I confronted this diagnostic problem on my daughter's Saturn. The car is a 1998 Saturn SC2 with a 1.9L engine and 45,000 miles on the odometer. The car has been running fine and has required no repairs to date. But recently the car has experienced a no start problem. It cranks normally but refuses to start. The problem only happens occasionally, usually after it has been driven, shut off and allowed to sit awhile.
The main challenge with an intermittent driveability problem is getting the car to act up. You can't diagnose the fault if the car starts and runs fine. So you have to try to duplicate the conditions that cause the driveability problem to occur.
After driving the car for several days, I finally got it to act up for me. I was running some errands, came out of the store, tried to start the engine and it wouldn't start. Okay, now I could finally do some quick checks to figure out why it wasn't starting.
When an engine cranks but won't start, there can only be three basic causes: no spark, no fuel, or no compression.
Compression obviously was not an issue here because the car ran great. The Saturn engine also uses a timing chain rather than a rubber timing belt so there is less risk of the cam drive breaking or jumping time.
That left fuel or spark as possible causes of the no start. My initial hunch was a fuel pump that was going bad. Sometimes a fuel pump with a worn armature will not spin when the ignition is first turned on because the brushes are not making good contact with the armature. Sometimes if you shake or jiggle the car, the sloshing fuel in the tank will move the pump just enough so the brushes make contact and the pump starts to spin. My daughter said she tried shaking the car and sure enough it started after several attempts.
If it was a bad fuel pump, Saturn dealers get $300 for a new pump and $200 labor to install it. I called around to some local auto parts stores and only found one who carried a pump that would fit the Saturn. Their price was $158.
Another possibility for the no start might be a weak fuel pump relay. When the ignition is turned on, the computer energizes a relay that routes voltage to the fuel pump inside the gas tank. If the relay fails to close, the pump receives no voltage and does not run. On this car, the fuel pump relay is located on the fuse panel (which is on the passenger side of the center console under a plastic panel).
A new fuel pump relay from Saturn costs $17. Given the cost difference between a new relay and a new pump, I thought I'd try replacing the replay first -- and if that didn't solve the no start problem, then I'd replace the fuel pump. Or, I could simply the fuel pump relay on the fuse panel with the one right next to it (both are identical and the other runs the blower motor) to see if that made any difference.
Now that the car would not start for me, I had a window of opportunity to find out why. I opened the hood, disconnected the number one spark plug wire and attached the end to a spark plug tester (a spark plug with no side electrode and a ground clip so it can be mounted to the engine). Using a spark plug tester is a lot easier and safer than trying to hold a plug wire near a ground while the engine is being cranked.
I got back in the car and cranked the engine to see if it was getting any spark. If it had spark, then the problem had to be the fuel pump or the pump relay. If it had no spark, then the problem had to be ignition-related.
Guess what? No spark! I could now rule out the fuel pump or relay as possible causes of the no start.
When I cranked the engine a third time, it suddenly started. The intermittent was gone and it was running normally again. I drove the car home, hooked up my scan tool to check the computer for codes -- and of course, found NONE! No help here.
I figured the problem was probably a bad ignition module. Why? Because if it was a bad crankshaft position sensor, surely the OBD II system would have detected no crank sensor signal while cranking a set a code. Since there were no codes, I assumed the fault was probably in the module or the wiring between the module and PCM.
The Saturn 1.9L engine has a "waste spark" distributorless ignition system. A pair of ignition coils are mounted on a module block that is bolted to the front of the engine. Coil #1 fires cylinders 1 and 4 simultaneously, while coil #2 fires cylinders 2 and 3.
If the ignition system had a bad coil, it would affect two cylinders but not all four. The engine might start but would only be running on two cylinders. Since the engine ran fine, I reasoned the coils were probably okay. Just to be sure, I checked the each coil's secondary resistance between their two towers. The spec said the coils should have 8k to 15k resistance. Both tested just under 8K. Close enough.
Next, I checked voltage and ground connections at the ignition module.
Ignition to the module with the key on was at battery voltage, and the ground connection was good. I also checked for any sign of corrosion or looseness in the connector. Everything looked fine.
