That was my initial thought - that is why we had the injectors all cleaned, mini filters replaced (inside the injectors), and completely calibrated and tested. All good now (and didn't fix the problem, unfortunately).

-Wayne

I've got a FLIR camera that's like one of those IR temp tools on steroids. I ran out of time yesterday to look at it - the plan is to do that tonight. The air-fuel mixture ratio is equivalent to the CO reading out of the tailpipe - the LM-1 gives you the exact number so that you don't have to guess at it. Still, I don't have a pure CO or HC meter, as I suppose that would indeed show an unburned (fault) condition.

Tonight's test plan:

- Smoke test the exhaust (as Steve suggested)

- Install new O2 sensor (as everyone suggested)

- Take an IR photo of the exhaust to check for variations (as J Walker suggested). Frankly, I'm slightly disappointed that I haven't already thought to do that - I didn't get the feeling that one or two cylinders in particular were down. The FLIR camera wasn't really cheap (like $500), so I try to use it every chance I get (useful for finding small animals in the backyard at night!).

Thanks, I will keep everyone posted -> heading to the garage to start the smoke test right now.

-Wayne

Don't waste your money on that now, a 911 3.2 engine should run OK (no missing) with the O2 sensor disconnected, although a little rich.

My feelings exactly.

Monday Update:

Okay, so I did the following:

- Smoke tested the exhaust (as Steve suggested). I found a few small exhaust leaks at the muffler gaskets (tightened those), but none that were significant to make a difference. Nothing came out of the exhaust manifolds that I could see. So, there's probably not an exhaust leak there that would be interfering with the proper operation of the O2 sensor.

- Take an IR photo of the exhaust to check for variations (as J Walker suggested). Unfortunately, the heat exchangers make it nearly impossible to see / tell what is going on. I also can't directly see the exhaust ports either, making it very difficult to take accurate measurements:

http://forums.pelicanparts.com/uploa...1588658629.jpg
http://forums.pelicanparts.com/uploa...1588658629.jpg

- Install new O2 sensor (as everyone suggested). The old one looked fairly new, and I think it only has a few hundred miles on it (despite being 12 years old). I'm not sure if these age over the years without any use (I don't think so), but I figured replacing it anyways would check this off the list. Bottomline - the new sensor made no difference at all (what Dave and JWalker said already). I was hopeful, but I really didn't think the O2 sensor was the problem either. The car acts exactly the same, unplugging it makes it run better (not perfect but better). This is beginning to not make any sense whatsoever.

Here's a video, pretty much the same as yesterday (multimeter and oscilloscope show the O2 sensor, the LM-1 shows the mixture - car runs fine to start up, then terribly, and then I disconnect the O2 sensor and it runs better).

<iframe width="560" height="315" src="https://www.youtube.com/embed/mPsZpW1RKG0" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>

Steve W. wrote that my cylinder head temp resistance values didn't look quite correct. He was / is correct. However, I discovered something odd tonight. The values that I have been measuring have been on the breakout box *with the DME* connected and running. With the DME connected, the values for the sensors are different than when the DME is disconnected. This doesn't seem quite right to me, but I'm not an expert on this. I measured the cylinder head temp sensor this time, using the breakout box, and unplugged from the ECU. These values seem to match closer the values that are shown in the factory manuals:

107 ohm after being run for about 10 min or so
158 ohm after sitting for 3-5 minutes off
170 ohm after sitting for 5-7 minutes off
230 ohm after sitting for 10-15 minutes off
83 ohm after running again, checked oil temp at oil filter, 180 degrees F.
85 ohm after turning off the car, two seconds after
86 ohm after turning off the car, four seconds after
87 ohm after turning off the car, six seconds after

This is testing it with the ECU unplugged. With it plugged in, the value is 4.59 mega-ohm. Odd, but I don't know what's normal.

Ambient air / intake air temp sensor was the same thing:

1.856 ohm @ about 75-85 degrees F.

This seems fairly correct, as per the factory manuals. Again, the values I was recording yesterday were on the breakout box with the DME connected. I guess to get accurate values while the car is running I would have to place an inline diode inbetween the sensor and my multimeter.

