01 Mazda Millenia Remaining Bugs

[semaphore1999]

Not related to O2 sensor:

I got a thermostat dial in from junkyard/ebay seller. This was a piece with a dial that fits into the dashboard. The trick here was that the numbers had to line up exactly in order for the part to function properly with the set of options that my particular Millenia has. There was only ONE of these dials available for sale in the entire cyberverse, and I got it. So, hopefully the thing works, right? I pllugged it into my dashboard as my dashboard was all in component wire-dangle mode and to my delight and surprise, the thermostat set temperature changed before my eyes! Now I can actually control whether to blow hot or cold no matter what the temperature is in and around the car. Fantastic! So from there I toggled out of dashboard component wire dangle mode, opting for reduced rattle fixed control/indicator position mode, which is the default mode most auto manufacturers use for new vehicle delivery to customers. Seems the customers want their dashboards with no dangling controls…

So this was good news. Now for the less good news, which brings us to the O2 sensor related concern:

I have now tried three different O2 sensors plugged in to bank 1 sensor 2 position, all three of which have flagged trouble codes of varying severity. when the CEL lights up, the code is detailing a heater malfunction. At present, I have a brand new part in the bank 1 position 2 location, and so I am having to drill down into actual understanding in order to sort this one out.

After going to O2 sensor school, thanks to wold wide web, I discovered that O2 sensors output lower voltages for less O2, and higher voltages for more O2, as backward as that may seem. Its ok, however, since at the end of the day, the ECU can get a measure of O2 levels, albeit a little bit backward IMO.

But what really struck me upon my O2 schooling was this concept that exhaust leaks are just as troublesome as vacuum leaks since they yield erroneous O2 sensor readings. So for OBDII cars like this Mazda, the ECU really does use the values that come out of these sensors to control the fuel mix ratio… This is important to me because according to my recollection, I just sorta slopped the exhaust back together when assembling the car. So I am starting to lean toward this notion that I may have an exhaust teak that is causing my O2 sensor codes. And I’ll also reluctantly admit that I have noticed a slight hesitation upon acceleration alongside a slight hissing noise I can hear with the hood opened.

OK, so I want to do an exhaust leak review on the car, as I believe this may be the last item I have to fix prior to inspection readiness (I pass all tests except for catalyst and evap, or at least I did prior to having all the O2 sensor troubles). This also makes sense in explaining why I never could pass catalyst, and all of these are likely prerequisites for evap.

So this brings me to my question for the auto experts: Now I had great success with the 'American Spirit" method I customized from Scotty Kilmer’s ‘Cigar Method’ for smoking out my vacuum leak. And I successfully used this method to locate and cure a vacuum leak at my throttle body, where I could see evidence where prior mechanics had squirted liquid gasket before. But I have concerns about using this method on exhaust. LIke I wonder will the smoke damage my catalysts, or maybe my sensors. Then, I wonder where I should insert the magical spirit smoke? at the tailpipe? Then it has to go through the muffler before coming to catalytic converters, then ultimately the headers. Is this a good idea? Is there a better idea?

Thanks all for the input! This project may actually end soon!!

[zmetzing]

LIke I wonder will the smoke damage my catalysts, or maybe my sensors.

I doubt that tobacco smoke, in the minuscule quantities needed for this test, will do any damage to any part of the exhaust system. Anything that is deposited out of the smoke will likely be burnt off during the first 30 seconds of real engine-on time.

[Raymond]

There should be very apparent air blowing outwards from either the exhaust manifold(s) or from the downpipe(s). If you’re sure you put on new gaskets and torqued them correctly then look for cracks where the tubes weld together to the single downpipe. It’s common for them to crack there and the easiest solution is to buy some cheap stainless steel headers to replace the factory manifolds.

[jast]

I am in agreement with Zach, that I doubt tobacco smoke would cause any problems. However, I have never and have never known anyone who has, used this method to detect exhaust leaks. Slight induction of backpressure buildup is usually enough to bring them right to the surface (rag over the tailpipe and apply slight pressure).
Though it is not outside the bounds of possibility, I also don’t think this is likely to be your problem, as the exhaust is largely pressurized while running. Also, I should expect any leak to be readily apparent, especially if it were so large as to cause this type of O2 infiltration.

