KINGTIGER

Recently been made aware of my check engine light on my STI V5 MY99.

I have took it to a garage who have said there is no bulb, so its hiding a fault.

Just had a diagnostics check and its a fault code 22. The knock sensor. Though the car is running fine and he said its most likely a dirty sensor. Lets hope.

Ive had car three years and have always run it on 99 super unleaded. Now maybe person before me may of put normal fuel in I don't know.

Anyways I've been given some bulbs to put in, but how do I remove the dials. I've removed the two screws on the trim but no luck.

Is there any more screws. Or do I have to remove the lower dash board lol.

scooby L

They've removed the bulb to stop the annoying CEL light from coming on rather than sorting the problem... oh my... that would worry me somewhat!


Reminds me of this

http://video.yahoo.com/watch/3401527/9504240

fast forward to -57 seconds

pinkypurkhardt

iTrader: (9)

have a look in the technical arcive mate there is a thread with the service manuals link which will tell you how to remove the dash step by step

rossi_p

iTrader: (26)

In order to remove the clocks you need to:

1 - Remove the lower dash on the drivers side. There is a plastic screw clip on side of the dash and a lower one underneath the lower dash section. Then one final one clip which needs a firm pull.
2 - Remove screws under the under steering cowling and take off the upper & lower pieces.
3 - Undo the two screws for the instrument surround.
4 - There are then 3 final screws for the clocks unit.

Splitpin

Whoever gave you that check has given you some not entirely good/complete advice.

The car may appear to be "running fine" when given a diagnostic in a workshop, or even a brief drive around. The trouble is that if your knock sensor has failed, the ECU has no way of knowing whether the engine is starting to knock (which it almost certainly will to some degree if you're running an unmodified JDM ECU on UK fuel). If it is unable to detect knock when it is happening, it will be unable to retard the timing or add fuel to correct it. The long term consequences of that process are not good.

There is also no way for this sensor to get "dirty". If the check engine light has come on permanently (or for a significant percentage of time) and code 22 is the only one you get back, either the knock sensor is duff, there's a fault in the wiring between it and the ECU, or, under some circumstances with the 99-00 cars (that I don't understand from a technical point of view, but reliable people have reported it), a failing mass airflow sensor can sometimes confuse the ECU and make it set the code 22 error state.

One way or the other you need to find out what the cause of that error is and fix/replace it ASAP.

That seems to be a bit of an unrelated point. If you're speculating that a previous owner putting something other than SUL into the tank could have caused or contributed to your error, the bottom line is that it didn't. Also, while you say you've always used Super yourself, don't forget that if your car is still running on the standard map, it is set up for Japanese-type Super, which is 101-102 octane.

The two screws to remove the trim, then three screws to remove the instrument pack. Don't forget to drop the steering column down as far as it'll go before trying to squeeze the pack out.

There is no need at all to remove the lower dash moulding, it won't help. As above it's the three screws you haven't yet taken out that will be stopping it moving, although even after you remove them it's a tight fit. You may find it easier to get the instrument pack out if you remove the steering column cover. You could also try removing the clear cover on the pack as well as the plastic angled moulding before you try to pull it out of the instrument binnacle. There are little tabs round the edge that need to be pushed in, and then the cover gently pulled off.

djc1

its really easy mate takes 5mins ,ive just put a new set in mine ,some sod smashed the led so the clocks were ****ed .
yours should just be a bulb though .

KINGTIGER

Ok so to solve that problem I should use a good Octane Booster like that NF or get a new ecu. Which I can't really afford.

I'm taking car in next week for them to have a better look at it. So I will go from there. I've managed to get the dash out but its just so tight I cnt get the bulb in so will give another go later on or leave it the mechanic.

He did reset the ecu and it picked up the fault straight away again.

Splitpin

To solve the octane/fuelling problem you mean? Yes, you can use octane booster, but you don't need a new ECU. You can get your own remapped - find an Ecutek agent whose reputation you trust. Money involved in that, yes, but a fraction of the cost of a mapped aftermarket ECU and you'll get a nice safe power increase out of it too.

