V8 starts on the nose,and idles fine, from cold. A couple of times the other day i "had fun" restarting when I'd just stopped for 5-10 minutes. Engine turned over no problem, but didn't fire up. After a few attempts it just about caught, and I had to keep it rev'd to 1500 for 15-20 secs for it to stop choking. As soon as I had it under load ie in reverse or 1st then it could idle at 650 again.

Does anyone recognise the symptoms?

I haven't yet cleaned up the IACV since i bought it, or fitted a new fuel filter, and i will shortly change the exhaust down-pipes and so clean the oxygen sensors at the same time. MAFs clean, injectors are ok, spark plugs and HT leads are new

Rob, plug nano in and check lambdas are working, mine did the same when one of them failed

Check the coolant temp is reporting to the engine ECU properly too. If the value is way off, it will mess up fuelling.

Chris/ Sloth, thanks for the suggestions, I'll check them out

If it’s a sensor, you’ll get fault code P0116 which is "Coolant temperature sensor - falling temperature fault".

I was supplied locally with a BMW part, which was pennies; No. VNE 39090 and it’s been fine since.

Is it on gas? If so, switch it to petrol a few minutes before you stop it, and see if it still happens.

Tnx for the tip George. No Brian, no gas...that would just give me another thing to fix. I'll get it on the nano this weekend.

OK, so no errors on GEMs in Nano. I completely replaced the exhaust system yesterday, and one of the o2 sensors was corroded in, so I'll be looking to replace that. I think the sensor was functioning ok beforehand though. I think the problem is that the air intake temp now reads anything from 1 to 8 degrees, and so at the restart (when the engine bay is now warm) the calculated fuel air mix is now wrong. I optimistically sprayed contact cleaner at the IAT sensor in the meantime - a new one's the better part of 100 quid

Unlikely to be the IAT as that only does anything when the ambient temperature in the engine bay gets too high. It has no affect at all at normal temperatures and is only there to retard the ignition timing if the intake air tempoerature gets over 55 degrees C. Unplug it and stick a 1k Ohm resistor in it's place and see if that makes any difference but I doubt it will.

This may be slightly unrelated, or it may not, but after replacing both 02 sensors for new (Lemark) I have a poor idle. According to Nano, the voltage on the sensors is hopping around ok ie 0-5V, but the trims to the LH bank are predominantly -ve, and for the RH bank are mainly +ve. Should there be a pre-defined relationship between the voltage and the trim - might this indicate one of my sensors is mis-reading? If so, which one? I still have one good one that came off, so could re-fit it, but ideally I would know which Bank to fit to.
I dug out some 6 month old screen shots of the 02 sensors, when the idle was running fine, and it was showing a similar relationship then -ve on the left, while +ve on the right, but obviously a screen shot was just a moment in time. The trim values seemed less extreme though. Also, my ECU is still learning, after a reset about 50 miles ago, but shd that make a difference?

There isn't a pre-defined relationship between voltage and trim... You've seen yourself that voltage can fluctuate the same on both banks while trims are different. It's unlikely one of the new lambda probes is at fault or one less compatible than the other if they're both the same spec.

The engine has a trim for each bank specifically to allow a different trim on each bank, there would be little point in having a different trim for each bank if trims were supposed to be exactly the same. That said, we usually expect trims to be similar - Trims are there to compensate for tolerances in sensor readings and mechanical aspects such as fuel pressure, injector flow rates, valve train (effecting airflow), even to compensate for the ECU's fuelling map not being quite correct. Some tolerances change over time as mechanical components and sensors age/wear. If everything was made to an exacting standard and remained the same as it ages trims wouldn't be necessary. We expect trims to be similar because tolerances such as fuel pressure and MAF readings will equally effect both banks of cylinders, wear on valve gear should be about equal, good injectors should flow almost equally. A small difference in trims isn't unusual, a large difference in trims could point to something being different between banks, since bank lambda readings are the average of the 4 cylinders on the bank something effecting just one cylinder can cause a bank trim difference - a leaky petrol injector on one cylinder might cause it's bank readings to have more negative fuel trim (rich), worn valve gear might cause a cylinder to pump less air resulting in the same more negative fuel trim (rich), a vacuum leak on one side of the engine might cause positive fuel trim on that side (lean).