In case it helps with the logic, O2 sensors actually produce voltage in the presence of O2 (as opposed to otherwise “measuring the amount of O2”), which is why their voltage is higher when O2 content is higher. Also, every car that has ever had an O2 sensor ACTUALLY adjusts the mixture based on its feedback. That’s why they were put into the loop (little closed-loop operation joke there). Actually, that’s a lie, and here’s why: With OBDII, twice the number of O2 sensors were implemented as necessary for closed-loop “lambda” control. Why? Because the second set test for catalyst efficiency. Thus, the first O2 sensor on each bank (pre-cat) are used for mixture control, while the second sensor (post-cat) in each bank is used for catalyst efficiency check. The first O2 sensor in each bank will show a lovely rich/lean cycle a few times a second, whilst the second O2 sensor in each bank should show a very lazy rich/lean cycle if the catalyst is functioning properly.

from here

All that aside, IF you’re going to try the magic smoke, the most likely place for insertion is the tailpipe…

[semaphore1999]

Thank you all for the input, gentlemen.

And while I was certainly hoping that I was nearing the understanding level necessary to tie this project off nicely, I am realizing that what I am seeing here is an open can, worms crawling around everywhere. I did some quick study, and realize that there is aplenty to consider when it comes to exhaust, O2 sensors, catalyst, etc. I’m going to need to spend a bit of time poking around and studying this stuff a bit more to try to get a handle on where my problem is coming from.

@jast
Your pictures depict warmed up state running say 2k RPM or maybe even slight load. Idle or deceleration looks different, correct? Also, what Mazda specs for appropriate catalyst behavior is taking a moment for me to wrap my brain around:

And so I think I want to do something like this: I have a model of a properly operating Japanese sedan from same era in the 99 Lexus es300. So, using it as a model, I will capture traces for the Lexus O2 sensors under various conditions, then capture traces of Mazda O2 under same set of various conditions. Perhaps this will shed some light as to where my problem may lie?

Oh, and one last thing… I need to increase my sampling rate in torque big time, else these plots are quite misleading I think…

[jast]

In case it helps with the logic, O2 sensors actually produce voltage in the presence of O2 (as opposed to otherwise “measuring the amount of O2”), which is why their voltage is higher when O2 content is higher.

I don’t know where this came from (actually, I do, but it’s a long story I thought I’d repressed)…
In a traditional O2S (like those on a 2001 Mazda Millenia) higher voltage indicates richer condition, lower voltage leaner. I’ll steer clear of the mechanics of this for fear of muddying the water (or saying it all wrong, again), but here and here are a couple of sites which do a decent job of explaining it (they introduce some information superfluous to this discussion, but interesting nonetheless if you find yourself working with later model vehicles equipped with wideband sensors)

Anyway. That document you posted boils it down to rudiment pretty nicely, if I’m reading it right. It effectively has you check the output voltage of each sensor, and then induce lean condition (deceleration) low voltage (0-.5) and then rich condition (acceleration) high voltage (.5-1.0). Notice the “rear” sensors stay steady in the range (0-1.0) regardless of rich or lean.

Your pictures depict warmed up state running say 2k RPM or maybe even slight load. Idle or deceleration looks different, correct?

Warmed up, yes. But that should be correct for no load <2krpm. Most cars I’ve worked with look like that at idle, as well, though there is an argument that idle falls off the near end of the spectrum.

I need to increase my sampling rate in torque big time, else these plots are quite misleading I think

I think I would try that. I recall the cycling of rich/lean being somewhere in the neighborhood of 2-4hz, but the graphs I’m seeing online (such as those on the page I linked previously) are showing more like .5-1hz.

this image of yours seems to show roughly 4 peaks in 10 seconds, making it about .4hz. I’d be curious to see the results from an increased sample rate. May or may not be useful, but probably easily tried.

Also, based on what I’m seeing in these graphs you posted, it seems like all 4 of your O2sensors are cycling pretty high and often, which appears to indicate a catalyst deficiency. I’m having trouble deciding if I think that’s reliable information or not…

I like your idea of comparing to “known good”. Although it’s a 4-wire sensor, the Lexus should function the same.

[semaphore1999]

Lets take about a five minute break from those plots. You see I just discovered that I had bank 2 sensors upstream and downstream switched. The car is acting a WHOLE lot different now that I noticed and corrected this.

