hillpc

Maybe this info can help both of us. My 97 A8 sat for 4 years in Canada, and now cranks but won't start. I had suspected the engine speed sensor, but it tests out OK according to the resistance specs in the service manual. When pulled out of the block or unplugged, it throws a code, and the engine makes absolutely no sign of wanting to start. When the code is cleared and the speed sensor replaced or plugged back in, the car seems like it wants to start every second or so while cranking. I have not pulled and cleaned spark plugs.

I tested spark by pulling a spark plug (and coil) and grounding it while someone cranked the engine with me watching for spark. First I got no spark. Then I got a yellow spark on another cylinder, which I assume is too weak. Then I got no spark on the same cylinder. Maybe I'm doing this wrong.

The ignition coil primary connections and voltage supply check out OK according to the service manual (seeing power at all 8 pins at the ignition coil power output stages). Here's something odd, though - seems that mice had made a nest right near the ignition coil output stages. I looked but did not notice any chewed wires, though. I think the next step is to figure out if the coils are being pulsed to fire by the ECM. I'll have to dig into the details of which terminals to probe unless one of you guys can tell me how to do this. I also do have a scope (Picoscope) available.

Then check the fuel supply. I drained all the 4-year-old fuel from the tank and put in fresh. (It smelled weird to me.) I can hear the fuel pump run every once in a while for a couple of seconds when I turn the key to "run". I hear that cracking a pressure fitting to look for spraying fuel all over the engine compartment is a way to check for supply to the injectors. The fuel pump fuse is good and a test light at the fuse pulses as I crank the engine.

Then the fuel injectors. I need to find the wire that is switched to ground to make fuel flow for each injector and see if it is actually being switched to ground or not.

Not sure if there's an easier way to go through all this. I have VCDS but don't know if that will help. The only code I get after clearing them all and then cranking the engine is "18032 - MIL Request Signal Active (Check TCM for errors too!) P1624 - 35-00 - ." But then again, I think the check engine light is supposed to be on while cranking.

Kneale Brownson

There are TWO sensors, the engine speed and the ignition timing, that are buried together on the driver side of the spot where the engine block meets the bell housing. You need GOOD signals (there's a resistance spec) from both to permit a start. One or both fail routinely on V8s because they're buried under the exhaust piping on that side and get exposed to heat.

hillpc

What's the second one called and numbered? The engine speed one (pointing directly at the side of the flexplate [flywheel]) is G28.

BrianC72gt

If so, clear them then crank the engine and see what new codes, if any, show up. There are no start conditions that will throw codes, and there are those that won't.

(**Edit ** you did clear the codes, just a Brake error now **)

BTW, I've never seen a crash detection error except the one time my ECU went for a swim. Mine was bad, the ECU box got flooded, and it created all kinds of short circuits, cooked my fuel pump, fried half the coils, and left the ecu with a permanent altitude correction error - caused by an internal air pressure sensor really didn't like getting wet. You may want to unplug the ecu and hit the connections with electrical contact cleaner., then connect it/disconnect it a couple of times to clean off the surface corrosion . It DID get wet, so that would be a good place to start.

Testing the fuel injector plugs for a switched ground while cranking is nearly impossible with a standard multimeter. The injector is switched on for milliseconds. A noid light or an oscilloscope are fast and sensitive enough to see it clearly. Noids are pretty cheap. Or you could use a mechanic's stethoscope against an injector body to hear it click. You might want to calibrate your ear against a known good working car while cranking so you know what it should sound like. Is the CEL behaving as it should? On in RUN and START until the car starts. OK, It might go out while cranking if the starter spins the engine fast enough, but it should stay on for a little while while cranking.

hillpc

Darn, there's only 1 sensor (G28 engine speed sensor, at the very front of the transmission) on my '97. OP fsrowsell - does your '98 have 2 or 1 sensors at the very front drivers side of the trans? Seems that at some point they had a second nearby sensor G4 that reads crankshaft position by sensing a rectangular pin on the flywheel. (The G28, as far as I know, counts flywheel teeth.) The thread at quattroworld.com Forums: Detailed Location and Size of the G4 Crank Position Pin has lots of info on the pin. The thread at https://www.audiworld.com/forums/aud...on-v8-2196802/ has a drawing that shows both sensors G28 and G4 screwed into a holder that bolts on; it seems to show the model year 1990, on a page from EKTA, whatever that is, with model listed as "V8". And https://www.audiworld.com/forums/aud...-info-2802889/ has more on the sensors G28 and G4.

Without the G4 on my '97, how does the ECU know where top dead center is? It seems the only other sensors that could tell that would be the camshaft position sensor G40, on my '97 at the back of the driver's side cylinder head (but only if the timing belt's not broken or slipped). But all the info I've seen so far says that a bad G40 shouldn't keep the car from running or starting. (I'm gonna check it anyway.) I'm puzzled now, unless the engine speed sensor G28 somehow also can read position.

silverd2

D2's only need the one sensor, G28 engine speed sensor ("crank position sensor"), which reads rpm, timing and the fact that the engine is turning while running and during starter operation. If G28 doesn't see the engine turning, the ECM cuts electrical ground to the fuel pump relay.

G28 reads off the flexplate, just like an ABS sensor reads off the slots on the CV joint ... a Hall Sensor.

The flex plate has a reference point "cut out" of the slots that is used for G28 to read rpms and engine timing...it knows from this reference point where TDC is.

Here >>

quattroworld.com Forums: Just 1 sensor (G28) that reads like an ABS wheel speed sensor. The starter gear/TC >>>

hillpc

Thanks, silverd2. I've seen enough of your posts that I believe you could definitely claim that you are, in fact, a guru (whatever that means). Is that snow in Tennessee in your picture? My cousin in Canada told me he'd routinely drive around stuck SUV's in (now) my silver D2.

Thanks again.

hillpc

The G28 engine speed sensor reads 150-250 rpm (on a VCDS tool) while cranking. (But the dashboard tach doesn't move while cranking.) 12V positive voltage is seen at all the coils' terminal 15. The power output stages are getting NO positive 12V pulses from the ECM, and the wires that are supposed to be carrying those pulses from the ECM are intact and not shorted to ground. The ground wire to each output stage is good. The ECM looks brand new inside, no obvious burnt components. Many posts claim it's extremely rare for ECM's to fail. And why would it go bad just from sitting for 4 years?

What would stop the ECM from sending the ignition primary pulses to the output stages? The fuses next to the ECM are good when I touch a test light to both sides of each fuse, and the relay near them clicks when I plug it in and out with power on.

Kneale Brownson

Sorry, I was going off my experience with the original V8s and 200s, and didn't know they had combined ignition reference and engine speed sensor functions into one as SilverD2 says. Ya learn something every day, eh? Lack of tachometer movement while cranking used to be a symptom of a failing engine speed sensor.

There should be a static test procedure for signal output from the G28 at its connector, I'd think.

hillpc

I suppose one could somehow put a scope on it, but the fact that VCDS reads rpm while cranking makes me think I need to look someplace else for the no-spark culprit. The resistance checks show good anyway, so I have 2 pieces of evidence that G28 is good.