Need some input -> Problem with 3.2 transplant going lean and misfiring after two min

[Marc Bixen]

[Wayne 962]

Originally Posted by brianlay

I scanned the whole thread and maybe I missed it, but have you watched the O2 signal on the scope throughout the warmup phase?

Also watched the injector pulse width through the warmup phase, to verify that the pulse width becomes noticeably shorter as the engine starts to run badly? This would confirm that it's the ECU causing it to go lean at idle after warmup.

Thanks Brian:

1- Yes, I did watch the LM-1 during the warmup period today. Starts at about 12.1 AFR which is rich, which I would expect during a cold start (from about 75 degrees - not really a "cold" Minnesota start but a Southern California "cold start"). I have a video of this, which I will post shortly.

2- I have not measured / watched the injector pulse width signal through this warmup stage, but that is a good idea. In general, it's very difficult to get a measurement on the "pulse width cycle" - the oscilloscope is a bit difficult to setup if one doesn't use it every day. For the injectors, I've found that I've had to mess with it for ten min to get it to read right. Of course by then, it's no longer cold! I guess I can set it up, save the settings, let the car cool down and then check... Or, I can just borrow an ECU from Steve W. and try it that way.

-Wayne

[Wayne 962]

Update from Wednesday:

I'm about three glasses into a bottle of Cabernet, so this will be brief:

I swapped the injectors from left/right. I found some of the gaskets missing in the wire harnesses which made them fit a bit loose. Made some new ones. Fixed the vacuum line. Ran the pump and checked for fuel leaks. Then headed to the store to stock up on some beef and chicken before they run out (they were all out of the good beef). Then came back and started up the car.

Bottomline - NO CHANGE. Still runs fine for two min, and then craps out. Swapping the injectors left / right made no change in the O2 readings on the right bank (cylinders 1-3). I raised the idle using the idle adjustment screw - it idled higher with the same crappiness. I have videos and pics of this, I will post later.

I'm running out of ideas. Tomorrow I will swap out these Magnecor plug wires for the BERU ones. If that doesn't make any difference, then I will take Steve W. up on his offer to loan me a working ECU and test it.

At this moment in time, I would start to suspect a mechanical (engine) problem like cam timing, or chain tensioners, etc. But the car continues to run very well for the first two minutes. So, even if there were something mechanically wrong, I would assume it would show up in the first two minutes. Looking down into the cylinder when I had the injectors out, it looked quite a bit "carbony" in there - the sign of a rich condition. When I removed the plugs a few days ago, I didn't really see much of that. But mucking with the thing over the past several days may have caused some build up. Not sure...

I've already lost sleep on this, and now I'm starting to lose my hair. This is not the most difficult car to work on (the BMW 700 and the 959 take that title, and this car has a ways to go to get to that level), but it's really starting to bug me. John Walker gave his "trouble car" to someone else for a fresh set of eyes to look at, I think I might do the same if I can't figure it out in the very near future. Testing a new / working ECU (thanks Steve W.) will add a tremendous amount of information to the equation.

I've been saying for years that the early Motronic (3.2) systems are my least favorite to work on because they don't store error codes and you need a shelf full of spare parts to test them out and troubleshoot them properly. Carbs are easy. MFI is straighforward. CIS can be slightly confusing, but it's all related to fuel pressures. Motronic is a $%^&**. The later versions (even on the 959) actually store some error codes. Boxsters, 993s, and 996s are a breeze because the computer tells you what's wrong. These 3.2 engines though are the worst! I mean, who actually is an expert on using an oscilloscope on these cars? Probably about .000001%. When the system works, it's pretty good and pretty reliable. But when something goes wrong and it fails, good luck with that.

Tomorrow is plug wire day. Swapping out the Magnecor wires for the BERU ones. I'm not hopeful that this will work, but at least it will eliminate a potential issue.

Anyone want to buy a used 914-6?

-Wayne

[blyguy]

When the car transitions from cold to warm (and runs poorly), is your o2 sensor connected? The o2 sensor, as you already know, doesn't come on line until its warm.

If the o2 sensor is disconnected (open-loop) and idle is crappy, then you need to trouble shoot the other inputs to the DME. There are only a few - whats the status of your barometric sensor? As far as I know that is the only (non o2 sensor) input that pulls fuel from the base maps. If possible disconnect it while engine is running poorly to see if it is leaning out the mixture.

