1995 164TS running very lean - or maybe not? Long post for the experts

[marco.marletta]

LIke many people, I guess: first post is a cry for help!

I have a 1995 164 Twinspark which has just passed its MoT (British inspection). However
the emissions were a little off compared to recent years: lambda 1.02, CO 0.2%,
HC 50 ppm. Last year it managed lambda 1.005, CO 0.05%, HC 30 ppm, and no
engine warning lights.

I'm getting a 1223 error code and a 1224 error code, and a ridiculously low voltage
from the lambda sensor once everything is warmed up - see below.

I've measured the voltage from the lambda sensor. When I switch on the ignition with a
cold engine I get 0.290V. Starting the car up, within about ten seconds I get readings
switching from about 0.250V to 0.850V, which is all very nice. However the readings
sometimes `stick', and I then get the engine management light coming on (1223 code).
When they stick, they can stick at any point from 0.2V up to 0.9V. Blipping the
accelerator, the readings start to switch again, and the engine management light goes
off.

Once the coolant temperature sensor says that the engine is fully warmed up, the voltage
readings drop right down: 0.05V, typically, giving a 1224 error code. However I don't
think it's the fault of the lambda sensor since hard acceleration will bring the reading
back up to something reasonable (0.2V-0.5V). As soon as I go back to constant speed
driving, however, the voltage will go back down to around 0.05V.

If I stop the car when hot, then turn on the ignition again without starting the car,
I see 0.05V rather than the 0.290V which is seen when turning on the ignition with a
cold engine.

Occasionally during warm-up the engine races at tickover (1500rpm instead of 900).
Blipping the throttle cures this. It never races once it's warm. In fact, once it's warm,
the idle can sometimes be slightly lumpy.

The AFM is almost certainly OK since I swapped it with a secondhand one which I've
had around for a long time and which worked perfectly when last I used it two years
ago, and this made no difference.

I've checked the usual suspects for air leaks: the trunking between the throttle body
and the AFM, the breather pipes, vacuum lines, the rubber gromet around the
idle control valve, even the line to the air horn. I haven't checked the evaporative vapour
canister but nothing changed when I pinched the line shut. If I cover the inlet to the AFM
with my hand the engine stalls, as it should.

There is a fair amount of oil mist blowing out from around the oil cap - enough to run
down the side of the engine after 50 miles of driving - but the cap feels tight. Anyway,
the problem started suddenly, and has not gotten progressively worse over time, so I
tend to suspect electronics rather than mechanicals.

If I pull the terminal off the idle control valve, there are 14V between the middle connector
and either of the side connectors. Disconnecting the idle control valve makes no
difference at all to the idle, but reconnecting it again causes the engine to take a
big gulp, almost stall, and then idle a little faster.

My thoughts at the moment are

** idle control valve problem;
** ECU problem;
** problem with some relay.

I've hauled out all the relays around the headlamps and the radiator support panel and
cleaned them with contact cleaner. I've also done the same on the relays up behind the
glove box (remember this is a right hand drive car). None of this has made any difference.

I believe the spark plugs are all sparking. I've had problems with missing sparks in
the past due to bad high tension leads, and the effects on performance were enormous.
The car is actually driving fine: it feels very lively on the road, with no hesitation or
surging.

Any ideas? Is there anything else I should check? Any further tests I can carry out?

Sorry for such a long post, but the car is driving me crazy. It's street-legal for a year,


Marco

[Craig]

G'Day Marco,

I don't know if there are any experts on this engine, it was limited to late 164s and early 155s neither of which we got in Oz.

I think the idle control valve sounds ok, that is I think it's working the way I would expect it (it's different to the earlier TS).

Grasping at straws, air leaks at the injector seals, sticking injector/s, blocked oil/air separator (does it have one?).

[marco.marletta]

Thanks, Craig: I'll try to squirt some engine oil around the injector seals
and see if it gets sucked in. I don't think any of the injectors is stuck because
the spark plugs are all the same colour; if one of the injectors was stuck then
the spark plug for that cylinder would look different. Actually, now that I
think about it, the same argument would say that *all* of the injector seals
would need to be leaking in order to make the spark plugs all the same
colour.

How about an exhaust leak, say at the manifold or before the cat? Would
that give me a really low O2-sensor voltage? The problem started when
I had a new exhaust fitted.

