car running lean sometimes

rage

rage

Lifer
having a strange problem with my engine.

rebuilt the engine and rebuilt the loom so there could be a little problem anywhere. just trying to narrow down the field a bit.

the engine runs fine and i set it up using a wideband and safcII
all the afr's were spot on and the car ran like a dream.

nxt day i start the car and she starts to stutter a bit on idle. a glance at my wideband tells me she is running afr 18!
give a bit of throttle and the afr shoots to rich. drove a little bit and the afr on steady driving was around 15, when giving it a bit of gas the afr shoots to 18 and then drops down to 12ish. but there is about 2 seconds of running very lean.

next day, start the car and she runs spot on again :confused:

why o why could this be? labda on it's way out? throttle sensor? ecu making a mess of things?
 
rage

rage

Lifer
will try running without safc and will check throttle sensor.
running a new walb ro pump. been in for just a few months now
 
rage

rage

Lifer
still can't figure this out.

when the engine is warmed up she runs fine, afr's are all good.
but on a cold start she still runs lean sometimes. really really lean. when i hit a spot of gas the afr's return to normal but starts to go lean again after a few seconds.

does the car run a diffrent map when cold? like a choke?
could ther be a problem in that circuti somewhere?

other odd thing is, when she runs lean and i get above 3k rmp it's all gone and runs fine. drop below 3k rpm and lean again
 
R

Rev

Member +
I read for various reasons if the lambda sensor continually reads high (rich), it will cause the engine computer to lean out the fuel mixture in an attempt to compensate for the rich reading. Lamda sensor functions for closed loop generally light throttle ecu changes to open loop at moderate to heavy loads say 3000 rpm and above.
 
Last edited:
R

Rev

Member +
rage said:
how can i check the labda sensor?
Click to expand...


I have seen a video on utube called -
How to check to see if your aftermarket O2 sensor is bad.flv

Also Wiki answers has how to test an oxygen sensor with a voltmeter.

> But the sensor may not be the problem - eg
Sometimes an apparent lambda sensor problem is not really a faulty sensor. An air leak in the intake or exhaust manifold or even a fouled spark plug, for example, will cause the lamda sensor to give a false lean indication. The sensor reacts only to the presence or absence of oxygen in the exhaust. It has no way of knowing where the extra oxygen came from. So keep that in mind when diagnosing oxygen sensor problems.

The lambda sensor is also grounded through the exhaust manifold. If rust and corrosion of the manifold gaskets and bolts is creating resistance, it may affect the sensor's output. To rule out a bad ground, use a digital volt meter to check for a voltage drop between the sensor shell and the engine block. More than 0.1v can cause a problem.

The safc2 should also be ruled out as said above especially if Lambda sensor is 2 wire heated type.
 
rage

rage

Lifer
well, turns out that it's not the labda sensor.

my wideband has a simulated labda signal so i hooked that directly to the ecu to see if that would solve the problem.

but no luck :( after 1 min of running ok she suddenly turned lean again.
i'm at a loss... what could cause this besides labda.
getting no fault codes, car runs well after 3k rpm, epic lean below 3k rpm.
after a short drive it's like the ecu finally switches to another map and she runs fine.
next day, same story, lean till 3k rpm, problem vanishes after a few miles.
 
rage

rage

Lifer
allright.. now things get interesting.

disconnected the labda signal from the ecu completely. so the ecu is not recieving input
started car, set the afr to ideal with my wideband.

car runs fine
i connect the labda signal.
car goes to fekking lean
pull the wire off
wideband drops back to ideal values.

so there is something up with the labda signal somehow.
but i tested it with labda signal and simulated labda signal from the wideband sensor.
so the fault should not be in the labda signal itself.

now the question, is it possible to drive the car without labda input and checking, setting the afr's myself through the SAFC
or will the whole thing mes itself up because the ecu doesn't get a labda signal
 
rage

rage

Lifer
yes, so running a double sensor. stock labda sensor to control ecu, wideband is completely seperate from the engine loom.

test drove a bit without labda signal. and the car runs fine!
able to input my exact wideband values

running steady around 14/14.5
flat out, 11.5/12
only having a spot of trouble with half throlle pull up. tends to overfuel a bit.
but beyond that, the car runs great without the labda signal!
not even a engine light on the dash...

so posted a wanted thread: ep82 stock ecu to check if the fault is indeed within the ecu itself. till then i'm running her in emergency loop. and with all the values on nominal i can't see any harm in that? atleast not as bad as running way to lean when cold
 
kieran.s gt

kieran.s gt

Member +
it will run fine without lamba but part throttle will be bit rich
try new sensor get a 1 wire universal sensor on ebay for cheap
 
rage

rage

Lifer
it's not the sensor, tested it by giving the ecu the simulated labda signal from my wideband sensor. and the ecu made the same fault, making it run lean.

so my thoughts are that it's a problem within the ecu.

and there is a sollution to the part throttle rich problem
floor it ^^
 
Last edited:
R

Rev

Member +
Have you set the part throttle map in the safc2 properly ?
Earth to ecu ok? ( used by Lambda )
Is the Lambda sensor dirty , carboned from times of overfueling ?

Not sure about removing lambda but remember the toyota ecu's have different states some rely on other sensors in combination with the Lambda - As per texxs post which also explains what sensors are used to establish load.
http://www.toyotagtturbo.com/forums/showthread.php?63903-intake-temp&highlight=intake+temp

Some info I found says you can have - Rich condition on Lambda that will cause ecu to LEAN out fuel or there can just be LEAN condition from lower voltage on Lambda -

> If the average voltage from the lambda sensor is running high (more than 0.50V), it indicates a rich condition, possibly due to a bad MAP, MAF or Air Flow sensor or leaky injector. If the average voltage reading is running low ( less than 0.45V ), the mixture is running lean possibly due to a vacuum leak or because the sensor itself is bad. ( 0.45v is "stoichiometric" fuel mixture of 14.7 )

> If the lambda sensor continually reads high (rich), it will cause the engine computer to lean out the fuel mixture in an attempt to compensate for the rich reading. This can cause lean misfire, hesitation, stumbling, poor idle and high hydrocarbon emissions (from misfiring).

> If the lambda sensor continually reads low (lean), it will cause the engine computer to richen the fuel mixture. Injector pulse width will increase causing fuel consumption and carbon monoxide emissions to go up. Constant rich fuel mixture can also cause the catalytic converter to overheat and it may be damaged.
 
Last edited:
rage

rage

Lifer
have checked for most of the above. my initial thought was also that there was a problem with the lambda readout. But it's odd that the wideband sensor reads just fine and when i give the ecu the simulated lambda signal direct from my wideband it leans out the mix.
and it does the same when i give the ecu the signal from the normal lambda sensor.

problem is, there is no rich condition as the wideband reads fine, but the car leans out anyway.
checked and measured most other sensors and they are all fine. cable connections are all fine aswell.
if i could see that the car was running rich and that that is the reason that the car leans out a bit then i could understand it. but one minute the car runs fine and the next it's leaning out a lot! afr's of 19 and up are quite dangerous for the engine.
and the thing i still can;t figure out, why does the problem vanish when the car is propperly warm (5 miles or so)
 
rage

rage

Lifer
read the text that texx posted. interesting read :) i'm supprised at myself that i'm actually understanding all that stuff now :D

all those sensors are still intact and working fine. so the engine should still recieve enough information to calculate injector timing etc.

as far as i can find out the lambda is only really used till you hit 3k rpm, that is when the system goes to open loop (or was it closed loop, correct me if i'm wrong here plz)
so doesn't do much with the ox signal.
 
You must log in or register to reply here.
Top