john banks

Remapped a MY99 today with just a SS backbox. The owner had taken it to a rolling road and it was detting at 5000 RPM.

The lambda voltage was about 860-870mV on full load (usually about 910-925 on a UK car) and when you got heatsoak into it it retarded 5 degrees on full boost. Not just the odd spike, but big troughs.

The MAF voltage was about 0.2V lower than you would expect. However, the owner had already changed the MAF and lambda sensors as a precaution after the detonation.

Swapping the MAF sensors back the result was the same.

Since the results were consistent and we were running out of time and laptop power, we remapped the MAF sensor calibration and a few of the load headings until the fuelling and timing came back into line and then just took away the last traces of knock in the individual zones. It ran great then.

Didn't have chance to get to the bottom of it, but the owner will return to the rolling road and check it out again, and will also use a Knocklink.

Theories:

Resistance in the loom from MAF sensor to ECU?
Dodgy ECU?

If there are any further issues we'll come back to it and look into these.

However, the worrying thing is the car felt really good on the road initially for the boost it was running because it was lean (and advanced beyond the knock limit which the ECU had to pull back).

An AFR meter, Delta Dash or a Knocklink would all help you out here before a disaster could happen.

I am convinced this guy's knock sensor has saved his engine, otherwise it would be detting all the time not just on a heatsoaked rolling road at 5000 RPM (where a lot of cars would first poke through with det).

Trout...

John,

I had this problem on my MY99 Sti when I first got it. I was driving about in it and after a few months took it PE for a RR run - it was detting it's nuts off. Was easily most powerful car - very briefly - but no run was complete.

We changed sensors - cleaned the one we had - no joy.

It was a voltage offset in the loom. Once the this was fixed - car ran fine. Took an age to bottom out the problem tho.

Trout

EvilBevel

Trout, do you remember how it was fixed ?

John, I am so glad that I'm not the only one who saw such a strange phenomena with MAF's anymore, you couldn't believe it.

After receiving the 2 spare MAF's, someone from the Dutch BBS asked if I could help out, as his MAF was broken and dealer had no spares. So as superman I came to his rescue, and we fitted the new MAF.

Started car, and it idled like sh*t... utterly surprised, we took it off and fitted that MAF to my car... car ran A-OK. Being clever, I forgot to take my DeltaDash, so never knew what happened.

I was very confused and didn't understand at all. He then went to another dealer that had sourced the MAF-only deals, and ... it idled roughly again with yet another one, but after driving it for a while, it all went perfectly & measured normal on the DeltaDash. Dealer said he'd seen that happen quite a few times.

Is there something in the dark holes of the JECS that needs to "adapt" to a new MAF ? I know it sounds crazy as a question, but then I still am utterly puzzled about this one.

Claudius

Why dont you use MAP sensors?

john banks

No resistance in my loom. Bosch knocklink sensor throws a CEL. Have swapped the positions of the sensor and will fiddle more tomorrow.

Would love to use MAP but JECS uses MAF and I use JECS MAF does have some advantages.

[Edited by john banks - 8/17/2002 5:46:07 PM]

AlanG

I'm missing something here....

The car detted on the rolling road from 5000rpm. Fair enough.

The car was also detting on the road i believe albeit it was faint. Fair enough.

You say
What does that mean?

So how come the ECU didn't reduce timing until it stopped detting and you had to go in to the programme to do so?

You also say:
but it sounds like if he had carried on the way it was, it would have resulted in possible damage being done to the engine since he could hear it quite readily and the ECU wasn't getting rid of it.

Or is there more to it?

