andyamf

hello, i wonder if anyone can help.
its a 1999 uk turbo 2000
starting a few weeks ago i started to get a hesitation on constant speed and very light acceleration. it is most noticeable in top at 70 to 80 mph.
fuel consumption is up too.
recently it has had new plugs, ignition leads, and maf sensor.
i have reset the ecu and no improvement.
in fact while doing the black and green plug ecu reset, the problem was worse.
so my next idea is a lambda sensor.
so are there 2 on my car? if so which on do i start with.
i looked at the first one before the cat and it has 3 wires coming out of it, some places refer to a 4 wire sensor on my car, i am confused.

any advice would be appreciated

thanks

jura11

iTrader: (7)

Classic have 3 wire Lambda sensor,New Age Subaru have 4 wire Lambda sensor.Generally lambda sensor for classic is cheapest than New age ones(from main stealer you are prepare to pay 300£)
At first check for error code,and after this check eBay for lambda sensor(but i think here selling Aztec Performance for great money also,but double check)

andyamf

thanks, no error code unfortunately.
still not sure how many sensors i have though

jura11

iTrader: (7)

I'm not sure but maybe 2 lambda sensor(front and rear),but this is just guess.
But without proper diagnostic is expensive buying sensors and error still last after all.

Splitpin

There is only one O2/lambda sensor on your car - the three wire device you've already seen at the top of the downpipe.

One thing you could try prior to shelling out on a speculative replacement is resetting the ECU and immediately afterwards, unplugging the O2 sensor and going for a drive in a manner you would expect to provoke the misbehaviour. Unplugging the sensor immediately after the reset is important.

If your O2 sensor is knackered enough to be causing the symptoms, running under these conditions should result in a significant improvement (or at least a significant alteration) to the engine's behaviour. If it still acts up in exactly the same ways with the sensor disconnected, this would be a strong suggeston that the cause of problem is elsewhere.

midnight

iTrader: (10)

If the lambda sensor is knackered,after you have reset the ecu simply connect the 2 black wires under the dash and look to see if it throws up an error code 32. As the sensor runs on such a low voltage it will rarely throw up a cel in normal use, but it will show using this method. I had similar symptoms on my old 99 uk turbo and car ran ok albeit a little rich. Incidentally I bought a used one off a breaker for £30.00 and no probs it the last 10 months

andyamf

thanks for the help, i tried your idea splitpin and the car drove much smoother, then reconnected it and it drove worse than before the test.
time to order a lambda sensor i think.

Splitpin

No, it usually won't. This is bad advice, for the following reasons:

1. As you say yourself, if the sensor is on its way out, but isn't knackered enough to satisfy the activation criteria for the code 32 error under normal usage, it usually won't fail the check performed during the reset procedure either.

2. The suggestion that you should reconnect the black wires and check for errors immediately after performing a reset procedure is completely pointless. If any component fails a reset procedure check, you would know about it immediately - because the CEL will flash up the error number(s) rather than the on/off/on/off "reset complete/OK" code.

As such, by definition, if the reset procedure gives you the "OK" flashes, you know that all the internal systems have successfully passed checks. The practical problem in the case of the O2 sensor (and the MAF) is that the conditions required to activate the error states are so lax that a sensor can be heavily degraded but still appear, as far as the ECU is concerned, to be functioning normally. That's why "indirect" tests, such as disconnecting the suspect sensor and looking for changes in the engine's behaviour, are often more useful.

Yep, that would be a reasonable move now. If you don't fancy paying main dealer prices, you could try Import Car Parts for a Subaru Genuine one, or alternatively a Bosch sensor (the "Subaru" one is a Bosch design made under license in Japan). I believe (though check to be certain) that the LS02 is the appropriate replacement for the MY99-00 part.

Glowplug

iTrader: (8)

These are exellent. A Bosch unit for under £50 .
Don't be tempted by the ebay universals, they last about 100 miles, believe me, i've been there.

andyamf

i did some shopping around earlier,

dealer - 190
import car parts - 138
camskill (aftermarket with plug) - 90
bosch - 50

i went for camskill, should get it in the morning, i'll let you know if it improves things.

midnight

iTrader: (10)

Sorry splitpin not trying to offer anybody any bad advice,and maybe i should have explained it a bit clearer,When my lambda sensor failed i did not even know about it,as no cel flashed up,i had similar symptoms to the op and the car was running a little rich ,but i put this down to the forge vta. It was only when i was having my car mapped that bob rawle told me it was knackered.

