2004 Berlingo 1.4i - Unstable fuel and ignition

[Pilot65]

VERY low power at the bottom of the valley, have to shift down 1-2 gears to get anywhere.
Both Lambdas are new.
TMAP checked OK.
Injectors checked OK.
Lambda signals checked OK.
It keeps hunting like this even if the throttle body is disconnected. Look at the fuel trim and ignition advance!

If I disconnect both lambdas the hunting is gone, and there is plenty of stable power, but then its in failsafe open loop mode, AFR was then measured to below 6 (off scale) and fuel consumption is up 200-300% and the car will obviously fail the upcoming MOT.

Have a look at my engine problem and live data here:

https://www.youtube.com/watch?v=CrD_BjSCabg

[Sol]

Is your fuel pressure fluctuating? I'd suspect the rail sensor, hp fuel pump, tank vent blocked, tank strainer blocked, in that order. you need Lexia to see the fuel pressure.

I'm a diesel guy so not up on petrols as much but it looks like cyclic fuel starvation.

[Pilot65]

Is your fuel pressure fluctuating? I'd suspect the rail sensor, hp fuel pump, tank vent blocked, tank strainer blocked, in that order. you need Lexia to see the fuel pressure.

I'm a diesel guy so not up on petrols as much but it looks like cyclic fuel starvation.

Havent measured the fuel pressure yet but will do it soon. This system does not have a fuel pressure sensor. The in-tank pump & regulator unit was replaced in 2019 by the previous owner (dealer job).
I have scoped on the coil return / monitoring pin and the signal was pretty messy... lots of noise, missing spikes, variyng amplitude etc. But it may be ok, as I have never checked such a signal before. But this engine problem doesnt seem like a coil issue anyhow.

I think this cyclic behavior looks a bit like as if the ECU is searching for the correct stoic signal from the lambda, but not quite catching it and overshoots from both the rich and lean side.
Have tried injecting an artificial lambda signal into the ECU (0-2V controlled by manually turning a potentiometer). The ECU / engine reacts by leaning / riching as the voltage passes through about 0.4V.
This was what I expected so I guess the ECU is ok. Also checked the heater signal from the ECU by placing a 12V/1W lamp over the heater wires; the light comes on for a few secs, then start flashing, indicating the heater driver circuitry in the Sagem ECU is ok.

[Sol]

Sounds like you're well on with the testing. Ok, so if it wasn't in closed loop mode, during cold start and warm up, it wouldn't care what the lambda reading was, so does it still act the same? (Hunting etc) and what does it do when you disconnect the MAF? Is the IAT and CTS working and showing sensible values? I know the IAT wouldn't make it hunt, just wondering about the ECU knowing the correct temps for air and water jacket...does it actually go closed loop at all....

Do you have Lexia? It's live sensor readings and logging functions are good for this stuff.

[Pilot65]

Sounds like you're well on with the testing. Ok, so if it wasn't in closed loop mode, during cold start and warm up, it wouldn't care what the lambda reading was, so does it still act the same? (Hunting etc) and what does it do when you disconnect the MAF? Is the IAT and CTS working and showing sensible values? I know the IAT wouldn't make it hunt, just wondering about the ECU knowing the correct temps for air and water jacket...does it actually go closed loop at all....

Do you have Lexia? It's live sensor readings and logging functions are good for this stuff.

Starting from (near freezing) cold the engine runs stable for some 30-60 secs, then starts hunting.
I assume thats the point when the ECU sees the lambda sensor is up to temp, and starts using the signal from it. At this point the logger shows transition form open loop to closed loop.
IAT and CTS data are correct and rock steady at all times.
When I disconnect the TMAP connector the reading shows slightly above atmospheric pressure (1200mB), and I obviously loose IAT (did not notice what value it defaults to), the engine is still hunting but in a somewhat diffenent way.

Yes I have Lexia3/PP2000 but havent installed it yet. Must dig up an old XP laptop first  
Also have an Innovate LC-2 wideband kit, today I will try to wire the narrow band output from this into the ECU and see what happends.
Strange that a 3 week old OEM compatible lambda sensor would be deffective, but it may be that it is not compatible with this engine, and that it NEEDS to be a true OEM device...
The LC-2 test may shed some light on that.

