Car failed today at Nottingham
Main reason was the fast idle emissions. Few other minor things but they could've been sorted there on the day if it wasn't for the emissions.
Emissions are fine at normal idle but the CO increases to around 2.5% as soon as the revs are raised to 2.5-3krpm.
The engine has a remapped ECU and M50 inlet manifold and passed the emissions in the donor when MOT'd but I'm pretty sure they didn't do it at 'fast idle'.
The obvious thing to do is replace the ECU with a std one (which I have) but that has the BMW EWS security and I'd have to rejig all my wiring
Bit of a bummer, but I'll work it out and I'll be back!!

Not too bad.
Can the ECU not be mapped to suit requirements?

quote:

---

Originally posted by loggyboy
Not too bad.
Can the ECU not be mapped to suit requirements?

---

Just been doing a bit of testing of the voltage from the lambdas (black signal wire).
They seems to start around 5v, works their way down to around 0.5v as it gets fully warm but then they both start fluctuating before ending up back at 5v. I've tried 4 different sensors with similar results, surely they're not all dodgy??
So, are the sensors reading correctly and the mixture is fluctuating due to something else? Or, are they giving a false signal leading to a rich mixture? Bit of a chicken and egg one this

You should have max 1v (assuming a narrowband zirconia type sensor), getting 5v at the signal wire indicates the lambda sensor is leaking current from a faulty heater circuit, or there is an issue with the sensor wiring.
If the sensor is disconnected while the engine is hot and running it will still output a 0v lean 1v rich signal across ground-signal pins.

If you have a narrowband titania type sensor fitted, it is 0v lean 5v rich using a 5v reference supplied by the ecu, the sensor normally has a red or yellow signal wire to indicate the different sensor type.

Have you got the obd socket (or bmw round 20pin diag socket )wired for use in your setup?
That would be the next port of call for both codes and live data, to check the recieved signals are correct and any fuel related codes that may be relevant.

What is your donor car and engine code, this may help in narrowing down the fault.

Dave

Thanks for the input Dave.
Sounds like I have the latter 5v type. 4 wire sensors with black, yellow, red and white wires IIRC. I've been measuring the black one as I'd read this was the signal wire from the sensor, but might be wrong.
When I first started the engine it had an issue with the ICV. It would rev really high on a cold start (which I hate as it can cause engine wear) and the revs were really slow to drop when you blipped the throttle. I cleaned out the ICV and the internals were moving freely but the problem remained. So I tried disconnecting the plug and the problem went away completely and the engine seemed to run sweet so I left it like that. Now, is having that disconnected causing it to go into 'limp mode' where it just dumps a set amount of fuel in at anything over tickover? BMWs of this generation didn't have a CEL (check engine light) so I have no indication. I could reconnect it and see if my lambda voltages stabilise. If it does then maybe I should block off the pipes to the ICV but leave the electrics connected? Or'try a new one? From what I understand the ICV is mainly there to compensate for the extra load on the engine from the likes of A/C and PAS, neither of which my car has obviously. It does cold starts too but tbh it's fine from cold and sits happily at 800rpm.
I've ordered an 20pin BMW adapter to see if my OBD reader will pick up any fault codes, should be here early next week.

Thanks again
Nick

Good to see the IVA tests appearing as the summer looms. Hope you manage to sort it and pass !

BTW- Donor was a '96 BMW 328i, M52B28 engine.


[Edited on 11/3/17 by Nickp]

Have you checked to make sure that the lamda sensors are the right way to around. Each one looks at a bank of three cylinders and the wiring is not interchangeable. The engine will run either way around, but it's rough if they are the wrong way

quote:

---

Originally posted by SteveWallace
Have you checked to make sure that the lamda sensors are the right way to around. Each one looks at a bank of three cylinders and the wiring is not interchangeable. The engine will run either way around, but it's rough if they are the wrong way

---

I also had a lot of emissions issues when I had the 328 and spent a lot of time trying to sort this. I think part of it was having to extend the loom so the sensors would reach - I did a poor job of this initially which really hurt emissions. Also the engine had nikasil issues which caused to burn oil with obvious results on emissions - hopefully that's not your issue though! One thing I realised when searching the internet - the US 328 engine used a different, incompatible type of lambda sensor, so be sure you have the right type.

You can test the lambdas off the car with a blowtorch, that might be worth a go.

On the car at idle, They should normally fluctuate from 0 - (I think) 1V at around 1-2 Hz. The engine loom is labelled 1 and 2, plug 1 is the front (nose) of the engine.

