Mondial 3.2 decides not to idle

A couple weeks ago, getting off the freeway at the end of a ~50 mi drive home, the engine died when I disengaged the clutch at the first stoplight, but started right back up seemingly normally. Half a mile later, it does the same thing at the next stoplight, and again half a mile after that at the final stop sign before my house. This seemed so much like it HAD to be a an idle switch not closing, that I let confirmation bias get the better of me; hurriedly probing the pins on the idle switch and not seeing continuity, I went ahead and spent $50 on a new idle switch (BMW cross-reference, so not Ferrari pricing, thankfully), installed it, and drove the car again (without testing the old one off the car to realize that I'd botched the first measurement). Got where I was going with no problems. Then, on the way home (20 mi, mostly freeway), it does it again! Dies at the end off the off-ramp, restarts just fine, dies at subsequent stops over the last mile to my house. (Although upon the final re-start to pull into the garage, it was apparent that it wasn't actually idling entirely normally—idle speed definitely a bit low.) So what else can possibly cause this behavior? (Note that the car is very low on fuel at the moment—probably lower than I've ever had it [except the time we drained the tank and refilled it with just a gallon and it died on the steep downhill between my house and the gas station], but I figure this can't possibly be related, because the car obviously had ~40 mi worth of additional fuel in it the first time this happened!) Heat soak seemingly must have something to do with it, but what's being affected that could cause this particular peculiar behavior? Would this be the expected behavior if the K-Jetronic ECU just died entirely (i.e. frequency valve not operating, mixture super lean)? An ignition problem? (Note that I recently replaced the ignition modules, distributor rotors, wires, extenders, and plugs.)

Thanks in advance,
Joe

 

Maybe you're on to something; the plug I pulled is definitely darker than I anticipated. That said, I'm not sure it's dying-rich-black, and I wouldn't call it fouled, per se:
http://joeshouse.2y.net:8080/~jelliott/IMG_7524.jpg

I just went through the records again and I don't see any evidence of the fuel distributor ever having been rebuilt. Is it really plausible that an internal leak in the fuel distributor would only manifest itself when things are really thoroughly heat soaked (i.e. fine after a ~30 min freeway drive, but only problematic at the end of a subsequent ~30 min return trip)?

The car is a USA-spec '86, so it has a catalytic converter and oxygen sensor. I replaced the oxygen sensor earlier this year, and checked the duty cycle on the frequency valve to satisfy myself that the system was working properly and hadn't been mis-tuned by previous owners ...and now I can't find my notes with those measurements, but they seemed reasonable (for example, there was no indication that the system's authority was saturated in the lean direction as though it was fighting an erroneously rich mixture).

Car starts instantly, cold or hot. (Note that this was not always the case before I replaced ignition components earlier this year.)

 

Yeah, I'll definitely repeat that measurement next time I drive the car, but I'm still struggling to imagine what could be different as a result of a few hours parked, such that it does it on the way home, but not on the initial half of the round trip (which was long enough for things to be thoroughly warmed up, such that I'd expect a warm idle problem to manifest itself when I got off the freeway on the first leg of the trip, rather than exclusively on the way home). (And, for what it's worth, I found what I'm pretty sure are my notes from when I performed this measurement in May—47% closed loop, 51% with sensor disconnected, 59% with WOT switch shorted.)

Funny that you should mention that—my fuse box is, indeed, heat-damaged in that area, such that I was frankly surprised it ran at all—I've since replaced the whole connector (and affected pins) and added a relay for the pump so the fuse panel is no longer carrying all that current. But I would expect any issues there to manifest themselves when the car is at speed—at idle I imagine there's enough fuel in the accumulator to keep the engine running for a few seconds if the fuel pump is intermittently cutting out.

Thanks for the tip; if I can think of a non-messy way to depressurize the fuel system first, I'll try this both cold and warm. (Do you happen to know the correct torque for reattaching the fuel line to the fuel distributor?)

Here's a photo of another plug:
http://joeshouse.2y.net:8080/~jelliott/IMG_7526.jpg

Did you notice any kind of weird temperature and/or heat soak dependence when your car had similar symptoms? That aspect (i.e. it only does it on the way home, not on the first leg of the trip), plus its willingness to restart every time, is what's really baffling.

Thanks!

 

Well, after a surprise business trip and a number of other distractions, I've finally gotten back to this, and performed the check of the fuel distributor as recommended—fuel did NOT rise in the open outlet port of the fuel distributor when the pump was running (at least not with everything stone cold). The fuel level was sitting maybe 8 mm below the port. A couple minutes after shutting off the pump, however (i.e. after I walked away from the car to get new copper washers for reassembly), the fuel level had risen to the top of the port—is that normal? How plausible is it that everything works properly when it's cold, but behaves so differently when heat soaked and fully warmed up a second time that it dies rich at idle following decel?

I hope to drive it Saturday to re-check the frequency valve duty cycle to see if its authority is saturated in the lean direction (or otherwise abnormal) after it's fully warmed up.

