Ed Scherer

I cleaned both of the rear grounds at the back of the V during the intake refresh. Nice and clean for sure. But... that was just the terminal-to-bolt/washer surfaces, and as Chris points out, there's still the possibility of a bad/corroded crimp on one of those ground terminals. FWIW, I've never seen any evidence of or experienced (to my knowledge) any corrosion on any of the connections on the car, with the exception of some ignition wires a couple of times (probably due to water intrusion).



OK, some more information from this morning.

I took a closer look at Temp II sensor. It looked like the connector might not have been seated absolutely all the way, so I disconnected it, took some measurements, and reconnected it. First, I checked its metal body to chassis ground (crossbrace mount), just to make sure there wasn't a grounding problem: 0.2 Ω; good. Then measured each pin to same chassis ground: 2.71 kΩ and 2.62 kΩ, respectively, at ambient air temperature of approximately 60 °F. Reconnected the sensor.

Started the engine cold (O2 sensor still connected, ARM1 still connected). It ran past the magic 75 second period, but stalled after maybe about 10 to 15 minutes (I was away from the car and timer at the time). While it was running, it was surging a little. Perhaps this was due to the fact that it hadn't gone through an idle adaptation since the last power disconnect. I noted that during this warm-up period, the ARM1 was not hunting and was showing full lean most of the time (at least during the first few minutes when I was monitoring it). (Note: there's a slim possibility that the ARM1 wasn't properly reconnected, as I had removed and ultimately replaced the quick-connect terminals last night). I didn't supervise it the whole time, so I can't say too much more about that. And I can't remember whether that (running lean during warm-up with no hunting) is normal or not.

Restarted the engine and a timer. 75 seconds later, it stalled. And again... 75 seconds later, it stalled. And another time or two. On at least a couple of these stalls, I was watching the ARM1 and it was showing hunting cycles from full lean to/from full rich at reasonable intervals (a second or two; I didn't time it). One of the stalls clearly happened when the ARM1 was indicating rich.

The stalls have an almost "electrical" feel to them: it's just like something commands "stop!" Very abrupt, and with no warning (other than the fact that I've got a timer that's about to hit 75 seconds ). The thought "alarm circuitry?" crossed my mind, but there is (and never has been) anything after-market added, and I'm not aware that a '90 928 has anything alarm-related that will actually stop a running engine (does it?) And even if it did, I don't know what that would have to do with temperature.

I should have (but didn't; needed to get to work!) measured Temp II signal at the LH connector with the engine at operating temperature; maybe there's something intermittent going on (like a ground starting to float) that only happens at higher temperatures.

This 75 second delay (which only seems to apply at higher temperatures), though, has me baffled. It seems like something is kicking in (or out! ) at that point, and I'd really like to know what it is! How long does the O2 heater run? Under what circumstances does it heat? Every start, even if engine is at operating temperature?

I'm torn between just sending off my LH for replacement or continuing to do testing. I'm not really in any huge hurry or anything; I've got another vehicle to drive if the 928 is down for a few days.

Anyway, at this point, it seems to me like the two most relevant observations are that temperature appears to play some role and that there's a 75 second delay after startup involved. That 75 seconds is so consistent and so not random that I really need to figure out its relevance.

Unless someone has a better idea, I think I'll take a close look at the input signals at the LH connector (and maybe EZK, while I'm at it) with the engine at operating temperature next time I get a chance (tonight or tomorrow).

John Speake

It's logical and normal for the time to fail from cold would be 15mins or so at idle speed. The O2 loop isn'ty activated until the engine water temp gets to about 80degC (this value varies somewhat depending on MY)

When the engine is warm there's no delay until the O2 sensor is used for the O2 loop that I am aware of.

You have confirmed that by observing the O2 voltage hunting back and forth well before the majic 75sec.

I am sure Louie will loan you one of my rebuilds as before, for cost of postal charges. Then you won't need to pay rebuild price out if not required. At least now you seem to have a repeatable method to check whether fault is present or not.

I must say it doesn't sound like a failure of the O2 loop circuits in the LH, that usually shows up as a faulty A/F ratio (i.e. not ruinning closed loop, like the example Bill Ball posted a link to)

There could be a thermal problem with the ISV loop circuits in the LH I suupose. A sudden glitch of the ISV could easily kill the engine at idle rpm. Do you have anyway of datalogging a few signals ?

Ed Scherer

Not really, unless I break out the old breadboards and A/D converter chips that I've got stored away somewhere...

I do have a SharkTuner II and DT999, if those would help. And an oscilloscope (non-storage) and DMM that can log one signal; probably not too useful for correlating stuff.