At this point, I'm beginning to completely run out of ideas. I've tested nearly everything. I don't have a spare ECU or a spare air flow meter (I really don't think it's the meter based upon the tests that I've done). I'm going to try to borrow an ECU from a local shop, perhaps tomorrow. I suppose there could be an issue with the wire harness, but I don't think so, as I've done all the tests at the breakout box, which should eliminate any harness issues.

Recap: The car runs terribly at idle once the O2 sensor is plugged in - the mixture goes lean. If it's unplugged, then it runs "okay" - not perfect, but "okay". With the O2 sensor plugged in and raising the rpm to about 2000, the mixture is spot on at 14.7, but it's still flubbing again. Car runs perfect on startup - no missing, no nothing. After 1-2 min it starts to run poorly. Note: this period of "running well" in the beginning two minutes is indeed dependent upon engine temperature. If the engine is completely cold, it runs for about two min before flubbing. If the engine is warm, it will only run okay for about 20 seconds or so. So, this "warm-up" period is indeed temperature dependent based upon the temp sensor (presumably the head temp sensor). That would seem to hint at proper operation of the head temp sensor?

I pulled the ECU out of the car. It always amazes me how bad these always look if you're buying a used one. They really look like they've been left outside, etc., and/or kicked around the parking lot a few times. The only ones that I have seen that don't look this way are the spare ones I have for the 959. Those look nice. This one doesn't look mint (but none of them do, especially the ones for sale on the forums / eBay). It doesn't appear like it's been opene up or messed with, but I can't get for sure. If someone did open the case, then they did a very nice / careful job making it look like they didn't open it up. Usually you can see scratch marks all over it, etc.

http://forums.pelicanparts.com/uploa...1588658629.jpg
http://forums.pelicanparts.com/uploa...1588658629.jpg

I'm definitely beginning to run out of ideas. This probably would have been easier to diagnose if it wasn't a transplant into a 914. I.E. - it took me about 45 minutes just to find the test plug that is used to set the idle (hidden and tucked under the batter tray - had to trace back the wire harness, which is not easy to see in the 914 engine bay). I think the next step would be to try to swap out the 3.2 ECU, but I don't have a big pile of them under my desk!

Thoughts?

-Wayne

Wayne did you try to play with the fuel enrichment setting on the computer ..there is a back side plug and settings 1-7 turns....

Ivan

I have this on my list for tomorrow. I think I need to go over all of this again, it all started with a surging idle and now is something much worse. I’m going to go back to basics and reset the idle mixture and the idle speed again tomorrow.

Wayne

Wouldn't it be weird if it had carb phenolic spacers under the intake runners like that last problem child I had. No slots for injector spray. Found eventually when the injectors were out. You could see the blockage through the injector port. Like if a screwdriver was poked in there it would contact the spacer.
It did the exact same thing when started cold. Ran great for a minute or so, then deteriorated.

Does your exhaust mount the O2 sensor where it only reads one bank of cylinders? If so you can try swapping the injectors left to right and see if the results differ, either the whole batch, or two cylinders at a time.

Use a timing light to check each cylinder (around each spark plug wire) by watching the flashing light as the engine is revved for mis-fires.

I'm going to mess with the car this afternoon and try some things (more on that in a bit). If I don't make any progress, I will pull off an injector rail and take a look. Good suggestion.

-Wayne

Steve - I'm getting very close to opening up the ECU and taking a look inside (and then sending it to you for testing). I think you live like walking distance from my house!

As for the injectors, I pulled all of them a few weeks ago and sent them to RC Fuel Injection for a complete overhaul. So, I'm pretty sure that the injectors are good, I also did check them manually using the breakout box to make sure I heard them clicking on and off when triggered (as per the manual). Here are the results from RC:

http://forums.pelicanparts.com/uploa...1588709576.JPG

-Wayne

Yup. Good tip. While I didn't do that with a timing light, I used an inductive sensor (same one used on the timing light), and plugged it into the oscilloscope to check the spark pattern. It looked fairly good to me, which led me to believe that there wasn't an ignition problem. Since the car starts fine and runs fine for the first two minutes, that would indicate that the ignition system is okay. If there were a problem with the wires (the only thing I haven't replaced yet), it would theoretically show up 100% of the time, not just in the warmup stage?

