behavior and it can only be a hydraulic problem. The key off time is as brief as possible, but allowing for the tachnometer needle lagging the engine I'd say the engine is close enough to stopped. The transmission does have a low enough seriel number to have the oil pump housing flex and a number of other issues that were later fixed. I don't know if the metalized gasket and other band-aids were applied without checking the old dealer service paperwork. At this point my guess is that when I cleaned the control vavle assembly it restored the pressure regulation, and ended the delayed engagement, but there's another hydraulic circuit, possibly in the oil pump, that hangs up at full pressure. If I connected the pressure gauge to each circuit and recorded the behavior I might be able to isolate the current problem, but... Since its become certain that any repair will involve removing the transmission its become more expedient to swap it than to do any more work on this one. It seems to have reached the point where if you try to fix one thing at a time something else will fail so it needs to be rebuilt or replaced. Since I've torn my rotator cuffs I should have someone else remove and install it and for $350 for a used transmission they can do it all at once.

Yes... I unplugged a square connector with 16 contact cavities and 13 loaded contacts on top of the transmission near the firewall that probes out with all the correct resistances and grounded pins as the transmission and after unplugging it the transmission was stuck in a single forward gear. (The other connector near there is rectangular and has 12 contact cavities.) At this point I don't have any useful ideas. This transmission has a low enough number to have the oil pump housing flexes and blows out the seal problem. (and all the other housing related problems later revised). I don't know if it ever got the the various band-aids, but whether it did or didn't it might take a lot of time disassembling this transmission and figuring it out, and applying more band-aids and at the end of the process it will still have almost 200,000 miles on it. Unless anyone has had any similar experience it seems like its time to either put in another transmission or scrap the vehicle. Admittedly anything I can get from the scrap yard is going to be old as the hills, but if I can get a '93 or '94 it won't have half of the design bugs.

I know what "buck and sputter" means for a carburetor engine, its usually a little different for a closed loop engine. Its only in "open loop mode" while the O2 sensor is cold. Restarting the engine may go into closed loop very quickly. A dirty, or fouled, O2 sensor is often very laggy, and the closed loop control breaks down. Some of your description points that way. It might be worth visually examining the upstream sensor. The Bosch web site has some good pictures of what different kinds of fouling look like. You didn't mention the condition of the donor of the swapped parts. During a true cold start it should be not only open loop, but mapped very rich, to the point of a smelly exhaust. If not, check the coolant temp sensor. If you can get your hands on a cylinder leak down tester the answers are more useful than a standard compression check. You mentioned a radiator swap, and sometimes the root cause is a cylinder leak. Dirty coolant can be a hint, but not every time. You also haven't mentioned anything about the air filter, cleaning the throtle body, cylinder carbon cleanout (SeaFoam), or injector cleaning (Lucas, Mystry Oil, Techron, etc.) Any of those could cause the problems mentioned. In my personal experience, dramatic bucking and stumbling with a closed loop engine that idled well has been due to a bad O2 sensor, but I'm picky about what I call a good idle.

Well, I took it for a couple of rides just to be certain of what is happening. It takes a bit longer to begin disengaging, bit it still disengages. Without the TCU there is a prolonged tail-off as it disengages the clutches. It also still comes back after the briefest of key-off cycles. There is never any engine braking. There may be some information in how it behaved. If I understand Section 3.2.M.24 Fail Safe Function; Solenoid 1 & 2 off should put the hydraulic circuit in 3rd gear hold, but it drove like 2nd gear. Solenoid 3 off should lock up the overrunning clutch. The overrunning clutch was wide open - there was no engine braking. This is an unexpected result because the manual's description is just the opposite. The manual's description of how the function works doesn't seem to agree with the hydraulic schematic. Duty Solenoid A (Line Pressure) Off should be max line pressure - I didn't put the guage on it. Duty Solenoid B (Lock Up) Off should release the lock-up clutch leaving the torque converter to transfer engine power at all times. Duty Solenoid C (Transfer) Off should leave it in direct coupled AWD at all times. I didn't notice any unpleasantness while making a U turn at minimum steering radius. Right now there is about as little sand on the roads as there ever is. I expected something like when the differential gear oil has over-ripened, only more so. So... I figured I would try the stall RPM. Its 2700, (book says 2600 to 3000 is OK.) The test can't be run for more than a few seconds, but it stays good that long. Obviously when "disengaged" the stall RPM woul be very high. Well, its still confusing, but I suppose this lets the TCU off the hook for the moment.

In MA road salt dooms the heat shields, there's nothing in a scrap yard that looks any better, and they're hideously expensive from Subaru (a set would be $700 or so). Since's there's rarely dry grass to burn, when they fall off they don't get replaced.