I have just been out to my car and connected the 2 black connectors & the cel is flashing at steady 1 second intervals. I have then disconnected them and connected green & black and reset the ecu and driven the car around steadily. After a few mins the cel flashed up at steady 1 second intervals,which i take is telling me that most of the sensors are ok. I then pulled up disconnected the plugs and drove home. When i got home , i lifted the bonnet disconnected the lambda & removed it and fitted this:

which is a lambda i know to be knackered


Got back in the car and started it up and drove the car, NO CEL.After 10 mins ,pulled up and connected the 2x black wires,and switched on ignition ,straight away code 32 flashed up. I then disconnected black plugs drove home and fitted original lamda,and connected black plugs - steady 1 second cel flash.
I do see what you are saying that this method will not detect a failing lambda ,but it will show up a faulty/broken one.

andyamf

Nightmare, took out the old lambda, it came out with no fuss.
only problem is it brought out all the thread with it.
now i have a nice stainless sports cat with a useless hole in it and no lambda sensor.
doh!

Glowplug

iTrader: (8)

Unlucky! You any good with a welder?

You can buy a new boss easily/cheap enough or re-use the original if you can get it off the lambda.

Splitpin

Yep, I know. Trouble is that when someone does so inadvertently, the only way to make sure any subsequent readers can weigh up what they're reading is if someone who knows otherwise pops along and says so.

No, it won't. At best it causes a specific behaviour with your particular sensor which you already know is faulty, it is not a general rule that will work the same way for everyone. Moreover, as you have written further down, the entire premise of your "test" is based on a mistake.

What you're claiming is no different from that bollocks about disconnecting the MAF sensor with the engine running, and determining that the sensor is "good" if the engine stalls, and "knackered" if it doesn't.

That might have made sense to the person that first suggested it. It may even have been true in the case of the particular sensor that he was "testing". The trouble is, as a test that can be applied to any car at any time, it's total (and dangerous) bull. As well meaning as your theory is, it's in exactly the same category.

Not quite. If you get to the equal spaced on/off flashes, it is telling you that all (not most) of the internal systems have checked out according to the error definitions contained within the ECU and that the reset sequence has completed (and thus, both the current and historic trouble code memory sections in the RAM have been zeroed). The moment you disconnect the black plugs, the ECU returns to normal execution mode and will log any faults back to the trouble code RAM areas.

What you have just written is impossible Midnight. If you successfully completed a reset (and therefore reset the error memory), drove around for 10 minutes, read the error memory and found code 32, you did get a CEL during that 10 minutes you were driving. Whether you noticed it or not, whether for a tenth of a second or whatever, it did come on. The only way you can end up with a code to read via the black plugs is if the CEL has come on. That is how the firmware in the ECU works, and there are no exceptions to that scenario.

As such, the entire premise of your original post is wrong, and you've made incorrect assumptions on the basis of it. If you have reset successfuly, gone for a drive and found code 32, the bottom line is that your ECU has thrown up a CEL "in normal use".

If you want I can sit here all afternoon explaining to you how the ECU's error detection and reset sequences work at firmware level, but the bottom line will be the same. The "test" you've suggested does not "prove" what you think it does.

Splitpin

Doh indeed. As B13 has already said, and if you don't have the kit/expertise to sort yourself, the best/most cost-effective way to fix this would probably be to take the downpipe to your nearest tidy exhaust place and get them to cut out the old lambda boss and weld a new one in.

andyamf

new boss getting welded in tomorrow.
i hope the lambda fixes the problem after all this.
thanks for the advise chaps
i'll let you know if the fix works.

andyamf

hey guys, still not right.
picked up the car, reset the ecu, drove for 50 miles.
the car felt fine, couldn't fault it.
next day problem is back, same as before.
this same thing happened after i replaced the maf
after the ecu reset the car was fine then after a day or 2 it's back.
i have had the coolant temp sensor suggested as a problem and also the map sensor.
any ideas?
if i can't sort this out i will have to take it to a dealer i think.
it is getting to expensive just throwing parts at it.

