Too lean idle

Joined
May 17, 2021
Messages
30
1993 Mitsubish Expo LRV, 1.8L engine, 230,000 miles. I've had it the last 50,000 or so, driven gently. Gradually developed rough idle though not a single cylinder miss. Compression down about half the allowed amount and somewhat uneven; I put on a rebuilt head. Compression fine, idle somewhat improved but not okay.

02 sensor (there is only one) replaced.

Classic air leak problem, right? I went looking, found leaks at manifold flange, throttle body flange, throttle shaft -- all confirmed by spraying with starter fluid and/or sloshing with water. All fixed in the obvious ways and those methods no longer find anything.

Car runs fine off idle: This is now only an idle problem. It's fine when cold, equally okay when just warm enough for closed-loop operation, but when full temp -- say 10 miles on a warm day -- it's rough when you pull up to a stop light.

This a 'federal' car -- 93 had very little emission stuff. I've been all through the few vacuum hoses, nothing found. PCV valve replaced -- that improved things some but still not okay. I have two spare ECMs for this car; performance is the same for all.

I can read the OBD-1 and ECM data: The internal numbers for intake air temp, barometric pressure, and air flow are correct and/or in specs when at idle. Injector pulse width and duty cycle are in specs for idle. Fuel pump, strainer, and filter are all new. Injectors were replaced with used -- 'tested and flow matched.'

Fuel trims are all +5 to +8 %. Hummm ...

Best results (though not satisfactory) are obtained by backing off the mechanical stop for the throttle valve a bit from the correct factory setting. But the ISC steps number still tends to be low. These things suggest an air leak somewhere.

If you spray some starter fluid at the air intake the idle smooths right out for a minute or so. It really is a fuel/air mixture issue. but who is doing this?

I'm obviously missing something but darned if I know what it is. Been chipping away at this for nearly a year now ...
 
I would test any vacuum operated component with a vacuum pump, the hose may be okay the part it operates may not. Did you check the engine vacuum with a vacuum gauge? What is the reading?
make sure the vacuum hoses are all properly routed and non are mixed up (easy enough to do and it all looks okay).
 
I had a 92 with that same engine. As the miles pile up two things to watch, the throttle body will get dirty and stick. Keep that clean. The other was the EGR valve. I had mine replaced at some point. Both would exhibit a rough idle when they needed attention. On mine the throttle body got some fuel cleaner at each oil change in the later years.

The fuel filter might be another item to check.

That Mitsu 4G93 was a great little engine.
 
it is possible you're getting a lot of blowby when engine gets hot and it is pushing hot oily mess into the intake air. Try venting PCV to air and cap that intake connection see if it makes a difference.
 
I had a 92 with that same engine. As the miles pile up two things to watch, the throttle body will get dirty and stick. Keep that clean. The other was the EGR valve. I had mine replaced at some point. Both would exhibit a rough idle when they needed attention. On mine the throttle body got some fuel cleaner at each oil change in the later years.

The fuel filter might be another item to check.

That Mitsu 4G93 was a great little engine.
+1 ...my 02 F150 it was common for the EGR to get coked up and sticky. You could remove a cover and clean the passages using a coat hanger good as new.
 
I knew it was a good idea to post this question here!

Lazy, I stuck a digital thermometer into the intake air duct and got +/- a degree of what the ECM gave via the ALDL connector so I think that's not the issue. Pretty hot because when stopped the intake duct sucks radiator air but the numbers agree.

Kawl, I replaced the fuel filter within the last month.

The rest mostly require research which I'll probably do over the next week or so.
 
Everything after the MAF sensor is suspect for an air leak.

"Venting the PCV to air" is no good on a MAF system, it is exactly what you don't want-- unmetered air coming in.
 
Does it use MAF? I know mine got dirty on a 93 Escort and it would misfire when doing wide open throttle in acceleration only doing open looped air fuel control. At cruising it would be fine though.
 
Does it use MAF? I know mine got dirty on a 93 Escort and it would misfire when doing wide open throttle in acceleration only doing open looped air fuel control. At cruising it would be fine though.
Yep, it's MAF. There are some carbon deposits near the throttle shaft but slight, not enough to interfere with butterfly movement and no ports nearby. However in the next week or so I'll get in there and scrub clean.

MK I think the suggestion to 'vent the PCV to the air' is on the vacuum side, and plugging the vacuum hose. Which would show whether the air being drawn through the PCV at idle is the problem. It's an aftermarket unit "This fits your 1993 Mitsubishi ..." but who knows. I have one claiming to be the real deal (MD024719) on order. I will try the suggested test shortly.
 
Some further data. There are only three connections to the plenum: The brake booster, the PCV, and the vapor canister purge. I have now reproduced the problem with each plugged or closed with a finger. The PCV connection makes a slight difference (nothing like a fix) while the other two make no difference at all. There's vacuum at each but very little air flow at idle. This engine has no EGR.