I posted the output of the spark wire inductive sensor (did all cylinders, but this is just one) on page one of this thread, but here it is again:

http://forums.pelicanparts.com/uploa...1588489309.JPG

-Wayne

Okay, so I could not sleep well last night trying to figure out what I missed on this. After taking a step back, I'm now really confused. So, the original problem was surging at idle, which has now gone away. But surging is typically caused by a richer mixture - now it's super lean. So, something has changed, but I didn't really mess with anything - the biggest change was that I had the injectors rebuilt. Four of the injectors were "dripping" as per RC Fuel Injection's report on them - that would / could probably cause a rich condition and surging at idle. So, fixing them would presumably create better performance at operating range, but I suppose might make the mixture go lean at idle. But with the O2 sensor plugged in, the car *should* recognize the lean condition and correct. Hmm...

The car runs somewhat okay with the O2 sensor unplugged, and terrible with it plugged in. The LM-1 air fuel meter shows it running slightly richer with it unplugged, which is expected. When warm, it seems to be running pretty well, but I'm *very* picky and if it misses every once and a while, I'm not happy. With the O2 sensor plugged it, it runs super-lean. Which means that the signal from the O2 sensor is confusing the computer (thinking that it's running rich when it's really not). Since it's a new sensor, it may be a fault in the wiring harness or the circuitry of the ECU. I did remove the breakout box from the system and tried it without the box installed and the problems remained the same.

With all this in mind, I formulated a new test plan for today:

- Check the ground from the hood of the O2 sensor to the chassis. We have stainless steel headers installed, and the ground may not be that good.

- Also check the acutal values coming out of the O2 sensor without the ECU connected. This *should* match the output performance of the LM-1 meter. This is a brand new O2 sensor - so there shouldn't be any issues. The first time that I looked at the O2 sensor, I noticed that the wire was looped around one of the spark plug wires, so I fixed that. These are Magnecor wires by the way - one thing to perhaps think about would as to whether the spark plug wires were interfering electrically with some of the sensor inputs. I don't think so (as I scoped them on the breakout box), but I'm opening my thought process up to anything at this moment in time.

- I'm also going to plug in the O2 sensor that is out of the car, and stuff it into the tailpipe (the other side of the LM-1) and compare the values to the one that is installed in the pipe. I'm not sure if this will work, but again, it's more data. So, at this point, I'll have three O2 sensors being recorded by the oscilloscope so that I can compare them all. I don't expect to see any problems, but I want to check - this will also expose any ground issues.

- Kenny at Autowerkstatt offered to lend me a cylinder head temp sensor to check mine out. He mentioned that several times throughout the years, the sensor has 'tested' okay, but replacement has fixed problems. So, I might drive over there and try that later on today.

- I'm going to open up the DME and take some photos and also check the "fuel quality rating switch" on there to make sure it's in the proper position. (just did that, it was turned completely counter-clockwise, which is the default/correct setting).

- Check the alternator voltage to make sure it's in range while running (did this already, but I will do it again).

- I'll go over the setting of the mixture and idle speed one more time:
Setting base idle and CO mixture
1) Oil temp at 90C
2) All electrical consumers off (AC, heat, lights, ...)
3) O2 sensor unplugged
4) Adjust the CO screw on the AirFlowMeter for CO=0.8 or AFR=14.2 (clockwise=richer)
5) Now connect O2 sensor
6) Bypass idle speed stabilization by bridging B&C terminals on test port (left side near rear fuses)
7) Adjust idle speed with bypass screw on ThrottleBody, 84-85 cars are 800RPM 86-89 are 880RPM, set base idle 20to40RPM above target, if target is 880RPM set base at 900 or so.
8) remove jumper

I grabbed this image from The Samba - shows which way definitively to turn the idle mixture screw (which can be confusing as it's mounted upside down on the 911):

https://www.thesamba.com/vw/forum/viewtopic.php?t=356377
http://forums.pelicanparts.com/uploa...1588712529.jpg



- I don't like to do this, but I will test the intake gaskets by spraying carb cleaner on them. Spraying combustible materials into a hot engine compartment is not one of my favorite things to do (smoke test didn't show any problems, so I don't expect any here).