Either it's gone or the loom is then. Genuine JECS sensor £55 plus delivery here, and you can change it yourself in less than 20 minutes if you have a 12mm socket, 3/8" torque wrench and a couple of long wobble extensions.

KINGTIGER

Hey cheers for the help. The car is going in on wed so will go from there. I will have a look at Ecutek and see whats on offer. Probably should of done it years bak when I got the car lol

Gambit

iTrader: (6)

who was it you where getting your info off?

im pretty certain this is causing my intermittent error code 22, that is becoming fast becoming an every other day occurance - more so when the car is stone cold though it will flick on and off for at least 10mins. initially thought it was knock sensor but from checking it & wiring all looks sound, plus the car feels a bit rough on hard accleration so been tootling about till i change it tomorrow

i going to put the cars old MAF back in tomorrow (current 1 is under 2yr and a genuine part) and see does it cure it

is it me or was scoobynet riddled with 98-00 MAF issues in october - first cold snap has wiped out half of them lol

tjmatt

iTrader: (1)

Gambit what you have sounds like issues I saw a while back. Replaced it with an ebay sensor advertised as direct replacement and had similar issues after a short period of time. Kept trying loads of different things, then ended up buying a new genuine sensor and that kept the code away.

However on occassion it will still light up if changing from 99 to 95 ron fuel without a reset and normally just sat idling in hot weather. Then when pulling away there will be a hesitation while the ECU pulls timing and then it clears up.

What I'm saying is that the code 22 is "indicative" of a failing or incorrect sensor but NOT necessarily.

Gambit

iTrader: (6)

well i put the old genuine 'green stripe' MAF (it wasnt bust in first place i just replaced when i 1st bought the car) back in the car this morning, reset the ECU, only driven the 10mile to work so far but no error code as yet

car is always run on 97RON SUL (thats all u get in NI) but the issue only started appearing after i returned from holiday at end of sept where the car had lay up for a week.

so its a wait n see now if that clears it!

alcazar

iTrader: (2)

Or get a remap, which is cheaper than a new engine, trust me

Once the screws are out, the steering column is at it's lowest and the electrical connections are disconnected, it's just a case of wiggling the instruments about until they come free. It's NOT difficult, trust me.

And while they are out, replace all the lighting bulbs, they are W5W capless and sit in little black quarter turn bulbholders, the larger ones on the instrument rear panel.

Gambit

iTrader: (6)

well its not the MAF - so guess ill replace the knock sensor at weekend and try again!

Splitpin

Direct anecdotal reports from people who've had the 22 error, fitted a new knock sensor, not resulted in the CEL going away, and, whether via diagnosis or guesswork, replaced the airflow meter and that has sorted it. However, I know more than a small amount about the firmware design of the 99-00 JECS ECUs, including the exact criteria necessary to activate all of the diagnostic trouble codes. From that point of view, I don't see the mechanism by which a failing MAF sensor can create a set of circumstances where the 22 code can be set. The only thing that can set the 22 error is the voltage in the knock sensor signal line going outside preset limits for more than one second. There is no obvious mechanism by which an airflow meter acting up can make that happen.

So, on the one hand, it shouldn't be possible. But on the other, enough different people have mentioned it to make me believe that, somehow, it is. Until/unless I see it myself I can't work out in detail how it happens. Thinking about it I suppose it's possible for some types of MAF unit malfunction (say an internal short) dragging the ECU sensor supply rail down (and in turn dropping the supply to the knock sensor circuit below the error threshold), but I'd much rather see it than speculate.

The obvious thing to do under these circumstances would be to datalog the car. Fluctuating mass airflow output will be obvious under acceleration. Similarly, although my belief is that the publicly available data analysis packages won't do it, if you know how you can log the voltage in the knock circuit too. You can't log it at anything like high enough frequency to establish whether there's det happening, but you will see if it's acting up.

Was the old one removed for a reason or due to elective replacement?

Always seems to work like this round here. You'll get a spate of threads about a similar subject, then there'll be a spate about something else. Weird!