 

Well, I drove it Saturday (this time with a full tank of fuel), another case of a ~20 min freeway drive, sitting for a few hours, and then the same drive home. And sure enough, IT DID IT AGAIN, but only at the end of the return trip, just like before. But not as definitively this time–it didn't die at the end of the freeway off ramp, only at the stop sign a block from home (restarted fine), and then again as I was pulling into my driveway. But here's the frustrating part: after restarting again and pulling into the garage, I couldn't get it to misbehave!! I was all ready to go with my timing light to verify ignition functionality on all cylinders and my multimeter patched into pin 15 on the ECU to check the frequency valve duty cycle, but after that second restart, it just kept idling fine! I obviously couldn't simulate load/decel in the garage, but I revved the engine a bunch and closed the throttle abruptly, but it carried on idling just fine, and the frequency valve oscillated right around a perfect 50% duty cycle closed-loop, and was steady at 59% when I shorted the full-throttle switch to force it open-loop. (And, interestingly, the idle was no worse or maybe even a bit stronger when I forced it open-loop, which suggests that I'm dealing with something other than an excessively rich mixture.) What the heck? The ability to instantaneously (but intermittently) cure itself suggests an electrical issue, in my mind, but what?

Given that my mixture seems nominally perfect (i.e. that the Lambda feedback system runs the frequency valve at exactly 50%), I'm that much more reluctant to play with it. But first I'll have to figure out of the Phillips-head screw jammed in my mixture adjustment is just meant to be an ersatz dust cover, or if turning it will actually turn the adjustment (and, if the latter, I'm afraid I'll never be able to get the adjustment back to where it was, after removing the stupid screw)!

 

Yeah, any kind of fuel delivery problem would manifest itself more obviously at high engine load as opposed to decel or idle.

I'll definitely take a look at this, but if the vacuum of decel opens up a leak in that hose, the timing would be retarded to the full-throttle advance curve, which I wouldn't expect to kill the engine or affect the idle too much—on older distributor ignitions we set the static timing at idle with the vacuum advance disconnected, after all, and I don't recall the idle changing noticeably when doing so.

Oh gosh—unintended EGR from a busted air injection check valve—is that a known failure mode?? I might expect that to be more apparent under load (pressure in the exhaust manifold higher), but maybe not, given the intense vacuum that's present in the intake manifold on decel. I should probably take a look at this too, but this is one system I'm not terribly familiar with—is it really as simple as it sounds—electrovalve opens, and the exhaust scavenging effect is so strong from Ferrari's flat-plane V8 that the little transient suction pulses in each exhaust manifold branch are strong enough to overcome the intake vacuum and draw little gasps of air into the exhaust?

Yeah, certainly no white smoke or unusual smells. And I spent enough time standing behind the car, idling in the garage Saturday afternoon, staring at "50%" on the multimeter and cursing myself that the problem went away just as I got home, that I'm pretty sure I would have noticed!

Yeah, nothing worse than an intermittent problem! I'm curious if my symptoms are consistent with the ECU just dropping dead (and the engine thus going super lean due to lack of frequency valve actuation). Things did get quite hot back there when I had my single-bank ignition failure back in March, and I'm pretty sure it was only after that, that I found and repaired the peeled-back bit of insulation in the ECU compartment, so I was actually a little surprised that my ECU wasn't totally fried then! So I can't help but wonder if it did suffer some damage that only becomes apparent after heat soaking. (Or maybe the culprit is that dang Porsche-branded relay that powers it—the fact that restarting the car [i.e. turning it off and on again] seems to briefly cure it seems entirely consistent with a failing relay, and I know that the identical relay that powers the ABS in my Porsche is one of only two things that's failed multiple times in the 16 years I've owned that car.)

Looks like that last sentence got cut off—the suspense is killing me—were you about to tell me what's actually wrong with my car?

 

Ah yes, the classic Bosch "damper that looks like a regulator" trick. But what's less clear to me is whether you're still saying there's a failure mode associated with it that I need to rule out. Related question: I was pleasantly surprised to discover that there are still a number of purpose-built K-Jet fuel pressure test kits available via Amazon, etc.—is there a particular product that people recommend?

Thanks for the cross-reference! I'll buy a couple of those right away, if only to keep on the shelf!

What would the symptoms of a failing protection relay look like?

Already did that!

The fact that the symptoms go away (and don't necessarily recur, based on last weekend's experience) when I turn the car off and on again is hard to reconcile with internal leakage—is there anything that could be binding up in the fuel distributor that would be effectively 'reset' when the fuel pump is turned off and on again (bearing in mind that the system is designed to retain pressure with the pump off, so it would almost have to be something flow-dependent).

Where did you send your distributor for overhaul? Or did you do it yourself?