- I have the build photos from when it was put together, and there is a photo of a replacement harness being spliced together. I there is some ground wire or something taped to the outside of the harness, I will investigate this further. Since this is a transplant, who knows that the heck may be going on. It's quite difficult to figure this out when it's been patched together - it took me about 30-40 minutes just to find the test plug (hidden under the battery tray).

- I'm going to check fuel pressures again while running - I didn't take a photo of the pressures while the engine was running, just when the pump was running while I jumpered it at the DME relay.

-Wayne

Before doing all of that, I took apart the DME. The unit had "85 3.2" written on it in sharpie in the build pictures (faint traces of that can be seen on the cover, so it's the same unit as in the build photos).

- PN 0 261 200 082, 911.618.111.20 - appears to be from a 1987 911.

- Appears to have the correct circuitry for the O2 sensor (as expected)

- EPROM shows a 24-pin socket, which is consistent with a 1987 DME.

- Conformal coating is flaking off in a few spots.

- The soldering around the fuel quality switch looks a bit "different" than the rest of the board.

- There is a slightly black mark in the center part of the board, but it almost looks like it was black ink - not burn marks...

If I can't get this resolved, I think I'll send it off to Steve Wong for inspection, and perhaps a chip put in (perhaps ROW chip that won't use the O2 sensor).

-Wayne



http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg
http://forums.pelicanparts.com/uploa...1588713455.jpg

Loren is right around the corner.

Thanks for the suggestion John.

On June 28th, 2012, after about 5-10 warnings, I gave Loren a five-day timeout from the forums. This was based upon his tone in the forum thread "Performance Tuning - The Myth - ". It was a difficult situation for us because he was one of our vendors (rebuilding CD boxes). But the number of complaints had reached a level that resulted in semi-daily emails from forum members in my inbox. I was way more patient with him than with anyone else on the entire forum, in the history of the forum, because his knowledge is / was a tremendous asset.

After that 5-day timeout, Loren never posted again on the forums, and has never returned a single phone call from anyone at Pelican.

I have no ill will towards Loren, and he certainly seemed like a nice guy every time I spoke with him. However, I'm not going to try to chase Loren down when there are other qualified people (Steve W., Ingo, etc., also near by).

-Wayne

deleted sorry..i was a little harsh....afterall it was 25 years ago....people change;-)

I deal with Loren for everything electronic. nice guy, never an issue. ymmv I guess.

Once the engine is warm, about 3-5 minutes, the temp sensor can be bypassed with a paperclip.
After the temp sensor reaches less than about 300-500 ohms, it has basically no effect on the injector pulse width.

The key point being made here, in addition to the injector spray, is that you need to eliminate a potential mechanical problem
in the engine, e.g. a weak/missing cylinder (valve seats, valve adjustment, etc.). By disconnecting either an injector or spark plug,
you can quickly tell how each cylinder contributes to the engine's running. All cylinders should have an equal effect on how the
engine idle drops when the cylinder is shutdown.

Did you pop the airflow meter cap and look at the track yet?

I didn't take off the air flow meter to take a look at it, but I did test the flapper box using the scope, and it tested perfectly. I did take a photo of the results on the scope - it's on the first page of the thread - I will take another look in case I missed something. I can also retest very easily with the breakout box.

Here's the image again. Basically tested the voltage coming from the air flow meter over time. Pressing on the flapper makes the voltage go up. Letting it close makes it go down. Basically what the factory manual says to check for:

http://forums.pelicanparts.com/uploa...1588489309.jpg

Here's a neat thread on repairing and testing the meter. My scope readout is not as pretty as his is because I had to reach all the way in the intake and open it by hand (so it wasn't consistent). But there are no voltage drops, so I'm 99% sure that the meter is fine. It supposedly only has 65K miles on it, so it shouldn't be bad either way.