[Sol]

That could be the case. I haven't had much luck with non OEM sensors I have to admit, not just on PSA cars. BMW, Vauxhall, etc from reversing sensors to cam position, they were all crap and when I bought OEM, they worked straight away.

It will be interesting to see what the wideband kit shows...I had near identical behaviour from an m59 HDi, but the other way round. On open loop it hunted like a mad thing and blew black smoke from the exhaust like the red arrows. As soon as it transitioned to closed loop (60 secs) and started using the lambda values, it went smooth as butter with not a puff of smoke, even at WOT in high gear from mid rpm.

I tried everything, IAT & CTS were spot on within a couple of degrees and rock solid too, I only ever found disconnecting the EGR stopped the weird behaviour in my case.

I suspected the common rail pressure sensor at one point but since it worked perfectly after it went closed loop, couldn't be that. But when the EGR made a big difference I decided it wasn't actually closed during the open loop cycle when the ECU thought it was, and was sending the engine into a hunting action from the actual airflow not matching calculated values, so it was overfuelling. Oxygen levels would have been way below what was required for stoic, so it overfuelled badly and smoked until it was closed loop and began to take account of recycled EGR gases.

It would be most useful in your case to know the AFR during closed and open loop, fuel pressure and measured airflow as well I guess eh. The wideband kit should show if it suddenly jumps out of stoic when it goes to closed loop, but then that would point you back to the lambda sensor?

[Pilot65]

Getting closer: https://www.youtube.com/watch?v=6-NqhL0E3AE read video description.
Its still not perfect but a whole lot better than with the narrowband sensor.

Before reaching out for a new and different O2 sensor I will get Lexia running and get into the ECU to reset the auto-adaptive settings for O2 and throttle body.

[Sol]

Interesting. Looks like it thinks it's overfuelling as soon as it goes into closed loop, going by the fuel trim, yet if the LC-2 gauge is showing AFR then it's reading lean? (If I'm reading things correctly) and the timing jumps all over the place, yet should be something like 9 to 12 degrees btdc I'd guess and should be steady? I'm really not familiar with the 1.4i - does it have single point or multipoint injection?

Camshaft timing sensor?

Edit: One thing I have read on another Citroen forum is the 1.4i is prone to head gasket failure or cracking of the head causing similar symptoms. Have you carried out a cylinder compression test? I'd pull all 4 plugs and check the compression before doing any more. Also seems not to show the usual signs of water and oil mixing if the head cracks but the compression test shows up bad readings.

[Pilot65]

Yes the AFR is all over the place, I guess its because the sensor is so close to the ex ports. I've used the LC- 2 in the tailpipe on non-cat cars and there the AFR is stable and very readable around 14. I guess the exhaust is better mixed and evend-out as passes through piping and silencers.
Looking closer at the video (pausing) there is a clear correlation between negative fuel trim and high AFR. parameters are still fluctuating a good deal but not nearly as much as in the 1st video.
This is a simple engine, no cam sensor, no fuel pressure sensor, no fuel return, no EGR... only the bare essentials.

No oil/water mixing found. Will do a compression test after the holidays... a leaky in/ex valve could cause lots of trouble for a ECU controlled engine.

[Sol]

Yeah, it's really weird to get negative fuel trims but very high (lean) AFR.... Could be the post cat o2 sensor is reading very low oxygen when it isn't low, after all it can't be >22 before the cat then <14 after it, so if the lc-2 is reading pre cat AFR then the post cat sensor, which should be very stable is maybe shot and not detecting the high AFR that's actually there, so the ECU reduces fuelling making it run very lean, and pre-ignition occurs causing it to be very lumpy?

Burnt or sticking ex valves would be putting unburnt fuel into the cat on the compression stroke...so the AFR should be low at the lc-2? I'm thinking the head could be cracked between adjacent cylinders causing mixing of exhaust gas and unburnt fuel, it would behave like a bad EGR on an engine that had one.

Ignition timing bouncing around so much is also really weird, how is the ignition timing & triggering done on this engine ?