O2 sensor failure and lambda testing

[barrie]

Using a lambda meter, I get a reading that is essentially stable (minimal needle movement) at '5'. This suggests that the O2 sensor is not doing its job, yet the O2 malfunction lamp does not indicate a problem.
Can someone (stevebfl ?) explain what condition and threshold triggers the malfunction lamp? Is it possible to have a failed O2 sensor without triggering the indicator?
Looking forward to your input on this one. Thanks,
Barrie
86 300E

[stevebfl]

Through the years the MIL (malfunction indicator light) has been activated for many reasons. From your signature I presume we are talking about an 86 300E.

That model was the first to have a light that actually reacted to the way the car was working. I do not know the exact enable criteria for that light.

I will give you a good idea but let me explain the characteristics of the O2 sensor circuit. The O2 sensor is a voltage generator in this application. The ceramic material used for the sensor becomes conductive at around 300deg C. If there is a difference in the amount of O2 on either side there results a voltage potential of as much as 1v. This differential is used for closed loop fuel control. It compares the O2 in the exhaust to the O2 in the atmosphere.

The control unit usually sends out a 0.5v signal to the sensor so that an open circuit can be distinguished. If the sensor is unplugged the light shiould come on after at set time maybe after the engine gets to a certain temp.

If the sensor is in the circuit the control unit will also look for the start of activity of the sensor. Once it reaches some threshold it starts modifying fuel control. If this doesn't happen within some time interval after reaching some temp the light will also be turned on.

I suspect that you haven't heated the sensor or it is old and slow. Open the throttle and hold at 2000rpm for as long as a minute and you should see activity. Once heated it should work at idle. That car should have a sensor heater which should shorten the time and kept it hot at idle. The early models weren't very critical. If you want to see if the system is working disconnect the green sensor wire under passenger side carpet and drive it five miles. The light should come on.

------------------
Steve Brotherton
Owner 24 bay BSC
Bosch Master, ASE master L1
26 years MB technician

[barrie]

Thanks Steve, for your extremely knowledgeable and articulate reply! As always, your thorough explanations and thoughtful suggestions are extremely helpful and much appreciated.
I'll get on the job right away!
Barrie

[Benzmac]

If you have a meter that is capable, I have written how to check lambda with a FLUKE 87...You can search and find it. You are looking for a reading of about 50% at idle with less than a 10% deviation at 2,500 rpm

------------------
Benzmac:
Donnie Drummonds
1992 500E (very soon I hope
1981 280GE SWB
ASE CERTIFIED MASTER AUTO TECHNICIAN
SERVICE MANAGER FOR 14 BAY FACILITY
MERCEDES SPECIALIST 8 YRS
PARTNER IN MERCEDESSHOP.COM

[barrie]

The measurement can be made with either a Lambda or duty-cycle meter. However, the criteria cited for a good sensor (a 50% duty cycle at idle with less than 10% deviation at 2500rpm) will also pass a BAD sensor. If the fuel mixture is correct and the sensor is bad, you will get a reading of 50% with a variation of perhaps 2% or less at 2500rpm. Also, a good sensor can easily have deviations greater than 10%, thus failing this test!
Mis-diagnosis of sensor operation is apparently common, prompting Mercedes to issue a service bulletin reporting that 60% of sensors returned for replacement from dealers were not defective, and admonishing them to use the correct test procedures. When my mechanic advised that my sensor wasn’t working properly and needed replacement, I decided to research the thing and test it myself.
Thanks to MB manuals/service bulletins, other research, and the resources of this great site (Steve and others), here are the recommended tests and diagnostics that I used:
----------------------------

Condition: suspicious O2 sensor

1) Retest (with duty-cycle or lambda meter): heat the sensor by revving to 2500 rpm for 2-3 min; test before engine has idled for more than 1/2 minute. Measure duty cycle between pin 3 and ground, observing the reading while varying engine rpm. If reading does not vary MORE than 5-10% you may need to slightly depress the air sensor plate to force a response. If you still can’t get a good reading, don’t replace your sensor just yet! And even if you do get adequate meter movement, you may want to skip to step #4 to check the general health and integrity of the sensor.
2) If the duty-cycle reading is not very responsive(and the O2 malfunction light is OFF) then use Steve's test (above) to see if the ECU recognizes the sensor and its basic operation.
3) If the O2 malfunction light does not turn ON, then problem is with the ECU, connections thereto, or the warning circuit; if test is OK then sensor element and heater should be tested. These tests are easily performed using only a multi-meter; wiring is located under the passenger-side carpet.
4) Testing Sensor Element: measure the output voltage of the sensor ceramic (black wire) under full-rich conditions; create this condition by shorting the input to the ECU (green wire). Specified tolerance is 0.45 - 1.0 VDC. Mine measured 0.48, suggesting that the ceramic element is getting weak.
5) Testing Sensor Heater: the KE-Jetronic system uses a heater (3-wire sensor) that should also be tested. Current draw (through either of the white wires) should be between 0.5 – 1.3 A (mine was a healthy 1.0).
6) There’s also a more sophisticated oscilloscope test for the sensor, which fortunately I didn’t need to perform. Mine is working fine, for now!

I’m not sure what was more satisfying – saving $250 on a replacement sensor, or getting it figured out and being able to test each stage of operation. Thanks to everyone!

Barrie