Still, I will re-run the test with the scope again tonight (takes five min).

-Wayne

Just retested again. No problems, the waviness is me not being able to stick my hand in there consistently each time. http://forums.pelicanparts.com/uploa...1588740397.jpg

Wayne,

Difficult to see the 'discontinuities' on your scope trace for the AFM. I think the best way to test it for it's IDLE condition is to put your breakout box in, check the regulated power supply to the AFM (should be 4.95V - 5.05V coming from the ECU) and check that it returns a stable 0,24 to 0,25V. Measure while running at idle (measure voltage, do not try to measure ohms with ECU connected).

Here are my values, measured on the bench with the AFM open after re-tracking. I had a 'significant' discontinuity when coming off idle (jumped from 880 rpm to 2300 rpm straight away) and it was relatively hard to see on the traces, had to select the right timescale. When opening you could see it was worn down to the PCB !

Anyways, if it gives you that stable 0,25V output at idle (or 5% of Uin), you are fine for debugging your idle condition. No need to open the AFM for that.

Pin 6 (ECU) = GND for AFM
Pin 9 (ECU) = Reference voltage for AFM
Pin 7 (ECU) = AFM return voltage
Pin 22 (ECU) = AFM air temp sensor

If you want to check the AFM air temp sensor (also easily done without the afm being opened):
- disconnect ECU
- measure the ohms from pin 22 against the pin 6 GND

0°C/32°F : 5K - 6.2K (tolerance 10%)
20°C/68°F : 2.2K - 2.8K (tolerance 10%)
30°C /86°F: 1.5K - 1.9K (tolerance 10%)

http://forums.pelicanparts.com/uploa...1588745780.jpg
http://forums.pelicanparts.com/uploa...1588745780.png

Maybe another point to check in your list : are you running a 87 chip or an earlier chip (as the ECU seems 87). As Steve already mentioned, the chip programming determines the idle rpm to be used when you are going to reset your mixture following your procedure.

If original :
As of 87 : 880 rpm
Before : 800 rpm

Tuesday update:

Today was a weird day. I'm even more confused now.

Jacked the car up so that I could get a clear thermal photo of the exhaust headers. All six were the same temperature, which leads me to believe there are no particular issues with any one cylinder. When I replaced the plugs the other day, they all looked pretty much the same and fairly good. As I mentioned previously, we had the injectors completely rebuilt and tested, so I have faith that they are functioning correctly too. If we had a bad cylinder or bad injector, then that would show up as a cool spot on the thermal camera (I've used it many times before to check and find issues like dead cylinders on other cars).

The next thing I did was disconnect the O2 sensor from the harness and plug it into the scope to measure the output. I also hooked up 12V to the heater. The O2 sensor was reading a consistent .88 volts which corresponds to a rich condition. This doesn't make sense to me, since the LM-1 wideband O2 sensor tool sticking out of the tailpipe was reading closer to 14.7 at this time. I ran this test and checked it many times, and determined that the O2 sensor was reading correctly. To double-check, I took the LM-1 wideband sensor and screwed it into the O2 sensor bung (on the right side of the car), then started the car up to check the mixture. Sure enough, the sensor was reading a mixture in the low twelves. So, both oxygen sensors (the new one from yesterday and the wideband one from the LM-1 tool) were saying that the cylinder 1-3 bank was running rich. I suppose that yesterday when the car was reading really really lean with the LM-1 sensor, it was because the O2 sensor was reading rich. That part makes sense.

What doesn't make a whole lot of sense is to why the LM-1 tool sticking into the tailpipe is registering a much leaner condition than when it's installed in the O2 sensor bung. I have a feeling that it has something to do with the two-pipe sport muffler, and perhaps one of the sides of the muffler is sucking in fresh air from the outside inadvertently, which is leaning out the mixture. I tried sticking the sensor in the other pipe, and got slightly different results, but this needs more attention tomorrow.

