Can't get Lambda leaner, 1983 380SL - PeachParts Mercedes-Benz Forum

erubin · #1 02-12-2006, 06:06 PM

1983 380SL.

Background: Car failed smog. I tracked down that the O2 sensor was no good. Replaced it and now the car can go into closed loop. Lambda is very low so...

I am adjusting the lambda by turning the 3mm adjuster CCW = leaner. lambda increases and stays steady up to 36%. If i continue leaning it out the car starts hesitating and finally stalls. The leanest i can get it at idle is 36%. Does this suggest an air leak?

I sprayed carb cleaner at all the vacuum connections i could see as well as the intake manifold rubber rings that connect the upper and lower manifold but could not detect a leak. Would you still be convinced of an air leak or could something else cause this. Car runs well (smooth idle, plenty of power) but i don't want to bring it back to smog untill i understand why I can't go leaner than 36%. At idle the duty cycle is very stable.

O2 sensor is oscillating as expected.

BTW, when I measure Lambda at diagnostic terminal #3 with key on engine off the duty cycle shows up as 0% (0 VDC was also measured). I'll just assume that this year and model does not put out a fixed 70% or 85% with KOEO.

I've spent many hours reading over all the great info on this site but still am stumped.

Duke2.6 · #2 02-12-2006, 06:48 PM

If you read any of my posts you should understand (at least somewhat) how to use the gas concentration data on your test report to analyze your emission test results.

It would help if you post the numbers and your analysis.

Why are you leaning it? You want the idle duty cycle to average near 50 percent. (It should fluctuate, and the "average" is about halfway between the high and low meter reading fluctuations.) I like to set the idle average mixture slightly rich (meter reading range fluctuation between 50 and 55 percent, for example) since cold starting is better, and this is okay if the 2000 rev duty cycle average is within ten percent. Fifty percent means that the base mechanical mixture is right at stoichiometric, so the control system spends as much time richening as leaning to maintain a stoichiometric average.

If you get the duty cycle too far from 50 percent the system runs out of control authority to maintain an average stoichiometric ratio, and there won't be enough free O2 in the exhaust for proper catalyst reactions.

Too lean a base mixture can fail emissions just as surely as too rich a mixture.

Duke

erubin · #3 02-12-2006, 09:43 PM

Let me clarify:

The highest lambda value I can adjust to at idle is 36% duty cycle (positive lead at term 3 and neg lead at term 2). I can lower the duty cycle (screw CW) but if i try to raise it (screw CCW) to get the duty cycle near 50% the car hesitates and stalls when i go above 36%. I will stop calling this enriching or leaning the mixture since that is the wrong way to describe it.

Duke, you said to target arround 50% but I can't get there. Can you tell me the likely cause(s) for this scenario?

I checked for air leaks and found none. Should I keep looking for air leaks or are there other likely suspects if the engine can't be raised to 50% lambda?

I assume my smog values are not very useful since it was smogged before i replaced the defective O2 sensor.

FWIW, here are the smog test results at 15 mph (25mph values are not much different) with a defective O2 sensor.
CO2 5.56%
O2 3.89%
HC 538ppm (gross polluter)
CO% 11.69% (gross polluter)
NO 92ppm (pass easily)

stevebfl · #4 02-12-2006, 10:03 PM

Not sure exactly what you are looking at but the worst problem dealing with duty cycle is point of view. When looking at this properly a lean compensation for a rich mixture has the duty cycle oscillating lower than 50% say 35 -40%.

Another absolute is that the mixture in closed loop is always the same whether the correction necessary to achieve it is 30-35% or 70-75%. SO! If you are in closed loop you really are never changing mixture as the control system takes what you are giving and stays the same. That is what is meant by closed loop.

I'm afraid we my have some comments here that are looking from a different points of view. We all know that 30% duty cycle is also 70% duty cycle. It depends what point of view one is taking.

raymr · #5 02-12-2006, 10:12 PM

Does it run better with the new sensor installed? I give my cars the subjective 'sniff' test. If the exhaust smells like clean hot air, they have always passed emissions.

Regarding duty cycle, I thought I read that earlier models are something like 30/70. I haven't looked at mine in a while, but I'm pretty sure its not at 50% and it passes the state test very easily.

erubin · #6 02-12-2006, 10:18 PM

Thank you all,

<<"When looking at this properly a lean compensation for a rich mixture has the duty cycle oscillating lower than 50% say 35 -40%">>

So my control system is compensating for a rich mixture. Does this mean something is not working properly. It stays at 36%, hardly any fluctuating at idle.

When I turn the lambda adjusting device screw CW the duty cycle on my digital meter goes down. When I turn the screw CCW the displayed duty cycle goes up. When i raise the duty cycle above 36% the engine stalls. This part does not confuse me. What does is that I'm told to target around 50% but the engine stalls if i try to adjust higher than 36% (by turning screw a little more CCW). Any suggestions?

With KOEO, duty cycle is 0% and the voltage on term 3 is battery +.

Exhaust smells fine.

Does the fact that I can't adjust the base setting higher than 36% at idle indicate a problem that should be explored? If I can't adjust higher than 36% what should i adjust it too?

Raymr, what does 30/70 mean? Are those target duty cycles or something?

stevebfl · #7 02-12-2006, 10:27 PM

I would suggest calibrating your tool.

There is a very simple way. Disconnect the O2 sensor. Start the engine and you will be reading 50%. If you aren't fix that first. I'm not sure what you are reading but I'm pretty sure it isn't closed loop control.

