Xantia V6 broken exhaust (update: and rough low rpm running)

[addo]

Stevie: if you seriously believe that in a modern car, erratic conditions as you describe are principally the result of pump fuel variations, I have some news for you.

You're wrong.

[lexi]

My V6 Vel with Nissan engine states minimum 95 Ron in Dialogys. I use Shell V Power. That may be 98?

My thinking is to keep engine clean and maybe combat Ethanol content. I also find it a great icebreaker at parties, asking everyone their fuel RON.
I do remember Honda tech saying that to run an engine with higher octane than it needs is a waste. They stated that there would be unburnt deposits that engine was not igniting and using. Mmm...

[Hell Razor5543]

However, cars that are designed to run on the higher octane fuels (such as the Mitsubishi Evo 10) do not run well on lower octane fuels.

[Mandrake]

My V6 Vel with Nissan engine states minimum 95 Ron in Dialogys. I use Shell V Power. That may be 98?

Shell V-Power is 99 RON, BP Ultimate is 97 RON, BP Unleaded is 95 RON, and I think Shell Unleaded is 95 RON as well.

My thinking is to keep engine clean and maybe combat Ethanol content. I also find it a great icebreaker at parties, asking everyone their fuel RON.
I do remember Honda tech saying that to run an engine with higher octane than it needs is a waste. They stated that there would be unburnt deposits that engine was not igniting and using. Mmm...

Can't say I've ever heard that running on a higher than optimal RON fuel is a problem - except to the wallet.

Running on a lower than optimal RON forces the ECU (on cars with knock sensors) to reduce timing advance below optimal under load, especially at lower RPM so there is a loss of torque (more so at low RPM) and a decrease in MPG, both of which come from the change in timing. Less advance means less responsive throttle at low RPM too.

The Xantia V6 is tuned for maximum timing advance and performance/efficiency at 98 RON - any higher RON theoretically won't make any difference. The minimum it can cope with is 95 RON, which will be limited by how much the ECU is willing to reduce timing advance in response to detected knocking. If you tried to run it on 91 RON (yep, 91 RON is still available in NZ AFAIK ) then even with minimum allowable timing advance it may still knock which wouldn't be good for the engine long term, and it certainly wouldn't perform well.

Another factor though is lean running - if anything is causing it to run lean under load (as mine appears to be doing, see next post...) then for a given octane and timing advance it will knock more, triggering more knock retard to the timing than you would normally get for that grade of fuel. Or looking at it another way, a higher octane is required to prevent knocking at the same timing advance as before. So lean running effectively increases required RON to maintain good performance and avoid excessive reduction in timing advance. (Again, that seems to be what's happening to mine)

[Mandrake]

Update time.

A little bit more detective work is starting to unravel what is really going on I think, putting a scope semi-permanently on the oxygen sensor so I can glance at it when I notice the performance isn't what it should be has been very helpful.

Performance hasn't been as nippy since I changed the injectors over and there has been more of a tendency to "bog down" at low RPM and high load. The scope on the O2 sensor has been showing me that the bogging down correlates precisely with lean running in open loop mode, or at least a lean O2 sensor reading. (I can't entirely rule out a misfire causing a false lean O2 reading - but in that case is the misfire the chicken or the egg ? If there is a misfire it could be due to a lean mixture)

One thing I did revert when I swapped the injectors over was I re-instated the canister purge valve line and it has been nagging in the back of my mind that this was the reason for the regression in performance, and especially the somewhat intermittent nature of the poor performance. So on Friday I blocked it off again:



Again, I was careful not to disturb anything else, the car had been driven quite a bit earlier in the day with the less than stellar performance evident (both observed performance and lean episodes on the O2 sensor) and was still warmed up when I went back out for a drive again. The difference was immediate.

A lot more eager off the line, (in fact now able to spin the wheels on dry surfaces) a lot sharper throttle response below 2500 rpm and especially below 2000. In fact overall power below 2500 was considerably better. Although "bogging down" under wide throttle still occurred in the 1500-2500 rpm range it was much less frequent, and when it did occur it "recovered" much quicker. (Also shown by the O2 reading which went less lean and recovered to a rich reading a lot quicker) Again any bogging still correlated perfectly with the lean O2 reading.