I then adjusted the idle mixture screw so that the reading from the LM-1 (screwed into the O2 sensor bung) was reading 14.7 (approximately). At this point, I thought this would fix the problem - it didn't. The car started to run very poorly again, yet the meter says that the AF ratio is 14.7. I called John Walker at this point and he suggested I head to my friend's shop and have it tested with their CO and HC meter. He said (correctly so), that the A/F mixture will tell you if the adjustment is correct, but it won't tell you about other combustion problems. Still, the thermal camera from underneath the car shows that the intake headers all look pretty even, so I'm not sure if there is a cylinder-specific problem. Perhaps there is something amiss with the distributor, but again, the car runs nearly perfectly during the initial 1-2 minute warm-up and then starts crapping out.

I also checked (on the scope), the injector pulses for both channels. They match what the expected pattern is, as shown in the factory manuals.

As I mentioned in the previous post, the air flow meter appears to be working correctly. Even if it were "stuck" open, then that should make the car run richer, not leaner.

Fuel pressure - I checked the fuel pressure according to the factory manuals. 2.0 bar running, and 2.5 not running (with fuel pump jumpered). In addition, when I disconnected the vacuum line to the throttle body, the pressure increased back up to about 2.5 bar (disconnected). Connected it reads 2 bar. All of this behavior is consistent with what the factory manual states is normal. I of course broke off the little T-fitting at the back trying to pull it off (I felt it was old and I knew it would break!). Part number: 928-573-727-02 if anyone needs it.

I also ordered new OEM Beru ignition wires. This car currently has the Magnecor wires, and I've never quite been happy with the fit of the connectors or the fact that they feel a bit flimsy. I prefer the OEM ones in this case. Maybe (fingers crossed) they will magically fix something there. The current ones are at least 12 years old. I believe that the BERU ones have resistors in the spark plug leads. Not sure about the Magnecor.

I'm running out of ideas. I guess tomorrow I can take out some of the injectors (check to see if the insulators are installed properly), and swap some of them around. When the injectors came back from the shop, they looked completely different (newish). They looked so good that I thought they may have given me someone else's, but I checked the BOSCH part numbers and they were the correct ones.

Maybe tomorrow I will change out / test the cylinder head temperature sensor, or test it when it's warm by putting a paperclip across.

I don't get the following:

- Why the air fuel LM-1 monitor in the tailpipe would read differently from the one in the bung. Maybe getting some outside air sucked in?

- If the readings are / were off, then when I adjusted the mixture properly to 14.7, the engine should have run well at idle.

- If this were truly a spark / ignition problem, then shouldn't the issue be prevalent from startup?

I think I need to sleep on this, a lot of this isn't making any sense. What is really annoying / frustrating is that the engine is running perfectly fine for the first two minutes, which basically rules out any mechanical issues (which is good) and focuses the attention on the fuel injection / delivery.

Thoughts?

-Wayne

Photos:

This shows the reading of .88 volts out of the stock Carrera 3.2 narrow band O2 sensor. I grounded the multimeter directly to the sensor to verify that there weren't any grounding issues:
http://forums.pelicanparts.com/uploa...1588747387.jpg


With the LM-1 Sensor mounted directly into the bung, this shows 12.6 AFR:
http://forums.pelicanparts.com/uploa...1588747387.jpg


Showing the LM-1 O2 sensor plugged directly into the right side (1-3 on the 914) of the engine):
http://forums.pelicanparts.com/uploa...1588747387.jpg


Injector pulse width patterns on the scope - looks similar to the pattern in the factory manuals, and is also identical on both channels:
http://forums.pelicanparts.com/uploa...1588747387.JPG


Fuel pressure of 2 bar while running:
http://forums.pelicanparts.com/uploa...1588747387.jpg


Video of the car running with the LM-1 installed into the right-side (factory O2 sensor disconnected). This is after I set the idle mixture to 14.7. Car runs like crap, as you can see:
<iframe width="560" height="315" src="https://www.youtube.com/embed/Bwebs3LiHms" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>