I'd be willing to pick up a new diagnostic device, if it's called for. Can never have enough of those.

John Speake

If you have PEMs in both LH and EZK, you could data log from hot start to stall. Use highest speed option on datalog.

This may give some clues ...

Ed Scherer

I haven't put the PEMs in yet, as they prevent me using the DT999. But now that I think I've gone about as far as I need to with the DT999 (which seems to be much better for engine off testing than engine on testing), maybe it's time to put the PEMs in.

Hmmm...

Maybe I'll give it a try. Feel free to recommend exactly what/how I should log; I haven't fired up the SharkTuner yet, so I'm not that familiar with it. If it's obvious, I'll figure it out when I read the manual again (it's been at least a month since I skimmed the manual). Feel free to tell me to just RTFM.

And thanks for your help so far... you know while asleep way more about the engine controls than I do while awake.

John Speake

Fit PEMs, load new files appropriate to your model (S4) to LH and EZK. Afterstarting the car and admiring the System Monitor Screens for both LH and EZk you can then just go straight to the datalog screen, select all paramters to be logged at 20 samples/sec, then just select "start logging".... and "stop logging" after stall. You can then observe the results in tabular or graphical form. You should be using ST2 software Ver. 6.0.13

Maybe worth trying the test with/without O2 sensor plugged in..... and observing any noticable differences.

Ed Scherer

OK, John, I think I'll give this a try within the next few days (unfortunately, my time is really limited until tomorrow evening).

You sent me ST software v6.0.13 back in August, so I'm good on that.

Oh... and I'm trying to remember... are the PEMs preloaded with anything? (since you mentioned loading new files)

And I'll hoist a single malt in your honor while admiring the System Monitor Screens.

borland

On your 90' S4, you can bypass the factory alarm control unit's EZK relay disable feature. All you need is to install a jumper wire between two connectors on the CE panel.

Ed Scherer

Today's update: this morning before work, I installed the SharkTuner PEMs in the LH and EZK brains. Discovered that my crappy old garage notebook computer only did USB 1.1 and wouldn't talk reliably to the SharkTuner II and my wife had taken her (very new and fast) notebook computer to work, so I decided to wait until this evening to try it out.



Once I had the fast computer with USB 2.0 to work with, everything went smoothly with the SharkTuner software. Updated the ST2 firmware to the version needed by the latest ST2 app software and the ST2 app and ST2 were happy.

Loaded the baseline S4 maps (which I'm assuming should be similar or identical to what was in my EPROMs before) with no tweaking at all.

And... now I can't get the car to stall anymore.

The only two anomalies I noted:

1. After returning from a couple of test drives (with ST connected and data logging), the idle stuck at an unusually high rate: 970 RPM or so (while in park). Then I'd restart, and it'd be back to normal, e.g., 650 RPM or so. And after the restart, I'd let it sit and idle for awhile... the same conditions that caused it to consistently stall after 75 seconds before. And it just kept running.
2. If I stabbed the throttle, it would hesitate before revving. I don't think it did that before, but I'm not sure. Based on what I read in the ST manual, I can probably tune that problem away.

And, for the record, I did change one other thing: while I had the LH and EZK brains out, it was really easy to pull the fuel pump relay (it's hard to get out when blocked by the brains), so I replaced it while it was convenient.

So... once again, I'm not sure what to conclude from this.

I'll drive the car to work tomorrow and see how it does.

Two questions:

• I don't think I can use the DT999 with the PEMs installed in the LH, so how do I invoke an idle adaptation? I don't fully understand all the ST-based tuning yet, so perhaps some of the fuel/ignition parameters you can set via the ST are the same things as what otherwise gets set by idle adaptation.
• I don't yet have a WBO2 sensor installed yet (a TechEdge 3A1 is on the way, though), so what is the ST logging for the A/F ratio (nothing is plugged into its WBO2 input)? If you just feed the LH from the NBO2, can the ST log what the LH is seeing from the NBO2?

I've got a lot of reading to do on ST-based tuning, obviously.

Bill Ball

Earlier you had other symptoms besides stalling, such as low idle, bucking and some throttle issues, so I discounted relays as a cause. Lately, at least in the last few days prior to the PEM install, all you mention is stalling and you describe it as an immediate OFF. Did the low idle and bucking go away?

I see now the stalling is gone but you had some high idle and some hesitation. On a stock motor, you should not need to SkarkTune out hesitation. It shouldn't hesitate if all is well.