Performance was back to at least as good as it was before I did the injector swap, probably slightly better, and I still have the smoother idle and lack of cold start misfire that the injector swap seems to have cured.

So what's happening ? Clearly disconnecting the evap system has made the car run less lean (but still not rich) in high load low rpm situations where it was previously bogging badly due to a lean mixture.

Here's the thing - I DON'T think that the evap purge system is faulty. It's normal for the canister purge solenoid to affect the mixture a small amount when it opens, and it does open fairly wide under wide throttle conditions. (This can be seen on a Lexia)

When the canister is empty of vapours after some period of driving the only thing that the purge solenoid lets through is extra air (basically a small but expected vacuum leak) whereas when its full of vapours after the car sitting for a while it will potentially richen the mixture slightly or at least not cause it to go as lean as an empty canister.

This helps to explain the seemingly random fluctuations in performance that come and go while driving - the canister purge operation is somewhat random as you drive (it's not of course, its under ECU control, but the logic the ECU follows on when and how much to open it is somewhat inscrutable, so it seems random to us humans) and the leaning effect when its open will be WORSE after the car has been driven for some time and all the stored vapours have been consumed, hence better running initially and gradually sagging performance the longer you drive as the leaning out gets worse.

As I said, I don't think the evap system is faulty - its doing what it should do and a small effect on mixture is normal and expected. In closed loop mode the O2 sensor can compensate for the purge valve opening, and in open loop mode you should be running MUCH richer than stoichiometric, where a small fluctuation in mixture won't matter and wouldn't be noticed.

Also if it was faulty blanking it off would cure the symptoms completely - it doesn't, it just lessens their severity. So it's a clue and a partial workaround, (it certainly runs a lot better) but not a real cure.

What the real problem is then is that the open loop mixture is TOO LEAN below 2500 rpm, for reasons as yet unknown. The research I've done suggests that for a NA engine in PE (power enrichment) mode, which is basically whenever you accelerate hard enough to go from closed loop to open loop, (clearly visible on the O2 signal) the mixture should be at least 13.5 to 1, with 13 to 1 typical at lower RPM dropping to as low as 12 to 1 at maximum RPM / full throttle.

It's hard to tell what the mixture actually is with a narrow band O2 sensor - a full rich reading occurs at something like 14.2 to 1 and full lean at 15.3 to 1 with a neutral 450mV signal at 14.7 to one. The point is in PE mode the O2 sensor should read flat line rich CONSTANTLY until you either let off the throttle enough to allow closed loop to resume, or let off the throttle completely to allow injector cut-off. (which instantly swings full lean)

What I was seeing under heavy throttle at low RPM when it bogs down was an initial rich flat-line lasting maybe half a second then as soon as the engine started sagging I'd see the O2 voltage dropping with a noisy fluctuating line heading down towards lean - sometimes it wouldn't go all the way and would linger fluctuating between the two extremes (eg approx stoichiometric) other times it would go right down to a full lean reading and stay there for a number of seconds until I either eased off the throttle slightly or until it recovered. If its flat-lining lean on the O2 it must be about 15.3 to 1 or LEANER, when it should be at least 13.5 to 1 or RICHER! No wonder the poor thing is bogging!

Of course running it so lean under a heavy throttle load will cause knocking, the knock sensor will detect this and yank the timing back to protect the engine, the reduced timing will also cause a massive loss of power. So you have a one two punch of lean mixture and retarded timing.

So there is no doubt in my mind that the fundamental problem is lean running in open loop PE mode, and that any timing retard is a secondary effect in response to actual knocking caused by that. (Although potentially causes more of the total power loss)

Could it be a vacuum leak through the cam box breather ? To test that theory I unplugged it from the throttle body and blocked it off. A very slight further improvement but nothing compared to the canister purge line so that isn't the cause either. Probably just a very small, normal air leak there to draw the fumes out of the cam boxes.