I will attach plots under various circumstances in the event this was the silver bullet, and of course I’ll include my codes and grievances if it is not… lol…

[semaphore1999]

OK. I know what happened. @jast, those plots are indeed correct that you were looking at, except that bank 1 sensor 2 is dead. When I fixed it (the first time, I did so by playing musical chairs with the other new sensors so I could get sensors with the longer wires where needed. Evidently, I did a poor job at musical chairs, because evidently I swapped some wires. So the plots look reasonable to us because they are. However, any plots after my ‘fix’ should have bank 2 reversed.

[semaphore1999]

And now I am going to throw in something that may or may not be correct. I note that at no time has my ECU thrown a catalyst code, nor has it passed them as inspection ready either, but I have not let it run with 4 good sensors hooked up properly yet (and never mind the fact that I found my air filter was leaking in air (American Spirit testing…lol) and also that there was NO form of exhaust leak at all anywhere.

But I think my cats are OK, but what do I know?

Hear me out, nevertheless, because I think this may have merit. OK, so a lot of auto configurations include O2 sensors pre cat, and a single O2 post cat, like waaaayyy down the pipe near the muffler even. I would expect a situation like this to behave as you guys have been describing.

However, the Mazda has pre and post cat O2 sensors both immediately before and immediately after the two cats. So while Lexus and Toyota stuff usually has 3 cats on similar sedans with the post cat on the combined pipe, the mazda is four sensors two per pipe prior to combining. Now think about how one side is longer than the other, combination of gasses out of phase maybe 180 degrees or so. Hmmmm. I would expect 3 sensor system to behave as you have described. However, I would expect 4 sensor system like mine to do a reduced amplitude sort of thing, which is what I think the Mazda doc is implying (and what I have seen mine do when hooked up correctly).

Am I making stuff up, or is this a bit reasonable?

[kbraby]

Don’t think about it as flow based. It is based on the active adsorption capacity of the catylitic converter. Replace the cat with a straight pipe, and the two will follow close together. Put a good cat in, and you get smoothing based on the surface of the cats storing reactants. Put in a bad cat, and it is somewhere between the two.

[jast]

So while Lexus and Toyota stuff usually has 3 cats on similar sedans

I think you mean 3 O2S on similar sedans…

Now think about how one side is longer than the other, combination of gasses out of phase maybe 180 degrees or so. Hmmmm. I would expect 3 sensor system to behave as you have described. However, I would expect 4 sensor system like mine to do a reduced amplitude sort of thing, which is what I think the Mazda doc is implying (and what I have seen mine do when hooked up correctly).

I don’t follow. The amplitude will be affected by O2 levels in the exhaust stream and little else. The Mazda doc appears to be implying exactly the behavior the other docs are describing: pre-cat will see 0-.5v for lean conditions, as what happens in a coast down, and .5-1.0v for rich conditions, as when under acceleration. Post-cat should be 0-1.0v regardless of situation.
I’m not clear what you’re seeing. From the graphs posted, you appeared to have (had) a “dead” O2s, at about 0.0v. They should be at .45v “base” if they’re working. Then the other graph you posted with all 4 cycling makes it appear the cats are deficient, as noted by the cycling of the post-cat sensors. They should be noticeably slower to cycle, and lower amplitude, as the O2 levels exiting the cat should be fairly steadily low (as the O2 is consumed in the reactions the catalytic converter supports), but with upticks as the stored oxygen is consumed and replenished in the trim cycles. This should be true regardless of where they are hooked up to the ECU (though it certainly might confuse the ECU if they are switched, front to back especially).

Am I making stuff up, or is this a bit reasonable?

I think you’re making stuff up.
I don’t think the exhaust paths are different lengths, or not significant if so. I also don’t think whether it’s a 4 sensor or 3 sensor system will change the way each O2S should function given its location in the system: post cats’ll always do what post cats do, and pre-cats’ll always do what pre-cats do, regardless of how many of them there are (which DOES leave me wondering, at this point, WHY they choose to use a sensor per bank of cylinders on a single cat system when a single sensor just before the cat would probably work equally well, but that’s really another topic, I think).

[Raymond]

I found my air filter was leaking in air (American Spirit testing…lol) and also that there was NO form of exhaust leak at all anywhere.

If extra air is seeping around the filter it’s only an issue if that is between the intake sensors and the engine. If the sensors are between the air filter and the engine then it’s not a problem for the ECU.

[semaphore1999]

OK, I would like to cite general connectivity goofiness as a prime suspect in the previous postings.