What sort of ECU is this with? Subaru or Apexi/else? As above the OEM ECU cannot illuminate the CEL (or set the 22 error) in response to what it perceives to be knock activity, only in response to an extended overvolt/undervolt on the knock sensor signal input. If you're getting enough legitimate output from the sensor to keep the voltage outside the limits for a full continuous second with the engine idling, there must be something very strange up with your car - more than just a tank of low octane fuel.

Gambit

iTrader: (6)

the old one was removed as a routine service when i first got the car. it had pretty big hesitation while driving and people where saying it could have been the MAF, while others saying plugs & leads. so i replaced them both.

there was never a CEL light came on before replacing, or after. and that was nearly 18mths (May 09) since initially replaced. light only started after i went on after i went on hols for a week (Sep 10) and car lay outside for a week. hence why i found it odd, i have nearly convinced myself that water has just run down onto it and got into it giving the duff signal

just one other question regarding the CEL, when the light is on (no matter what the error) will the ECU always go into limp mode or a restricted mode?

Cheers,
Craig

Splitpin

It'll certainly be interesting to see whether the hesitation returns after refitting the old sensor. However, I would caution you against giving it beans with a suspect unit installed.

The big problem with the error detection system (on the classics at least, not sure whether FHI/Denso have improved it on the subsequent ECUs) is that the conditions required to set the code 23 mass airflow sensor error are so loose, the CEL is basically useless. It will pick up a unit that has completely failed (i.e. is not providing an output at all), and it will pick up a shorted loom, but that's all.

What it will not effectively detect is a sensor that is consistently under-reading, or is suffering intermittent signal drop-outs, which is a problem as one, the other or both is what airflow meters tend to do for some time before actually failing completely.

On the 99-00 sensors, the signal drop-outs get longer and more pronounced, and most drivers only notice them when the behaviour becomes so bad that it affects the engine's behaviour at idle. The problem is that it's usually happening (worse) under boost - but because the initial symptoms of this are leaning of the mixture, during the early stages of sensor breakdown it actually makes the engine faster, so the driver, if he notices anything, will believe the car is running "great" rather than realising it's the early onset of a problem.

It's only once the sensor degrades past a certain point that the error correction system in the ECU can't cope and you end up with lean spikes so bad that it starts to feel like fuel cut/hesitation, or else the ECU detects so much knock that it pulls huge amounts of timing out which causes what feels like a dead spot.

What part do you think water has got into? The maf unit is environmentally sealed all round, almost impossible for moisture to get somewhere important. The only likely point of ingress would be the sealing round the connector.

No. Only some errors cause that. Others won't directly restrict the engine's behaviour but you lose whatever functionality the sensor would've provided, which can cause the engine to feel weird. Off top of head the errors that cause limp immediately are cam, crank position, mass airflow sensor, pressure sensor, throttle position sensor and neutral position switch.

I can't remember whether the speed sensor causes full limp or not. It may simply be that it feels crap because the ECU thinks the car isn't moving, and so switches off all the stuff that normally happens only when the car is in motion.

If, on the other hand, the knock or O2 sensor errors are activated, the engine will drive as per normal but you won't get any modification to its behaviour that would otherwise be made. So, no learned ignition/fuelling adjustment (other than what is already stored in RAM), and no air/fuel correction - just whatever A/F learning is already in the memory.

KINGTIGER

Well I've had me car in to garage again.

He has checked the knock sensor and its in a real bad state. Its corroded and has several bad cracks in it. So I'm getting a new one from Jap Imports. The garage deals with them so thats a help.

I just hope it is that but will be a week to find out. He also went to replace the bulb
my bubls. But its the newer dash with the Led's. They had put ktichen roll in to blank the light.

Which is annoying as I could of had severe damage to my engine. I won't be buying from that particular dealer anyways, though I've heard they gone under anyways.