I think I have a spare 328 ecu, Ews etc if you want to try going that route. Also spare set of lambdas.

quote:

---

Originally posted by rodgling
I also had a lot of emissions issues when I had the 328 and spent a lot of time trying to sort this. I think part of it was having to extend the loom so the sensors would reach - I did a poor job of this initially which really hurt emissions. Also the engine had nikasil issues which caused to burn oil with obvious results on emissions - hopefully that's not your issue though! One thing I realised when searching the internet - the US 328 engine used a different, incompatible type of lambda sensor, so be sure you have the right type.

You can test the lambdas off the car with a blowtorch, that might be worth a go.

On the car at idle, They should normally fluctuate from 0 - (I think) 1V at around 1-2 Hz. The engine loom is labelled 1 and 2, plug 1 is the front (nose) of the engine.

I think I have a spare 328 ecu, Ews etc if you want to try going that route. Also spare set of lambdas.

---

What is the ECU seeing as the coolant temp? Is the coolant sensor playing tricks?

Could also be a bad MAF, you can test this by seeing if emissions improve when it's unplugged. Emissions etc should be ok but not as good as closed loop with it unplugged (not sure if it would be good enough to pass)

quote:

---

Originally posted by owelly
What is the ECU seeing as the coolant temp? Is the coolant sensor playing tricks?

---

quote:

---

Originally posted by owelly
What is the ECU seeing as the coolant temp? Is the coolant sensor playing tricks?

---

quote:

---

Originally posted by rodgling
Could also be a bad MAF, you can test this by seeing if emissions improve when it's unplugged. Emissions etc should be ok but not as good as closed loop with it unplugged (not sure if it would be good enough to pass)

---

quote:

---

Originally posted by Nickp

quote:

---

Originally posted by rodgling
Could also be a bad MAF, you can test this by seeing if emissions improve when it's unplugged. Emissions etc should be ok but not as good as closed loop with it unplugged (not sure if it would be good enough to pass)

---

Ok will do, gonna try the ICV first but tbh not sure if this has any effect on fuelling.

---

quote:

---

Originally posted by rodgling

quote:

---

Originally posted by Nickp

quote:

---

Originally posted by rodgling
Could also be a bad MAF, you can test this by seeing if emissions improve when it's unplugged. Emissions etc should be ok but not as good as closed loop with it unplugged (not sure if it would be good enough to pass)

---

Ok will do, gonna try the ICV first but tbh not sure if this has any effect on fuelling.

---

It shouldn't do (unless the hoses are leaking), it's just an extra throttle body really. I probably have a spare if you want to replace it though.

---

Sorry if you already know all this: lambdas are *supposed* to fluctuate at idle. If they report oxygen in the exhaust stream, the ECU will enrich the mix slightly, and if they don't, it will lean it out. So it constantly cycles just slightly rich/lean, around twice a second.

If you unplug the MAF, it goes open loop (i.e. doesn't use the MAF or the lambdas, just estimates fuelling based on RPM & throttle position) - this is less accurate so it runs a little bit rich to be safe. So, you see a fixed value (i.e. slightly rich) when you do this.

The problem seems to be that your lambdas start out cycling (which is what you want) but then end up a bit rich? It's odd that you get 1.5-2V - as someone else said, normal output should be (I think) 0 - 1V. Is this the same for both sensors? If it is it would be a bit of a coincidence for both sensors to fail the same way.

Excess CO (as opposed to CO2) on your emissions readings means incomplete oxidisation of the fuel because there wasn't enough oxygen, i.e. it's too rich. Which is also what the lambda sensors are telling you - so that doesn't really suggest that the lambda sensors are bad. Also you've seen consistent results from four sensors... so it's not the lambda sensors.

Could be the MAF, if it's reporting more air than is actually present then the ECU will obviously over-fuel to match the amount of air it thinks is present. In my experience buying third party MAFs didn't go well... I'd go OEM on that one if you need to replace it.

Do you have the exact numbers from the emissions test?

I think I'd be tempted to point the finger at the remapped ECU at this point but that's just a guess.

When I say they fluctuate, I don't mean slightly, I mean from 0.3v all the way back to 5v. All over the place tbh. All sensors react in a similar manner. As soon as I disconnected the MAF they stabilised at 1.5-2v. Seemed to point to the MAF to me. I should probably wait to try and read the codes but want to get it retested ASAP. I don't mind paying for an OEM one but only if it's definitely the problem. Might have to try a cheapo one for now just to see if it changes things.