Could it be a fuel delivery problem, eg low fuel pressure under high demand ? I'm not sure about that either, because I've done quite a bit of testing at higher RPM too, and from 3000 rpm to 6000 rpm I can floor it up a steep hill and the O2 reading will stay full rich constantly (as it should) and the performance is good too.

Of course with a narrow band sensor I can't tell HOW rich it is, it should be about 12 to 1, but it might only be 13.5 to 1 for example and still give the same rich O2 reading. So while it seems ok at higher RPM it might still be marginal there too.

So at the moment I know what the problem is (lean running in open loop, possibly only below 2500 rpm) but I don't know what the cause is. In another post I'll speculate on a few possible causes and what I can (or can't) do to try to pin it down.

In the meantime Steve, could you try blanking off the canister purge line as I did, and see if it improves the performance and more importantly makes it a bit more consistent ? If it does then it would suggest your problem is also mixture related, even though the evap system is not the actual cause of the problem. (The mixture is just so marginal than the small change that the purge solenoid activation causes is enough to make it even leaner and cause the engine bog down even more)

[addo]

When you're logging that lean condition, what's your MAP reading doing?

[Mandrake]

When you're logging that lean condition, what's your MAP reading doing?

Good question addo, that's what I intend to find out, if I can find enough bits to make a second back probe lead that I can run into the cabin to monitor the MAP voltage on channel B on the scope while I keep the O2 on channel A.

I'm starting to suspect that this ECU is old enough and primitive enough that it does NOT apply learned long term fuel trim to high load (open loop only) cells, and thus can't compensate in open loop for fixed errors such as a lower than normal injector flow rate or a lower than normal (but steady and well regulated) fuel pressure.

What I mean by this, is that more modern ECU's will use oxygen sensor feedback to learn long term fuel trim for each load/rpm cell, and those learnt readings are extrapolated into the higher load range where the car always runs in open loop - eg near full throttle.

So if it learns that it needs a +10% fuel correction at 2500 rpm at 50% load in closed loop, that it probably needs +10% correction at 2500 rpm at 80% load in open loop mode as well. So this type of ECU would adapt to a fuel pressure 10% too low over time even in open loop.

But after watching the O2 sensor under many conditions especially closed to open loop transition I've come to the conclusion that the MP7.0 system can't do this and that high load open loop fuelling is entirely pre-set and determined only by other inputs eg RPM, MAP, TPS, Coolant temperature and Air intake temperature. Eg long term fuel trim from the oxygen sensor is never considered for high load open loop fuelling thus no adaption ever takes place.

Meaning that while in closed loop it has adapted the mixture correctly to compensate for eg low fuel pressure, but as soon as you increase the throttle just enough to push it into open loop it switches back to its pre-set tables which are (due to some problem such as low fuel pressure) causing it to fuel lean, thus a hesitation/stumble/sag in power at that transition.

To try to prove whether its an actual lack of fuel pressure or whether the ECU isn't fuelling enough I thought about connecting a potentiometer to the output of the MAP sensor so that the output can be "scaled". Connect the top of a 1k pot to the 5v reference through a 1k resistor, the bottom of the pot to the MAP sensor output pin and the wiper of the pot to the ECU.

At one extreme of the pot the signal would work over its normal range, as you turned the pot the voltage would be "scaled" towards the higher load end and should force the ECU to increase fuelling in open loop mode by tricking it into thinking the load is greater than the MAP sensor reports.

[Mandrake]

When you're logging that lean condition, what's your MAP reading doing?

Didn't get a chance to check MAP readings today, the weather closed in on me so maybe next weekend.

Meanwhile here are some O2 sensor captures I made to try to demonstrate what I'm talking about.

First a normal closed loop reading at idle:



Nothing wrong with that, at least once the O2 sensor has been warmed up by driving.

Here is closed loop at 2500 rpm, which is where the emissions test is performed:



Again, nothing wrong with that.