Video of the car with the injector pulses on the oscilliscope:
<iframe width="560" height="315" src="https://www.youtube.com/embed/a3e4XukfOvo" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>

RC Fuel Injection is a great place, in fact it's the first place I recommend as the go to when owners need their injectors cleaned and tested. However I worked with more than a handful of owners who still had injector problems even after they came back from cleaning with a good bill of health (no particular injector service facility in particular) that only replacement injectors solved. That said you may have an injector on the bank that is not closing completely, or sputtering on low duty cycles, such as near idle, that will cause that cylinder to idle excessively rich or lean, and if it's on the bank where the O2 sensor is, it will send a signal accordingly to the DME of the average mixture of those three cylinders. This I'm assuming your O2 in your exhaust is only reading either the left for right bank. The LM-1 at the tailpipe shows a different reading, because it's displaying the average of all six cylinders. So this is why I previously suggested if there's a suspect injector, to try swap them side to side and see if the car's O2 sensor displays differently. However you've seemed to confirm that left and right bank are different, which indicates there either is an injector problem, which again you can try swapping them side to side to see if the readings swap, or possibly the the left and right cam timing is different. When a cam is retarded on a 3.2, it tends to lean out the overall mixture noticeably. That's not to say there couldn't be an individual cylinder mechanical issue such as a bad valve or spark, but that's getting a bit beyond the basics.

You could try disconnecting one injector at a time to see if you can discern the one cylinder that makes the overall idle behavior different than the other five, but you'd want to do that with the O2 disconnected and the idle control valve also disconnected so they don't try to compensate. A 3.2 idling on five cylinders almost sounds normal like all six firing, so you may or may not be able to detect a difference. You can also observe your LM-1 readings for any differences. You could also check the harness connector for the injectors to make sure there is no corrosion on the pins, as left and right bank are separate at the firewall plug, but common inside the DME unit. Your problems don't initially seem like a faulty DME unit, but if you want to borrow a good tested unit, just let me know.

I would start by checking the ground connection of the O2 sensor that is made through the tread?
That's were it is measuring it's 0,88 V against (very very rich condition)?
Is the thread full of anti-seize ? One of the wires for the heater element normally goes to ground point MP XII on the intake runner. Check for a clean 0 Ohm connection between this wire end and the ground point.

Just ruling out any electrical issues with the O2 sensor first.

UPDATE - forget the above, your external O2 sensor has it's own power independent of the car, so the ground connection will not be the issue

Thanks for the useful data dump on the AFM. Just to be 100% sure, I will run the tests again on it, moving the vane slower to see if there are any drops in the voltage. It's already setup from last night, so it should be easy. As you mentioned, the AFM is reading low voltage at the start, so that wouldn't be affecting this running / idle issue. At this point, I do not want to remove the meter because I sealed all of the boots (smoke tested before and after), and I'm confident that there are no vacuum leaks. I'm also reluctant to start opening up the AFM and messing with it, when all of the data seems to point to it being okay. I broke the wide-open-throttle switch removing it the other day - any time you start messing with stuff (especially in the tight confines of the 914 engine bay), there's plenty of room for disaster.

I've got new plug wires coming today. I will pick them up later on this afternoon.

As Steve W. suggested, in the meantime, I will remove the injectors (again), and take a look down into the cylinders to see if the problem that John Walker had is indeed the issue. I will also swap the injectors left / right and see what the mixture looks like after doing that.

I forgot to mention that I did spray carb cleaner all over the intake gaskets yesterday - no changes detected in the engine. Spraying some down the air cleaner made the engine rpm increase (as expected). So, there does not appear to be any leaks there.

When I swap the injectors, I'll take the car out in the street and jack it up so that I can check the header temps with the thermal camera. That in the past has been my number one tool for identifying cylinders that are weaker than others. It's best to use the tool when the car is very cold as the differences in temperature tend to even out as the whole engine gets warm. Yesterday, when it was warm, the thermal photos (I haven't posted them yet) seemed to indicate no problems.