At the IVA it had 0.2% at tickover and 2.5-3% at fast idle (2.5-3krpm)

quote:

---

Could be the MAF, if it's reporting more air than is actually present then the ECU will obviously over-fuel to match the amount of air it thinks is present.

---

I think you have titania sensors on that engine, apparently output voltage for these can be 0.5 - 4V or wider range, so fluctuating in the 0.3 - 5V sounds like normal behaviour. Your multimeter might be a bit too slow to keep up so it may not be obvious if it's cycling in a regular pattern.

I don't think it does indicate the MAF particularly. With the MAF removed, it will be slightly rich, so you would expect the lambdas to have a stable rich reading (although 1.5 - 2V is strangely in the middle which seems odd). It could still be the MAF but this doesn't help prove it IMO.

[Edited on 11/3/17 by rodgling]

quote:

---

Originally posted by rodgling
I think you have titania sensors on that engine, apparently output voltage for these can be 0.5 - 4V or wider range, so fluctuating in the 0.3 - 5V sounds like normal behaviour. Your multimeter might be a bit too slow to keep up so it may not be obvious if it's cycling in a regular pattern.

I don't think it does indicate the MAF particularly. With the MAF removed, it will be slightly rich, so you would expect the lambdas to have a stable rich reading (although 1.5 - 2V is strangely in the middle which seems odd). It could still be the MAF but this doesn't help prove it IMO.

[Edited on 11/3/17 by rodgling]

---

quote:

---

Originally posted by gremlin1234

quote:

---

Could be the MAF, if it's reporting more air than is actually present then the ECU will obviously over-fuel to match the amount of air it thinks is present.

---

could also be too high fuel pressure

---

quote:

---

Originally posted by Nickp
Also, when it's ticking over and i disconnect the MAF it improves the idle. When I plug it back in it stalls.

[Edited on 12/3/17 by Nickp]

---

quote:

---

Originally posted by rodgling

quote:

---

Originally posted by Nickp
Also, when it's ticking over and i disconnect the MAF it improves the idle. When I plug it back in it stalls.

[Edited on 12/3/17 by Nickp]

---

It's probably your MAF then. Typically it will stutter but then carry on when you plug it back in. Also if it idles significantly better without the MAF that's another hint.

---

MAF fitted and initially I thought it was sorted as both Lambdas were reading around 0.5v. But just tested again today with it fully warmed up and they now seem to be alternating between 0.5v and 5v, each taking it in turns to be 5v (full rich). Then started it again and they're both sat around 2-2.5v!! All very random. I got an adapter lead for an OBD2 reader but it won't connect. A mate's sending his working one over for me to try, so fingers crossed that'll give me a clue. I suppose there's a chance my cheapo MAF could be dodgy

Nick,
Here's a few things that may help you in tracing doen your fault.

What was the lambda value during the fast idle test?
If this value is 0.980-1.030 the mixture is correct and your catalyst may not be capable of oxidising the CO being produced during closed loop running (this could be due to carbon fouling from extended idling or aftermarket cats that are incapable of oxidising enough CO under test conditions). If it is in the 0.900-0.970 range the engine is running too rich for the catalyst to operate correctly and further investigation is needed.

If the engine is rich, it is worth checking is the oil separator valve and pipework. If you remove the oil cap with the engine idling and the idle speed drops significantly/stalls or there is noticable pressure buildup , the pressure regulating part of the valve has failed. There should be 10-15 mbar vacuum at idle in the crankcase with no osv faults present, with a failed osv anything from manifold vacuum to 0.5+ bar pressure may be present depending on how the valve has failed.
The osv failing will mess up all the long term fuel trims and may be the root cause of your problem, as well as causing excessive oil usage.
It is likely that the pipework may require replacement if there is a problem, as the pipes get very brittle and tend to break upon removal, the dipstick tube can also block up at the oil separator connector but is cleanable.
If this is part of the problem, you'll have codes 227 & 228 , but they can also occur from faulty maf readings.
I don't think you'll be able to read obd codes unless you have a us market ecu fitted (or possibly us firmware flashed), for eu ecu you'll need an inpa clone to work with the 20pin adaptor to get codes, or someone friendly to read the codes/live data for you.