Also, what AFR is causing your running condition? Are you lean or rich, and why is it differing from the spots you are measuring from?

While you have it idling, you can turn the screw on the AFM (NOT the idle set screw) to lean or richen the mixture. This may give you the truest sense of which side of 14.7 AFR you are on, if the engine running improves.

Don't give up - I think, with all due respect Wayne!! that you are making this more complicated than it needs to be. The Motronic system really is straightforward when you step back and look at it. The DME tells injectors how long to stay open and when to fire the ignition, based on load and RPM. The injector time is modified from the base map by oxygen sensor in closed loop, and baro sensor ( and in 964's knock sensing) At idle, load and RPM are disregarded (and the idle switch is telling the computer it's at idle). I think there is also a signal for an AC compressor to raise idle when it's turned on, but I don't believe it affects injector time/mixture.

Assuming ignition timing is ok which can be checked with a timing light ---> your injectors are not delivering the correct amount of fuel.

You've troubleshot the fuel system - seems like normal pressure and assuming correct injector operation. It would be nice to see what the acutal signal (pulse width) the injectors are seeing. If it's a correct signal, it's an injector problem, and if not it's a DME problem. Without seeing that pulse signal, you are using AFR (by way of o2 sensor voltage or LM-1 wideband sensor) as alternative to determine injector opening time.

Something seems weird there, with the various AFR readings you've got. Just like with carbs, engines idle more smoothly with richer than 14.7 AFR than with leaner than 14.7 so my guess is you are lean. You've got to eelctronically 'turn the mixture screw' like you would on a carb to see if you are lean or rich. Your sensors may be reading incorrectly and this would be a way to confirm or disprove that.

As a lot of posts have mentioned, you may have a mechanical problem with a dead cylinder, bad plug etc thats masked by rich running at start-up/cold idle. I think the old trick to pull one plug wire at a time to see how it affects running would help there.

Good luck ! If you have problems like this, where's the hope for the rest of us?

[Wayne 962]

Thanks blyguy. I've gotten deperate and I've started to mess with nearly everything and now the car runs much worse. One should not do this haphazardly, but I've now dialed in and out on the mixture screw, and also the idle air valve in and out. The idle goes up and down, but the car still runs like crap. I'm thinking of taking off the sport muffler and trying to measure the air-fuel ratios out of the open pipes (but my neighbors would not like that one bit).

I just went out and checked all the cylinders with a timing lamp - they all looked okay. I pulled each plug wire. The problem is that the engine is now running so poorly, that simply pulling one cylinder didn't seem to make much of a difference. #2 and #6 made the "most" difference if I had to really guess at it. Same thing with the injectors - unplugging one injector at a time really didn't make it run any worse than it's already running.

On top of all this, my LM-1 meter started recording impossible numbers - 3.64 lambda and 53.5 AFR. I have four (4) spare wide-band sensors so I can swap them out, but I just saw this about 20 min ago and tossed my hands up in the air and came in to have some coffee.


blyguy - I'll answer your good questions in detail:

1- O2 sensor has been disconnected for the past day or two. The LM-1 external air-fuel monitor has been plugged into the O2 sensor port.

2- The barometric sensor is a good guess - I've checked that three times now, I will check it again however. Even with that not operating properly, I don't think the car would run this crappy.

3- Good question. Steve W. thinks the mixtures are off left-to-right. Taking off the muffler and measuring it at the tailpipe would probably reveal more info, but I really can't do that here at home (I'll have to take the car to the shop and do it there).

4- I've messed with the screw so many times now, indeed. The mixture does change, and the car continues to run like crap.

5- "you are making this more complicated than it needs to be" - yes, I agree. At this point, it's such a rabbit hole where I'm just messing with stuff. I've made it much worse, and now have changed and messed with so much stuff that the "change a single variable and measure" techniques are out the window.

With the breakout box I'm checking all of the inputs to the ECU, and they all look good on the surface. With the breakout box, I can test the wire harness, and that is okay too. So, perhaps the problem is with the connector to the ECU or something like that (cold solder joint?).