Indeed, this new information from yesterday complicates things slightly. When the car was surging (at the very beginning of this mess), I jumpered the test port and set the idle to be just a hair less than 900. I had to crank the idle adjustment screw quite a bit to reach that number. Now I'm beginning to wonder if the culprit perhaps lies there - although that wouldn't explain the rough flubbing / missing that was happening when it was raised to 1,500 or so rpm...

-Wayne

I found this photo from the build of the motor - looks like the proper spacers and gaskets are in there (so not the same problem that John Walker had, although this only shows one side):

http://forums.pelicanparts.com/uploa...1588794962.jpg

-Wayne

Wayne,

I would only check for a stable 0,25V from the afm at idle. That is the only thing that matters for your idle condition. And the temp probe ohm range as the ecu uses both values to try to calculate air mass from the flap (volume) and temp probe.

Both should be stable at idle and in range. You can test both from your breakout box, no need to move the flap.


Sent from my iPad using Tapatalk

Thanks Frank. Both have checked okay.

-Wayne

Indeed, agree - this would only be a condition that would occur at idle, as previously, the LM-1 in the muffler was reading 14.7 pretty much spot-on. Unless perhaps the LM-1 was picking up ambient air from the outside. Yup, the next test will be to swap the injectors. I will do that now.

Another thought - reading the Bentley manual, I noticed it said that the wire harnesses from the two position sensors and the O2 sensor were "shielded". While I'm not 100% sure that is correct, it got me thinking that the whole big mess of wiring in the engine compartment (routed differently because it's a 914) may be picking up stray signals from the plug wires, etc. (the O2 sensor wire was inadvertently wrapped around one of the plug wires previously). My gut is that this is not the issue, but I'm starting to think way outside of the box now.

-Wayne

I doubt that it would be interference because your external wideband sensor confirmed the .88V reading, presumably not using the internal wire harness ?

You are using the original cable harnass with the shielding in place I assume.


Sent from my iPad using Tapatalk

It's the original harness as far as I can tell, but it's been slightly modified to work in the 914 (modified by the builder, I have photos of this).

I agree that there shouldn't be interference on the O2 sensor, but my thinking was that there might be dirty signals floating around on perhaps some other lines. Again, starting to grasp at straws.

Taking a break while the spilled gas from the injectors airs out...

-Wayne

Photos of the harness (from the build):

http://forums.pelicanparts.com/uploa...1588797214.jpg
http://forums.pelicanparts.com/uploa...1588797214.jpg

-Wayne

Another straw to grasp at - I've been running it without the K&N air filter on - just open to the air. That should not make any difference either.

It also has these two boots that have been "merged" together to create a 180 degree turn on the air flow (to make it fit in the 914 engine compartment). I did something similar on the Boxster transplant (996 engine -> Boxster) for the Boxster 101 book - worked great! I don't anticipate anything amiss with this (no leaks either), but again, grasping at straws...

http://forums.pelicanparts.com/uploa...1588797368.jpg

http://forums.pelicanparts.com/uploa...1588797541.jpg

-Wayne

A good friend of mine who I've worked with for many years had a BMW of a customer that had similar problems. 1988 535i - they replaced *everything*. The only thing (last thing) they replaced was the fuel pressure damper, and that fixed the issue. I'm not sure if too many people have replaced those on these 3.2 Carreras, but it would seem like it might be something to check out. I do know when they fail, that gas gets sucked up into the throttle body, and I'm not seeing that right now. I disconnected them yesterday and fuel did not spray everywhere, nor was there any fuel coming out of the vacuum hose either.

I removed the injectors and the garage is airing out right now. Something odd though - injector #1 has the paint scratched off of it, and when I pulled the injector rail, the injector stayed in the head, and did not stay with the rail. This means that the clip that attaches the injector to the rail was loose. Gosh, hard to say if this would cause problems like we're seeing, but any little thing that is amiss will be looked at carefully.

I'm headed out now to pick up the plug wires. I will reinstall the injectors (not the wires) when I get back and see if there is any differences.

-Wayne

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.