With your engine code you definately have titania sensors and your ecu should be a motronic ms41.
Your described behavior for the lambda sensor feedback sounds as expected for titania sensors (as it is a narrowband sensor the only accurate value is lambda=1 at 2.5v, either side is rich or lean with no accurate indication of actual ratios hence the system cycles rich/lean to maintain lambda=1 at the exhaust).
Disconnecting the maf is putting the engine into "limp mode", if there is a fuel trim issue it will run better but is not a guarantee of a fault with the maf.
Going by tech data, at ecu pin 8/maf pin 2 yellow wire you should have 0v with ignition on engine off, 0.6v at idle and 1.5v at 3000rpm, this will be correct with the engine at normal operating temp. maf pin 1 is engine earth, pin 3 is 12v supply and pin 4 is ecu supplied earth.

You need to discount any codes that have been caused by disconnecting maf etc, clear codes and run up to temp before rechecking codes.

Have a look here for further info:

https://www.picoauto.com/library/automotive-guided-tests/titania-lambda-sensor

Some further reading on interpreting gas analyser lambda values:

http://www.austincc.edu/wkibbe/lambda.htm


Dave

Blimey, good answer!!

Yeah it was rich at anything over tickover, around 2.5% which then affected the lambda reading.

So I should be looking for 2.5v then? Seems to be a lot of conflicting info, most says 0.5v is the optimum.

BTW I have an M50 manifold fitted and the associated plumbing under it to the OSV, if this makes any difference?
Also, the engine revs up too high on start up but I thought that could be the ICV, I've got one of those coming too.

Update- if anything the idle rises slightly when I pull the oil filler cap.

MAF voltage appears OK too, 0.6v at tickover and rises as expected with revs.

Does the ECU reset and clear fault codes when you disconnect the battery?

[Edited on 15/3/17 by Nickp]

quote:

---

Originally posted by Nickp
Blimey, good answer!!

Yeah it was rich at anything over tickover, around 2.5% which then affected the lambda reading.

---

quote:

---

So I should be looking for 2.5v then? Seems to be a lot of conflicting info, most says 0.5v is the optimum.

---

quote:

---

BTW I have an M50 manifold fitted and the associated plumbing under it to the OSV, if this makes any difference?
Also, the engine revs up too high on start up but I thought that could be the ICV, I've got one of those coming too.

---

quote:

---

Update- if anything the idle rises slightly when I pull the oil filler cap.

---

1st test was CO 2.34 % 0.931 lambda
2nd was 3.22% 0.905 lambda

It also says that it failed the normal idle test too at 1.33% but I'm sure this was OK before the fast idle was done and was 0.2%

I'll do some more testing in the morning and hopefully my new ICV will arrive tomorrow to fit. A friend is fetching me another OBD reader tomorrow night to try too.

With the ICV connected the engine revs too high on cold start and also the revs are very slow to fall when blipped. Disconnect the ICV and it's much crisper, if that gives you any clues?

Yes M52 TB and ICV.

[Edited on 15/3/17 by Nickp]

Hi Nick,

You're welcome to borrow my unmapped DME if it helps you at all? It's had the ABS and EWS deleted. I'm a little way from needing it...

Cheers,

Chris

quote:

---

Originally posted by CTLeeds
Hi Nick,

You're welcome to borrow my unmapped DME if it helps you at all? It's had the ABS and EWS deleted. I'm a little way from needing it...

Cheers,

Chris

---

Those lambda values are a good clue.
That is rich enough to stop the catalysts from working fairly quickly, hence when you returned to idle the CO was a lot higher.
The catalyst should start to work properly once the fueling is corrected.
It looks like the fuel trims were getting richer while the test was being carried out which leaves a few options.

-The engine management is interpreting the lambda feedback as lean and so was richening the mixture to compensate - this could be caused by air getting into the exhaust system somewhere between head and lambda sensor positions. It would take something like a pinhole or leaking gasket to cause this kind of problem, but should be noticeable as a blow.

-There is a fault with 1 or both lambda sensors, causing the same richening as above, comparison between sensor voltages and gas analyser measured lambda may help here. I don't think this is likely but would not totally discount it.

-It could also be caused by the maf overreading and fueling to match, but this should be initially rich and lean off as the fast idle was held and the lambda feedback cut back the fuel trims, which i think is unlikely with given data.

