3.2 flooding on cold engine start???

[Aprilia]

First time posting, so please accept my apologies for any protocol errors.

I have recently acquired a 1986 911 3.2 with only 90k miles on it. It was a garage queen but in great physical shape. Now that it is on the road, I can understand why it spent so much time in the garage. The beast is a b**ch to start and runs so rich that it only gets around 9 mpg. I have worked my way through a number of postings but have yet been able to find anything that seems to help my situation. I am hoping that there are others that may have experienced something similar and been able to correct it. So far, I have performed the following actions:

- New injectors (actually rebuilt and flow tested)
- AFM refurbished (IAT checked and verified, Vp linear across flapper angle)
- ICV refurbished (22k Ohms for each side, push/pull operation verified on bench)
- New O2 sensor (doesn't come into play for cold start problem)
- Reference and Speed Sensors verified (Ohm and pulse tested using O-Scope)
- CHT sensor verified (Ohm tested across temp range and matches Bentley specs)
- Idle and WOT sensors verified
- Fuel pressure checked (29 psi during operation, 25 psi after 30 mins)
- Vacuum leaks checked (engine drops in rpm when oil fill opened)
- New spark plugs (NGKs gapped at 30 thous)
- Coil resistance and voltage checked
- Distributor cap checked and cleaned (no cracks or carbon traces)
- Distributor rotor checked (resistance was only 50 Ohms instead of 1k Ohms, new one on order)
- Spark plug wires checked
- Spark plug end cap resistors checked (3k Ohms across all units)
- Engine and body grounds cleaned and checked

As previously stated, I am at a loss as to why the car is running so rich, which I assume makes it so hard to start. It takes around 15+ seconds of continuous cranking for the car to start (outside temp is 60 degrees). When it does, it chugs and coughs for another 15 seconds while the engine tries to burn off the excess fuel. Once it finally catches, then the rpm raises to around 1200 until it warms up a bit and then slowly drops to around 900 after a couple of minutes. The car seems to run well and strong, but gets worse mileage than my brother's deuce and a half.

Any insight that could be provided would be greatly appreciated!

[Hunt3R]

Did you retune the base A/F ratio after replacing the AFM and injectors? Does it have a stock chip in it? If so then the idle should settle down to 800rpm, not 900. The base idle should be reset too after setting the base A/F ratio. Did you check that the Fuel Quality Switch on the back of your DME is set to the stock setting? Does the car start right up if the engine is warm and if so how long does it sit before starting becomes an issue?

[Aprilia]

The base air/fuel ratio has not been changed yet. It is currently 2.75 turns out from fully seated.
I believe that the DME has the stock chip and fuel quality setting, but will verify next.
I tried setting the base idle to 800 rpm using the standard method, but the engine would start an erratic rpm oscillation and then die during idle when fully warmed up. I turned up the rpm by leaning out the mixture to compensate (turned base idle screw out about 1.5 turns from 800 rpm base idle).
When the engine is warm, it starts back up immediately. If I let it cool for over 1 hour, then I start getting the hard start problem and lots of unburned fuel during the cranking process. It is exhibiting symptoms almost like there is a cold start injector stuck open or an accelerator pump dumping fuel into the intake. Once it finally starts, it doesn't seem to fire on all cylinders but chugs and spits out black and blue smoke for a minute before it settles down.
I have verified multiple times that there is spark during the cranking process, but is it possible that not having a 1k Ohm resistance across the rotor is actually reducing the spark duration on the plug and they are not able to burn all the fuel in the mixture?

Thanks for the quick response and the constructive input.

[Chuck.H]

Have you checked things like CHT resistance right at the DME pins? That plus air temp are the factors in setting cranking fuel, maybe you have a wiring harness problem.

Does it fire up quicker if you hold the accelerator pedal to the floor? That should verify your 'too-rich' hypothesis.

It's possible the fuel dampner and fuel pressure regulator have diaphram leaks, which may let fuel into the manifold.... have you seen the system hold fuel pressure after shutdown?

I'm in ATL for the week, let me know if you want a hand. I stay in Midtown near the Marta.

