O2 woes- what the heck?

Uuuuug. I'm still getting a CEL for the O2 sensor on the 1984 Celica. I had to extend the wire for the header. Tonight I bought a new sensor, made a new extension cable. The sensor I bought has the factory connector (which is just a quick disconnect in a big hunk of plastic). So I made a wire with connectors on each end and hooked it up. So I didn't cut the wire on the new sensor- it's just as Bosh made it. Still getting the CEL.

I don't know what to do next. I guess I need to check the wiring harness back to the ECU. If I hook my DMM to the wire would I be able to tell if it's getting an appropriate signal?

What is "CEL"

Check engine light. Some piddly thing is fubar.

What he said.

That should take care of the light.

O2 sensor should rapidly and predictably oscillate between about .3 and .6 volts, IIRC. Slow changes or different voltages indicate a problem. It's been a while since I did this; I may be off on the exact values.

Why do you think the problem is the O2.

I have the check engine light, and when I run the codes, I get an O2 sensor code. So it's something with the O2.

And the car is lean at idle. I have a wideband in it and the gauge says it's 17.5 or 18 at idle. Then it gets pretty darn rich at WOT. Like almost down to 9. I'm guessing it's all related, but I'm not sure. Figure I need fix what the car is telling me is wrong before I start messing with other stuff.

I assume it is just a single or even double wire o2? So that means it is unheated....You have moved it farther away now with the header, and you have changed the heat retention properties with the material change to get the o2 to temperature to work.....

'84 would surely be a single wire, since the O2 sensor has barely even been in production by then.

How's the exhaust? As in is it properly grounded through the engine and through the ground strap? Being a single wire, it uses the exhaust and engine as the rest of the loop for the circuit. That, and it needs to be nice and warm to work, so if your header moved the sensor downstream, it's probably taking more energy to heat up, and stay hot- so make sure to drive the car above 35mph (just a guess for heat) to warm it before jumping too hard on an O2/circuit problem.

Also, check the sensor voltage. Running that lean at idle is pretty odd, but running that rich at WOT isn't. Odd, just not that odd. The system doesn't use the sensor at WOT....

Exhaust is brand new, custom made by a good shop and no leaks- I just went through everything and replaced every gasket from the head back with Remflex. Last round with this, I ran a ground strap from the mounting bolt on the sensor to the body, so it's grounded. And, yes, it's a single wire. Good news is these things are only $20 now 'days. I didn't realize the old ones were so cheap now. It's a new Bosh put on yesterday.

This morning I drove my son to school in the Celica. I decided to see what happened if I drove with the sensor unplugged and, as I kind of suspected, it's exactly the same. I hooked my Fluke to the extension wire I made (with it still unplugged from the ECU) and I get about a half volt. So I think the sensor is working and theres a problem in the wiring somewhere before the ECU. I'm burning some vacation days and have today off, so I'll drive it over to the workshop and check continuity on the wire from the plug back to the ECU. That's got to be the problem. I can't think of anything else.

Man, I hate stupid problems like this, but I guess it comes with the territory. Need to get this sorted. I've been stalled on this project with this problem for way too long. Too busy at work to spend the little time to diagnose it properly and it's all I can think about- so the little time I do have has been wasted tinkering on it. It's getting solved today if at all possible.

(edit) Also of note- the header worked fine and there were no codes when I had the last engine in it. This has only happened since I put the new engine in.

The simple fact that a car of that age has an O2 sensor code does not mean the O2 sensor is wrong. All it means is that the sensor is not cycling through its range. That could mean the sensor is bad, or it could mean it is always reading lean, or it could always be rich. Put a voltmeter on the wire, run the engine, what voltage are you getting? .5 volts constant, the sensor is open. If you start the engine and the sensor starts at .5 and trends towards .9 constant, its too rich. 0 constant, its too lean. Its also possible for the sensor to be biased rich or lean on its own, but its not real common.

How poor did the engine run before replacement??? Almost sounds like historically, the ECM is confused on how to run now. I assume new=used? Could it be a calibration difference? Shorted ECM internal circuitry?