I memory serves grey was for California emissions and red was for everywhere else. You may need to swap all 4 so they have compatible timing delays. You may also be better off swapping the whole fuel rail so you don't break the injectors while extracting them from the fuel rail.

The suggestions so far are good tune up ideas. If they don't resolve the problem, the problem could be either the idle switch or the idle air control valve, and it could also be the coolant temperature sensor. When the engine control unit senses that the throttle is closed via the idle switch (part of the throttle position sensor) it uses the idle air control valve to adjust the amount of air entering the engine to set the idle RPMs. The ECU uses the coolant temperature sensor to decide if the engine is cold and if so, it runs the engine fuel-rich to allow it to start and idle despite the poor fuel vaporization. Below 30F it can be very difficult to start the engine if the coolant temperature sensor is defective and at 0F its all but impossible. A defective coolant temperature sensor should cause a stored code but may not always light the MIL (malfunction indicator lamp or check engine light). Its a fairly common part failure and difficulty starting a cold engine and poor idle when cold are typical symptoms. BTW - The temperature gauge on the instrument cluster has an independent sensor, that never seems to break. Unfortunately the ECU can't read it. The part is usually somewhere between $20 and $30 and its reasonably simple to change. The idle switch is the top two contacts on the four pole connector on the opposite side of the throttle body from the throttle cable. To test it disconnect the wiring harness, and using an ohmmeter verify that the contacts open and close whe the throtle is moved. You shouldn't have to test this manually... If this were defective, the MIL (malfunction indicator lamp aka check engine light) should be lit and a code should be set. The idle air control valve gets gunked up pretty easily, but its also fairly easy to clean. All you need is a can of throttle body cleaner. The idle air control valve gets air through a small hose that taps into the large air intake hose that runs from the air filter to the throttle body. With the engine off, remove the air intake hose from the throttle body open the throttle and spray all that clean and reassemble. Next, detach the small hose that leads to the IAC (right next to the throttle body) from the air intake hose. Spray throttle body cleaner down the hose until it just about drowns and reconnect it. Start the engine. You may have to open the throttle a little to get it started, and it may die a couple of times as it pulls the cleaning fluid through. I find it takes two to three cycles of drowning it in cleaner and starting the engine to get the IAC clean when its sludgy. You'll know its wokring correctly when the engine idles correctly again...

One of my two 1992 Legacy wagons has a transmission problem that has me stumped. After its started cold (well, it summer here) it operates well for a couple of minutes and then ... it just seems to disengage the clutch. It doesn't re-enagage while the engine is on no matter how long its left in Neutral or Park. Turn the engine off, not even long enough for it to come to a complete stop, and turn it on again and all is well again for about 20 seconds, and then it disengages the clutch packs again. This can be repeated just about endlessly as far as aI can tell. Oh, and its exactly the same for forward D 3 2 1 and Reverse and with or without Manual engaged. Before this began there had been some badly delayed enagement on shifting into gear. The valve body was removed disassembled and cleaned and the solenoid and accumulator o-rings were replaced as were the upper and lower seperator gaskets. The transmission was flushed using the cooler lines. The delayed engagement problem disappeared and it seemed to be running very well and then about 10 miles later this new problem started. The only description I could find on-line of a simiilar problem is purportedly from a transmission tech training manual and it says that the problem is the fluid thinning as it heats and the solenoid valve strainer being clogged and unable to dump enough fluid. So out came the control valve assembly again and all the solenoids got cleaned and tested. There was some sludge but now they work great. I also double checked the 9 check valve balls and the micro strainer and the whaterever it is on the tiny spring as I reassembled it. But all that didn't solve the problem. Neither did a bottle of TransX. (I'm not proud of trying "hope in a bottle" but I'll fess up since someone was bound to suggest it.) I did the shift-lever macarena to get the TCU codes and the transmission control unit has neither any current error codes nor any stored eeror codes. I also tested the wiring harness and the resistance values for all the sensors and the solenoids, the ATF temp sensor and the throttle position sensor and the idle switch. Everything was at spec. I could put the transmission gauge on it and watch the clutch pressure drop off as it disengages but that won't put me any closer to solving the problem. I can't think of a mechanical problem that would be reset by turning the key off for a second. I have no ideas for how to test the TCU outputs while its operating. I don't think its likely the solenoids since they have plenty of "OFF" time in N and P and the very briefest off/on power cycle resets the problem... for 20 seconds. It also doesn't make sense that the a defective TCU would do this as opposed to staying dead. I'm not a huge fan of the idea of grabbing a transmission from a scrap yard. There's one available about an hour away for $350, but I understand seperating the transmission from the engine involves muscling it around for an hour or so. I have a pair of torn rotator cuffs, so that sounds like more pain than I'm willing to sign up for. (anyone have an idea what a shop in the Boston, MA area would charge to do the work?) So... I'm completely stumped... has anyone seen anything like this before... anyone have any ideas? ... ANY ideas ....

