I'm not sure I would want to beat on it with only 5 cylinders operating. When it is warmed up it should start on 5 cylinders if this theory is correct however. If you read through this thread I think 2 or 3 have said injectors solved this problem.

I don't think the DME relay should be getting hot. I don't know what current the DME, injectors and idle air control should draw. I think for periods of time it could also be negative (injectors in hold mode). Does the relay only get hot after it fails?

The 12 volts with the key off is hard to explain if it is not the relay stuck. However the DME is pretty dumb and shouldn't care when it got power.

Again the voltage falling off at the fuel pump (shares a common path in the relay) is interesting.

Rick,

That kind of feedback about running on 5 cylinders is what I was looking for, thanks.

The biggest current draw on the DME relay is the fuel pump and since we put a relay on that it should no longer be an issue. I'll research the other items.

I didn't see the voltage measurement dropping off at the fuel pump, that was Scott's mechanic and since there are several variables there, I don't know what to think of it. Such a voltage drop is a possible cause but when it quit on us on Saturday I confirmed the voltage at the fuel pump (voltage on red/green wire from DME relay) was fine: 11+ volts while cranking the engine that wasn't starting. We got it running again I think by pulling the DME relay - Scott wanted to show me how warm it was. We also found the negative terminal on the battery was loose and tightened that. But as soon as he floored it, it died again and it was during that failure that I noticed the voltage on the injector harness while the key was off. After that weather moved in so we hightailed it home when we got it running again.

Russ

I found this from the other thread: http://forums.pelicanparts.com/porsche-911-technical-forum/694968-3-2-voltage-injectors-key-off-position.html.

Not sure what to make of the 12V on the injectors with key OFF.

My heart goes out to you and your mech. I was really trying to follow this, forgive me if I missed something.

Have you swapped the injectors (all 6 at once)? Consider sending those out?

I tend to agree with Tippy - you should try to instrument the DME, as hard as it sounds, to see what DME pins 14 and 15 are doing before and in failure mode. On the Dyno so you can have the scope in the car.

Pins 14 and 15 provide a gnd pulse when it's working and it is too quick for a DVM. (This is the grey/x wire side of the injector) If you use the scope, it needs to be done in-circuit.

For anyone, with the injectors disconnected, what should the DME pins 14 &15 read to gnd? Some floating value? I am not sure that measurement is of any use when the injectors are disconnected. You want to test your theory that the injector driver stopped, and this can't be done if the 12v from the injectors aren't connected to 14/15 (in circuit)

If you don't want to run with 5 injectors to failure, then drive to failure first and then discon' an injector, one at a time to see if she starts w/o it. (is the 15 minute break reliable?)

other than that, I am leaning toward the DME, but the fact that this happened with 2 DMEs throws us off.

if the FP is really ON (in failure)?, that should elim the speed/ref portion of the DME.

disregard with apologies if you mentioned these before - it's a longish thread.

best of luck to you.

Voltage on lines that shouldn't have voltage on them always brings me back to a bad ground someplace.

I have not seen a conclusive end to this one. So here is my advise:

Once the issue happens you need to measure with a portable oscilloscope across the injector and and across the coil while cranking (no start). Each measurement will tell you whether there is real current going through the devise. For each you expect a flat line with pulses. If interested I can post pictures of typical scope traces you would expect during cranking.

Then while you have the no-start measure each side of the coil and injector with reference to GND (at the battery). You expect to see nice clean +12V permanent on one side of the coil and injector and the GND pulses on the other.

This data will tell you which of the four signals isn't there or is corrupted:

- clean +12V to one side of all injectors
- GND pulses with PWM pattern on other side of injectors
- clean +12V at one side of coil
- charge pulse with spike on other side of coil

Follow the missing signal and you will find the root cause

Off topic but then very similar: I had my ML430 dying randomly on me. It ran weeks and then shut down cold, came back within an hour, ran days, died, etc. Drove me nuts. Turned out to be a bad ignition switch. It was just bad enough so that switched +12V to the coil packs dropped to +4V during dwell time and as result spark was weeeeeak. The rest of the switched +12V goes through a relay that is controlled by the ignition switch and a logic module. So it was only for the millisec dwell time of the coil charging where the T15 dropped. Not enough to affect the rest of the terminal 15.

A test light on the spark plug wire showed spark and it took me days to recognize my error (believing I had spark). The highly intermittent nature didn't make matters easier.

Another thought.... Just for completeness you should unplug the oxygen sensor heater. That could short as it gets hot. Of course this would not be load related as implied before and the injectors would be.

