I failed my MOT on emissions last Friday - everything was good except lambda which was 1.2 (should be 0.97 - 1.03 IIRC), which is pretty bad.

As I understand it this often means exhaust leak (air gets sucked into exhaust causing the probe to see excess oxygen). This would make sense as my fuel economy is rubbish (lambda sensors see the extra air and rich up the mixture).

So I took my exhaust apart and found a few scorch marks on the gasket, at the same point where I also found that one of the nuts holding the primary on had become loose:

http://imgur.com/a/8CAGv

Any thoughts, does this look like a big enough leak to explain my emissions problem? New gaskets are on order, I plan to put it back together very carefully with a lot of high temperature RTV sealant in addition to the gaskets - good idea to use sealant or not? It is supposed to be lambda-sensor safe.

stopping any leak on an exhaust will help your readings, dont know if thats all of your problem though. Not a good idea to use rtv on joints or gaskets . May sound daft but use copper slip grease (will seal joints as dries out with heat and will not damage cat eithr)

stopping any leak on an exhaust will help your readings, dont know if thats all of your problem though. Not a good idea to use rtv on joints or gaskets . May sound daft but use copper slip grease (will seal joints as dries out with heat and will not damage cat eithr)

If that fails try disconnecting the lambda - it should send the ECU to default mode and make it run lean, not recommeneded for long, but fine for the 5-10 mins of MoT running time.

Lambda >1 = lean, and does indeed occur if you have an exhaust leak. It'll also trash your fuel economy if it's pre-lambda sensor because the ECU will fuel to drag it back from lean. And yup, those leaks are plenty big enough to cause the issue.

Thanks for the comments all.

I was pretty happy to find these leaks as I've been chasing fuel economy for ages looking for exhaust leaks but finding nothing (never quite got as far as removing the primaries from the block). Hopefully this will get me through the MOT and above 20 mpg...

Update: it didn't. No improvement at all, lambda still reading around 1.2. Not really sure where to go from here...

you might have to get it on a rolling road

First assumption then is that you have tuning issues. Second option is you may have valve issues.

A little bit of progress... got it through MOT fine by unplugging the MAF. This improves the idle and gets lambda down from 1.2+ into the pass region.

Without a MAF, the BMW ECU will estimate airflow off TPS, IAT and RPMs instead, and (I think) doesn't go closed loop.

This means that fuel pressure etc is fine, valves and all other mechanical bits are OK, etc - I presumably just have a problem affecting closed loop mode. This could be:

- bad MAF (or wiring)
- bad lambdas (or wiring)
- intake leak
- exhaust leak
- bad ECU

Anything I've missed? Any experts on the BMW engine management here?

Old/poisoned lambda can sit in one direction but normally the ECU should spot this, depends how far out of whack it is. MAFs are notoriously feeble. Get a fast multi-meter on the output of your lambda, after it's warmed up give the throttle some blips - if it flies between the two extremes of 0 and 1v you'll /probably/ find that's fine. AFM can be more difficult to read depending on the type. Do you have an OBDII ECU or is it pre-OBD, diag system still in tact? Unplugging the MAF would normally throw an error and fail your MOT.

The MAF and lambdas are both new which makes me think it may be an intake air leak.

I will test them both anyway (thanks for the tip on testing the lambdas, I've also tried using a blowtorch in the past which seems quite effective but more hassle). I do have ODBII so should be able to get some live readouts of the MAF value which is a start. I guess multimeter is the other way to test it but wouldn't know what values to look for.