I have measured the engine vacuum: 20.5" Spec. is minimum 18".

All the usual parts causing 'runs lean' problems have either been substituted or replaced: MAF (air flow, temperature, and barometric pressure) numbers are correct in the ECM and substituting another unit made no difference. Substituting two other ECMs made no difference. I've been all over the area with water and starter fluid; there were leaks at the TV to plenum joint, throttle shaft (both sides), PCV valve (stuck somewhat open -- even after cleaning, a new one was better).

The head was a reman; it definitely made things better and compression checked perfect.

Disabling one injector at a time they all had about the same effect on idle.

The problem came on gradually over a couple of years.

Last thing to change is the intake manifold itself. There's no leak on top or upper edge of head/manifold joint but no good way to check the bottom of that joint due to my lack of tolerance for shooting either flammable liquid or large quantities of water up over my face while lying under a hot engine. I'm halfway through the process of swapping, well along with the PITA of cleaning the gasket surface on the head. Really an easy job on this car ('93 Expo LRV with 4g93 engine) except for the cleanup.

Should have results today if I don't get rained out.
 
If the compression readings comparing all cylinders are uneven and are over ten percent from highest to lowest, you will not get a nice smooth idle out of it. Generally vacuum leaks of any significance will raise your idle speed above specs.
 
Progress ... Replacing the intake manifold made no difference. I thought that eliminated the last unreplaced/twice tested part that could cause lean idle. But I was wrong.

I started cycling through the spare ECMs again, this time not just doing a short functional drive test at a time when I still had unlocated real air leaks but sitting there looking at the scan data on a laptop. Ding-ding-ding ... instead of an injector pulse of about 1.5 ms (I can't find that note right now) it was 2.5 ms or so when hot. And sure enough the no-load idle is much better with those two ECMs. And all the other scan data makes good sense now -- the fuel trims are near zero, the O2 sensor jumps around 0.4V, etc.

So the original problem was both too much air and too little fuel at hot idle. I have had ECMs that claimed a good pulse width on the scan data but didn't deliver it to the injectors due to a failing driver transistor (find that with a 'scope at the injector) but never one that simply got the answer wrong.

This seems like a processor failure -- it's getting the table look-up wrong. Anyway that ECM is off to a repair place and I'm running better with one of the two spares.

Not running well because this ECM doesn't seem to be quite right either: When the A/C is on the idle is unstable and looking at the scan data I don't see consistent info about either the closed throttle switch or the A/C switch. The CT switch is mostly okay; when that one is wrong the idle air control valve doesn't -- it'll just sit at a wrong RPM. (But an ohmmeter on the switch says it's okay.) The A/C switch signal just seems dead which would cause the idle instability because the ECM doesn't know the correct number of IAC steps and keeps trying to go back to a normal idle setting, then discovering the RPM is way too low and opening up again ...

I'm awaiting the repaired ECM -- week or two I suppose. Probably in the spare time thus provided I'll change the oil pump and timing belt.
 
More progress -- Perhaps the end of the line for now.

On further study the various ECM switch values that appeared to be wrong were indeed wrong because the Evoscan program was looking at the wrong memory locations for my car. Correcting that made the idle position, power steering, and A/C switch bits correct in the Evoscan display.

I went over the intake system with one of the modern electronic 'mechanic's ear' microphone/amplifier gizmos. Lots of sounds but nothing like the hiss/whistle of an air leak.

I dragged out the oscilloscope again and looked at the injector drive pulses directly: All were correct. I pulled the spark plugs, they were about the same and arguably a little whiter than I'd expect. #1 was slightly darker than the others on one side -- that's where the PCV connection is.

The discussion of idle adjustments in the shop manual is correct but suffered somewhat in the translation from Japanese. Once I realized that these are perfectly normal field adjustments and reset both the throttle stop screw and the "fixed speed adjusting screw" (a separate screwdriver-adjust air passage) the idle improved. I had to crank the SAS down by over a turn. Car may have had several previous owners so who knows what that means?

Then I discovered another part of basic (early) closed loop fuel injection system theory: During prolonged idle periods there's not enough exhaust gas to keep the 02 sensor operating so the engine returns to open loop operation. That means that any attempt to study problems by prolonged idle is useless -- you have to drive it and then look. Might be five minutes or so before O2 fuel trim goes to '0' indicating open loop operation. And in the last part of that period the 02 sensor output is low making it look lean.

The most accurate read is while idling right after driving a few miles: Under those conditions the 'low' fuel trim was +9.4% which is probably the max the ECM will do. So -- it's lean to the limit of what the ECM will try to deal with when idling. 'Mid' and 'high' fuel trims were also + but under 5%.