The theory here is the unfused heater draws the voltage in the DME relay distribution land pattern down to the 6 volts observed and a bunch of stuff in the DME stops working.

This sounds like a road map for our next session. Please post the scope trace pictures. I've got a dual trace scope so I can put it on each side of the coil and the injectors and verify that the high side is staying at B+ while under load. The scope is battery powered though I didn't buy the battery for it, so I'll have to sort that out.

Thanks,

Russ

OK, I will take some scope shots tonight. I'll set the RPM to 200 to simulate cranking on the bench tester. For an in-car measurement it actually doesn't matter whether your scope is battery-powered or powered through mains or a charger.

When I work on my bench-tester I have to ensure I don't short out anything with the oscilloscope GND clip. That's why I prefer my battery-powered portable scope for that.

Below is my bad ignition switch in the ML430. Note there the 4 millisecond pulse with a voltage drop of 8 volts. At the end of the dwell time you can see ringing meaning there is a spark event of sorts. But it is so weak it won't run the engine. The measurement is taken between terminal 30 (perm. +12V battery) and terminal 15 (switched +12V from ignition switch, ON and START) during cranking.

http://forums.pelicanparts.com/uploa...1344968424.jpg

If this would be simple to do with your test bench I wouldn't mind seeing that either.

ischmitz - If I understand your scope picture, if your ignition switch had been working correctly, there would be no voltage drop across it so the voltage at the hot side of the coil would not vary as the coil was charged and so would the scope show the ringing or just nothing? My guess is nothing, like the scope probe was connected to B+ and the battery would just sit at some constant voltage. Of course, during cranking, B+ probably varies quite a bit but with the engine running, it should be pretty smooth.

So I am guessing scott's engine will start right up and once it does, idling in the driveway, I'll have the A channel of the scope from coil hot to ground and the B channel from the low side of the coil to ground. The A channel should show around 14v pretty steady, the B channel should show pretty regular spikes and ringing, right?

If I do the same thing across the injectors, picking up the hot side of the injectors on one channel and the DME CPU side of the injectors on the other, again the hot side should be pretty steady and the CPU side should have the wave form that Rick referred to earlier.

If we get that far, we drive the car and watch the scope and see what changes when the engine turns off. We are going to look at injectors first and spark second.

Sound reasonable? I've yet to see the DME unit under the seat but I am assuming that I can get access to the necessary injector wiring somehow from there so that I don't have to try to run scope probes into the engine compartment.

Advice welcome.

Russ

Guys,

Please check my DME relay operation understanding here:

1. Ignition key in start or run position sends B+ out terminal 15 of ignition switch through red/white wire to fuse box fuse 7 which is jumpered to a post between 7 and 8 and to fuse 8.

2. From the hot side of fuse 8, B+ flows across a black wire to DME Relay terminal 86, through the diode, through the resister/coil of the DME power relay, out terminal 85 and through a brown wire to ground at MP XII.

3. This causes the DME power relay to close which conducts power from B+ on terminal 30 of the DME relay to terminal 87 and through a red wire to a junction that provides B+ to:
- the two banks of injectors
- the idle positioner
- pin 18 of the DME CU
- pin 35 of the DME CU
Pins 18 and 35 appear to power the DME CU itself.

4. Under conditions that aren't clear to me (engine RPM over a certain amount, e.g. cranking?) the digital electronics in the DME CU send point ADV7 high which causes the transistors at T480 to conduct to ground which in turn causes pin 20 at the DME CU to conduct to ground which takes the black/white wire going to terminal 85b of the DME relay to ground thereby closing the fuel pump relay and feeding B+ from DME relay terminal 30 to DME relay terminal 87b.

5. B+ flows through the red/green wire from DME relay terminal 87b to fuse 3 and then through the red wire to the fuel pump.

Have I got that right? If the engine dies and the conditions in step 4 that I don't understand are not met, the fuel pump should shut off, correct? For example, if the car is idling and I stall the engine and leave it sitting there with the key on, the B+ on the red/green should go away, right?

But if the engine dies but is still being spun by the momentum of the moving car, I assume that the conditions in step 4 are still being met and pin 20 will still be at ground and the fuel pump relay will still be latched. Right?

If I put a meter or a scope on pin 20 or DME relay 85b, I would expect to see 0v in the key off position rise to B+ in the run position and then drop to near 0v when the engine is cranked. Right?

And DME relay pin 85 should be ground, hard ground. If that isn't right at 0v, that could be a problem. Right?

