CIS goes lean when warm

[Won]

Background: 83SC, stock, WUR adjusted perfectly and was running with O2 sensor connected.

Symptom: Once the oil temp goes past ~ 80C, the O2 sensor reading goes berserk. Drops from ~0.76V region down to 0.5V at idle (O2 disconnected now), and sometimes goes all the way down to no reading at all (same when accelerating, deccelerating and idling). I pull over, idle for a bit and the reading goes back up. Basically it randomly jumps around between 0 and max voltage.

Recent changes: Car ran fine, even with O2 connected (the reading swinging around 0.5) and WOT switch worked which brings exhaust CO up to about 3.5% as set with a Gunson. The only thing I've done lately was to put the original muffler back on (cat bypass still present), from a Dynomax 1 in 2 out can (mixture with Gunson was set with one side capped actually). I did not see much difference between mixture from this change. Mixture was richened for colder temperatures.

Been driving in the snow so the bottom was scraping a bit, but as far as I know never drove over hard snow or ice, fresh stuff only.

I've read before that lean when warm condition is related to broken head studs. Unfortunately I don't have the tools with me at the moment to go out and open the valve covers to verify. Is this true?

[john walker's workshop]

CIS leans out when hot. that's the way it is. folks around here unplug the O2 sensor and set the hot CO to 3.5%. and leave it unplugged.

[Won]

Thanks John. I understand that the system leans out with temperature. I think this is a different issue though as the system leans out completely -- 0.0V reading on O2 -- and I can smell the difference in mixture, kind of a sweet smell.

Even when the O2 was connected and working (swinging between 0.4V and 0.6V) it couldn't compensate for the mixture leaning out. Actually I thought it was the lambda circuit that was causing it, due to problem with either the sensor, frequency valve or the brain. Now I see the same ultra lean condition with the O2 unplugged, so it's not that.

I'm assuming water in gasoline will cause lean condition since part of fuel volume will be replaced by water? The tank is nearly empty and we have had an unusually high amount of precipitation for the last few weeks. I'll fill it up and see if the problem goes away. This might explain the random occurance of the problem as well.

The CO is already set to 3.5% hot as per your recommendation, John, but I do like the improvement in MPG from the working lambda system.

[MatthewBrum]

It might be a good idea to check for vacuum leaks, if you already have not done so

[don gilbert]

won, not all of us discontect the 02, i like getting 26 mpg. what are your control pressures?

[psalt]

Hello Won,

My experience is you can not accurately measure AFR with a narrow band sensor. They are really just a switch and they only need to provide a steady, rhythmic signal over the rich/lean trigger point to do their intended purpose. The system goes open loop and ignores the sensor signal after 35% throttle, so your mixture is not going lean under load when it matters. If you want to really know what your AFR under load is, you need to use a wide band sensor. CIS lambda is not a bad system if everything is workings as designed. My advice would be to replace the sensor, if it is old or suspect, and hook up a dwell meter and view the duty cycle cold open loop and hot closed loop and (remotely) the default duty cycle under WOT while driving uphill in 3rd. If the system checks out, trying setting the hot closed loop dwell to 30-40. This will generally give you a good AFR when it matters.

[kodioneill]

at what terminals do you measure the frequency?

[psalt]

There is a test plug behind the plastic cover on the driver side of the engine compartment. The dwell meter coil wire goes to the green/white stripe terminal of the plug, +12volts at the heater fan fuse, ground to the engine. Dwell on the 4 cylinder scale is 90% of duty cycle.

[1982911SCTarga]

not all of us discontect the 02, i like getting 26 mpg.

CIS lambda purist here, too.

Brian

[Won]

A fresh tank of gas seems to have fixed the issue, so it must have been water. I shouldn't let the tank get too empty to avoid this problem in the future.

Paul, once again thanks for your input. I was wondering which terminal I use to measure the duty cycle. I have a digital multimeter with a duty cycle function. It reads around 65% from dead cold which I think is normal, but when lambda goes into closed loop it reads 70% or so and does not change regardless of change in idle mixture, i.e. manually raising or lowering the metering plate arm. When I move the arm the idle will stabilize after a short while so the system must be working. I guess the multimeter is not compatible with the lambda output.

[kodioneill]

salt psalt is offline
Registered User

Join Date: Mar 2006
Posts: 228

There is a test plug behind the plastic cover on the driver side of the engine compartment. The dwell meter coil wire goes to the green/white stripe terminal of the plug, +12volts at the heater fan fuse, ground to the engine. Dwell on the 4 cylinder scale is 90% of duty cycle.

[psalt]

Hello Won,

You need an analog dwell meter, anyone who has old cars usually has a few lying around. If the needle does not fluctuate back and forth when the temperature switch goes to closed loop, your sensor is bad.

[Won]

Thanks guys, I will find a dwell meter if I decide to further look into the frequency valve operation.