The expected / normal fuel trim isn't necessarily zero, an ECU's base map (with zero trims) may intentionally be set to provide slightly lean or rich mixture before fuel trims are learned (even if the engine, sensors and fuel system components are all exactly in the middle of tolerances, bang on spec) but the ECU is always expected to provide correct mixture when fuel trims have been learned.

Before changing the exhaust it would start OK from cold but had problems restarting when warm... Now we're talking about trims and poor idle but no mention of if it starts ok when warm now. Anything else to note after changing the exhaust/sensors? If the engine seems good and drives/idles well enough, if trims aren't miles apart I would drive it for a while and see if the idle improves / trims come more in line / restarting when warm is any different.

Dunno, maybe the O2 sensors are a red herring then.
The car starts fine from cold still, and will run warm/ under load with no problems, albeit idling (at about 625) seems a bit rough, but for example, today I had stopped for a few minutes and couldn't get the car to restart. Turned over no probs, just wouldn't fire up. I had to reset adaptive values and then it restarted (unhappily) and after 10 seconds' rev'ing ran stably again - second time I've had to pull that trick now in the space of 2 weeks, but other than a few times it re-starts normally - there's no obvious pattern to the failed starts, they've all been after short hops of c. 5 miles. Obviously I don't want to be resetting the values, as it puts me back to square 1 again. But something else is influencing the trim at the re-start, if it's not the O2 sensors.
Coolant temp is being recorded at 97-100 degrees in the GEMS ECU, so fine, I'm guessing. Next on my list is probably the IACV, except the M4 bolts seem to have about a 5.6mm head (or 7/32 ?) and I don't have anything to safely remove those for now. Injectors were all tested with identical resistance 6 months ago, plugs and HT leads are all new/ the same, MAF got cleaned (with contact cleaner), IAT sensor got cleaned (although it still reads a bit off - like most of my 21 year old thermistors tbh). I have good sparks on all cylinders. I have no o ring on the MAF sensor, not sure if it really makes a difference, but I'll get one anyway. Then only other thing I did in the background was drain the coolant, flushed the sytem and refilled with red coolant (which was what was there before)
Is this re-starting issue not sounding familiar to anyone?

Sounds vaguely familiar but on a cold or hot start if the MAF is diconnected. Cranks for quite a while, eventually starts with a bit of pumping on the throttle and after a few seconds of holding the revs up will idle and run OK (ish). Never seen an O ring on the MAF, can't find one listed on Microcat and the only time I've ever seen one mentioned was on the other side. Could it be an iffy connection on the MAF or the wiring to it?

NTC3354 for the o ring

I'll take another look at the MAF, although it was doing fine before, and after cleaning the idle responded by dropping from 675 to 625, so seemed happier at that point. Previously if I disconnected the MAF the engine failed to start. If it is the MAF though,why would that be isolated to the odd restart?

Seems like something temperature related?

Had to do another Nano job today to get it restarted. It's looking as if the TPS voltage settings/ position settings are diverging ie today I didn't do a full reset but forced a 0.63 setting on the TPS, and 30 steps on the IACV. Still didn't sound wonderful at start-up, but better. Maybe I just need to go on a much longer journey to let it learn, it seems that short hops are confusing the hell out of the ECU. I should be getting my 7/32 socket soon, so can try and remove the IACV

I've seen similar problems with them but it's been a while. I don't have a Nano.

If it idles in closed loop mode with correct mixture (flicking lambda voltage) but idle is slow it points to IACV or TPS,

if it runs in closed loop mode with correct mixture off idle but reverts to open loop (and probably poor mixture) at idle I've found it can help to run the engine at fast enough idle to keep in closed loop mode, gradually reducing idle speed using the throttle as closed loop mode becomes active at lower rpm. And dare I say it, disconnecting the battery before all this seemed to help.

Seen missing Orings on MAFs on a lot of them that run fine so doubt that would be the problem.

Odd restart.... Is that randomly or cold engine versus warm engine?