Glowplug

iTrader: (8)

Connect the 2 black wires and turn the ignition on, see what codes it throws up now.

andyamf

no fault codes
ok, i have just reset the ecu again.
gone out driving, problem best noticed at 40mph and 3000rpm in 3rd at constant speed
got home, connected black wires again, no faults

Fuzz

Stop with the suggesting checking the CEL every five mins.
If the light hasn't been on since the reset then nothing will be stored.
Both Lambda and MAF sensors do not necessarily throw up a code when they are failing as the voltage output is a legitimate voltage for the ECU to think the sensor is OK.
For e.g the lambda can and often does fail around the 0.33v point
However this could also mean the ECU just thinks the engine is running a little lean.
You will see this after a reset by looking at the data from the ECU as it starts to add extra fuel to "trim" it back to Lambda 1.00.

Fuzz

double post bollox

Glowplug

iTrader: (8)

CTS would usually give you poor starting first thing in the morning. MAP sensor seem to be just good or bad, hence give a CEL or fault code.

What dump valve do you have fitted, OE, aftermarket recirc or vta??

Fuzz

Err it usually fails "open" so would be permanently telling the ECU it's cold.
So would actually start in the morning just fine.

andyamf

it is the standard dump valve
the cts, if it fails in cold position could it be responsible for bad fuel consumption, always too rich? but would hot starts be affected, they are fine

Fuzz

If the ECU has enough "trim" available then you would never know as the lambda sensor would be saying "I'm rich, I'm rich and the ECU would trim all the excess fuel out.
Obviously it has to be working correctly to do this.

Just because it's "drinking more fuel" doesn't mean it's specifically running rich, just running inefficiently for whatever reason.

What you need to find out is if it IS running as it should be and it sounds like you dont have the equipment to test anything.
Don't take it to a dealer, take it to a trusted independent garage, preferably one that knows a Subaru inside out..

andyamf

ok, here is the final result.
fuel consumption is back to normal, since fitting the lambda sensor.
still have a slight hesitation, so i took the car to scooby clinic.
they put it on the dyno, and took it for a test drive.
they found no problem anywhere.
they recognised the hesitation and put it down to a glitch in the '99 - '00 ecu, they all have it to some extent.
what it is, is the point where the ecu is switching in and out of closed loop.
they said they could see it on the computer they had plugged into it at the time.
sounds exactly the same as bigfarlz is talking about in this post

https://www.scoobynet.com/general-te...ut-lambda.html

so i will live with it, at least i know there is nothing wrong that could cause big problems.

thanks for all the help lads.

andy

Splitpin

They might want to check their facts a bit on that one, the "explanation" they have given you doesn't make too much sense. They don't "all do it", there's a considerable difference between the original ECU fitted to MY99 cars and the second version that was introduced part way through the model's production.

In this case it might be worth you taking a few minutes to pull your front passenger footwell carpet (and the cover plate) and check whether your ECU is an AE800 or an 801. The 801 is built on a heavily revised codebase and does run a fair bit smoother.

Bigfarlz has an MY00 (and thus the AE802 ECU). His problem also occurs on steady throttle cruise. Your earlier post also said that you were getting it during constant speed running.

If by "constant speed", you also mean light throttle, low load, Scoobyclinic's explanation doesn't make internal sense as, under constant throttle low load cruise, the ECU will be in closed loop mode basically all the time, and thus, even if we were to assume that an open loop to closed loop transition caused a degree of hesitation, as long as you're sitting there at constant throttle cruising gently, there shouldn't be any such transitions occurring in the first place. The ECU only switches off closed loop fuel correction when the engine is either accelerating or running above a certain speed/load threshold.

If it feels inconsequential and Scoobyclinic have determined that there isn't anything serious wrong, living with it is a practical option. As above though, if you check your ECU and find it's an AE800, you will probably find that picking up a used AE801 from eBay for a handful of quid will improve matters if not fix it completely.

If you look and find you already have an 801, chances are there's something very minor and inconsequential up somewhere. Under these circumstances, about the only thing you can do is live with it and see if it gets more noticeable/identifiable.

dj219957

the 801 runs a tad more power as well doesnt it? would he have to re-code keys as well?
in either case id live with it now that the fuel econ is back.

andyamf

yes it is an 801 i have

Splitpin

Than what, an 800? No, power/torque on AE800 and 801 is the same, because the map data is the same. The difference between them is that the 801 has a newer (bug fixed) program code.

It's the 802 (MY00) UK/Euro ECU that produces less power than the other two.

And yes, I'd live with it too under these circumstances. However, seeing as you asked the question, no, on MY99-00 UK/Euro cars, unlike other model years it isn't necessary to do anything with the keys when swapping ECUs. The OE immobiliser is a separate unit under the steering column, so even if you swap the engine ECU, the relationship between the keys and the immobiliser stays intact.