Okay, so it's not an O2 problem...
 

Okay, in reference to this thread a couple weeks ago: https://www.miataturbo.net/general-miata-chat-9/help-troubleshooting-o2-sensor-problem-11400/ , after running my battery dry leaving my lights on at work, I jump-started it and it threw a CEL code 17, O2 sensor value not changing on the way home that day. Replaced my O2 sensor (that was only a couple years old), same deal. Thought maybe a wire was grounding/shorting or something, ran only the O2 sensor's signal wire to the ECU, thought that fixed it, but I was wrong...same deal. Output signal on the O2 sensor just goes up and up until it maxes out around ~1.0V when the engine's warmed up. Sometimes, when revving the car quite vigorously, you can get the output voltage to go down/change, but not by a whole lot, not nearly as much as it should.

For the heck of it, tightened my AFM flapper door a couple clicks to see if the behavior changed, to no avail. According to my enthusiast's shop manual (British book), the code 17 refers to the "feedback system," presumably the whole ECU-going-into-closed-loop deal. Explains how to use the Mazda self-diagnosis check tool, which I don't have of course, and then goes on to say "check the fuel injection system," "check the spark plugs," and "check for a short in the wiring between the O2 sensor and ECU terminal." Finally, "have your Mazda dealer check the ECU and replace it if necessary."

So, tomorrow I was planning on throwing in the stock, BK6RE-11 plugs, instead of the current BK7RE's, just in case something's going on there, but I can't imagine what. I have not checked the TPS or water thermistor but would think I'd be getting a code 12 or 9 if that were the case. I guess another thing to try would be to disconnect the Bipes ACU and run without it, but since if that were shorted/grounding, I'd be getting AFM or CAS signal problems and the car wouldn't run, so I doubt there are any issues there.

So, do you think it's possible something in the ECU got fried/shorted when I let the battery drain that day and jump-started the car? I suppose anything's possible, but it seems unlikely...however, I'm pretty well convinced now that something happened that day that caused this...I had thought previously that perhaps it had happened some time ago, and I just never noticed the CEL, but with the frequency that the CEL has been going off and the insane drop in gas mileage, there's no way I'd not have noticed.

Does anyone have a spare stock '90-'93 ECU they could send me to test with? Perhaps, say, one of you that's upgraded to the Megasquirt?

EDIT: Wow, just realized how fucking long that post is.

I've got a spare ECU if you want it. BUT you would have to pay shipping both ways from Canada :)

Damn, I thought you had it fixed.

Sounds like the o2 sensor itself does work. Are you positive it's hooked to the ecu? I would check that wire. When I had a similarish problem with my Link, I opened the ecu case and probed the 02 sensor trace on the Link board itself with the DMM to ensure it was receiving signal. In my case it was, but the Link wasn't doing anything with it. The fact that you were in closed loop last week, and your sensor seems to work, makes me wonder if the o2 signal isn't getting to the ecu's terminal.

If the ecu gets signal, we need to go through the things that cause closed loop. The obdI ecu is pretty primative, and is known to not trigger CELs even though problems exist. I would probe for voltage (or is it resistance?) at the coolant temp sensor.

Ah, something quirky I just remembered. When I was running the stock ecu and FMU, my car always liked it if I let it warm up for a minute before driving. If I just turned the key and took off, I almost always got the CEL for o2. If I gave it a warm up first, I NEVER got it. I don't know what the problem was. Easy fix was allow warm up time.

You can borrow my ecu if you want to cover shipping. But you probably would be able to borrow one locally.

Okay, I'll guess I should bust open the ECU case and test the O2 and thermosensor traces, as you mentioned...the probes on my Harbor Freight multimeter are too wide to really hit the ECU pins effectively.

Thanks for the ideas, Ben, and I may actually have to ask to borrow your ECU down the road, as I've never seen a Miata in our junkyards and the only guy from the forum I know of is Beerslurpy, and he's got a later model car. Actually, I could probably get a hold of one of the autox guys, come to think of it.

Did you check the grounds? A voltage differential of even the smallest amount could be causing you problems.

Well, I would think the only ground that could really cause the ECU to have closed-loop problems would be the ground wire in it's own harness, right? I mean, there are several ground locations I'm aware of, two or three in the engine bay, and I seem to recall one on the PPF.

But if the ground at the ECU trace/harness is reading ~0V, it should be good, right?

I'm no electrical genius, so any ideas on how to test grounds, etc. would be welcome :)

Clean and secure them. When I installed my link I had to replace the ground strap from the body to the engine on the rear driver's side. Until I did that I was getting really wonky map readings ... and that's 0-5V, no the more sensitive 0-1V that the NB sensor uses. 250mV difference is very significant to a NB sensor.

Just a thought and likely a long shot but cheap and easy. I made a new ground strap out of 8ga Amp sire and a couple of connectors.

Are you using a 4 wire sensor?

Yes, I know that at least two of the three (or is it four?) grounds in the engine bay are good because I was just disconnecting the grounds from the 4-wire O2 sensor at those two spots while troubleshooting this issue last week. So right now, my 4-wire O2 sensor is acting as just a 1-wire, bypassing the O2 clamp, just in case, but none of that seems to have been the problem.

I'm just now about to go out and check some stuff.

I tested several traces at the ECU tonight with the car idling and found the water thermosensor, TPS, AFM, ignitor, CAS and grounds to all be within spec once the car was warmed up.

The O2 sensor signal had this behavior: while cold, it showed 0.0V, when idling, the O2 would barely register, around 0.02V, when I plugged the O2 sensor heater circuit wires up to 12+V and GND, the signal almost immediately started rising...about 1/10th of a volt per couple seconds, I'd say. It just kept going up and up, .5V, .6V...all the way to ~.94V, presumably the end of it's range, and stopped. This is what the old O2 sensor did (keep in mind, this one is brand-new), and it would do the same thing unplugged from the heater circuit, as long as you drove it long enough to get the O2 heated up.

So this is exactly the behavior I would expect of a car not going into closed-loop...just stays in uber-rich open-loop mode at all times....it's like the car is not coming out of warm-up mode or something.

And it still has to be related to that fateful day I ran my battery dry and jump-started the car. I guess my next step is to swap out a known-good ECU.

Okay, so the ECU swap happened and the problem returned...I'm thinking it's neither ECU nor O2 at this point. I guess I'll try running a brand-new signal and ground wire for the O2 sensor, first, then I'll go back to testing sensor inputs at the ECU.

I still can't understand why this happened after jump-starting the car...if it had happened before then, I would've noticed the CEL that comes on during every drive, and/or the miserable gas mileage.