A

john banks

Sorry Alan, put it badly. What I mean is that the base map was too advanced (for the under-reading MAF setup) - ie beyond the knock threshold and the ECU was pulling timing when the heatsoak got significant. Without a working knock sensor (like my car presently) I think it would have been an audibly detonating nightmare. With big heatsoak it got to -5 degrees but was not audibly detting. This was on a pretty warm day. I never heard it detting today. On the rolling road it would possibly detonate at 5000 RPM because the conditions are so much harsher on a hot rolling road with poor cooling compared to the road. In road use I think the knock sensor probably would have saved his engine and stopped it detting since the timing would be learned. A sudden change to a harsh rolling road environment probably made it hit the retard and enrichment limit, and I expect if the detonation continued it would have zeroed the wastegate solenoid duty cycle. Maybe a hot trackday would have given trouble too, but it was holding its own on the road.

The noise at 2000 RPM that the owner was hearing was not detonation - I think it was a noise that was focussed on because everyone was listening for det. It did not sound like detonation, did not disappear when 5 degrees timing was removed (and later about 10 from the map to make sure), did not get worse with less heatsoak, and was a continuous evenly (but very briefly spaced) run of almost pneumatic drill noises but higher frequency. What it is I don't know but with a car that was so active on the knock sensor actually advancing in this region I am happy that it is not detonation otherwise I would not map the car. Will be interesting if his Knocklink picks it up which he installs on Monday. He never heard the 5000 RPM noise, neither did I on the road - just not enough heatsoak, and just within the ECU's abilities to sort it.

I wish I had logged the knock correction - initially very little, then as the heat built up it went -1,-2 on boost, next time -1,-2,-3, then down to -5 (big troughs on boost running very active correction). Restored my faith rather in the ECU knock correction - if it is working it seems pretty good to me. On my car it seems caput - no resistance in the loom, sensor works with knocklink. If it was working I would not be seeing much on my Knocklink. I look forward to getting a setup back with active knock correction - then I can run more timing. If you go for a quick burst of power it is really good. If you then track the car it might retard a bit. Knock sensitivity adjustment would be nice too - Steve has it in prototype and I am going to crank mine up to see if I can get it going again with the manifold.

[Edited by john banks - 8/17/2002 10:24:43 PM]

carlos_hiraoka

John, have you tried connecting the knocklink with the oem knock sensor (in its std. location) ..... I know this might throw a CEL but that way you could find out if it is a ECU related problem.

Carlos H.

[Edited by carlos_hiraoka - 8/17/2002 11:35:48 PM]

WREXY

John,

Very intersting and glad you put up these problems you experience so that we can see what different problems can arise. Thanks m8!

What would cause resistance in the loom or a voltage offset, (same thing), in the loom such as Trouts car?

I like Theo's question too! Why did it start working OK all of a sudden? Very



Trout,

Was anything done to the car and it happened or was it like that from factory?


Theo,

Does the dealer who sources the MAF only bits in The Netherlands get the same part number as our ones or does his have PE at the end of the part number, like this, 22794AA010PE ? I've seen a few that have written on here from the UK sourcing these MAFs with PE at the end of the part number. Would this be the same MAF sensor or maybe a stronger MAF? Anyone know?

Cheers,

Wrexy.

john banks

Carlos the std sensor works with the knocklink. The loom and the ECU are fine. It is a manifold related problem I reckon. Have asked Ecutek for a prototype version of software which can adjust sensitivity of knock sensor.

carlos_hiraoka

John, since you have had the opportunity to test both the BOSCH and OEM knock sensor with your knocklink, which seemed to work better ?????
I have used both, and still think that the OEM sensor was more sensitive.

Carlos H.

john banks

They both work fine with the Knocklink on simple testing by tapping. Not tested on the road as there is a CEL because I don't have the OEM sensor connected to the ECU.

[Edited by john banks - 8/18/2002 10:17:12 AM]

EvilBevel

Wrexy, that dealer alledgedly got an email from an anonymously Belgian based Subaru owner asking if his system would allow him to order the part number for sensor-only

It did, and so he did LOL

But I have no knowledge about what the PE stands for.

Anyway, I'd *really* like to know what happened when we changed over the MAF's (and when he did the same at the dealer) ...