You actually don't even need to start the car but the fuel pump relay has to be jumped otherwise as lambda control is taken off one terminal of the FP relay.

erubin · #8 02-12-2006, 10:46 PM

Steve,

I started up the engine with the O2 sensor disconnected. I read 22% for the duty cycle not the 50% you wrote. My tool is not user calibrated so i guess I need to get another tool or track down the car problem causing this even with the O2 sensor disconnected.

If the tool is miscalibrated and the offset is linear i could add the difference between the 50% expected and the 22% measured to future duty cycle measurements.

stevebfl · #9 02-12-2006, 11:05 PM

Sounds like you are reading about half proper reading.

Try looking at the reading in volts. You have to understand what you are looking at. If you could watch the action on a scope tat would slow it down, you would see that one one side of the frequency valve you have battery voltage. The other side is going to pulse at a given frequency. Say the frequency was 1hz (once a second). If you looked at the control side you would see 13v (suppose 13v battery voltage), then zero v as control is pulsed. It wouldcycle 13v, 0v, 13v, 0v, 13v. Now we have to decide how long at 13v and how long at 0v. With a one second cycle, .5sec at 13v and .5sec and 0v would be 50%. A .3sec at 13v and .7sec at 0v will be 70% duty cycle. It is 70% ON because 0v signafies the time the current is applied.

So if you look at this with a voltmeter one will see 6.75v at 50% because voltmeters average. A 70% would be 30% of 13v or 3.9v. Check it out.

erubin · #10 02-19-2006, 12:58 AM

Steve,

You inspired me to focus on the meter (or the way I used it) as the root of the problem and that was it. I was considering getting a used computer but hesitated since they probably rarely go bad and are often unneccessarily replaced.

My friend, a former MB shop forman lent me a home made device that he used on the early 80s cars. He recalled needing that devise set on 100. His more modern tester didn't work but his older devise (a black box cryptically labeled 100, 101R and Test) did. The unit's LCD displayed numbers that were reasonable and varied up or down as expected when adjusting the Lambda screw so i adjusted to 45. At higher revs (~2500) the Lambda stabilised within +/-5 the targeted 45.

On a side note i never could get either of my digital V-meters (don't have an analog) to vary from battery + voltage on pin 3 at idle and never got a 50, 70 or 85 reading on my friend's mysterious meter with KOEO. Maybe the lambda computer on cars such as an 83 380SL do not generate a "fingerprint" duty cycle with KOEO.

The new O2 sensor, a universal Bosch single-wire, splice-in part definetly made the difference for passing CA smog. Unlike the old sensor the new one moved above 0.9V with a squirt of propane into the throttle and below 0.1V with an induced air leak when disconnecting the booster hose.

Thanks for everyones help.

stevebfl · #11 02-19-2006, 10:25 AM

The old K-jet w/lambda cars didn't creat a value for the x11 connector, it was an actual hook-up to the control circuit of the frequency valve. As I pointed out above one needs to run the motor and pull the O2 sensor to view the 50% control duty cycle. It also can be had with the engine off if one jumps the fuel pump relay.

erubin · #12 02-19-2006, 02:28 PM

Yes steve, you did say that.

With the O2 sensor disconnected I got these lambda readings at idle:
Engine cold, fluctautes between 56-59
Engine warm (btwn 40-80C), fluctautes btwn 47-49
engine hot >80C, fluctautes btwn 47-49

Well close enough to 50 i guess? In any case i got lambda set right i believe with everythinng connected and the car past smog.

Thanks again.

stevebfl · #13 02-19-2006, 02:55 PM

Yes, during cold running the duty cycle will be 60% till a certain temp, then 50% untill O2 control.

fpharryc · #14 03-05-2006, 11:08 PM

Erubin, I am curious. Since your duty cycle is at 22% O2 sensor disconnected, why don't you just adjust the mixture to 22% with your original meter in close loop? Older MB apparently is different in terms of fuel mixture design, using a "calibrated" equipment certainly entailed guessing, I think.

I don't know how to disconnect my 85 500sel's O2 sensor, I couldn't find the connector under the passenger side carpet. Can you tell me how?

Same as your finding, I couldn't lean the mixture above 46% or it stalls after that. I couldn't read duty cycle KOEO, either. The voltage is 0.05v by measurement at pin 3. I am very intrigued after reading various posts including yours. After some experiments, it seems for our car, the duty cycle somewhere around 22-25% is the sweet spot.

My car runs great. I don't think there is anything wrong with it. What do you think?

erubin · #15 03-05-2006, 11:55 PM

I read 22% at a certain point but it was with my Actron set in duty cycle mode. It was a false reading I believe. When i finally hooked up with my friends meter i got reasonable values that moved with the adjustment screw as expected. I don't know what year car you have but on the 83 380SL the actron meter set to duty cycle didn't work. Furthermore, i had no luck coming up with an intermedaie voltage reading (somewhere between battery + and ground, where 6.5V would be about 50%) with my digital voltmeter set to measure DC volts. Maybe an analog or a differnet digital meter would have given better results when extrapalating duty cycle based on the average voltage.

My friend with the mysterious meter said that on the earlier lambda controlled SLs he always used his homemade meter. I'm sure MB had the right tool available at the time but for me i finally got it when i abondoned my actron and used his older of two tools. His Fluke meter (which he set on frequency) didn't work (it oddly measured 9, units unknown) although he said it worked on the newer cars when he was a tech he finally recalled that the older cars needed the other tool.

my advise to you is to focus on the advisae and strategy Stevebfl provided and maybe try different meters.

Look under the car for the O2 meter and trace the wire. the connector should be where you looked but maybe you'll find the connector under the car? I put in a new O2 meter and had to splice it in (it was a generic Bosch spec'ed for my application but minus the MB connection).