A voltmeter is just not enough test equipment to troubleshoot this problem. The O2 sensor is simply one part of the ECU system, which works in closed loop mode. As stated, when the system is working properly, the sensor voltage cycles between 0.2V and 0.8V. It functions basically like a switch, 0.2V when the mixture is lean, 0.8V when the mixture is rich. But it only does that because the ECU controls the pulsewidth to the fuel injectors to run lean-rich-lean-rich .... The goal is to achieve perfect average mixture.
Imagine driving a cruise control that had no 'Set' function, only 'Accelerate' and 'Coast', to maintain a desired speed you would watch the speedo, and continuously cycle between the two switches.

An intake leak is invisible to the ECU, so when it commands 'rich', but the O2 sensor still says 'lean', well, it assumes the sensor is faulty. Could be a header leak for that matter, air is air.

I assume that your sensor is mounted in the header collector?
As stated above, they have to be hot to run correctly ( over 600F).
Consider using header wrap to keep the exhaust gas temperature hot at the collector.

The O2 sensor is like a weak battery, it generates 0.8V when Oxygen is in the exhaust gas. You should be able to read the O2 sensor voltage directly (ie even unplugged from the harness) with a digital meter. It's a weak high impedance signal, analog meters will degrade the signal. An oscilloscope is best, will show you exactly what's going on instantly.
(what, doesn't everyone have an interoscitor in their attic? ;)

I'd beg/borrow/steal/buy a wideband sensor to help troubleshooting, knowing what's really coming out the tailpipe will help a lot.

erohslc wrote: The O2 sensor is like a weak battery, it generates 0.8V when Oxygen is in the exhaust gas. You should be able to read the O2 sensor voltage directly (ie even unplugged from the harness) with a digital meter. It's a weak high impedance signal, analog meters will degrade the signal. An oscilloscope is best, will show you exactly what's going on instantly. (what, doesn't everyone have an interoscitor in their attic? ;) I'd beg/borrow/steal/buy a wideband sensor to help troubleshooting, knowing what's really coming out the tailpipe will help a lot.

That's backwards- it will show a high voltage when rich/lack of O2, and low voltage lean/more O2. It uses the difference in O2 as the battery, and the "outside" is seeing fresh air, and the inside is seeing exhaust. No or little difference = 0 to low voltage, large difference, high voltage.

But the range people are quoting is spot on 0.2v lean to 0.8v rich for most engines on a warmed up sensor.

For this age of tech, a volt meter should be fast enough- it will swing, at fastest, 1hz, but probably slower than that.

What's interesting is that it's reading a constant 0.5v. Even under the best conditions, it's not all that easy to maintain .5v exactly- it's a very steep part of the a/f to voltage curve, so that part is odd.

If its reading .5 volts even hooked up, thats usually the bias voltage from the ECU. You will usually get .5 volts on the wire from the ecu with the sensor disconnected.

I have a wideband sensor on it. The engine runs lean at idle, but while cruising at steady throttle it's always pretty good - in the high 14s. If I ease on the gas, it gets a little richer.

I checked the continuity of the wire and I'm convienced that the signal is getting to the ECU. I checked the voltage again now that the car is completely warm. It should have been at operating temp before, but it was a chilly night in Denver. I'm not getting about .3 at idle. Wideband says I'm at right around 18. What's weird, though, is the CEL comes on only when I'm cruising on the highway. And if I look at the A/F gauge when it comes on, it's always right around 14.7. At least in the neighborhood. This has been the case now with two different O2 sensors (narrow band for the ECU) in place. So I don't think it's the sensor and I now don't think it's the wiring.

The engine runs well, except at idle where it's rough. I can restrict flow through the return line and it will smooth out. I can't see the gauge when I do that, but I assume it's forcing it richer.

I'm very frustrated with this. Right after I built the engine I decided it must be an intake leak. I took the whole intake off and put new gaskets on. No change. I really don't think I have an intake leak. The air meter is the same one I used with the other engine and it worked fine before, but who knows? Could be the problem I guess. I have a new air meter, but it's not the direct replacement- it's a bigger one from a Supra, but it is a direct plug and play replacement. I wanted to get all of this sorted out with factory parts in place before I thew that kind of variable in to the equation, but I guess it would only take a few minutes to plug it in and see what happens. I'm just concerned with allowing more air on an engine that's already showing a lean condition, albeit only at idle.

