I was just outside, vacuuming the inside of my car, while the seat is out (in preparation for the testing Russ and I will be doing later, where we need access to the DME), and I discovered the DME relay is hot, just as it is after the car stalls. What's interesting is the face the car is not turned on, and the key is not in the ignition.

Is this normal?

I did not read this whole thread, but FWIW, I've seen 4 or 5 different cars that strangely cut out at different rpms as if the rpm limiter was kicking in early. It did not happen until the engine warmed up. In each of these cases it was traced down to bad injectors. On the dyno, the air fuel ratios get very erratic and inconsistent from run to run, and the engine would buck randomly or rev limit at 5000 or 6000 rpm. I don't know if this would be your problem as the bad injectors in these cases cause bucking and cutouts, but not for the engine to shut down.

Old and faulty injectors get sticky or have difficulty switching in a controlled manner at the extreme ends of their duty cycle, such as at a high rpm full throttle range where the injectors are called to pulse open >90% or at a very low duty cycle such as revving with no load on the motor at >4000 rpm where the injectors are called to barely switch open at a high speed at a <3% duty cycle. When they get heat soaked their coil resistance rises further which would make an already marginal injector harder for the DME to control.

No it is not normal. Time to get out the voltmeter, schematic and probe around in the DME relay socket.

But I doubt this is the main problem.

You can slide back the relay a little ways to expose pins. If I'm not mistaking, only one pin should have power. If more than one, I wonder if your ignition switch is staying on when turned to off?

That would explain hot relay and power to injectors.

As well as providing power to your DME (that is apparently turned ON).

Greetings folks,

We just finished for the night. Russ, Mark and I first attacked the issue of the DME relay and DME always being on. It turns out there was a short in the DME relay connector, which caused this issue. Once we identified the issue, we fixed the short, tested it, and put it back together, all is fine there. Then we took the car out for a test drive. The car died as usual, thus we began our testing. Note, we did not hook up the Oscilloscope yet, that will happen tomorrow night. When the car died, we checked power at the injectors and fuel pump - the volts were fine in both spots. Then, for some strange reason, while the car was dead and wouldn't start, we pulled the connector off the middle fuel injector on the drivers side. The car started right up. Then when we re-connected it, the car died and would not start. Then we tried that same test again, this time pulling the connector off the injector closed to the rear bumper, still on the drivers side. The car started right up, then when we reconnected it, the car died and would not start. Thus after the car died, we were able to start the car on five injectors plugged in and it died when we attempted to re-connect the sixth injector. Of course, after a bit of time, the car restarted with all six injectors connected. I'm not sure if this leads us to bad injectors, but it may. We plan to test the injectors tomorrow night with the o-scope.

Any and all thoughts are welcome. More to come soon. Thanks for the suggestions and hanging in there.

Scott

More details to follow Scott's post (and a few corrections):

First, we solved the problem of the voltage on the injectors with the key off. Turns out there was a short between the DME relay terminal 30 wire and the terminal 86 wire in the DME relay socket. This meant that the DME power relay was always on and the DME CU was always powered. We couldn't see it but the wires tested as shorted and removing all the wires from the socket and carefully reinserting them resolved the short. We may revisit that one because it may cause problems in the future. We also found that the black wire on DME relay terminal 86 was wired to the wrong fuse position so we fixed that though the short masked the effect. So we have all that wiring back to as it should be (the wiring harness that goes between the engine, the DME CU and the fuse panel has all be replaced so that may explain why these issues have cropped up).

We also did some road tests but didn't get a chance to open the DME CU and put the scope on anything. However this is what we found in the road test (all done after fixing the DME relay wiring):

1. We tried slowly accelerating the engine to 5000 RPM in first and second and it didn't die. Then we shifted to 3rd and jumped on it and it died.

2. When the car dies the power on pin 87 of the DME relay is constant, it doesn't cut out. So this means that the injectors have power and the DME CU has power.

3. When the car dies, the power to the fuel pump is constant, it doesn't cut out until the engine stops turning and then the DME CU shuts it off.

4. When the car dies and won't start, the coil has 12v. The fuel pump has power while we are cranking the engine, the injectors have power (e.g. there is power on DME 87) and yet the car won't start.

5. The injectors each tested as 2.7~2.8 ohms when cold. After the car died we tested the middle injector on the driver side and it was 2.7 ohms.