Many thanks,

Marco

[AR67202]

Quote:

Originally Posted by marco.marletta I've measured the voltage from the lambda sensor. When I switch on the ignition with a cold engine I get 0.290V. Starting the car up, within about ten seconds I get readings switching from about 0.250V to 0.850V, which is all very nice. However the readings sometimes `stick', and I then get the engine management light coming on (1223 code). When they stick, they can stick at any point from 0.2V up to 0.9V. Blipping the accelerator, the readings start to switch again, and the engine management light goes off. Marco

[AR67202]

Quote:

Originally Posted by marco.marletta How about an exhaust leak, say at the manifold or before the cat? Would that give me a really low O2-sensor voltage? The problem started when I had a new exhaust fitted. Many thanks, Marco

[Tifosi]

Quote:

Originally Posted by AR67202 - on the car, when idling in closed loop, the voltage should flip from 0.2v to 0.8v or back at least once per second.

[AR67202]

cars do nowadays... in fact, not diling between 0.997 and 1.003 for any post 92 cars is an immediate fail here on the MOT equivalent...

[Tifosi]

.997 and 1.003?

On a narrowband sensor that is roughly the equivalent of 11.5:1 ~ 12:1 A/F ratio, way richer than the target .4-.5V ~ 14.7:1 target of closed loop.

On a wideband sensor, (which you'd never find on anything that old), those numbers would be impossibly lean for idle.



Failing MOT because of not hitting the #'s you quote is because the O2 sensor has failed/ceased to function and thus will make for nasty emissions when tooling down the road as the ECU will never actually be able to get into closed loop. (no sensor to tell it what the current state of the exhaust is)

...

[AR67202]

1.000 equals 14.7 afr, or rather, actual AFR divided by stoich AFR.... you're talking about oxygen sensor output voltages, I suspect.

When using a different fuel, you'd still have to be in the .997-1.003 range, though your actual AFR would be different. Only exception are Diesel engines and Direct injection petrol engines, which would indeed have a broadband lambda fitted.

EDIT: the Lambda figure is thus independant of what sensor is fitted, or what fuel you are using, it's merely the allowed deviation from stoich....

[marco.marletta]

Thanks for all the help, guys.

The heater circuit resistance is 4.6 Ohms cold, 6.5 Ohms hot.

When I put the sensor in a propane torch it works: I can go all
the way up to 1.001V, and I can cycle the voltage up and down
quite quickly by moving the tip in and out of the flame.

So is the problem that the heater circuit of my O2 sensor is fried?
I'd have thought that the sensor would eventually get up to operating
temp. even without a heater.

Or do I have to look somewhere else? I couldn't find any upstream
exhaust leaks, though it's a bit hard to check around the manifold
as it's so &**@! hot. But if it were blowing there, then the car would
surely sound like a bag of spanners.

The cat on the car is not getting very hot. I momentarily touched
the flange where it bolts to the front pipes, just with a finger tip,
and did not burn myself. And that was after a 20 mile drive.

Marco

[AR67202]

if the heater has such a high resistance, it's definately no good. The placement of the probe is too far downstream to ensure it's getting hot just from exhaust gasses

[marco.marletta]

According to this article by an Associate Professor of Automotive Technology
at Southern Illinois University,

Engine Performance

a heater resistance of between 4 and 7 Ohms at room temperature is normal.

This contradicts what Steve said. Who is right?

I tried a different ECU in the car and that made no difference. The sensor
still returned sensible voltages until it was warmed up, then it went a bit
crazy.

At one stage I managed to capture a voltage of 1.895V (yes, that's volts,
not millivolts) with my multimeter. So I disconnected the sensor from
the ECU in case a crazy voltage was somehow coming from the ECU,
and tried revving the car again, and managed very briefly to get 1.564V.
The switching was ridiculously fast, going from over 1.5 volts down to less
than 0.1V in the blink of an eye. The heater circuit was still connected to
the ECU at this stage; I'm starting to wonder if there is a short from the
heater circuit once the probe is warmed up.

Marco

[marco.marletta]

The solution at last!

It was the ECU.

A secondhand ECU is now installed courtesy of an AROC UK member who
sold it to me for next to nothing. The O2-sensor voltage now cycles
beautifully and the performance is completely transformed. In particular,
above 4500 rpm the acceleration is really impressive.

Many thanks to everyone for the helpful discussions.

Marco

[AR67202]

so where did you measure the lambda sensor voltage before? I mean when the sensor was connected to the car...

[marco.marletta]

I spliced a couple of wires into the O2 sensor using insulation displacement
connectors from Maplins (UK equivalent of Radioshack) and measured the
voltage that way, with the sensor still connected to the ECU (or sometimes
not). The crazy readings which I saw literally just for less than
a fifth of a second were caused by my multimeter struggling to cope with the
rapid voltage changes when I put the engine under serious load.

Marco