Just thought I'd follow up as where I ended up so far. Now that I can test the transmission cold (25F or -3C). The pump pressures when very cold are more than double the hot values, and the time delay is negligable for all shifts. This is slightly less meaningful than it sounds, because the idle RPM is 1500 immediately after the cold start, and drops to 800 long before the transmission fluid is hot. Comparing pump pressures at 1500 RPM cold and hot gives a ratio more like 1.35 or 1.40, which matches the changing fluid viscosity. If I track it as the RPM drop and fluid warms up, any excess time delay is solely a function of line pressure. As the line pressure drops below 20PSI delay becomes apparent, and below 15PSI the delay is just about infinite for all shifts. Just as I was about to open the ATF plug and drop the AT pan and remove the control valves and look for any obvious easy to get at o-ring or seal failures - the car lost its hanger queen status and was pressed back into regular commuter duty. Since TransX is an extremely thick lubricant, I removed a like volume of ATF and dumped the bottle in. The only effect on line pressure was an immediate 15-20% increase that persisted. Its almost enough to raise the line presure enough to eliminate the shift delay - but not quite. What he driver experiences is that the minimum RPM to get the AT to shift was dropped from 1100 to 950, just barely above idle. I'll add a follow up if anything useful becomes apparent when I get inside the AT. My guess is I won't see anything that's an obvoius culprit and I'll change all the rubber and paper and have no way to know if the problem was rubber or torque..

To start - if it looks like a total rust crust I brush on acid,. either HCl (Muriatic) or Aluminum Brightener (Phosphoric) and rinse it off. I have patience for gloves and goggles but not for Naval Jelly. Next comes Liquid Wrench (penetrating oil) Then put the impact socket on and tap it around with a ball-peen hammer. Now find something else to do for a few minutes to let the oil it do its job. I just use an 18" breaker bar and give a good sharp yank, but if your hands dont need the abuse, you could slip in some cardboard to protect the radiator and swat the breaker bar with a big rubber mallet. (If you don't hit it straight...) Putting the pulley back on is no biggie. You can hit the torque click before it turns over without jerking it. If it does turn it too easily to turque the puley, there is a compression problem...

In the hope that the "slipping transmission" pundit will either show some expertise or back off a little - one of my two 1992 EJ22 AWD Legacy Wagons has a similar symptom. It drives generally correctly, but while the engine seems to perform the same as its twin, the power doesn't get to the wheels quite the same if you floor it up a steep grade, and either matching the tachometer to speedometer of the highway fuel economy indicate that the torque converter isn't really locking up at highway speed. (For history buffs, I bought both of these cars new on the same day late in November 1991. They have 145,000 and 165,000 miles, the problem is the 165,000. The transmission filters have been changed in the last 20,000 miles, and they got a full fluid purge that time rather than just the 4 quarts in the pan. These had 100,000 mile warranties and Subaru dealer service to about 105,000 and ran most of their miles in the 20th century before becoming the family's extra cars. From looking at old records, they both had regular transmission fluid changes and at least 2 transmission filters each. Neither vehicle has an external transmission fluid filter.) I did the transmission self-diagnostic shift lever line dance and the flashing oracle swore everything was good to go. Clearly, the ability to deny reality isn't restricted to intelligent life... I don't have a subaru select monitor, so I can't give you any of the values or verify the TCU is really using the speed sensors, etc. The Stall Speed settles at just about 2750 RPM for all selector positions. (that's the one with the left foot on the brake, and the right foot on the gas). The Shift Delay test is obviously well over 1.2 seconds (20s cold, 3 thereafter) for D and well over 1.5 seconds for R (20s cold, 2s thereafter, slightly faster than D). So I attached a transmission testing gauge set and hooked it up just as seen in Section 3.2 to check the transmission line pressure. It was impressively low. Something close to 15PSI in all positions at 750RPM, and between 94 and 110 PSI at stall speed. The stall speed numbers were varying by selector position, with R highest at 110, then 3 at 104, 2 at 98 , and 1 and D both 94. The service manual indicates those would be good numbers at idle. It enough of a hassle to At this point I assume the problem is either 1) worn oil pump vanes, 2) Duty Solenoid A which mudulates the secondary pressure regulator, or 3) a blown out O-ring at line pressure perhaps on one of the little cross-over pipes that has to come off to change the transmission filter, and probably not one leading to the R circuit. Conversely, I think its unlikely the problem is anything to do with the truly expensive slipping disasters - the torque converter, or one of the bands, or worn gears, etc. So... does anyone have any experience doing anything about this other than A) adding a quart of "hope in a bottle" or getting another 4EAT from the scrap yard. (BTW in MA 92's seem to be down to just 1 yard east of the CT river)