Now some taken when applying wide throttle between 1500-2000 when the engine really starts bogging badly:



To interpret this (and the following) graphs properly you have to realise that this scope doesn't support a "scrolling" mode, it operates in wraparound where it draws from left to right then starts at the left again and overwrites what was previously there. So in the above capture the vertical line under AC50mV is where the capture started/stopped.

Wide throttle had just been applied around 2200 rpm, the engine started bogging badly as the TC locked up and you can see the O2 reading drop like a stone (about 2 seconds from rich to lean) and then go flat-line lean until the end of the capture.

Another example, this time you can see the momentary full rich when opening the throttle that in just over half a second gives way to a rapid fall to mostly lean:



Unlike the previous example it doesn't completely go lean so is probably relatively near stoichiometric, but should still be rich under wide throttle.

Here are a couple of examples of bogging where the bogging fluctuates up and down rather than being a steady downwards slide:



Looks almost like its trying to switch to closed loop in that one, but its not as the throttle is open about 80% which is well past a "maximum load" MAP reading at low RPM. (The ECU would never be trying to operate in closed loop at full load)

One final one where its bogging and the fluctuation in bogging and O2 reading are all over the place:



Now I'm no expert but apart from the first two showing good closed loop response I don't think ANY of the others are "normal". What I don't know the answer to either, is why the graph is so "noisy" when in open loop and the engine is bogging, when in closed loop the waveform is nice and clean, likewise when its working properly in open loop it shows a nice clean switch between rich and lean when pressing and releasing the accelerator.

But somehow when bogging the waveform is very noisy. I'm wondering if the "noise" on the waveform is a sign of a misfire causing turbulent pockets of oxygen to pass the O2 sensor ?? Just a guess there. Maybe its just a result of the extreme sensitivity of a narrowband sensor to mixture - the tiniest little change in O2 level near stoichiometry results in a large voltage swing, hence tiny fluctuations are exaggerated. However away from stoichiometry even large changes (like 12 to 1 to 14 to 1) make no difference to the voltage.

So do I have an excess air flow problem, (vacuum leak etc) air flow measurement problem (MAP etc) causing mis-calculation of open loop injection times, or a fuel supply problem. (injectors/fuel pressure/fuel pump) That's the million dollar question.

[Mandrake]

Almost forgot, my new oxygen sensor arrived yesterday. The correct OEM type this time... Looks quite different, measures very different, has the right colour (brown not black) and keying on the connector.

The heater on the one on the car at the moment measures 6.6 ohms, the heater on this one is 2.2 ohms. Bit of a difference!!

Assuming about 13.8 volts at the connector that's 86 watts compared to 29 watts, so its no wonder the current one doesn't warm up by itself in a timely fashion. Bearing in mind that's cold resistance, once warmed up the wattage will drop back quite a bit to help regulate the temperature.

Will be interesting to see how quickly this new, correct type warms up, (I'm guessing fully warmed up at idle in under a minute) but it will be about a month before I get a chance to swap it over.

[Mandrake]

What I don't know the answer to either, is why the graph is so "noisy" when in open loop and the engine is bogging, when in closed loop the waveform is nice and clean, likewise when its working properly in open loop it shows a nice clean switch between rich and lean when pressing and releasing the accelerator.

But somehow when bogging the waveform is very noisy. I'm wondering if the "noise" on the waveform is a sign of a misfire causing turbulent pockets of oxygen to pass the O2 sensor ?? Just a guess there.

Looks like I might have made a pretty good guess stumbling across the following article I seem to have answered my own question:

http://www.motor.com/magazine/pdfs/102011_05.pdf" onclick="window.open(this.href);return false;

The noisy, furry oxygen sensor signal does indeed indicate a misfire due to brief pulses of excess oxygen from the misfiring cylinder(s) interspersed with normal cylinders. Rather interesting!

How does this help my diagnosis though, am I just getting one or two cylinders on the verge of a lean misfire as a result of a lean mixture (eg the misfire is a symptom of the real problem) or is there something else causing a misfire that is then causing a lean reading once enough cylinders misfire at once ?? Not sure how to tell, at least without having it put on an exhaust gas analyser under load.