WREXY

Quote by EvilBevel


LOL

Wrexy.

Trout...

Well - what we found was the MAF wire had been accidently cut and instead of soldering it they had reconnected it using crimp on bullet connectors - useless - I guess over time the decay in signal quality was gradual but enough to create an offset in the MAF signal.

We stripped it back and resoldered the connection properly. Fixed the properly.

Trout

WREXY

Thanks for that Trout.

Cheers,

Wrexy.

Sam Elassar

hi carlos

you are right mate , the standard knock sensor is more senstive than the one that comes with the knocklink.

carlos_hiraoka

Hello Sam, glad to see you posting here ..... like you said after some testing I also found that the OEM knock sensor worked better.

Carlos H.

john banks

Do you keep a second one for the ECU?

mutant_matt

John,

You say it is an MY99. Do you know which ECU and if the car was an "early" 99 car? I ask becuase perhaps it's actually a late registered 98 car? A dealer recently told me that there were some occasional problems with some UK MY98 cars with detting (which IM seem to have kept fairly quiet) so your description just got me wondering.....????

Matt

john banks

It was an early 1999 registered MY99 with an AE800 ECU.

mutant_matt

John,

So it's a possibility then? I'll see if I can find out which ECU was suspect....

Matt

iainalpine

hi guys, as john has told the storry with my car i dont need to say to much. i would love to get to the bottom of this. i never like to work round a problem if i can help it. i have a strong background in diagnostics. i have worked for Mazda UK as a diagnostic technician for 11 years and have never been so stuck with a fault as i am at the moment. if any 1 can point me in the write direction i would be very grateful. the rx-7's didnt even come close to this and i always thought nothing could be any worse than a fault on them. i would like and take in any advice you lot can give me on these cars. havnt had the car long and have never had the joy of working on them. i no john is as lost with this 1 as i am. hope you can help me out.
cheers, iain macleod

Pavlo

Have you checked the wiring of the MAF sensor? In paticular resistance between ground and the return, to make sure it's not pulling it down.

Paul

iainalpine

hi paul. if some one has a wiring diagram for the car and could email it to me that would help me along a bit. havent checked that, but will in the near future.

iainalpine

have any of you guys had a problem with the earth system on these cars. i remember someone talking about it b4 but cant remember what was said.
iain

john banks

The next plan was to check for offsets in the loom. Will mail you wiring diagram Iain.

mutant_matt

OK John,

I have got confirmation from a dealer that there was a known issue with *some* AE800 ECUs which would result in det. The fix is to upgrade to an AE801. IM know about this and will fix under warranty (sometimes ) including any repair bills for new engines etc...

Hope this helps,

Matt

Pavlo

doesn't explain the lack of voltage from the MAF though. but it's not stated whether voltage measurement was from delta-dash or a voltmeter. If the error is ECU related, it could well be giving delta-dash erroneous info.

My thoughts are bad wires, bad earth, or earth leakage

Paul

Sam Elassar

hmmm.


does not make sense to me though. if it is hardware issue, surely swaping the ecu to a different one should show the problem straight away.! but i am sure JB would have done this to eliminate a hardware fault.

ian

have you got your car sorted out yet? was the car detting as standard ? if it is not, then it is a mapping issue! if it was detting with the standard setting, and it is not a ecu hardware problem, then it has to be a menchanical problem. getting a tek 2 or 3 or 10 will only go around the problem and the more you actually raise the game of the car the more the problem will be pronounced.


do you hear the det from inside the car while driving on full load? 5th gear for example?

if no one is going to help you, you have to have a list of possible causes and go through a process of elmination. or just get the ecu back to standard and take it to a dealer and ask them to fix it, instead or all this hassle. at the end of the day very few poeple on the public forums really know what they are doing. and most of the info is just passed on from on person to another , it is free for all if you look hard enough. so if it has not been on teh board before, you are not going to get an answer in here!


sam