 

UPDATE:
I drove the car again today, this time with a multimeter in the car, wired to pin 15 of the K-Jetronic ECU to monitor the frequency valve signal. And sure enough, the system seems to saturate in the rich direction (3% on my meter, which I presume is actually 97%, since full-throttle open-loop operation shows 40% rather than 60% on this meter), suggesting that it's reacting to excess unburned oxygen in the exhaust. (So this explains the slightly rich-looking plugs, if it's reacting to something other than an actual lean-running condition.) But I could not get it to stay running long enough in this state to get it into the garage, and upon restarting the car it's fine!

So, upon realizing that the inductive pick-up lead on my timing light is just barely long enough to reach from the engine compartment to the passenger compartment, we set out again, this time with the timing light (attached to the aft coil wire) and my girlfriend in the back seat. After a few minutes of driving around, we got the issue to reoccur, but with no change to the timing light signal. So we moved the timing light's pick-up to the forward coil wire and, bingo, erratic timing light flashing corresponded to the 97% frequency valve duty cycle! What's interesting is that she said it never stopped flashing entirely (until the car dies at idle, obviously), but rather flashed erratically when the issue was occurring, as though it's not a complete failure of the ignition on the forward bank, but rather something causing a weak spark or loss of spark on less than four cylinders.

Of course now I'm regretting not replacing the distributor caps when I replaced nearly everything else in the ignition system six months ago, but my next step is going to be to switch the coil assemblies between banks and see if the problem follows the coil. (Note that the coils themselves are the other thing that wasn't replaced.)

 

Well that didn't work. It seems like the erratic timing light behavior on the forward bank correlating the 97% frequency valve duty cycle was a fluke. When we got back on the road with the coils swapped, at first it seemed like the problem followed the coil, but then the timing light seemed to flash erratically in a way that didn't correlate to anything when we swapped it to the other coil wire. The lead is only long enough to reach cylinder #4 individually (with the timing light in the passenger compartment and the hood closed), but when it was hooked up thusly there was no interruption of the light when the car started acting up.

And then we realized that the way it was coming and going, jumping from 50% duty to 97% and back to 50%, we would have noticed an instantaneous kick of acceleration or deceleration every time it happened at steady state cruise on the freeway, but we didn't. That—and the rich-looking plugs on both cylinder banks—make me think that it almost has to be something in the fuel ECU, i.e. the change in duty cycle isn't a response to unburned air in the exhaust, but rather erroneous behavior of the fuel injection system itself. (And not the protection relay, either, because—if I can trust my meter—that would produce a 0% duty cycle, not 97%. I also confirmed that when it's acting up on the freeway, flooring the throttle makes the duty cycle jump to 60%, as it should, and reverting to part throttle went back to 97%, in this case.)

So now I think my next step should be to wire up a switch to short the full-throttle microswitch, and next time it happens, if I can keep it idling while it's misbehaving (which I was able to do a couple times today, by tickling the throttle), I'll flip the switch: if it continues to try to die, we'll know that there's something else wrong and that the rich duty cycle was just a response to unburned air in the exhaust; if the idle recovers, we'll know that the rich duty cycle is the actual reason it wants to die. In the latter scenario, does that necessarily mean I have a bad ECU, or would a fault in the oxygen sensor wiring produce this behavior? (I know it's supposed to detect a disconnected sensor and revert to an open-loop mode of operation in that case, but I seem to recall that if you short the oxygen sensor wire directly to ground, that will produce like a 95%+ duty cycle and make it too rich to idle...hmm.)

 

Success! Shorting the full-throttle allowed me to prevent the car from dying at idle, i.e. the super-rich mixture is the problem, not a system response to another problem. So I could just drive it around like that indefinitely, but if I'm going to go that route, I should really try the other open-loop mode, i.e. disconnect the oxygen sensor, because that should make it run open loop at nominal mixture, whereas what I've done is making it run open loop in the richer-than-nominal mode intended for full throttle mixture enrichment and super cold starts.

But, of course, the question remains as to what's broken—as much as I'd like to think that Amazon sold me a defective oxygen sensor earlier this year, I'm increasingly fearful that it's the K-Jet ECU, since turning it off and on again temporarily resolves the problem. My cursory research suggests that my ECU is not the same as an earlier 3.0 car (even though all it has to do is modulate the same frequency valve in response to the same oxygen sensor input), limiting my opportunities for troubleshooting or inexpensive repair. Does anyone know if the car will run properly with a 3.0 QV 0 280 800 063 ECU? Anyone in the Seattle area want to loan me a correct 3.2 ECU for troubleshooting?

 

I actually haven't confirmed the Bosch part number for my ECU (Ferrari part number 127479, pretty sure that should be Bosch 280 800 068), but the ECU from a Swiss car, especially a later Swiss car with KE3, would certainly be very different.

Connector has been inspected/cleaned/reseated. I've never heard of an analog Bosch ECU failing either, but I experienced an ignition failure on one bank earlier this year, which ended up with the exhaust system glowing red hot... and only subsequently did I discover and repair a bit of displaced insulation in my ECU compartment. So it's safe to say that those 35-yr-old circuit boards were exposed to some pretty excessive temperatures four months ago.