I also checked the feedback voltage signal from the module back to the PCM with the key on, and then again with the engine running. With the key on, the voltage in circuit 633 measured 5.03 volts which was within specs (4.5 to 5.5 volts). With the engine idling, it measures 2.95 volts (the specs say 2.0 to 4.0 volts). Again, okay.
I was getting nowhere fast with my diagnosis. Maybe it was the crankshaft position sensor. So I crawled under the car, located the crank position sensor above and behind the starter on the backside of the engine, and removed the sensor. I checked the sensor's resistance and found it measured 790 ohms (the specs allow 700 to 900 ohms), so it appeared to be okay also. Bad connector or wires? I checked those also and found no problems.
The crank position sensor (CKP) has a magnetic tip surrounded by a coil. The sensor produces an alternating current signal when slots in a ring on the crankshaft rotate past the tip of the sensor. The PCM monitors the frequency of the crank signal to determine engine rpm and the engine's firing order. If there's no signal from the crank sensor, the PCM won't trigger the ignition module or the fuel injectors, and the engine won't start or run.
Having done all of the above checks, I assumed the problem had to be a quirky ignition module. Years ago I owned a Ford Escort with a TFI ignition module mounted on the side of the distributor. The car started and ran fine, but occasionally would not restart after being driven. The problem was a poor module design that would overheat and stop working -- usually after the engine was shut off. If you let the car sit awhile and cool off, the engine would start and run fine. The Saturn problem seemed like the same problem to me, so I assumed the Saturn must also have a bad ignition module.
A trip to the local auto parts store and $180 later I had a new ignition module. I installed it and assumed the problem was fixed.
The original spark plugs had never been changed and were showing some wear, so I also replaced all four spark plugs with brand new AC plugs (same as the original), and all the spark plug wires, too ($34 for Bosch premium magwire), just to be safe -- even though the original wires appeared to be in good condition and tested within specifications (1.5K to 15K ohms).
This should have been the end of this story, but it wasn't.
A few days later the car would not start again. I was running fine, but when I went to restart it it cranked normally but would not start. After the car had sit for about an hour, I tried again and it started right up and ran fine. No what?
In frustration I checked AllDataDIY.com for any technical service bulletins that might pertain to this problem. I found Bulletin No 98-T-49A that described a similar condition. I followed all the diagnostic checks recommended in the bulletin and essentially came to a dead end. Everything checked out fine.
So I put up a post on the International Automotive Technician's Network (www.iatn.com) describing the intermittent no start problem and asking if any other technicians had encountered the same problem. I received back about a dozen responses and all offered the same advice: replace the crank position sensor.
As it turned out, they were right! They said Saturn has had a problem with bad crank position sensors. I replaced the crank position sensor ($14 from the local auto parts store) and it solved the intermittent no start problem for good!
But why? The original crank sensor had tested fine -- or so I thought. The problem was that I was testing the sensor at room temperature, not when it was hot.
I wanted to confirm I had really fixed the problem, so I took the old crank sensor and dipped the end in a pan of hot water on the stove. I hooked up my ohmmeter to check the sensor's resistance as it heated up. Sure enough, when it got hot, the resistance suddenly shot up and then it went completely open. It stayed open until it cooled back down to room temperature, then it returned to normal and again tested within specifications. That would explain the no start problem when the engine was hot.
Apparently, the sensor would stay within specs and deliver a normal crank signal while the engine was running. But when the engine was shut off, the crank sensor would absorb heat, get too hot and quit working. It would stay open and produce no signal when the engine was cranked until it cooled down. Then it would work normally again and generate a good signal to the PCM.
With no signal from the crank sensor, the PCM would not trigger the ignition module or the injectors. In retrospect, what I should have done was also checked an injector to see if the PCM was sending it a pulse signal when the engine would not start. If there was no injector signal and no spark, that would have told me the PCM wasn't picking up a crank signal.
Another mystery in all of this is why the PCM never set a code for the crank position sensor. If the PCM was not receiving a signal from the crank sensor while the engine was being cranked, it should have set a fault code but did not. I guess the engineer who programmed the onboard diagnostics on this particular application didn't take that into account when he set up the perimeters for setting a crank sensor code. Thanks fella for the curve ball. I've always said if the engineers who design this stuff had to actually diagnose and repair cars for a living, onboard diagnostics, accessibility and repairability would all undergo some drastic improvements. Maybe someday, but I wouldn't count on it...