[stevieb]

Stevie: if you seriously believe that in a modern car, erratic conditions as you describe are principally the result of pump fuel variations, I have some news for you.

You're wrong.

I'm sorry, but I disagree. I had a MASSIVE improvement in performance, mpg, start-times, throttle response, and general ease of driving just by cleaning up the corrosion on the fuel pump terminals, presumably allowing the pump to receive its full voltage, thereby supplying the pressure expected at the injector rails. I don't know how much headroom there is between the pressure the pump is capable of producing and the pressure required - if there isn't much spare and the pump isn't working to full capacity because of low-voltage/wear-and-tear then it stands to reason that performance is going to suffer. I live in a city built on seven hills - so the slightest power loss is VERY apparent.

Of course, if fuel pressure wasn't an issue then there wouldn't be so many people talking about cleaning fuel filters to ease the flow...

I think the fact that Simon has witnessed the pump drawing in air when the fuel level was below a given point, and my own car was running like a dream until my fuel dropped low - I felt and heard this change at 50mph on the motorway, there was no mistaking it - seems to add to the assertion that there's potential there for problems. I think the return of water vapour in the exhaust after a long journey is an indicator that something is amiss - I'd say head-gasket if I was losing coolant, but I'm not.

It's possible something could be happening elsewhere in response to the fuel level dropping low (I still wonder if a low fuel state causes the ECU to drop into a "low power mode" in the same way limp-mode kicks-in under certain circumstances - when mine got louder and slower, the mpg improved quite markedly on the same stretch of flat road), but it seems odd that mine hasn't completely cleared despite the addition of half a tank of fuel .

Adding that half a tank of fuel last week (Shell Regular) did improve things a lot, but it's still not right. However, I felt a change on Sunday that was a lot like a part-reversal of the power-drop I had the week before, so who knows? Maybe the air is shifting. I've seen a number of videos online where fuel rails (or parts of them) can become vapour-locked on a warm engine if the fuel pressure is too low or air is being introduced somewhere - my thoughts are that the symptoms Simon and I have experienced with slowly deteriorating power is something to do with this (tiny air bubbles in the fuel supply finding corners in the fuel rail/injector unions to bunch-up and create restrictions in the flow), but I don't have the equipment to test the theory - just a growing urge to change the car instead.

As for fuel, I've had a number of occasions in the past when I've had "bad fuel" - where a car has run like a dog when I've used fuel from a particular garage (usually a supermarket garage), and this is why I use only branded fuel now. As Simon says, the ECU runs different settings based on the octane used, so clearly the fuel grade makes a difference to the engine's performance. I also know that some garages are inclined to add some of their unsold high octane fuel into the tanks of the regular fuel right before delivery day (it's all to do with making sure the delivery tankers go home empty), and this means real octanes could actually vary anywhere between 95 and 99 on any given day. I sometimes wonder if the minor "power loss" I sense is actually nothing more than getting unlucky and getting some straight 95 RON fuel, rather than the mish-mash I get on other days. I often add fuel at my local Shell garage after the tanker has been - they close the forecourt when it's there, usually just when I need to go and top-up

Of course there WAS a case in the UK some years ago of bad fuel being dispensed at supermarket chain forecourts. Apparently this fuel was destined to be "blended", but was somehow mis-directed to the pumps instead. You have to wonder whether this mis-direction happens more than we realise or the "blending" varies a little more than we realise. This IS rip-off Britain after all...!!! http://news.bbc.co.uk/1/hi/6405701.stm

As for my girlfriend's car, her power loss was horrendous. On a trip down the motorway with a break at the services, it was happily teasing her to break the 70mph speed limit on the first leg, then on the second leg struggling to get much above 50mph with her foot in the carpet. The only changes the previous day were the oil and filter (potentially wrong grade oil, but unlikely at the main dealer - I'd bet most modern Toyota's use the same grade anyway), some injector cleaner, and a tank full of Shell fuel. My main thought is the injector cleaner, but that coincidence with Shell has got me wondering...