Thanks LPGC.

The failed restarts are all on a warm engine, but not all the time.

I just ran the car for 20 odd motorway miles and it threw an open fault on Bank 1. Bank 1 always used to work (ie wiring should be good) prior to fitting the two new Lemark sensors/ exhaust. I suppose there are a few ways to go: 1. Remove the Lemark sensor clean it up as well as the plug make sure there's no contamination (shouldn't be, but you never know) 2. Swap the Lemark sensors and see if the open fault switches sides (but presumably that will require a reset and relearn beforehand) and/or 3. fit my one good/ old sensor to Bank 1 and see what happens.

Does anyone know of an off vehicle test for an O2 sensor, in case I end up having to return it?

This morning we're back to closed loop on Bank 1, with some sensible trims, so I guess I'll just let it do its stuff. Patience is a virtue etc.

Is it idling OK now? Closed loop at idle on bank2?

Probably more than just coincidence that the bad starting and poor idling started just after you'd changed lambda probes and I've seen same/similar happen before. When I've seen it happen a few possible explanations have occurred to me. One theory goes that behind the scenes there are two types of open loop mode which I'll call OL1 and OL2. OL1 uses only the ECU base map no fuel trim adjustments applied at all (but still adds cold start enrichment and high load enrichment), OL2 works the same as OL1 but also applies the long term fuel trim. In a no fault situation OL2 is used and gives best results (drive-ability and emissions) - the fuel trims have been learned to provide correct mixture when the engine was warm and in theory when emissions equipment was working properly, the lambda's don't work properly until they're warm enough and are also incapable of measuring the rich mixture necessary during early stages of engine warm up... But still if we want to run as close to correct mixture as possible we should apply the long term fuel trim, or if we want to run a 10% rich mixture we should apply the long term trim + 10%. Or put in another way, if the engine has been running correct mixture which happens to be when LTFT is at +10% and then we cut the wires to a lambda probe while the engine was running we'd expect mixture to remain correct if LTFT remained at +10% or to run lean if the ECU suddenly stopped applying fuel trims because it had detected the fault. Just before you swapped the lambda probes the ECU might have been running in OL1 mode because it had detected a problem with important emissions equipment (e.g. lambda heater circuit) and OL1 mode could give decent results, but before switching to OL1 mode when probes first went bad fuel trims could have been learned incorrectly. By fitting the new probes the ECU sees the heater circuit fault has gone away and switches back to closed loop mode but applying the incorrect fuel trims. There may be an element of when the ECU has first switched back to closed loop mode after running OL1 the long term trim that applies to idle is re-based to be same as an off-idle fuel trim. Off idle fuelling/trims are less fussy than idle fuelling/trims, the volumes being measured (air) and metered (fuel) are far greater and easier to measure/meter within a percentage tolerance range, also the volume of exhaust gas flowing past the lambda sensor is greater so the lambda voltage more quickly responds to exhaust oxygen content (mixture) which is the main reason ECUs respond to lambda quicker with increasing engine load (faster lambda voltage flick). Open Loop and Open Loop (Fault) may correspond with OL1 and OL2 but not necessarily in that order and not under all conditions.

As an aside, not helpful to you because you don't have LPG... My last post I described a way I've got around the running OK closed loop off idle but running poor open loop fault mode at idle problem. Another method I've used is to adjust LPG calibration to make mixture OK for idle on LPG and gradually steered LPG calibration to bring fuel trims back to a range where mixture is also OK for petrol. The idea in both methods is to keep the engine running closed loop with correct mixture instead of reverting to open loop fault mode (which can happen if fuel trims min/max out for long enough) for long enough that fuel trims are steered, don't let lambda voltage hang high or low for long enough to cause the ECU to interpret this as a fault and become latched in fault mode with poor running.

Thanks LPGC. It's idling a bit better, although not brilliant yet. Bank 1 and Bank 2 are back on closed loop, and the sensors are now generating some sensible values ie +/-5% trim. I wonder whether new O2 sensors need a bit of a chance to run in ie go through a few heat cycles. I removed the IACV this morning to clean it up, but it really wasn't too bad.