After roughly 1994 Mitsubishi adopted heated 02 sensors. Those come online much more quickly and stay functional during idle.

The above steps improved drivability considerably but with 'low' fuel trim still at the limit and idle still so-so the problem was not gone. And -- except for a possible non-obvious issue in the ECM (I'm awaiting a repaired one) I was out of ideas for things to look at.

Well ... WTH. The standard injectors on this 1.8L engine are MDH-210, probably 210 cc/minute flow at some standard pressure. I had a set of MDH 275 from the 2.4L engine and swapped those in. What a difference. Runs all but perfectly. Fuel trims after a test drive are (L, M, H) -4.6, -0.8, -1.4. All in the reasonable range.

I have three more of this model car -- another 1.8L, two 2.4s. All of them run and idle normally. This is not some model-specific oddity that shows up in old age.

So -- there is a problem and it's not fixed or even diagnosed, in fact I still have no real clue what it is. Lots of small stuff is fixed. The problem is bypassed -- the car's perfectly fine for ordinary use. I'll check out the repaired ECM when it arrives, who knows?
 
Still more data.

The 'repair' ECM came back 'CPU dead -- unrepairable.'

After several trips another look at the ALDL info with EvoScan showed all fuel trims -- L, M, H -- at -4.6% This is with 275/210 = about 1.30x the fuel. So it was short by perhaps 25% at all speeds/loads. In other words not an air leak any more. Could it be greater displacement than 1.8L? The Expo LRV was also fitted with a 2.4L engine but that one is so different you can't miss it.

Can only make wild speculations: It's clearly a 4g9-series engine; the only such engine with a displacement over 1.8L is the 4g94 which is 2.0L but although some U.S. Mitsubishis in the early 90's are said to have been fitted with that engine none seem to have been sold here with it so it would have had to come from elsewhere.

The engine is marked of course but none of the markings (VIN or engine numbers) on the block is where I can read it without taking off stuff I'd rather not. Or conceivably the engine is stroked. I can measure the stroke and I'll do that when changing the plugs in a few days. The history of this car (like most bought used when almost 20 years old) is totally unknown.

After straightening out the mixture problem the idle suddenly got bad enough that the car couldn't be started with the A/C on -- it would stumble for a moment and die. And it had a miss on acceleration at speed. Sounded like an ignition problem so I swapped the distributor+cap+wires from another car same model -- problem gone. After the usual search process the problem was pinned down to the cap or wires for which replacements are on order. I'll replace one at a time and keep the unneeded new part as a spare.

The usual ohmmeter tests and visual inspection show nothing -- that's rare.

Finally, after nearly a year of part time work the thing runs perfectly. There's still the mystery of the seeming need for more than the factory-correct amount of fuel (the numbers were unaffected by ignition fix) but a mystery that causes no problem is tolerable on a '93 car.
 
Final Edition.

The fact that with modestly oversize injectors the engine runs perfectly and fuel trims (L, M, H) are all -4.6% indicates that the engine pulls more air than the ECM expects at all RPM. That's not an air leak -- it's more displacement than the ECM is designed for.

While replacing the spark plugs I measured the stroke: 94 mm. The spec is 89 mm. This engine has been stroked. That's enough to raise the displacement from 1834 cc to 1937 cc. When I had the head off a couple months back I looked for a ridge at the top of the cylinder: There was none. The car has nearly 240,000 miles so likely the engine was rebored and stroked not too long before I got it A 1 mm rebore would be about 3% increase or ~60 cc. That would take it to right around 2000 cc.

I cannot find a current stroker kit for the 4g93 engine but I found hints that they used to be out there. It's also possible that the 4g94 (2000 cc) internals could be used. The 4g93 head gasket fits (and 4g94 wouldn't) so it's probably a 4g93 block.

What you get when you buy a 15-year old used car ...

The purist would install the injectors from a 4g94 but the -4.6% fuel trim is well within the +/- 10% the ECM can handle and the thing now runs perfectly -- gas mileage in suburban driving is mid-20's. I'll leave it as it is.
 
With fuel trims at -4.6% that means you are running ever so slightly rich. The ECM has cut back on fuel which is why your trims are on the negative side. At less than 5%, it is acceptable.
 
keep in mind those pcm's were notorious for bad caps.
and also some had the rom mask programmed so no editing was possible.
who knows how many times this car was modified over the years.
sounds like you have it running ok.
replace the caps before they do serious damage to the board.
 
it is possible you're getting a lot of blowby when engine gets hot and it is pushing hot oily mess into the intake air. Try venting PCV to air and cap that intake connection see if it makes a difference.
Seafood used to make a cleaner with a super long plastic 'straw' that would slip down the Mass air flow and allow you to spray
 
Back
Top