So now, O2 sensors are hooked up correctly, and I took the vehicle, now generating no faults, no codes, on a warmup drive. Then I pulled in to gas station, pulled out torque. This is how the O2 sensors are responding at idle in drive now:

[jast]

Are your readiness codes now gone as well, or still waiting to see if those are happy?

Those traces make it look like your cats are lazy (the green trace for bank 1 sensor 2 is too active) and the trace on bank 2 sensor 2 is oddly high, but if the readiness goes green, and it passes inspection…

[zmetzing]

Have you considered doing a leak-down test to check for any head-gasket funniness?

1999 Jeep Wrangler, 4.0L Misfire Code - but found low compressio

Ya just never know. My BiL had a dead hole on an otherwise flawless car @ 80k. Fixed a burnt exh. valve and it worked another decade + and another 100k+miles. As always, Eric the Car Guy is more thorough (and more accurate) than I: https://www.youtube.com/watch?v=WgrfT0LFMhc

[semaphore1999]

I am not getting codes or faults anymore, and I have driven a significant amount, so this is a first. The car runs like a dream now that I got my little wiring concern figured out and have no vacuum or exhaust leaks. I sealed up the oil pan leaks using a felpro gasket, and when I did I changed the oil. The slightly used oil was a beautiful color I would expect to see from a brand new engine. No evidence of any coolant in it, however I did note a slight amount of sparklyness which I presume has to do with the graphite shavings when I scraped head gasket #1 off. I am quite sure the head gasket is sealed nicely due to the way the engine is behaving alongside the color of that oil. The lack of smoke or smells from exhaust is a big clue as well.

The difference is night and day between the felpro MLS in the car now with the prepared surfaces, and the previous head gasket attempt. It clear I’m on path to success this time, just need to get the state inspection to complete now. So I’m just sort of driving around right now waiting for the computer to complete its check cycles. I’ll advise should we see an issue, and my take some more screen shots for you guys to illustrate system health.

[semaphore1999]

So I had installed a bogus O2 sensor. It was the fat one you see at the auto parts store for cheap. I went back to replace it, and i wanted one I didn’t have to wait for ebay to deliver, so I went to the stores. They were out of these sensors at every auto parts store I tried in Dallas, and internet searches indicating no local stock.

Then I stumbled upon these NTK sensors (formerly a little known company called NGK). They were of solid top shelf Japanese construction and not well advertised on the auto parts stores websites. The person helping me didn’t know much about them, but I was able to go to the shelves in a certain store and there was a whole bunch of these just sitting there. And what’s more, is that they even had the correct plastic connector for my car on them. The wires were too short, but the connector was correct. The pinout, however, was not, but thats an easy fix given I have to lengthen the wires anyhow. Finally, upon careful examination next to one of the OEM sensors, I was able to surmise from the markings on the unit and the exact match mechanically to OEM sensor, that these sensors are in fact the very ones used by Mazda during assembly in Japan.

I installed, and reset my codes. Shortly, the sensor got flagged, but I am experienced and reset again. Now I have been driving for a good bit and I am not getting any fault or CEL code at all. All the diagnostics pass, and I am down to three remaining state inspection tests to pass.

Attached is another Torque plot of the O2 sensors I have currently installed. All of them are now correctly installed and of the OEM type.

[semaphore1999]

And if that last Torque shot was not enough proof that the Catalyst is good, how about this one I just got in. Notice that the catalyst and all O2 sensors have passed. In the past, I have seen EGR pass with the hardware on the system, but it takes quite a bit of driving around. I would think that it might pass sometime today, at which point I will make a b-line for the inspection shop, solidify things.

The car runs better than ever now. It seems that once I installed that NTK (formerly NGK) O2 sensor, any hesitation I was noticing upon acceleration went away. Furthermore, I had cause, or should I say desire, to go ahead and press the accelerator on down deep from a full stop. Let me tell you, the front wheels spun like a drag racer would his rear, unweighted tires. This went on until I realized my acceleration was minimized due to reduced coefficient of friction, so I let off for the tires to grab and me to take off. One tire is brand new, one is < 10k miles. I think this result is pretty giood for the ole Milly!

