The car is a 1986 300E, with 103 engine (103 983 12 037 507) and VIN -293626.

I'll start with the questions, but follow with the whole diagnostic tale.

Q1. When are errors actually reported via duty cycle from the ECU? Is it at key on engine off, replacing the signal that otherwise says whether your ECU is capable of reporting errors, or is it only when an error happens, like in my first set of tests when I went to a fixed 29.3% when the engine warmed up?

Q2. Is a 9.4% duty cycle at KOEO even rational on a California car?

Q3. Is it normal that I get the exact same duty cycle reading whenever the throttle is fully closed (idle) with the engine running, and that it is the same value that I get at with key on engine off?

Here's the whole story.

The car flunked its California smog check as a gross polluter, with high CO and HC at 15 and 25 MPH. The test was at a reliable shop I've used before, and the engine and cat were plenty hot after a long freeway run. I know this means it's running really rich, so I need to sort it out.

The numbers:
At 15 MPH CO was 8.48%, way over the max of 0.79% and over Gross Polluter 2.29%
At 25 MPH CO was 7.94%, way over the max of 0.59 and over Gross Polluter limit 2.09%
At 15 MPH HC's were 339 PPM, over the 126 PPM max and even over 301 Gross Polluter limit
At 25 MPH HC's were 322 PPM, over the 101 PPM max and even over 251 Gross Polluter limit

My first thought was that it was doing extra enrichment for warm up so (being in a hurry) I jumped ahead in the factory KE-Jetronic test procedure to the coolant temperature tests and too hastily condemned the sensor because I read the resistance wrong (by reading across the two pins of my two pin sensor, it turns out they each go to a common ground and have to be read separately) and mistakenly thought it read too cold. Put in a new temp sensor, but of course no joy. Still ran the same and smelled rich. The new and the original sensor were probably both good. Miscue on that one.

I knew the car had closed loop lambda feedback, so my next stop was the O2 sensor that hadn't been replaced in the many years I'd had the car. Checked the O2 sensor signal with the engine at operating temperature with my averaging voltmeter and my scope, and the output wasn't very good, not getting over two or three tenths of a volt, and very, very short duration, very irregular spikes. I expected it should range higher, up past half a volt anyway, and expected longer and more frequent signals. So, put in a new O2 sensor. Great signal now. With the new sensor, at operating temperature at various RPM's its output ranges from 0.1 to 0.6 volts, with longer duration spikes.

Figured I had it then, but the exhaust still smelled real rich, and oddly enough under some conditions the car was harder to start (needed pedal where none was needed before to start) and odder still, unplugging the new (presumably good) O2 sensor would make the car easier to start, quick start, no pedal. Two strikes.

I decided to see if I was entering closed loop or not, or whether there was a detectable error, by watching the duty cycle. Armed with my trusty Craftsman duty-cycle capable meter (set to percentage, not frequency, with the positive lead in pin 2 (ground) and the negative lead in pin 3 (signal) order to reverse the polarity and thus read the "off" duty cycle percentage directly on my "on" duty cycle meter without doing the 100-n math), I did a set of tests from cold start through warm up and operating temperature. The first tests were interesting, and showed signs of life in the lambda system as the engine approached 80C like this...

Condition ..... O2 Sensor Volts .... Duty Cycle at Idle ..... Duty Cycle off idle
KOEO .............. 0.44 ..................... 9.4% ........................ N/A
Cold ................ 0.82 ..................... 9.4% ....................... 49.4% fixed
Near 80C ......... 0.1 - 0.6 ............... 9.4% ........................ 42% - 47% varying
80C ................ 0.1 - 0.6 ............... 9.4% ........................ 29.3% fixed

I interpreted this to mean that after trying to run in closed loop, the ECU had given up because it couldn't keep things in range or had detected a fault, but I didn't know whether the 29.3% was just the fixed operational mode or an interpretable error value like the post 4/86 ECU's have (I don't know how to get my month of manufacture out of my -293626 VIN, but it seems like non-error capable ECU's have no (or 100%) duty cycle at KOEO, so I think mine is trying to tell me something). Assuming it was the "30%" error code indicating a coolant temperature sensor issue, I double checked the new coolant sensor, even unplugging the N3 connector at the ECU and doing what tests I could from the factory procedure without having the breakout box to be able to read the signals with the ECU plugged in (no "twin socket" available).

