I couldn't answer this question yesterday as at the moment I was in the midst of having a 83 380SL totally whip my butt idling at 900rpm. I began to doubt all my knowledge and experience.
I gave up last night and couldn't look at the car this morning. I went in and sat down with my electronics guru brother and began explaining what I had done and how the conclusions kept telling me that I had a bad idle controller, but I already had a new idle controller and it was doing the exact same thing.
Now I'm embarassed as this is the height of simplicity in the world of electronic control, we are talking a system that is over 20 years old and I was already an experienced MB tech when it came out and I've been tangling with this one off and on for two days.
This is how it works: the idle valve is held to 650 rpms by 400ma of current. The MB book says a little more maybe. My brother about 15 years ago built us a transistorized power control for these idle valves as a test tool. I hooked it up and at 400ma the idle was at 650rpm. Don't you wish you could check your valve so easy. BUT, the darn system is running at 250ma.
First conclusion is that the circuit is poor due to high resistance or bad OVP relay. Voltage drop tests show the power and grounds to be OK.
Since the current was low and the power circuit good we got a controller. Same story. So now we need to look closer at whats happening. It is noticed that when the trans is put into gear the current drops quickly to 150ma maintaining the same 850-900rpms. I thought about how the speed signal could get misinterpreted, so I compared the square wave prepared speed signal (its on one of the pins of the diagnostic connector) to the primary signal at the coil. The signals totaly overlaped running at 57hz (855 rpm).
After watching this thing I become convinced that the car is idling at the low temp setting. It is strong and determined to run at 850-900rpm which is the cold (below 16deg C - 60deg F) range.
The idle control box only has 9 wires on it. It looks at the idle switch. When grounded the idle switch places the box in the control mode. Open moves the valve to a fixed current holding the valve farther open than at idle. This is because if the system worked otherwise, the valve would shut as the engine sped up. If the throttle was shut rapidly the engine would die as the valve couldn't open quickly. The unit views temp by looking at the 16deg switch on a wire that also carries the same signal to the lambda sensor.
Very early in testing I found that the 16deg circuit was grounded. Seemed the answer. I was checking the system at the controller under the dash. To isolate the problem, instead of finding the source of the short, I decided to just eliminate it for test purposes by pulling the pin from the connector. At this point there was no ground at that terminal since there was no wire at all on it. The circuit is only hooked to various grounds (both the full throttle switch and the snap acceleration vacuum switch also can ground the circuit). I probably should have paid more attention to the fact that plugged or unplugged there was no change.
At this point I was through (5:30 yesterday). This morning I talked the whole thing through in the A/C with my brother and we absolutely decided that the system had to be operating at cold setting. Since the wire was shorted we found the problem - the snap accel vacuum switch was a hang-on this car; it was an 83 and they were installed in production starting in 85. The harness ahd been tied to the airpump and was shorted internally. I unplugged the factory 16deg oil temp switch connector from the harness and plugged it back to the original point on the oil switch under the alternator.
The problem was solved. This should have been easy, I found the short quickly, but didn't believe it because nothing happened when I disconnected the short.
Here is the answer: normally a control unit expecting a ground signal has a small biasing voltage that gets pulled to ground by the ground. Evidently the idle controller is so small that they left the circuitry out and rely on the circuit in the lambda controller for the biasing. SOOOO. When I disconnected the short the voltage stayed the same. What should happen is that when the switch to ground (short in this case) goes away the biasing voltage of the lambda controller is seen (since it isn't grounded).
So after working on this system for 20 years I have now found a new way it can fail ( I am also wondering now if I have seen this before). The short I knew about, but a break in the wire between the two controllers will simulate the cold response since the idle controller is actually looking for a bias voltage that comes externally from the lambda controller.
I hope this makes some sense.
Anyway, finding this, is more fun than money. If one has never chased a problem to a solution one can never know how much fun it is. This makes the fifth time I have told the story. Everyone one of my techs has heard it.
Bosch Master, ASE Master, L1
33 years MB technician