More Lambda Problems

blackbox · 22-07-2014 11:15pm #1

My car failed NCT for high lambda 1.09 at high idle
CO = 0.04%, HC= 11 ppm

- so it looks like the exhaust is somewhat lean. Previous test had lambda 1.00, CO=0.1 and HC 32.

I've gone to a lot of trouble to check for an exhaust leak, even as far as blowing smoke up it to see if it would come out anywhere, but have been unable to find one.

I've looked at OBD2 output for the lambda sensors, and I'm not really able to tell much from the data. Both banks are varying up and down and appear to be behaving similarly (though different patterns at different times)

Does anyone know if the problem can be diagnosed short of replacing the lambda sensors and taking it for a retest? I'd be somewhat disappointed to spend the money on these only to have it fail again.

If I took it to a main dealers, what would they do differently?

blackbox · 24-07-2014 9:57pm

It appears that they are not expected to last forever, and the car has nearly 100k miles up, so I've taken the plunge and ordered a pair of lambda probes.

I'll let you know how I get on.

- if you're thinking on purchasing lambda probes, it pays to shop around - huge range of prices for identical branded items.

wriordan · 25-07-2014 2:32am

Are you sure the lambda sensors are at fault, cause if there working correctly then replacing them won't change a thing. Your car could just be running lean and could be caused from something else entirely such as clogged injectors or a leak in the intake system.

blackbox · 25-07-2014 11:34pm

Unfortunately I'm not sure as there is no easy way to test them.

As it's a closed loop system, it should compensate for any issue upstream of the sensors, such as a leak in the intake system or clogged injectors (up to a point). The car is not misfiring and does not appear to be noticeably down on power.

As far as I can find out, the only likely cause apart from the lambda sensors is a leak downstream from the sensors. I can't be 100% certain that there is no leak but I haven't been able to find one.

I suppose it's possible that the control unit is at fault but I'd expect things to be much worse if that was the case.

PADRAIC.M · 26-07-2014 11:29am

Lambda sensor doesn't compensate for any leaks up stream. What car, model, year?

blackbox · 27-07-2014 10:04pm

PADRAIC.M wrote: »

Lambda sensor doesn't compensate for any leaks up stream. What car, model, year?

Car is a 2002 Toyota MR2 Roadster.

Why does it not compensate for an upstream leak? I thought if a little extra air got in after the MAF, the sensor would detect this as a lean mixture and the control system would add extra fuel via the injectors to compensate (i.e. to bring it back to the target Lambda value). Obviously it could not cope with a large leak, but if that was the case wouldn't it be running badly?

...new lambda probes not expected for a few days.

dougie-lampkin · 28-07-2014 4:29pm

blackbox wrote: »

Car is a 2002 Toyota MR2 Roadster.

Why does it not compensate for an upstream leak? I thought if a little extra air got in after the MAF, the sensor would detect this as a lean mixture and the control system would add extra fuel via the injectors to compensate (i.e. to bring it back to the target Lambda value). Obviously it could not cope with a large leak, but if that was the case wouldn't it be running badly?

...new lambda probes not expected for a few days.

You're correct, the system will compensate the A/F ratio based on the lambda, this is why there's a long term (and short term) fuel trim value, which you can also see with your code reader. Without doing this, what would be the point of a lambda sensor at all? At WOT it runs open loop.

A correctly functioning lambda sensor will fluctuate between rich/lean like you saw, that usually means it's working. A dead lambda would be a flat line or have small fluctuations. In closed loop mode the ECU is constantly switching between lean and rich, trying to keep an average of stoich. The lambda reading you took is showing this switching.

High lambda and low CO/HC would definitely point towards an exhaust leak, and probably fairly sizeable. See if you can block the exhaust outlet with something (preferably metal

) and listen for the exhaust escaping elsewhere. It could also be an inlet leak as you've suggested, but I'd be more confident it's the exhaust.

blackbox · 28-07-2014 8:23pm

dougie-lampkin wrote: »

You're correct, the system will compensate the A/F ratio based on the lambda, this is why there's a long term (and short term) fuel trim value, which you can also see with your code reader. Without doing this, what would be the point of a lambda sensor at all? At WOT it runs open loop.

A correctly functioning lambda sensor will fluctuate between rich/lean like you saw, that usually means it's working. A dead lambda would be a flat line or have small fluctuations. In closed loop mode the ECU is constantly switching between lean and rich, trying to keep an average of stoich. The lambda reading you took is showing this switching.

High lambda and low CO/HC would definitely point towards an exhaust leak, and probably fairly sizeable. See if you can block the exhaust outlet with something (preferably metal ) and listen for the exhaust escaping elsewhere. It could also be an inlet leak as you've suggested, but I'd be more confident it's the exhaust.

Thanks Dougie,

I could see the short and long term trim values, but I don't know enough to understand what the numbers are telling me. I've read that lambda sensors can gradually deteriorate before they flatline or generate a fault code.

Exhaust leak was my first suspect too, but I've checked the exhaust as far as I reasonably can. I was nearly going to post pictures of the procedure that involved generating smoke from sticks of incense and pressurizing exhaust with compressed air. If there is a leak it is absolutely tiny. It would have been better if I had denser smoke - I may have to build an improved smoke generator if it fails after lambda sensors are installed.

Sensors have been despatched...

blackbox · 09-08-2014 9:37pm

Sensors eventually arrived (Amazon). They took eight days by post from UK.

You need a special socket to change them - cost EUR 18 in motor factors, or EUR 10 on internet plus EUR 10 P&P.

With the right tool it's just like changing spark plugs. As I bought an exact match rather than a universal type they simply plugged in.

One of the old ones was noticeably whiter than the other, so I suspect that I would have got away with replacing only one.

Then off to Arklow for the NCT re-test.

PASSED!.

I have to say that the guys there that dealt with me were very pleasant and helpful.

More Lambda Problems

Comments