Coincidentally, the use of injector cleaner by a main dealer at a 10,000 mile service maybe speaks volumes, don't you think? Once upon a time we were told injectors never needed servicing. Not any more it seems...

Anyway. I'm planning on running mine low(ish) before slinging in some non-Shell fuel to see if anything changes, but for now my mpg is still above 30 and my power-loss certainly isn't in the league of Simon's past problems (it's the extra noise that annoys more than anything else), so I'm going to live with it for now.

[stevieb]

What I don't know the answer to either, is why the graph is so "noisy" when in open loop and the engine is bogging, when in closed loop the waveform is nice and clean, likewise when its working properly in open loop it shows a nice clean switch between rich and lean when pressing and releasing the accelerator.

But somehow when bogging the waveform is very noisy. I'm wondering if the "noise" on the waveform is a sign of a misfire causing turbulent pockets of oxygen to pass the O2 sensor ?? Just a guess there.

Looks like I might have made a pretty good guess stumbling across the following article I seem to have answered my own question:

http://www.motor.com/magazine/pdfs/102011_05.pdf" onclick="window.open(this.href);return false;

The noisy, furry oxygen sensor signal does indeed indicate a misfire due to brief pulses of excess oxygen from the misfiring cylinder(s) interspersed with normal cylinders. Rather interesting!

How does this help my diagnosis though, am I just getting one or two cylinders on the verge of a lean misfire as a result of a lean mixture (eg the misfire is a symptom of the real problem) or is there something else causing a misfire that is then causing a lean reading once enough cylinders misfire at once ?? Not sure how to tell, at least without having it put on an exhaust gas analyser under load.

I don't think you'll have any definitive answers until that oxygen sensor is swapped.

Step back from it for a moment. You're looking for a solution to a problem which on its own could be caused by the wrong oxygen sensor. There are a whole bunch of tell-tales that you've revealed, but they're being revealed either by that wrong sensor or by other sensors working alongside the wrong sensor.

Put it another way. You could barely complain about the ECU running an incorrect mixture if you knew for certain that the CTS was giving it false temperature readings, could you?

[Mandrake]

Stevie: if you seriously believe that in a modern car, erratic conditions as you describe are principally the result of pump fuel variations, I have some news for you.

You're wrong.

I'm sorry, but I disagree. I had a MASSIVE improvement in performance, mpg, start-times, throttle response, and general ease of driving just by cleaning up the corrosion on the fuel pump terminals, presumably allowing the pump to receive its full voltage, thereby supplying the pressure expected at the injector rails. I don't know how much headroom there is between the pressure the pump is capable of producing and the pressure required - if there isn't much spare and the pump isn't working to full capacity because of low-voltage/wear-and-tear then it stands to reason that performance is going to suffer. I live in a city built on seven hills - so the slightest power loss is VERY apparent.

Of course, if fuel pressure wasn't an issue then there wouldn't be so many people talking about cleaning fuel filters to ease the flow...

Before you get too hot and bothered Steve I think you're misinterpreting what addo was trying to say. He's saying the erratic performance you were seeing isn't going to be a result of variations in the quality of the fuel from one filling to another, he doesn't mean as the result of variations in the fuel PUMP in the car, eg fuel pressure. I'm not sure that I agree though as I have sometimes seen significant changes in behaviour with different fuels.

I think the fact that Simon has witnessed the pump drawing in air when the fuel level was below a given point, and my own car was running like a dream until my fuel dropped low - I felt and heard this change at 50mph on the motorway, there was no mistaking it - seems to add to the assertion that there's potential there for problems. I think the return of water vapour in the exhaust after a long journey is an indicator that something is amiss - I'd say head-gasket if I was losing coolant, but I'm not.

Yep same here - I've sometimes wondered if I might have a partially blown head gasket, but since I replaced my leaky coolant expansion bottle I've had ZERO drop in coolant even with the pressure cap on tight. (which also suggests my heater matrix is ok too ) I would have thought a blown head gasket would use coolant relatively quickly when the system is under pressure. (I don't have any direct experience of failing head gaskets though - so far I've gone through life without any!)