Remember that while the short block is new, this car is running with heads that were reconditioned ala myself utilizing knowledge/wisdom of DMS experts combined with a few great web reads. I did go outside the norm by following the youtube guy’s suggestion to resurface heads with large sandpaper on flat surface. Yes, I concur, this works quite well, and I was unable to fit .002 inch feeler under any portion of either head (once all surfaces carefully cleaned of all particles). So while I realize that my victory speech may be short lived, and I may pay the price for my arrogance with unexpected short engine life, I’m feeling pretty good about things at the moment, and not seeing signs of trouble at the moment.

As for most recent fixes, I spent a day earlier this week where I flushed the radiator 3 times using distilled water only. Then I refilled with 50/50 Zerex. This was because I noticed that once I had used the heater, the core let loose its stored sludge fury upon my unsuspecting rebuild, and I saw the temp guage rise during a test drive. Lucky I noticed before it went up much, as I know that overheating is the MOST dangerous scenario for head gasket installations due to the extremely large cylinder pressures.

And finally, I have patched up numerous oil leaks. That stuff just gets everywhere. I had to remove my oil pan because the liquid gasket I had used seemed to soften in contact with the oil and was leaking profusely. I put in felpro gasket and torqued to 13 ft/lbs, forget that 6 I saw somewhere. Consistency is key here, IMO. The oil filter attachment to new block is particulary troublesome. I had to make my own gasket out of that overly thick gasket paper I purchased. And I had leaks at both sets of cams on the side by the timing belts. There are these seals that get stuffed in, and seals around the cams with timing belt pullies. I don’t think I have this one completely resolved, but I think there may be a way to replace these without removing the cams.

And finally, I notced that the bluetooth was not working on the stock stereo in this 01 Mazda. Just because bluetooth wasn’t big back then, and nobody had an android in their pocket is no excuse to omit this important entertainment and cell phone safety consideration. I could literally die without this, or at least that was my justification for the expenditure… of vitally critical importance you see… So I fixed this with a new aftermarket double din 300 watt stereo, which prompted the replacement of the rear speakers (total was about $120 for Pioneer 400 watt speakers AND stereo, can’t believe how cheap stereo stuff is now).

Oh yeah, here is the promising screen capture telling me that the end of my laborious rebuild is near…

[jast]

Then I refilled with 50/50 Zerex

You’ll be especially happy about this when the outside temp hits 20F in a couple days.

flushed the radiator 3 times using distilled water only.

You’ll be especially happy about THIS when you’re 5 years and 100k miles into this overhaul and your radiator hasn’t corroded to death.

And I had leaks at both sets of cams on the side by the timing belts. There are these seals that get stuffed in, and seals around the cams with timing belt pullies. I don’t think I have this one completely resolved, but I think there may be a way to replace these without removing the cams.

I’m curious about this, but yeah. You don’t have to remove the camshafts to replace the seals in modern engines (i.e. anything that’s not using the old “rope” seals, and even THOSE didn’t usually require complete removal of a component). There CAN be a “touch” to it, largely amounting to “don’t roll the lip” and “don’t sprong the spring”.

I put in felpro gasket and torqued to 13 ft/lbs, forget that 6 I saw somewhere. Consistency is key here, IMO.

No argument on consistency.

13ft-lb seems really high, to me. I would expect about 1/2 that, which is more in line with the 6 you saw somewhere (and which I have seen as well).
Here’s my reasoning for why that makes me uncomfortable:
these are usually “junk” bolts (i.e. no “class” or “grade” rating, or at the bottom of that scale, as is usual for automotive industry)
when I look for “standard maximums”, I find 14ft-lb to be about the maximum for the highest class of metric fastener in the size I suspect these are (6x1.0 or 6x1.25).
17.7 lb-ft according to this site
13 lb-ft here, pg 43
14.5 lb-ft here
(Sorry, can’t find my way into the Mechanical Engineer’s Handbook, which should likely be the standard here)
Also, these are threading into aluminum, making it the weak point, I would think, and repairing these threads is a royal pain in the ass.
And my final reason, that just sound high, to me.

But y’know what? If it’s working for you, call it a day!

Good luck on getting the next Readiness complete!

[semaphore1999]

Good point about the 6X1.0s I torqued to 13. None of them broke, which seems fortunate to me looking at those numbers you sent, but they didn’t and I don’t think its the thing to do to loosen them.

I may have gotten away with it this time, but it seems I remember breaking some bolts like this off on a job in a previous lifetime somewhere. Or was it a dream?

If a non-mechanic breaks a bolt, but nobody is there to witness it, is he in a forest?