Then something weird happened. We pulled off one of the injector connectors to measure the voltage to ground while cranking the car and the car started on 5 injectors. With the engine running we plugged the harness back on the injector (the middle one on the driver's side) and the engine quit and wouldn't start. We then unplugged that one injector and it started. We then shut the engine off, reconnected the injector and it wouldn't start. Unplugged it again and it started. Then we plugged it back in and it wouldn't start. Then we unplugged the front one on the driver's side and the engine started. Shut down reconnect and it wouldn't start. Unplugged again and it started. So the car would start if 5 injectors were connected but not 6. We discussed this for a while, 5 min or so, reconnected the injector, and then it started on all 6 injectors and we drove it home.

So, just after it has died, it has power to the coil, power to the fuel pump, power to the injectors, but it won't start with 6 injectors but it will start with 5.

The injector harness in the engine compartment has been replaced. The harness from the fuse panel to the DME CU to the engine has been replaced with a used harness. This means that all the wire between the DME CU and the injectors has been replaced. The grounds have all be cleaned up and redone. Steve Wong has checked out the DME CU as good and other DME CUs have been run in the car with the same engine cut off effect. So could there be a condition with these 6 injectors that appears under load, disappears with time, or disappears if one injector is disconnected? By the way, with 5 injectors connected to the harness we measured the resistance between the pins on the 6th injector harness connector and got about 0.8 ohms. This is in the neighborhood of what we would expect for 5 2.7 ohm loads in parallel. So there isn't an obvious short.

Next up, put a scope on pin 14 or 15 (injector signal) of the DME CU and see how the signal looks and what it looks like when it dies. Our hypothesis is that when it dies, and when it won't start, there are no pulses or at least no good pulses on pins 14/15.

Russ

It's acting like one injector is shorting.
Removal of any one of the injectors may be enough to raise the total R enough to permit the driver to work.

You need to measure the Resistance of each injector before and after it dies. Or swap them all.

Scoping now may not help, as the driver is common to all, and there'd be nothing to measure anyway when the driver is off (in fault mode).

Nice job on the relay.

As all six injectors are wired in parallel to the DME, if the overall impedance of the six injectors exceeds a certain threshold, the DME will not fire the injectors, and you will either get a no start or shut off condition. This can happen if the impedance of one of the injectors rises just enough so that the net impedance rises above this threshold. For example, the net impedance of six 3.2 injectors at 2.7 ohms each wired in parallel is 0.45 ohms. If just one of the injectors when hot were to jump to 7 ohms or higher, the net impedance of the six parallel injectors adds up to 0.50 ohm, enough for the DME to shut off the injectors.

Steve - we are looking for a rise in injector resistance? I thought we would be looking for a drop, which would draw more current, which would overheat the output stage of the circuitry driving pins 14 and 15 of the DME.

Rick or Steve,

Looking at Rick's explanation here:
http://forums.pelicanparts.com/3868545-post194.html
is there some part of this circuit that would "overheat" or shut down under some overload condition?

When the engine won't start, should we expect to see flat line on DME CU pins 14/15 but something interesting on pin 1 of that IC, in other words, its sending pulses but the output section is hosed? Or does the whole IC shut down? Maybe that's pointless.

Other than changing all 6 injectors, what would be a solid test for an injector who's impedance is going bad? A changing ohms measurement between pins 14/15 and ground? We get a baseline when the car is cold and then another measurement right after the engine dies? We can't measure the 6 individual injectors all that quickly so I am thinking of a test done from the harness inside the car.

Russ

Did the thread say you tried multiple DMEs and they all did the same thing? I wouldn't suspect the DME.

Why not just call Motorman (see the above thread) and get 6 new injectors. The ones I have are not the same part # as what was in there but they flow the same and work great. He has used tested ones for $30 each (do all 6 this route = $180??). New are $45

EDIT If you look at that peak and hold driver data sheet link you can get a better idea of what the DME is trying to do and how if the impedance went down it would screw it up.

EDIT AGAIN: Here is the link. Start reading application information on page 5.
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00000086.pdf

Rick-I,

Regarding the DME's, I have tried 2 other DME's (all known to be good) and both gave the same results. After that, I sent my DME to Steve Wong for a bench test and repair, he then installed my DME in one of his 911's and drove it successfully for 20 "spirited" miles as he put it, and it performed well.