Shoot, can't think of anything else to try, so might as well give it a shot.

I'm only testing the output of the sensor with it not hooked up. I don't have anywhere to check it with it hooked up. I could make up a pig tail to do that, but I don't think that's going to tell me much at this point. Dunno. The sensor is giving me a signal, it's brand new and the wiring I'm plugging it into has continuity to the ECU. I'm about decided that the signal is getting to the ECU and the problem is something else.

Older BMWs give the fault code 013-Oxygen Sensor anytime there is a misfire, rich, or lean condition. In other words, it's not the O2 at fault, that's just the sensor seeing a problem. Toyota might be the same, so I would find out if something else is causing the lean condition.

Do you have an air meter that's correct with the car?

Now that you've gone though all of what you did, I'm leaning toward what Cone_Junky just posted- something is causing it to not run right, and the results is an O2 code.

A larger meter will pass more air for a given voltage than a smaller one, so it may appear to plug right in and play, but if the output is far enough from the original sensor, it will run quite a bit different. So if you have an original sensor, it may be a good thing to try.

Basically, you are posting that everyting else is in good order- sensor, exhaust, grounds, wires to ECU- so it appears to be not a fautl as much as something causing it to run 'wrong'- and the one thing that you have posted that is different is the air meter.

Thanks everyone. I'd have beat on the O2 sensor all day.

Okay, I just tossed the bigger air meter on and fired it up. Immediatley, the idle is awesome. I kept it on just long enough for the wideband to heat and give me a reading- 13.0. So that's way, way better. I need to rig the air meter in place well enough to take it out for a test drive, but this is encouraging.

Alpha, I know what you're saying and that's exactly why I haven't done this already. The swap on a 22R to an '82 Supra air meter is well documented and has been done many times, but I'm like you- I wanted to get it running right with the right parts before I did it. I have a bigger throttle body to go with it, but it's all be on hold until I sorted this out.

I'll get this cobbled together well enough for a test drive, but it looks like I may be on the right track.

In reply to alfadriver: Yes, 'ya got me, I mis-stated effect of exhaust mixture O2 on Nernst cell voltage:

Outside O2 = inside O2 = lean = 0.2V
Outside O2 > inside O2 = rich = 0.8V

But I still assert that 1 Hz is simply too slow to reliably detect what's going on when the system is working in closed loop. Typical loop frequency is under 1 Hz, so Nyquist limit required to eliminate signal aliasing will be at least 2 Hz. And even if the meter can handle 2 Hz or faster, how fast can you accurately read those changes on a 3 digit numeric display? And that also ignores the effects of duty cycle.

'This age of tech'?
Please. AFAIK, laws of physics still apply in the day-to-day world. Been at this one way or another since 1962, one of the first things to learn is that your instrument readings can lie to you.

Anyway, like you said, seeing constant 0.5V is unusual, so it's time to distrust that observation, and try to validate it.

Carter

In reply to erohslc:

Age of tech on the car. '84 is very early in using O2 sensors, and they didn't really try to run higher frequencies until they had better controllers later in that decade. The '84 is probaly a very analog style computer- as in not much real programming vs a balance of the curcit to do what they want.

My guess is that at 35mph, it would not be switching much faster than 1hz. I still don't switch it much faster than.

I know reading the signal on a scope has been possible back, well, probably older than you have been born (back tracking 20 odd years from '62, if that's a decent guess).

It's more about how the car runs, and how far the computer changes the fuel switch vs. the O2 signal. Early fuel control means conservative gains, smaller steps, and slower switches, overall.

This sort of discussion makes me want to spend time watching eBay for a cheap oscilloscope to better monitor sensor outputs than an average DVOM can offer.

That era Toyota is rather infamous for problems with the sheath on the O2 sensor wire. It either opens the ground, or shorts to the lead, can't remember which. I'm not talking the short section of wire connected to the sensor, I'm talking about the wire in between that connector and the computer.

I tried finding the nice link I used to have about it, but I can't relocate it. Might try hitting up 4crawler and see if he knows what I'm speaking of.

And weirdly enough, check your coolant level to make sure it's topped up. That era does weird things when low.

And, check that your TPS is fully closing at idle. Many of them don't because of gunk. Can also do weird running things with them.