At this point, I'm starting to think mechanical proble, *but* my gut doesn't really point to that because during the two-minute warm-up phase, the car ran fine - smooth, no misses, etc, able to pull out of the garage fine (now it doesn't and stalls, since I've messed with the mixture and idle so much and have the O2 and idle control valve unplugged). That's where the car is right now - running like crap with the O2 sensor unplugged, and the idle control valve unplugged. I checked the spark plug ignition pattern with the scope and I've also checked it with the timing lamp -> seems okay.

I think I'll drop Steve W. a line and see if I can borrow the ECU he mentioned he had. If the problems persist after that, then I will perform a compression check on the motor. When I got the car back (from being stolen), it had only about 17 or so more miles on it and started up fine (after cleaning the tank and the fuel filter) - it just hunted up and down at idle.

Originally Posted by blyguy

Good luck ! If you have problems like this, where's the hope for the rest of us?

As you mentioned already, I may be overthinking it, *and* again, this is a conversion into a 914, so there's been a bit of weird stuff modified...

-Wayne

[proporsche]

still following you Wayne..it is a struggle ...can you think when this all saga problem started?
I have never ever experienced anything like this on Carrera engine....working on these beasts for a long time i had never have to use any electronics tests,oscilloscopes and e.t.c.....
can you go back to the basics?Somehow i`m thinking it is not mechanical problem..since it runs cc.2 minutes perfect...

Ivan

just red through it again i do not see anywhere mentioned the coil...is it black an original or blue or silver...
Also did you already put in the Beru spark plugs wires?

[3rd_gear_Ted]

When debugging complex problems, one approach is to have two persons working on identifying the problems each of them see or find in separate individual efforts unknown to each other. Upon completing their individual separate efforts, you bring the two together and see what you got then.

This is how they found a missing nuclear sub that sunk long ago. Many inputs from different areas of expertise.

[Wayne 962]

Just looked at the coil a mere two min ago and thought "I wonder if it's the coil". I have some spares at work (not here at home). But the spark is consistent and level - this doesn't feel like an ignition problem. I've seen ignition problems, they are typically easy to spot - one cylinder runs colder than others.

Originally Posted by 3rd_gear_Ted

Many inputs from different areas of expertise.

Indeed, that is the power of the Internet, and in particular these forums. I thought I would have had this licked by now.

I'm going to be borrowing an ECU from Steve Wong this afternoon (he lives around the corner = useful). He also mentioned replacing the DME relay. I've tested it a bunch of times, but I did buy another spare one yesterday so I will check that out. He also mentioned that there are two external signals from the DME that drive the tach - the rpm and the upshift light. He said that the signals are similar, but if you swap them (as often happens with a conversion), one can blow up some components on the DME. Go figure.

My tach does have a problem - it sticks sometimes and you have to tap the glass to get it working. This is a somewhat common problem I've seen with gauges before - my 911 has this issue with a temp gauge, and the 959 also has this issue with the water / coolant temp gauge. Sitting on zero for a long time, they may get stuck. Still, I found in interesting that he mentioned that - I'm going to check the tach connection right now (before I plug in his ECU).

Here are some photos of the (lack of) progress:


Fuel injector holes were not too clean. I scrubbed them with some isopropyl alcohol:


I discovered some of the fuel injection harness plugs were loose, they were missing the little gasket inside (top one is missing, bottom one has it):


So, I quickly made some out of rubber (this first one didn't turn out as well as the 2nd and 3rd one):


Installed:


Here' the car up on jack stands so that I can measure the exhaust temps as it starts up (they were all fairly even):

[Wayne 962]

Just checked the tach. Here's a photo of the back I pulled off of the forums here:



I tested the connections to the tach and both appear to be hooked up correctly (although I've never seen the up-shift light go on):

Pin 1 on tach ---> Pin 11 on the DME (RPM signal)
Pin 2 on tach ---> Pin 21 on the DME (upshift signal)

NOTE: the Bentley Manual is confusing / incorrect on this (pages 240-36 thru 240-39).

Useful info here: 3.2 motronic tach signal

-Wayne

[dannobee]

Since it's a 914 conversion, have you verified what they're using for grounds? Are they clean and tight?

[Wayne 962]

Originally Posted by dannobee

Since it's a 914 conversion, have you verified what they're using for grounds? Are they clean and tight?

I checked them all to the negative battery terminal and they all appear to be good. I will triple check the ground strap from the tranny to the chassis...