Chuck.H
'89 TurboLookTarga, 373k miles

[Aprilia]

All sensor tests performed at the 35 pin connector to the DME. Both the CHT and IAT matched the Bentley specs for resistance vs temp.
Sometimes it is necessary to go to WOT for the engine to catch. When that happens I get some blue smoke for about 30 secs, which I believe is caused by a fuel wash of the cylinder. I try to avoid this as it may damage the rings.
I checked the condition of the fuel regulator and dampened diaphragms using a vacuum pump directly attached to the top ports and they both held 20 in mg steady for 15-20 mins.

As an aside, when I drove the car home from work (nice and warm) I attempted to adjust the air/fuel ratio using the port on the AFM to try to lean out the mixture a bit. First I disconnected the O2 sensor and measured the voltage (0.45V) and then turned the adjustment screw out (counter clockwise) 2 full turns and measured the voltage at the O2 sensor again. I was expecting it to be around 0.3V but it actually went up to 0.68V. What is going on??? It should be going down as increased unmetered air is supposed to be getting into the manifold and the DME shouldn't be able to react other than cutting fuel to the injectors to lower the rpm to idle. Now I am really confused.

Chuck, would a DME from an 86 work on an 89? If you are around on Thur after work, maybe I could bring mine to you for a quick test to see if it starts your car.

[Hunt3R]

You're right it should go down as you turn the screw out to lean the mixture, although it's a narrow band sensor so you should see it bouncing around, right? You want to get it to fluctuate evenly between 0.2-0.8v. And 2 full turns is a relatively big adjustment so that definitely should have leaned it out noticeably, assuming everything were working correctly. Unfortunately, it sounds like swapping your DME into a known good car might be a necessary test.

[Aprilia]

I'm beginning to think that either the DME is bad or the fuel map EPROM has been corrupted.

I am still trying to determine the effect of the rotor not being 1k Ohm on the spark duration and subsequent efficiency of the fuel burn. Since the overall resistance between the coil and the spark plug has been reduced from 4k Ohms to 3k Ohms, I believe that the spark duration has been reduced by 25%. I don't know if that is enough to effect the burn efficiency. If it is, then that might account for both free hydrocarbons and O2 in the exhaust cycle which could be picked up by the O2 sensor which in turn tells the DME to make the mixture more rich. This speculation may account for the poor fuel economy that I am experiencing (only 9 mpg), but shouldn't affect the cold start problem as the O2 sensor doesn't come into play until the engine reaches a certain temperature.

Well anyway it looks like I'm taking the truck to work tomorrow because I ended up fouling the plugs while adjusting the a/f mixture and it won't start. I won't have a chance to clean them and get the settings bake to the original position until Thursday. In the meantime, I'm going to research the settings of the fuel map vs temp to get an idea on what should be happening.

If anyone has any ideas, then please keep them coming as I have just about exhausted mine.

[Aprilia]

I finally got a chance to remove the spark plugs and they were indeed fouled. I replaced them with new ones and then removed the DME for inspection. After removing the box top, I moved the top circuit board to get access to the 35 pin connector on the inside. I reconnected the opened DME to the cable and connected my oscilloscope probes to pin 14 and 15 to test the duration of the pulse width during the cold engine start.
The test showed that the duration of the pulse width was around 8ms. The details that I was able to uncover from other postings was that the pulse width should be in the range from 1.7 to 3.5 ms. That means that my DME is soaking the engine with over twice the amount of fuel it should be.
What I don't know is if this is normal for a cold start? Sort of like a cold start valve until the engine gets warmed up. Anyone have any ideas?

[mysocal911]

Originally Posted by Aprilia

I finally got a chance to remove the spark plugs and they were indeed fouled. I replaced them with new ones and then removed the DME for inspection. After removing the box top, I moved the top circuit board to get access to the 35 pin connector on the inside. I reconnected the opened DME to the cable and connected my oscilloscope probes to pin 14 and 15 to test the duration of the pulse width during the cold engine start.
The test showed that the duration of the pulse width was around 8ms. The details that I was able to uncover from other postings was that the pulse width should be in the range from 1.7 to 3.5 ms. That means that my DME is soaking the engine with over twice the amount of fuel it should be.
What I don't know is if this is normal for a cold start? Sort of like a cold start valve until the engine gets warmed up. Anyone have any ideas?

