UPDATE 20

The local mechanic managed to remove the old sensor, with lots of PB blaster and heat from a propane torch. Unfortunately, the new sensor being nearby (in the box but plugged in), the heat melted the plastic threads protector and the melted plastic was all over the sensor, he told me. He scraped it off with a razor but I'm afraid the new sensor is ruined. Putting the scope on it reveals the same trace provided in the last update. It doesn't budge from 0 V. Could the melted plastic have plugged up the sensor so it doesn't react at all to changes in O2? If this is the case, I'll have to get another one. What a lousy way to ruin a perfectly good new sensor. The low idle and stalling when warm still occur.

 

UPDATE 21

Bought a new O2 sensor and put it in. Just like the one I just replaced, the scope trace barely budges from 0 V. I saw some wiggle, of the order of mVs when playing with the throttle. I could never get the voltage to change more than about 5 mV. Is this normal? It is possible that I had more than one problem and the O2 sensor might have been the cause of the horrendous gas mileage, but the low idle and stalling when warm persist. Any further ideas?

 

Is this a secondary sensor, i.e. Sensor 2 (post catalytic converter)? If so, your voltage should be pretty steady and not fluctuating too much. If it does, that's an indication that the cat converter had lost its efficiency. However, voltage should be more than 0 v. Have you tested it on the cold engine or after a drive? Warming up cat converter and O2 sensors might make a difference.
Here is a good reference on O2 sensors. Perhaps it gives you some info we all are missing:
https://www.clrn.org/what-should-o2-...-on-live-data/

 

This is the one and only O2 sensor. It is the upstream sensor, just aft of where the L and R exhaust headers join. UPDATE 18 above shows the original sensor's output. It varied by roughly 1 V but had a DC offset of 1.8 V, which, I believe, was the cause of the horrendous gas mileage. The new ones don't produce any voltage at all. UPDATE 19 shows the output of the first one I installed, which I thought was bad because of the melted plastic. Unfortunately the new one behaves the same way. The first trace below is with the engine cold:



The next one is with the engine at operating temperature, just before it stalled. The only difference is that when lifting the foot off the accelerator when cold, the spikes stopped. Is this due to the throttle position switch? When warm, this doesn't happen.




Is it possible I got 2 bad new sensors? or maybe incompatible sensors (with the ECU)? The boxes and instructions sheets don't say what kind of sensors they are, presumably Zirconia for older vehicles, as the link says. Stamped on the first replacement sensor is this: "OZ-A686-EE16". They came from rockauto. I have not driven it enough to see if the poor gas mileage is gone. It is extremely annoying (and not too safe) that it stalls any time I stop.

 

That is a definitely a head scratcher... If I recall correctly, low voltage output on the O2 sensor means a Lean condition. Lean condition can come from several things: vacuum leak, bad MAF sensor, bad fuel pump, etc. Stalling every time you take your foot off the gas is kind of pointing in hat direction too. I'd check to see if you have a vacuum leak and if your fuel pump outputing fuel at the expected volume and at the right pressure. The O2 sensor might be just relaying that info to you vs being responsible for it.
Having very poor fuel economy would point in the direction of a vacuum leak. It's the easiest to check, so i'd start with that.

 

That is my recollection too: Low voltage output on the O2 sensor indicates lean condition. That's why I assumed that having a DC offset of +1.8V on the original sensor was causing super-rich conditions, causing the horrendous gas mileage.

As to the causes of a lean condition you mentioned:

I tested for vacuum leaks (see UPDATE 7) by spraying carburetor cleaner at all hoses at the top of the engine. No change in RPMs.

Tested that there is vacuum by putting a vacuum gauge on the hose that goes from what appears to be the "Purge control solenoid valve" to the throttle body. 10 in Hg at idle, about 900 RPMs, engine cold with lots of misfires. The purge control solenoid valve works when I place 12V across it. This has no effect on the idle RPMs or the misfires. Confirmed that it gets activated when the engine reaches the lower threshold of operating temperature, per the temperature gauge.

Per page 14-76 of the service manual, tested the "MPI-diagnosis" pin (and ground) with the oscilloscope and got the regular square wave indicative of "Normal state". If the MAF sensor is bad, the computer doesn't know about it.

As described in the first UPDATE (Jan 5, 2025) I was left stranded and had to be towed. The repair shop replaced the fuel pump, strainer, filter and fuel pressure hose. All this was unnecessary. Turns out I had just run out of gas. This prompted me to look at why I was getting 4 MPG.

Is there a way to check for vacuum leaks that I haven't tried? Is the vacuum I found what it should be? (10 in Hg)

 

Damn. You are very thorough and you've shot down all the ideas I had. One more possible place for a vacuum leak. Do you have a vacuum brake booster or electronic pump? I'm not familiar with older models?

 

The brake booster is of the vacuum variety. Only one vacuum hose coming out of it and it appears to have no leak, per the carburetor cleaner spray test. The brakes work well.

Finally did a compression test. In the order of cylinder numbering: 180, 180, 165, 170, 170, 165. So it looks like I don't have a blown head gasket or something radically wrong deep in the engine. The spark plugs have quite a bit of carbon build-up. Too much fuel? The only other thing I can think of is some bad injector(s), but the computer gives no such error. Any further ideas? Thanks for reading this far.

 

Thinking about what else could be the problem, I think I can rule out timing belt skip. The compression numbers I got seem consistent enough and both cylinder banks have a high (180) a medium (170) and a low (165). If there's a flaw in my reasoning, someone please let me know. Are these compression numbers reasonable?

 

Yes, those numbers are OK