-Wayne

[dannobee]

What about the clamps at the air valve? Are they tight? And since it's a 914, I assume there are plugs in the elbow from the AFM? Are they in place and tight? What about the vacuum fitting on the plenum where the brake booster hose usually goes? And the vacuum lines off of the throttle body are all nice and tight?

[Marc Bixen]

Originally Posted by john walker's workshop

Loren is right around the corner.

Hey Wayne, I know that you had an issue with Loren years ago, but as J.W. says, he knows his stuff, and he's right around the corner. I've been using System C (Loren) since 1985, and we've always gotten to the issue.

It's funny that you wrote this "I have no ill will towards Loren" on your post #57, right after you threw him under the bus. ???????

Also, why don't you just roll it down the hill to Callas? You guys have a relationship. Didn't he help you right your books, supply you with photos etc. Come on Tony, help a guy out.

[Wayne 962]

Originally Posted by dannobee

What about the clamps at the air valve? Are they tight? And since it's a 914, I assume there are plugs in the elbow from the AFM? Are they in place and tight? What about the vacuum fitting on the plenum where the brake booster hose usually goes? And the vacuum lines off of the throttle body are all nice and tight?

Ran a smoke test on all of that (three times). Found a very small leak on the first one, and sealed it all up. Should be good...

-Wayne

[Wayne 962]

Originally Posted by Marc Bixen

It's funny that you wrote this "I have no ill will towards Loren" on your post #57, right after you threw him under the bus. ???????

I didn't throw him under the bus at all. We tried to reach out to him numerous times again and again over the years and the response was *crickets*. This is a free country, and he's welcome to do business (and not to do business) with anyone he pleases. If I thought it would garner a call back, I would give him a ring I suppose, but after 12+ years of not getting calls back, one kindof gives one the message.

I learned that a while ago with some old ex-girlfriends!

-Wayne

[Wayne 962]

Originally Posted by Wayne 962

I think I'll drop Steve W. a line and see if I can borrow the ECU he mentioned he had.

So, I just did that. Steve W. was kind enough to loan me a loaner unit for the 3.2. Cut to the chase - no improvement. Both my ECU and his loaner ECU react the same. The car runs terrible. I'm beginning to question all of my initial assumptions and I think I'll start over by pulling all the plugs and running a compression check on it when it cools down. That and I will take a look at the plugs (which are new as of a few days ago).

Maybe I'm being fooled by my LM-1 getting affected by ambient air sloshing through this sport muffler?

I did pull the spark plug wires one at a time, and the engine did drop each time. #6 was less pronounced, I'll pay close attention to that on the compression check.

Again, this car was stolen, so who the heck knows what happened to it, maybe someone over-revved it while driving it away. A compression check would reveal that - time will tell if I should have started with that first!

-Wayne

[Wayne 962]

Compression check results:

1- 170 psi
2- 168 psi
3- 175 psi
4- 175 psi
5- 172 psi
6- 175 psi

These results look fairly good to me.

I think tomorrow I will head to work and grab the leakdown tester kit that is there. Then, at the same time doing the leakdown test, I'll do a valve adjustment and change the oil (get back to the basics).

Also:

- New plug wires (this still might be the problem)
- Post photos of the plugs (Some look better than others, I'll probably just use new ones)
- Check distributor alignment with TDC
- Check each injector harness

-Wayne

[blyguy]

How about adding a fuel pressure gauge to watch fuel pressure in real time? There’s a tap w male connector M12x1.5 I think on the drivers side.

Good piece of info to swap for a known-working DME, your problem is probably downstream then (fuel supply, fuel injector, less likely combustion chamber or spark since you have assessed those) Youre still in the dark if you are lean or rich- and those AFR numbers are definitely head scratching !

‘Upstream’ changes that modify the DME output AT IDLE are not dramatic like you said. Fuel injection timing based on the crank position sensor/TDC seems to be ok if you have good initial idle- might be worth checking those signals and sensors again to be sure. The baro sensor even at its worst will lean things out but not to the point of poor running. What happens when you rev the engine when it is running poorly? Stumbling, stalling or running rich?

[john walker's workshop]

When the valve covers are off, do a cam timing check. Much fun in a 914.

[creaturecat]

Loren helped me out, a few years ago.
he was most helpful - and a complete gentleman to deal with. he also did it for free ...... actually he was out of pocket a few bucks.
just sayin' : )