Gambit

iTrader: (6)

i was originally taking about water getting down into the knock sensor wiring harness seeing as its directly below the bonnet scoop - and being a 99 it has the later knock sensor with the electrical connector at the sensor end.

i picked up an earlier v3/4 sensor which has only the 1 connector at the harness end but i believe this are different and wont fit? (see https://www.scoobynet.com/general-te...-v3-4-5-a.html) didnt know this at the time which is a bit of a *****

so tomorrow the plan anyhow is to take out the knock sensor and give it & harness a good check/clean over and if poss repair any cracks in wiring casing - been told to coat in sticky back foil and then with insulating tape to prevent any signal intrusion/interference

after replacing the MAF to my original and resetting the ECU it made no difference at all, CEL back on after about 15mile of driving.

@KingTiger - sorry for taking over your thread as such but lot of good info in it so thought it better to continue it on - so for future peeps searching for similar issue

Gambit

iTrader: (6)

wee update.

light wasnt on last night on driving home even still this morning i took the knock sensor off but when removing it i found a rusty screwdriver bit (one of these) resting beside it in a shallow on the block. now i dropped this about 6-9mths ago when fitting newage WRX intercooler - would it cause any interference/rattle?

as for the sensor it looked in perfect nick, checked all over, no cracks or damage to the sensor or loom - cleaned it all up along with the surfaces plenty of WD40 all in round connectors and then refitted and torqued correctly.

on restarting the CEL light was not on, and hasnt been since my drive to shops for breakfast. though it was only moderate driving - guess ill just play it by ear now

and v3/4 sensor defo doesnt fit it

KINGTIGER

Ok no problemo its all interesting and ould help me out anyways lol

Splitpin

Put your wee in the toilet where it belongs.

It certainly shouldn't, although the actual knock detection algorithm used in the ECU is pretty rudimentary and can pick up false positives under certain circumstances. Again though, there's no obvious way by which something rattling around next to the sensor would drag the voltage up or down long enough to set the error.

Do you have (or know anyone with) a multimeter, or, even better, oscilloscope? Also note that as you've probably seen, even though the sensor is a two pin unit, only one of them actually runs through the link wire. It's grounded through the block to the sensor ground bolted to the inlet manifold. Check the bolt securing that ground is tight, corrosion-free, and also that the cable running from it is undamaged.

WD40 in electrical connections is not a good idea. Did you see any evidence of moisture ingress when you first pulled the connections apart?

If I was in your shoes I'd get a genuine JECS/Subaru sensor and fit it. If, as per above, you have the kit to test the sensor, you can test it and therefore know whether you can rely on it. If you can't, precautionary principle applies to MAF and knock sensors. Better to replace and know it's good than not be able to trust it.

IIRC the earlier sensor has a different sensitivity level so would have been no use to you anyway, even if the plug was the same shape.

Gambit

iTrader: (6)

dont be pedantic

i agree because i have the same cel

do you maybe realise that im asking because.....

i'm asking for information!?!? belittle....

yes you are correct it doesnt fit, as i have found now myself. yet im no closer to this error cause its still there. so do i just buy both then!?!??

sometime people look for an answer rather than pay people 'who are clued in' for the sake of a 'subaru' thing - im trying to learn as much as you are.

Splitpin

Come again? Are you accusing me of "belittling" you? You've asked for advice and I've offered it, freely, no "belittling" involved at all.

Both? You only need to buy the correct knock sensor, the car only needs one.

What? Again, I'm struggling to understand the intent there. You've asked for help, and received it. If you're going to try and bite the hand that feeds, you may just find yourself going hungry.

Gambit

iTrader: (6)

on re-reading what i wrote it does come across i bit aggro, for that i apologise, but i thought you where basically taking the hand out of me for taking out the knock sensor, inspecting, cleaning etc - which you werent

anyhow, what i meant in replacing both was that i dont want to buy a knock sensor to find out that its the MAF causing it, which inturn leads be to buy a new MAF also.

i have a friend who has datalogging software that i can run on my car, you had said previous that i could use this to view the behavoiur of both sensors. is it just a case of taking the car on a normal run and just logging it, or is there a certain driving i.e. full throttle etc that needs done to get a full spread of data?