An injector pulse of 8ms is a little too long for a cold start. There're a number of possibilities:

1. The CHT is open/intermittent giving a bad value, i.e. use a paper clip to jumper
the CHT connector. Note: The CHT when intermittent will test OK and then fail later,
e.g. at different engine temps.
2. The AFM is providing the incorrect voltage, e.g. wiper arm is stuck
at a large opening or the wiper spring is too loose.
3. The DME injector driver IC has been damaged causing too great of a injector pulse.
4. The AFM input of the DME computer's A/D converter is damaged.
5. The AFM's reference 5.0 voltage is too high.
6. The CHT input of the DME computer's A/D converter is damaged.
7. There's a problem in the injector final stage circuitry.

[Aprilia]

SoCal, you raised some valid points on possible causes for the symptoms that I am experiencing. Unfortunately, I have already performed bench tests on both the CHT and the AFM. They are both in spec with regards to impedence vs temp (for CHT) and Vp vs flapper angle (output voltage for the AFM after it had been refurbished). The reference voltage for the AFM is 4.5V and the maximum output voltage was 4.0V when flapper fully open (roughly 90% of reference, which is within spec). As a separate test I bypassed the CHT using a 220 Ohm resistor.

None of these changes made any difference regarding the duration of the injector pulse width during startup.

I agree with your assessment regarding the possible circuitry problems. I was hoping to borrow a known good DME to test the engine operation, but as I am new to the 911 community I have yet to be successful. I did order a rebuilt one from ECU Doctors in Florida. I hope to have it in hand to test by the end of the week. If a new DME fails to solve the flooding problem, then I have to carefully determine my next steps. Any advice would be greatly appreciated.

[Aprilia]

Quick update on the injector pulse width at idle:

I did some rough calculations and determined that the injector pulse width for cold start should be no more than 3 ms in duration.

It is important to note that during startup the volumetric efficiency of the engine is approximately 10%, which means that only about 1/10th of the engine displacement is actually getting air to be used for combustion. Also, the air/fuel ratio shouldn't below 10.0 or the mixture will not fire. The target a/f ratio is 12.

Below are the calculations used.

air volume per cylinder (195 ci/6 * 0.10) = 3.25 ci = 0.0018808 Cubic Feet
air mass (CF * 0.076 Lbs) = 0.00014294 Lbs

fuel needed for idle (air mass / a/f ratio) = 1.1911 E-05 Lbs

To convert the Lbs of fuel needed per cylinder at idle to injector pulse width depends upon the size of injector being used (fueling rate). Since I am using stock injectors, they are rated at 24 Lbs/hr. If we convert from Lbs/hr to Lbs/millisecond, we get

injector (Lbs/ms) = 6.6667 E-06

Finally, the ideal pulse width duration needed to get a target 12 a/f ratio is,

injector pulse width duration (fuel mass/injector rate) = 1.7867 ms

Since this assumes that the injector opens and closes instantaneously, we would need to increase the duration to deal with the injector opening slew rate of about 1 ms. When you add this to the ideal duration above, we get the final pulse width of

injector pulse width (ideal plus slew period) = 2.8 ms

What this means is that at cold startup the duration of the injector pulse width should not exceed around 3 ms; otherwise, you risk flooding the engine. As originally stated, I am experiencing a pulse width of 8 ms, which shouldn't be possible for the DME to produce for the low cranking rpm of the engine. The only logical conclusion is that the DME is bad. I hope to receive the replacement unit by this weekend and will provide an update following the DME swap.

[scarceller]

On startup the DME commands about 2ms extra pulse width added to the default calculated idle pulse of about 1ms.

I suspect a bad DME, any 84-89 DME will work. Can you find another car to put your DME in? It's always best to put your DME in a donor car not the other way around.

Given your description of the problem I don't see how you could possibly be running this rich. Also try unpluging the AFM and then try starting, it won't run like this but we want to know what's causing this richness. One more quick test is with key in RUN engine off what's the output voltage of the AFM? Should not be more than 1vdc.

[scarceller]

8ms pulse is way off base! Remember this is a batch fired injection where fuel is feed on EVERY crank rotation. Two shots from each injector per stroke.

8ms pulse on one crank rotation is almost at 100% load! like WideOpenThrottle! The DME actually has a fail safe where it clips at max pulse width of about 9ms!

You need another DME or car to put that DME in.

Originally Posted by Aprilia

Quick update on the injector pulse width at idle:

I did some rough calculations and determined that the injector pulse width for cold start should be no more than 3 ms in duration.