It's possible something could be happening elsewhere in response to the fuel level dropping low (I still wonder if a low fuel state causes the ECU to drop into a "low power mode" in the same way limp-mode kicks-in under certain circumstances - when mine got louder and slower, the mpg improved quite markedly on the same stretch of flat road), but it seems odd that mine hasn't completely cleared despite the addition of half a tank of fuel .

There's no "low power mode" on this ECU that I'm aware of. If you lost power but the MPG actually improved I would say your engine was possibly running lean. Lean running up until the point where it starts misfiring and/or knocking (causing timing retard) will give better MPG but worse performance. Normal steady cruising is stoichiometric in closed loop mode with the O2 sensor, but stoichiometric is not optimal for MPG, its only optimal for the CAT performance and lifetime.

Are you sure you don't have an intermittent fault with your O2 sensor ? A faulty O2 sensor does NOT trigger the engine management light on this engine and causes very little observable symptoms. If the O2 sensor fails the ECU runs in open loop mode all the time which will run the engine slightly rich all the time - so an unplugged / faulty O2 sensor gives worse MPG (and would fail an emissions test) but actually gives better performance. Unless you've measured the O2 sensor while driving like I have you would never really even know that it wasn't working properly sometimes.

Adding that half a tank of fuel last week (Shell Regular) did improve things a lot, but it's still not right. However, I felt a change on Sunday that was a lot like a part-reversal of the power-drop I had the week before, so who knows? Maybe the air is shifting. I've seen a number of videos online where fuel rails (or parts of them) can become vapour-locked on a warm engine if the fuel pressure is too low or air is being introduced somewhere - my thoughts are that the symptoms Simon and I have experienced with slowly deteriorating power is something to do with this (tiny air bubbles in the fuel supply finding corners in the fuel rail/injector unions to bunch-up and create restrictions in the flow), but I don't have the equipment to test the theory - just a growing urge to change the car instead.

Regarding vapour lock in the fuel rail - I'm not sure that its causing an issue for us (or me at least) while driving but I DO think that it's the cause of the intermittent near stall/surge cycles when starting a HOT engine that has sat for 10-30 minutes after being switched off.

This seems to be a relatively common problem on the ES9J4, I've certainly seen it discussed on here before - start a hot engine after it has sat for 10-30 minutes and it will struggle to start, drop below idle nearly stalling, surge up to about 800 rpm, drop below idle again, surge up again, and repeat this maybe three times before it finally settles down and becomes stable. Mine does this quite often but not always. I think where the intermittent nature of the problem comes from is that the non-return valve in my fuel pump is a bit intermittent. During my testing in the shed with my original fuel pump sometimes it seals and holds rail pressure for a long time after the pump is turned off (10-20 minutes or more) and sometimes it leaks away quickly in under a minute. I guess the ball valve is worn out.

If the pressure leaks away quickly after the engine is stopped and while the engine bay and rail is still hot the fuel in the rail and injectors can boil (especially if its a volatile winter grade fuel on an unusually hot day for the time of year) and cause a vapour lock condition. When you try to start it struggles to run due to the vapour lock which is not cleared until the fuel pump has circulated cold fresh fuel from the tank. If the pressure in the rail is held for a long period of time after switch off, until the rail gets time to cool down below the critical temperature, no vapour lock occurs and the car starts easily.

Thinking about it I now understand why the fuel pump is briefly activated every few minutes after switch-off for some period of time - to try to keep the fuel rail pressure topped up until the rail cools sufficiently and avoid vapour lock occurring! Of course if the pump only runs for 2 seconds once every 5 minutes for a total of 20 minutes and the pressure leaks away after 30 seconds then vapour lock will still occur.