Ed Scherer

Part of the apparent changes in reported symptoms or my emphasis on different issues might be because at times I was driving the car (like earlier this week) and noticed driveability problems with lots of different throttle input, and other times (like the last couple of days prior to this evening), I was mostly just idling it in the garage (with no throttle input) after noticing the 75 second delay stall issue. I got pretty focused on the stalls and the repeatability was encouraging (in the sense that I was hoping it would make diagnosis easier), but even when trying to diagnose that, I was noticing some minor idling problems like some surging during the warm-up period. The high idle I saw this evening is definitely something new. And in probably 45 minutes of run time (probably about half driving and half idling in the garage), it was not bucking or slow-idling like it did earlier this week.

Not having much experience trying to trouble shoot these kinds of problems (and after a time or two of thinking the problem was solved), I'm pretty reluctant to draw too many conclusions too quickly, especially when it's so hard to control all the variables. For example, I just don't know what to expect after removing all power, which clears the idle adaptation settings. What is the expected idle behavior after such a reset, and before running an idle adaptation? Does idle adaptation only optimize idle speed, or can it also fix idle surging? I also have zero experience (prior to today) with the SharkTuner, so I don't know if its presence (even with the stock maps loaded into the PEMs) has any side effects. And can I be absolutely sure that the maps and other fuel/ignition parameters in the LH and EZK now are the same as they were before I put the PEMs in? Well, some of the parameters are certainly different, since I reset the idle adaptation stuff by killing power for awhile, right?

I'd just like to get something consistent to happen for a day or two again, and then I can focus on whatever problems remain. It would help to actually understand why things are happening the way they are along the way. When I drive it tomorrow, it'll be with the PEMs installed and the SharkTuner disconnected. If I start noticing anything especially interesting (like bucking or stalls) happening again, I'll hook up the SharkTuner, turn on logging, and hope I can catch it in the act of misbehaving and get some data.

Ed Scherer

BTW, another thing that would be interesting is to see some ST data logs from a known-to-be-good car so I'd have something to compare to.

I'd be interested in seeing a log with most of the variables for an entire warm-up period.

In skimming the log from a cold start all the way to operating temperature, one thing that jumped out at me was that the O2 sensor adj [%] bottomed out for quite awhile during warm-up and then, for some reason, around 10 minutes in.



Remember that when I had the repeatable 75 second delay stall problem going on, disconnecting the O2 sensor and doing a power reset and then running without the O2 sensor made the stall problem go away. So is this logged data further evidence, perhaps, that there's something wrong with the either the O2 cabling or the LH's handling of the O2 sensor signal?


Oh... and that reminds me:
For the record, I did spend a few minutes just admiring those screens.


I'm done for tonight and will check back tomorrow.

John Speake

The A/F readings on the ST2 will of course not be anything but noise, as you have that flying input jack lying loose somewhere. However the O2 adjust you showed a data log graph of is an accurate depiction of what is going on with the O2 loop. So something strange appears to be going on there, certainly after 10 minutes warm up.

You are correct that you can't use your diagnostic tool to reset idle adaptation. But this should happen within a couple of minutes of a battery reset. It is the O2 loop adaptation that take 50 miles or so to adapt. The tool can't do that.

You can however observe "idle adaption %" on the Screens, so I suggest you look at that for clues. You can also switch off adaptation, which some people do on a permanent basis.

Ed Scherer

Drove the car some this morning, without any diagnostic equipment connected.

No anomalous behavior at all that I could detect (well, maybe with one small exception: read on). Good idle, even after I was done and let it idle in park for awhile; it didn't do that fast idle thing it did yesterday a couple of times after driving and then idling in park, with the SharkTuner still attached and logging. No bucking. No stalls or near stalls. No surging.

The only thing I could do that seemed not quite right was a quick stab from complete throttle closed to wide open; there's just a little delay (like maybe a .1 to .25 second) and hesitation before it starts revving.

I won't be convinced that all is good, though, until I get a few hours of driving in, which I'm unlikely to do until next week.


Entertaining observation for the day: remember back in post #101:

and my follow-up in post #104:

Well, I got my replacement air filter yesterday. Popped out the old (installed upside-down) one this morning, chucked it in the trash, and put in the new one, and then thought, "That's weird, I could swear the arrows on this one are different." So I retrieved the old one from the trash, laid them down side by side, and... well, check this out (older one on left, new one I just got on right, both oriented with "top convex side up"):



Exact same manufacturer and model number (Mahle micro-star LX 311, but arrows are reversed)! Nice! So the moral of the story is: don't try to figure out the arrows, just always put the convex side up!


The flipped filter probably wasn't exactly helping assure proper air flow, I'm guessing. I'll admit it was a pretty damn amateurish mistake.

Landseer

I can't imagine it matters much.

The porous media is highly likely, itself, to have directional tendencies.