It is important to note that during startup the volumetric efficiency of the engine is approximately 10%, which means that only about 1/10th of the engine displacement is actually getting air to be used for combustion. Also, the air/fuel ratio shouldn't below 10.0 or the mixture will not fire. The target a/f ratio is 12.

Below are the calculations used.

air volume per cylinder (195 ci/6 * 0.10) = 3.25 ci = 0.0018808 Cubic Feet
air mass (CF * 0.076 Lbs) = 0.00014294 Lbs

fuel needed for idle (air mass / a/f ratio) = 1.1911 E-05 Lbs

To convert the Lbs of fuel needed per cylinder at idle to injector pulse width depends upon the size of injector being used (fueling rate). Since I am using stock injectors, they are rated at 24 Lbs/hr. If we convert from Lbs/hr to Lbs/millisecond, we get

injector (Lbs/ms) = 6.6667 E-06

Finally, the ideal pulse width duration needed to get a target 12 a/f ratio is,

injector pulse width duration (fuel mass/injector rate) = 1.7867 ms

Since this assumes that the injector opens and closes instantaneously, we would need to increase the duration to deal with the injector opening slew rate of about 1 ms. When you add this to the ideal duration above, we get the final pulse width of

injector pulse width (ideal plus slew period) = 2.8 ms

What this means is that at cold startup the duration of the injector pulse width should not exceed around 3 ms; otherwise, you risk flooding the engine. As originally stated, I am experiencing a pulse width of 8 ms, which shouldn't be possible for the DME to produce for the low cranking rpm of the engine. The only logical conclusion is that the DME is bad. I hope to receive the replacement unit by this weekend and will provide an update following the DME swap.

[DRACO A5OG]

I concur with Dr Sal, place your DME into a donor car. We hate for your car to damage the replacement if your car is the culprit.

[scarceller]

Only other way I see something like this happening is if somehow the AFM signal line at the DME is shorted out to some voltage source. Meaning PIN #7 at the DME has more than 5vdc on it. But I've never seen anything like that before. Also see what the voltage is at the AFM.

[mysocal911]

"I reconnected the opened DME to the cable and connected my oscilloscope probes to pin 14 and 15 to test the duration of the pulse width during the cold engine start."

So next you need to bench test the DME ECM which should take five minutes to setup,
i.e. two powers, a ground, and the ref & speed sensors (developed by a signal generator).
Once setup, the DME ECM will generate both the spark & injection signals. You will then
be able to vary all the key inputs affecting injection time and view the results on your
o-scope.

[mysocal911]

"Meaning PIN #7 at the DME has more than 5vdc on it."

The A/D converter will not measure any voltage greater than 5 volts.

[scarceller]

Originally Posted by mysocal911

"Meaning PIN #7 at the DME has more than 5vdc on it."

The A/D converter will not measure any voltage greater than 5 volts.

Right, so anything at pin 7 greater than 5vdc will bury the input and cause the DME to think max air flow. This could happen if you have a harness issue and somehow pin 7 got shorted to a power pin.

But after more thought I think it's more likely the DME is bad, something like the TIMER0 interrupt that's part of the processor has failed. This timer is what calculates how long the injectors stay on. The software calculates the inj time turn the injectors on then kicks off the timer. The timer later expires and interrupts the processor via TIMER0 interrupt then the interrupt handler stop the injectors. If this interrupt fails the injectors stay on. Who knows but I do suspect the DME is the issue.

[ischmitz]

I once dealt with a case where the DME was modified for the MAF instead of the AFM (custom chip and different power supply). When that DME was used in a stock car it ran even north of rich....

Check and verify you have the correct stock EPROM installed and no funky wires internally going to the 35-pin connector shown below:



and go from there.

Ingo

[scarceller]

Originally Posted by ischmitz

I once dealt with a case where the DME was modified for the MAF instead of the AFM (custom chip and different power supply). When that DME was used in a stock car it ran even north of rich....

Check and verify you have the correct stock EPROM installed and no funky wires internally going to the 35-pin connector shown below:



and go from there.

Ingo

Ingo,
In my custom MAF setup I developed I do exactly this! I mod the DME so that pin #9 at the DME gets 12vdc instead of 5vdc. Pin 9 is the power feed for the AFM and if that car had a MAF in it you could be onto something.
With DME plugged in and key in run what's the voltage at pin9? It needs to be 5vdc. Also test it at the AFM it self, must be 5vdc.