Remember I saw 12v on the injectors while the key was off? Well that was terminal 87 of the DME relay and if the DME power relay was open, the only way to get 12v to that point is by flowing back through the DME fuel pump relay, e.g. the power was coming from pin 20 on the DME CU itself. Looking at the DME schematic, this means the power was coming from point ADV7 or from pin 4 of the DME CU which appears to be wired back to the start position of the ignition switch. Do I understand that correctly? Just trying to identify things I should test if we see this again.

Thanks,

Russ

I read the majority of the posts, and it recalled a problem a shop once told me:

While chasing a running issue on a 928, many hours into it, they discovered one of the boards the relays plugged into had a bad wire/connection in the back of it. I think it was one of the power legs.

I am not sure of the socket for the relay, (my '88 is 10 years sold) but is there a way to look at the back and verify no cracked power legs/wires loose?
If something was loose, this may cause a heat issue, explain why a relay swap solved the issue in the past...

If you are seeing battery voltage on the injectors with the key off it is probably not coming from the DME.
Earlier in this thread didn't someone add some wiring as a supplemental path?

http://forums.pelicanparts.com/porsche-911-technical-forum/431975-new-dme-relay-did-not-make-my-fuel-pump-work.html#post4198034

How is the scope working?

I removed that wiring (red/green from the DME to the fuel pump fuse 3 had been moved to another fuse - this resulted in the car continuing to run after the key was turned off). I don't think there is any more such wiring. An additional relay has been introduce for the fuel pump - the red/green from the DME relay is now powering this new relay which in turn connects the fuel pump directly to the battery. I have left that in place for now though I need to add a fuse in the new feed wire from the battery. All this was done before I saw the voltage on the injectors with the key off.

Note that the voltage problem is so weird that I am not sure I believe the result - there may have been "pilot error". So I will be looking to reproduce that.

We are going to try the scope tonight weather permitting.

By the way, getting access to all these wires while the car is doing 50 MPH down the road is making me think I should solder a cable onto the inside pins of the DME CU connector so I can access them freely. Scott doesn't think there is any way to get the DME CU and its harness out from under the driver's seat so I can get at it easily. Do you agree?

Russ

I'd agree. I gave up on building a DME Break Out Box. Non-working DMEs/wire harnesses are pretty expensive.

I thought the symptoms here were you could run it till it got hot and then it would quit. Can't you look at the signals when you first start it and them when it quits?

Since I'm getting a number of folks indicating the fuel injectors may be the culprit (something we're going to test later), I've decided to look into replacing them. Can someone point me to instructions on how to remove (and later install) the fuel injectors on an 85 3.2?

Thanks.

Doesn't seem to be heat related. Seems to be load. E.g. when the engine is making maximum horsepower, it shuts off. And note, this is not "stalling", its like you turned the key off. So this really seems to be an electronics issue - something that occurs at maximum load is causing the DME CU to shut the engine down.

We could run it till it quits and then check for weird conditions, we may do that today and then tonight I'll solder a break out cable on to it for future testing.

Russ

I just did this a couple months ago.

On the drivers side remove the heater duct
On the passenger side put the AC compressor on a towel on the fender. Remove the air filter.

Remove the compression nuts from the front and back of each fuel rail. Some gas will spill, let it sit for a while before dong this.
Remove the connectors from the injectors, they just pull off but some you may need to remove the metal clamp wire.
Take the bolts out that hold the fuel rails on and work them off. My injectors all stayed in the manifold.
Work the injectors out of the ports in the manifolds. One of mine was stuck real good.

Put it back together lubricating the new o-rings with gas.

Did I forget anything?

Which is why the injectors are suspect. When the manifold absolute pressure goes up the injectors have to squirt more gas.

Ingo, Is it just pulse width or does the frequency also change?

Here's how I built a break out DME. I got a spare DME, opened it flat and bolted it to a piece of plexiglass. Now all the pins are easily accessible using small hook type alligator clips. With the DME like this it's also easy to make software changes by swapping EEPROMs or using the Moates emulator.

I'm no SW or Iscmitz but I'm not bad at ignition/fuel tuning, plus I've made a few changes for faster starts.

I know what you're thinking, a spare DME isn't exactly cheap (I got lucky and found one cheap - it was broken but I fixed it) but if you're making the decision to work on these cars yourself, then take some of that saved money and buy spare parts and tools (like the o-scope mentioned above) to be able to help in diagnosis.

Chuck.H
'89 TurboLookTarga, 338k miles