-There is a possibilty that the remapper has turned off the lambda feedback or changed the conditions where open loop is enabled (so the lambda sensor feedback has no effect on fuelling) or adjusted the number of functioning lambda sensors(usually with turbo conversions), this may be tricky to detect directly but diagnostic live data may help here. If the lambda feedback has been disabled this may also mean fault codes have also been removed or disabled entirely. I have encountered before on other ecu's and can make things tricky to diagnose.

I would definately try a standard dme, it's worth getting some time with a friendly garage and their emissions tester, and bring working diagnostic equipment so if something is not right you can also see why.

Dave

There doesn't seem to be any blow on the manifolds. New pair of gaskets and nice and tight.
I've been interchanging my identical lambdas and get similar results with every combination. I've also confirmed that they're connected the correct way round as they work in 2 banks apparently.
I'm borrowing a std map (ews deleted) ecu but that won't be until next week. So I'll keep trying for now and see what codes if any I can read off it later today.

Thanks again
Nick

Dave,

Just done the rubber glove finger over the dipstick test and I definitely seem to have slight negative pressure as it sucks it in slightly.

Fitted my new ICV and straight away it didn't rev as high from cold and the revs seem to settle quicker now too, so hopefully that's one issue sorted.

I monitored the voltages as it warmed up and they were sat at a steady 0.5v with very little fluctuation. As soon as it was fully warm the voltages start to vary from 0.5-4.5v quite randomly. But saying that, one seems to flick straight from high to low while the other seems to climb and drop more slowly, not sure if this is relevant TBH. The picture above shows it cycling up and down continuosly, does this give an average voltage then? Ideally 2.5v? As my DVM isn't going to show this is it?

I really need to read any fault codes and get it on a CO meter now to see if I've actually still got an issue as it seems to be running much better now TBH.

[Edited on 16/3/17 by Nickp]

Just been looking at the live data. No codes coming up, which is a shame!! Lambda volts still all over the place but each set of 3 very different. One spiking up and down and the other much smoother. It has a rhythmical misfire to it now. Wonder if something is breaking down on ignition side? All plugs are black but none look any worse tbh.

We may have a breakthrough, and it may have been all my doing

After seeing it had no fault codes, all the plugs were black but the lambdas were responding very differently I thought about the only difference between the 2 manifolds - a 90deg adapter I'd fitted to one to avoid the bonnet.So I've now taken this out and the lambda voltages seem to follow a similar pattern now I'm not out of the woods yet as I still need to check the CO (Oh and trim the bonnet to clear the feckin Lambda!!) but I think I'm a big step nearer.

[Edited on 16/3/17 by Nickp]

quote:

---

Originally posted by Nickp
We may have a breakthrough, and it may have been all my doing

After seeing it had no fault codes, all the plugs were black but the lambdas were responding very differently I thought about the only difference between the 2 manifolds - a 90deg adapter I'd fitted to one to avoid the bonnet.So I've now taken this out and the lambda voltages seem to follow a similar pattern now I'm not out of the woods yet as I still need to check the CO (Oh and trim the bonnet to clear the feckin Lambda!!) but I think I'm a big step nearer.

[Edited on 16/3/17 by Nickp]

---

[Edited on 16/3/17 by Nickp]

quote:

---

but a bit miffed it was something so daft that could've been avoided.

---

That looks like it'll yield an improvement on the gas tester.
If the ethos/solus you've borrowed can display the long term fuel trim data, with the engine run up to temperature they should be in the +/- 5% is spot on , up to 20% should still give a pass on CO and lambda with no further work.
It is odd that no codes were logged, but the ms41 is an oldish ecu and may only record current persistent fault codes, so anything that is now fixed will not show up as a stored fault.

Dave

quote:

---

Originally posted by gremlin1234

quote:

---

but a bit miffed it was something so daft that could've been avoided.

---

yea and ALL other things you got RIGHT first time, building a road legal car ain't easy!

[Edited on 16/3/17 by gremlin1234]

---

quote:

---

Originally posted by obfripper
That looks like it'll yield an improvement on the gas tester.
If the ethos/solus you've borrowed can display the long term fuel trim data, with the engine run up to temperature they should be in the +/- 5% is spot on , up to 20% should still give a pass on CO and lambda with no further work.
It is odd that no codes were logged, but the ms41 is an oldish ecu and may only record current persistent fault codes, so anything that is now fixed will not show up as a stored fault.

Dave

---

Just popped up to a local MOT place and it's now scrapeing through the fast idle test with <0.2%CO and <200ppm HC.
Hopefully by the time it gets to Nottingham on Tuesday it'll be better still and pass OK. Fingers crossed!!