Here's the kicker. After doing those coolant sensor tests, with everything plugged back in, the results have changed. The car never seems to go into closed loop at all now, even for a minute, and instead runs at 49.4% fixed duty cycle like this...

Condition ...... Duty Cycle at Idle ....... Duty Cycle off idle
KOEO ................. 9.4% .......................... N/A
Cold ................... 9.4% ......................... 49.4% fixed
Nearly 80C .......... 9.4% ......................... 49.4% fixed
80C ................... 9.4% ......................... 49.4% fixed

I'm looking for some confirmation about what I am reading, here. I assumed my California car (if capable of reporting errors by "on/off ratio", as they say in the manual) would read 85% at KOEO, or 70% if it was actually a Federal spec car that found its way to California before I bought it. But my KOEO reading is always 9.4%. And although I could understand if I didn't get the polarity of the duty cycle right the 9.4% could really be 90.6% (almost the expected 85% for California) but I confirmed with my scope that the voltage on contact 3 of X11 is definitely low 9.4% of the time and high 90.6%, and everything I read here is that the "duty" being reported is the duty of the ground-controlled circuit, so low 10% of the time should be 10% duty cycle.

Thanks for any assistance. This forum, Phil's great on line parts supply, and a helpful shop and MB parts source near my office make keeping my two older MB's running a joy, except for these occasional forays into the mysteries of Bosch KE-Jetronic operation.

Well, never got any reply on how to interpret those duty cycle numbers, but here is the update.

After overall work of:
Replacing coolant sensor (probably OK originally)
Replacing plugs
Replacing O2 sensor
Repairing several vacuum leaks
Adjusting mixture by feel more towards lean

I passed my retest.

The numbers:
At 15 MPH CO was 0.35% (down from 8.48% on prior test)
At 25 MPH CO was 0.16% (down from 7.94% on prior test)
At 15 MPH HC was 123 PPM (down from 339 PPM on prior test)
At 25 MPH HC was 59 PPM (down from 322 PPM on prior test)
At 15 MPH NO was 915 PPM (up from 234 PPM on prior test)
At 25 MPH NO was 545 PPM (up from 160 PPM on prior test)
At 15 MPH O2% was 0.4% (up from 0.0% on prior test)
At 25 MPH O2% was 0.1% (up from 0.0% on prior test)
At 15 MPH CO2% was 14.6% (up from 9.6% on prior test)
At 25 MPH CO2% was 15.0% (up from 10% on prior test)


The new numbers are much better, of course, but the low speed HC and NO numbers were a squeaker to pass. My 123 PPM HC just squeaked by under the 126 limit, and my 915 NO wasn't much better against the 1097 limit.

Congratulations!

I have seen all sorts of remarks to the effect that the 103 engine is a polluter and difficult to get into compliance. Maybe, but it is not my experience. All the usual things that get an engine running right will reduce the emmissions.

Ours has 360,000 km (225,000 miles), poor compression on #4 and #5 (due to worn rings and cylinder walls) and still flies through AirCare like a new engine, with very low emissions.

In my case (1989 190e 2.6), the big thing was spark plug wires. That misfire really increases the HC levels. Of course, it gets new plugs periodically.

The other thing that was good for our car was new valve stem seals. It was burning a lot of oil, which stopped almost completely. My theory is that worn seals allow oil into the combustion chamber which then affects emissions in three ways:
(1) It carbons up the combustion chamber, increasing compression, and so increasing NOX production.
(2) I produces unburned hydrocarbons, which foul the catalytic converter, at least temporarily.
(3) It contributes to the high HC levels.

This just reflects my experience.

Andrew
1989 190e 2.6
1990 Jaguar XJ6

Where can I find the duty cycle numbers and procedures for my 560SL? Is there some handy/dandy chart or manual available somewhere? I've got a nine pin connector under the hood that I'd like to be able to use for its intended purpose.

Thanks.

__________________
Len
'59 220S Cabriolet-SOLD and living happily in Malta
'83 240D 351,500 miles original owner-SOLD
'88 560SL 41,000 miles - totaled and parted out
https://sites.google.com/site/mercedesstuff/home
'99 E300 turbo 227,500 miles
'03 SLK320 40,000 miles - gave to my daughter
'14 Smart electric coupe 28,500 miles
'14 Smart electric cabriolet 28,500 miles
'15 Smart electric coupe 28,000 miles