I'm pretty sure both my original pump AND the replacement second hand one in the car now have problems with the non-return valve intermittently leaking back, so I think other than a brand new expensive fuel pump the only way to solve this issue 100% would be to insert an external non-return valve in the line between the pump and the rail. (on either side of the fuel filter would be the ideal place) There are plenty available for not much money the only problem is trying to safely splice it into the fuel line without cutting and hacking the line. (Finding one that has the PSA snap lock connectors would be ideal - most just have spigots)

As for fuel, I've had a number of occasions in the past when I've had "bad fuel" - where a car has run like a dog when I've used fuel from a particular garage (usually a supermarket garage), and this is why I use only branded fuel now. As Simon says, the ECU runs different settings based on the octane used, so clearly the fuel grade makes a difference to the engine's performance. I also know that some garages are inclined to add some of their unsold high octane fuel into the tanks of the regular fuel right before delivery day (it's all to do with making sure the delivery tankers go home empty), and this means real octanes could actually vary anywhere between 95 and 99 on any given day. I sometimes wonder if the minor "power loss" I sense is actually nothing more than getting unlucky and getting some straight 95 RON fuel, rather than the mish-mash I get on other days. I often add fuel at my local Shell garage after the tanker has been - they close the forecourt when it's there, usually just when I need to go and top-up

But do you fill with 99 V-Power or 95 ? Are you suggesting they put a bit of left over 95 into the 99 tank before the tanker arrives, or the other way around ?

[Mandrake]

I don't think you'll have any definitive answers until that oxygen sensor is swapped.

Step back from it for a moment. You're looking for a solution to a problem which on its own could be caused by the wrong oxygen sensor. There are a whole bunch of tell-tales that you've revealed, but they're being revealed either by that wrong sensor or by other sensors working alongside the wrong sensor.

Put it another way. You could barely complain about the ECU running an incorrect mixture if you knew for certain that the CTS was giving it false temperature readings, could you?

I take your point, however the problem with this oxygen sensor is it doesn't warm up unless driving, however once driving at a reasonable pace it does warm up and seemingly work properly.

You're right, I can't be 100% sure, but I have seen this same lean under wide throttle issue before on the Lexia with my original oxygen sensor WAY back at the end of 2012 just a few months after all the problems began - and that was many months before I replaced the oxygen sensor. Of course on the Lexia you can only see a slowly updating number for the O2 signal which is far less informative than looking at a waveform on a scope.

I do think lean during acceleration in open loop mode is key to the problem - I'm now racking my brains trying to filter out unrelated changes that were made along the way to try to figure out what could have originally gone wrong with the car that would have caused lean running. For the first 6 months it ran like a dream with fantastic throttle response and tons of power even on 95 fuel, then in the space of 2 months it went from perfect running and passing the emissions test with flying colours to running like a dog and struggling to get up hills in a high gear... and apart from a couple of occasions its never truly been 100% for a long period of time since.

I might have to go back and read through the beginning of this thread (*gasp*!) to see if I can find some clues from my own long forgotten posts...(there were also a couple of other brief, earlier threads where I first hinted at problems beginning)

Edit: here is what I believe was my first postings at the time trouble began - the shudder at 1500 rpm that I complained of turned out to be the gearbox, and a number of oil changes solved that problem, but you can see engine running problems creeping in too:

http://www.frenchcarforum.co.uk/forum/v ... =3&t=41141" onclick="window.open(this.href);return false;

Another weird thing I've noticed only in the last week or two is the engine fluctuating between two different rev's with a constant throttle - for example doing 60mph on the motorway, steady throttle opening and I can hear (and see on the rev counter) the engine surging slightly and backing off about once per second... very odd. Until a couple of weeks ago the idle speed was very steady and the smoothness overall very good.

This surging only seemed to start after I had the fuse box out the other day, but I can't see any connection to that, other than the battery being disconnected...

I have noticed for a while that the car is sometimes extremely zippy and responsive, (it really zooms away with light throttle opening and responds quickly) at other times it feels a bit flat and sluggish, comparatively speaking.

[Northern_Mike]

Coincidentally, the use of injector cleaner by a main dealer at a 10,000 mile service maybe speaks volumes, don't you think?

Yes, it simply says that they can get away with selling mugs any old crap they are told they need. No more, no less. They probably buy it in bulk for a couple of quid a bottle, easy £11 made on every service.

Sent from my GT-I9300 using Tapatalk