Okay... 1990 300SEL, 49 state car. Same car from this thread:

1990 300SEL no oxygen sensor heater voltage

New Bosch O2 sensor. Repaired MAS.

This may be important: car O2 circuit operated perfectly, O2 heater worked fine, car went to closed loop mode reliably and never illuminated the check engine light (except when intentionally triggering it by unplugging various components during testing) until after going into a body shop for some work several months ago. While in there, they claim it stalled and would not restart, and they diagnosed a bad distributor cap and replaced it. After returning from the body shop, car would illuminate the check engine light as described below. Whether this was due to welding on the car, or shorting circuits / using an analog meter during diagnostic testing, I do not know.

Start engine, see battery voltage on O2 heater circuit. Monitoring O2 heater current shows about 2.5 amps falling to about 1.3 amps as sensor heats up.

With O2 signal wire and EHA valve disconnected from harness, EHA valve trim screw and air meter idle mixture trim set so O2 sensor signal wire shows about 0.7 - 0.8 volt at idle in park, at ~2,000 rpm in park, at idle in drive, and while driving under various steady state conditions from ~20 - 50 mph. Deceleration cutoff works, signal goes almost to zero when lifting throttle, comes back up to 0.7 - 0.8 volt under conditions listed above.

With O2 signal wire and EHA valve connected, on cold start and for the first mile or two of operation, car goes closed loop with duty cycle across pins 2 and 3 at X11 connector showing fluctuating values from 45% - 55%. Monitoring O2 sensor heater voltage during this time shows battery voltage on circuit. Monitoring O2 signal voltage during this time during steady state operation shows fluctuating values from ~0.2 - ~.8 volt going toward 0 volt on overrun when triggering decel fuel cutoff and toward ~0.9 volt under full throttle acceleration.

Within a few miles of operation, O2 signal voltage stops fluctuating and stays near 0.8 volt. Check engine light illuminates. O2 sensor heater circuit still shows battery voltage. Monitoring duty cycle across pins 2 and 3 at X11 connector shows 50%.

When car goes open loop and illuminates the CEL with 50% duty cycle across pins 2 and 3 at X11 connector, monitoring of EHA current shows a steady -2 milliamp at idle. This appears to be the "limp home" mode.

Occasionally the car will go open loop with CEL illuminated, then back to closed loop for a while with CEL not illuminated, but on longer drives the CEL tends to stay illuminated with car operating in open loop mode. Checking fault code when CEL illuminates invariably gives the 50% O2 code.

Monitoring O2 reference voltage from ECU with O2 sensor disconnected, do not see ~0.450 volt, but rather values ranging from ~ 0.300 volt falling to ~0.050 volt over period of several seconds.

Disassembly of ECU shows no obviously burned components. Tracing circuit on board from pin 13 to as far as I can follow it with a digital continuity tester before running into a chip shows no open circuits Testing of all relevant wiring in car shows good continuity and no shorts. Specifically, good continuity from O2 signal connector on passenger floorboard to pin 13 at ECU with no short to ground.

All sensors appear to be in good condition. Crankshaft position sensor, coolant temp sensors, air meter pot, intake air temp, idle switch, WOT switch, etc. All fuses cleaned and verified good with meter, installed in fuse box each verified good continuity between both contacts.

Damaged O2 circuit in ECU? Or am I overlooking something that would send the car into open loop mode with the O2 fault code when there's a good O2 signal reaching the ECU?

EDIT: Leaning towards the ECU. Was testing some other ideas, no luck. Tested O2 reference voltage from ECU again and got ~ 0.200 volt when initially turning ignition key to run position after letting car sit for several minutes, with voltage quickly bleeding down. Removed the ECU and took it apart again and went over both sides of the board scrutinizing it as well as my eyesight would allow, and with multimeter testing everything I could figure out how to test. Found nothing obviously wrong, but may have nudged a weak solder joint just so. Put it back together, installed into car, with O2 sensor signal wire disconnected and reading the reference voltage, got a steady 0.520 volt. Left the ignition key in the run position for several minutes, reference voltage stayed rock solid on 0.520 volt. Connected O2 sensor, started car, after short warmup it went into closed loop. Tested in shop while running for roughly 10 minutes, monitoring O2 signal voltage it jumped back and forth from ~0.2 - ~0.8 volt. Monitoring duty cycle across pins 2 and 3 at X11 connector, got fluctuating values between ~45% and ~55%, with expected spikes rich and lean during throttle application and overrun cutoff. We'll see how long it takes on the next trip for the CEL to illuminate.