I spoke to Motorman yesterday, and talked to him about the injectors. He's currently out of the 364's (0280150364), which have the metal tip, but has the 360's (0280150360), which has the plastic tip. He states both work the same and does not agree with what one person stated in another post.

I've been waiting til all tests are done, mainly due to the fact I've replaced quite a number of parts, in my attempt to resolve this problem. I'm all for ordering new/refurbished and tested injectors, I just want to make sure this solves my 11 month mystery. To date, I've done quite a bit of "preventive maintenance", which hasn't solved the problem.

Tonight we will continue our tests using the o-scope, and post results later. Russ's question above about the injector resistance rising or lowering during the overheat condition is still of interest.

Any thoughts? and thank you as always!!

"This can happen if the impedance of one of the injectors rises just enough so that the net impedance rises above this threshold. For example, the net impedance of six 3.2 injectors at 2.7 ohms each wired in parallel is 0.45 ohms. If just one of the injectors when hot were to jump to 7 ohms or higher, the net impedance of the six parallel injectors adds up to 0.50 ohm, enough for the DME to shut off the injectors."

Actually this is totally incorrect. The total injector impedance needs to decrease to cause
the injector driver stage to reach its limit current and thereby not provide enough current
for all injectors to open properly.

The problem occurring here can be caused by an intermittent fuel pressure regulator sticking
at the higher pressure after a hard acceleration where the fuel demand is great compared
to low RPMs & load. Therefore, the fuel pressure needs to be monitored to eliminate this
as a possible problem. The excessive fuel will cause the engine to die as experienced.

Bottom line: Let's not provide erroneous info and waste the car's owner time and money
using the shotgun approach to troubleshooting as it appears as has happen in this thread!

How does injector impedance decrease? Isn't it static to the coil?

Impedance and resistance are different things. The measured resistance of each injector is a function of how big the coil of wire is in the injector. The impedance is going to be a function of the *dynamic* signal being sent to the injector. As Rick explained in one of his other postings, when the DME CU first grounds the injectors and the current starts to flow, the rising current in the injector coil creates a magnetic field that pushes back on the current and slows the rise of current (I think that is how it works). This is called "back EMF" (EMF = electro magnetic field) and only occurs while the current is rising or falling. As well, the movement of the metal slug that controls the flow of fuel will further alter the magnetic field in the injector and the back EMF it causes. So I think this is all quite complex and if one of the injectors isn't working right, it may upset the performance of the circuit.

However, to produce the experience that we have of the engine just shutting off, it seems like the only possible failure is that the DME CU stops pulsing the injectors entirely. It flatlines. I can imagine that if too much current is drawn some thermal protection circuit in the Darlington transistor might cause it to shut down, or maybe the IC shuts down. Hence my question as to what we should look for in that output circuit. If the Darlington or T404 or whatever shuts down, does the IC know that? Does it keep pulsing?

Russ

I'm just going to provide a guess here now and electrically it is just a coil of wire.

I am thinking after 27 years a couple of coils may have vibrated together thinning the insulation and when it gets hot makes contact.

Since that coil will have less inductance the DME sees the target peak current too soon and goes to hold mode before the injectors open.

Many years ago I had a 911 stop running when it was given a meduim to full load on the engine. It turned out to be a restricted screen element in the fuel tank. It would run all day at light load and lower rpm, but not at full throttle. The best way is to either do a volume check thru the whole fuel system or take the screen out and inspect it. I don't know if this was mentioned before....there are alot of posts to read at this point.

Take the resistance measurements at each injector, not the harness, or use your mechanic friends spare parts.

While I'd never question SW, removing one injector be it good or bad is equivalent to increasing the resistance (and impedance for that matter).

And when you did this, it worked.


The constantly-stressed heat-sink-mounted T402 would get my vote.
A better question would be: "Would a shorted injector that equalizes the voltage across all injectors in parallel affect their operation?"

Ans:
Send them out or substitution method.
No
Yes - at each injector.
I would do what I could in the 15 minute window (and it sure sounds repeatable); and no, not at the harness.


your pain is felt bro
Swapping is free.
if it solves your problem, then send out your set.

Go ahead and send 'em out to put that part to "bed"....

Doyle