Change of diagnostic direction guys!
Well not exactly a change of direction, but revisiting something that I thought I had already ruled out a while ago...
Today has been sunny so being off work I was out tinkering on the car and made a bit of progress. I was trying to eliminate as much as possible the knock sensor so even though the ohm reading at the ECU connector was the same as Jim's out came the air box so I could inspect the green knock sensor connector again. Last time I looked at it I just gave it a squirt of contact cleaner and that was all...
This time I was more thorough. I checked the continuity of all three wires from the harness side connector back to the ECU connector - all below 0.5 ohms and nothing intermittent about them when wiggling the plugs/cabling etc. I then checked the reading of the knock sensor at the connector - slightly less than 5 Meg Ohm, and it too didn't change when wiggling the cable and connector. I sanded the male contacts just slightly, tensioned the female contacts slightly, used some more contact cleaner and reassembled it. Of course none of this would test for a broken shield wire on the knock sensor cable but short of stripping down the engine there is no way to test that...
I then went for a test drive and....nothing. No change at all. Still intermittent hesitation, the performance sometimes bad, sometimes good, but never great. Changing from good to bad and back again in just seconds as the driving conditions, selected gear, rpm and throttle opening changed. Feeling a bit deflated I glanced at some of the live sensor data and noticed something - during the lack of power and hesitation, under heavy throttle below about 2500-3000 rpm the oxygen sensor was reading constant lean for as long as I kept the throttle down. Wait, what ?
Those of you who have studied engine management systems in some detail will know that while the ECU will try to maintain a stoichiometric mixture in closed loop mode, which occurs during idle, cruise, and light acceleration, on a narrowband oxygen sensor system as soon as you go into heavy acceleration at any RPM it switches to open loop where it will deliberately fuel the engine rich (based on pre-computed map tables with corrections from long term fuel trim added) to maximize performance and minimize risk of knocking.
The exhaust should never be lean under heavy acceleration it should stay rich, the only time it should stay lean is during closed throttle overrun while coasting when the injectors cut off completely. The thing is I had observed this happening before a few months ago but thought I had resolved it, apparently not, or it has always been intermitent.
This got me thinking about the knock activity in a new light - lean under heavy throttle can only be due to fuel starvation, such as a faulty fuel pump etc... Lets say that there is a fuel delivery problem, the pump can achieve the regulated pressure with low delivery but not high, and lets say its a bit intermitent as well.
You put your foot down to accelerate, the vacuum diaphgram on the fuel pressure regulator should cause an instant increase in fuel rail pressure to match the increase in mainfold pressure but if the delivery of the pump is low the pressure will droop and cause a flat spot while the pressure recovers. If the delivery is really low then maybe the pressure doesn't recover at all under load and the engine then runs lean as long as the demand remains.
What does an engine under heavy load do when it runs lean, and its timed as far advanced as it can be ? It knocks of course! The lean mixture CAUSES the engine to knock/ping under conditions where it otherwise wouldn't. Even though I'm running on 99 octane a lean mixture can still cause it to knock, although not as badly as it would on 95. Running on 99 is allowing it to run on a leaner mixture without complaining as much.
Not only do you loose power from the lean mixture and potentially lean misfires occuring, when the ECU hears the knocking it retards the timing which reduces the knocking but causes further power loss! It takes a second or two for the timing to be retarded to the maximum possible and 5-10 seconds for it to be advanced again after the knocking subsides.
These time delays roughly match the delays I'm seeing where the performance sags under load and then recovers over a few seconds under a lighter load.
In other words yes the ECU is retarding the timing due to knock sensor activity, (I saw some of that happening today on the Lexia too) but its real knocking caused by running lean under load not a faulty sensor!
Ok what to do about it then...One thing I haven't checked before is the voltage at the fuel pump, so I ripped up the seat cover and pulled out the bung for the pump. At the top side of the connector there is 12.75 volts while idling (seems a bit low but not excessively) however there could still be a voltage drop across the connector so I unplugged it and gave it the same contact tweaking and cleaning treatment as the knock sensor and went for a drive.
Was there an improvement ? YES, there was! Not cured because it still swings lean and stays lean under wide throttle below 2500 rpm... but acceleration off the mark was a lot better, it even spun the front wheels on take off in 1st in the dry for the first time in a long time.
The connector didn't look bad so I suspect that its not the real problem, but that what I did may have reduced a small voltage drop allowing the pump to eek out slightly more delivery.
So what could my fuel starvation be ? It isn't the fuel filter because that's already been replaced and didn't make any difference. I think I've ruled out the voltage to the pump - 12.75v is a wee bit lower than I'd like but the wiring is quite thin so it could well be the normal voltage drop for that wiring. (Someone want to measure theirs ?)
Could there be a blocked fuel strainer on the bottom of the pump ? I've read about them getting badly blocked on diesels due to poor biodiesel, but I'm not sure whether its something that would happen with petrol ? The pump doesn't look easy to remove, how hard is it to lift it out and clean the filter ?
Faulty pump maybe ? The trouble is they're not cheap, and I can't see a good way to conclusively prove that the pump has poor or intermittent delivery with the facilities I have, and I'm not willing to replace it on a hunch.
Intermittent fuel pressure regulator ? Normally I couldn't see how the regulator could cause poor delivery as such, but it could certainly cause intermittent fuel pressure if there was a tiny crack in the diaphragm, and that problem may come and go intermittently with throttle changes due to the change in vacuum flexing the diaphragm...the pressure on mine is also a bit low - 2.7 bars instead of 3.0.
I DO actually have another fuel pressure regulator on hand, Stempy very kindly donated a spare 2nd hand but known good regulator he had off his V6 which he no longer has need for (much appreciated!) so I guess the logical thing to do would be to replace the fuel pressure regulator first, now that I have one ?
Obviously its a manifold out job, I still have one spare 2nd hand good manifold gasket (that was only on for a few weeks) however once the top manifold is out, how is the regulator replaced, does it require further disassembly ?
By the way the gearbox has settled down nicely now, I think it was a mistake to do the auto-adaptive reset, I should have just left it alone after putting in the additive. It's not as-new by any means, but there are no obvious signs of trouble at the moment, its quiet, doesn't shudder, gear changes are smooth, no banging into 1st from 2nd, kickdown changes are very quick but smooth, no sign of slipping or rpm flare up during up-shifts, basically it seems to be working normally.
So I'm very happy with the improvement in the gearbox, now if only I can nail this engine problem...
