Engine Starts and Dies.. Fuel Pump Loses Power

aether · Jan 14, 2012

This is my first time post on this forum... I've read through my manuals over and over, and I've searched around all the dsm forums looking for similar resolved issues to no avail. I'm a little confused right now and don't know if I should track down a new pump, another ECU, or maybe a CAS (it seems like those cause similar issues, though I think the ECU would throw a different diagnostic code).

Here is the problem and what I have done so far (everything is in chronological order):

The car starts up great every time, but dies after 5~20 seconds (usually closer to 5 than 20). It runs smooth and when it dies it does it smooth and immediately, just like turning the car off normally.

I tested for boost leaks, and couldn't find any.

I then tested fuel pressure by modifying the banjo bolt on the fuel filter with a swagelok fitting that can be capped when not hooked to a gauge. The fuel pressure went to spec (~27 psi) as soon as I started cranking and maintained proper pressure for a few seconds, but then started losing pressure quickly and then the engine just died like normal.

I then hooked a voltmeter to the fuel pump power wire at the fuel pump electrical harness in the trunk. As soon as I start cranking the fuel pump is getting 14V and it maintains 14V for a few seconds while running, but then the voltage to the pump drops to 0V just before it dies...

I then pulled the ECU and fuel pump relay. The relay tests fine, all the proper voltages and proper continuity/discontinuity when energized/unenergized. I opened up the ECU and it looks OK. I talked to my buddy who I got the car from, and he said he had the ECU caps replaced less than a year ago by one of the electrical engineers at our work.

I then put both the ECU and relay back in the car and hooked up a test light to read any diagnostic codes. It gives a code 42 (fuel pump).

Does the ECU have some way of determining power draw of the fuel pump? If the fuel pump is going out, or drawing too much power, or getting too little power, will it trip this code and cause the ECU to turn off the pump?

Any help or a nudge in the right direction would be much appreciated.

Thanks!

95eclipseguy · Jan 15, 2012

Im not sure if this will help or not, but it's what I would do for troubleshooting anyway if I didn't have a spare ecu.

Unhook the power wire at the pump and run a test wire from the battery through a switch of some kind like a toggle etc. to the pump If the pump continues to run and maintain pressure after your 20 second window at least you know it's not the pump. One less item to worry about.

It's a start.

steve · Jan 15, 2012

aether said:
I then put both the ECU and relay back in the car and hooked up a test light to read any diagnostic codes. It gives a code 42 (fuel pump).

Does the ECU have some way of determining power draw of the fuel pump? If the fuel pump is going out, or drawing too much power, or getting too little power, will it trip this code and cause the ECU to turn off the pump?

A 42 fault code means that the ECU thinks it's activated the fuel pump but it doesn't see battery voltage on the line to the pump. Pin 13 on the ECU is the sense line for the pump power. Pin 56 on the ECU is the output line to activate the fuel pump relay.

You should see pin 56 (which connected to the MPI relay pin 7, white w/red stripe) have battery voltage on it when the pump isn't running and go to 0v when the ECU is activating the pump. Pin 2 on the MPI relay is the output (black w/green stripe) to the pump and is connected to a black w/white wire that runs to Pin 13 on the ECU and to the fuel pump check connector under the hood.

While your cranking the fuel pump relay is activated by getting battery voltage on pin 9 of the MPI relay (black w/yellow stripe) from the starter circuit. Once the ECU starts receiving CAS pulses it pulls pin 56 to ground and holds it there while the engine is spinning. Every time it gets a pulse it resets a timer, if the timer expires (the ECU has seems any CAS pulses in awhile) it lets pin 56 float back up to battery voltage deactivating the fuel pump relay.

The simplest test is to ground pin 7 on the MPI relay and see if the engine stays running. If it does either the ECU or the wiring is damaged. Measuring the voltage on that circuit will also show if this is the problem. If you don't see the ECU pull the pin to ground or if you don't see battery voltage at the ECU when it's not activated the pump are both problems.

I wouldn't be surprised of it's the ECU. People changing the caps don't usually know how to fix the damage that takes awhile to show itself. The electrolyte is corrosive and unless it's all cleaned and neutralized it continue to eat at the circuit traces until they fail.

aether · Jan 15, 2012

Im not sure if this will help or not, but it's what I would do for troubleshooting anyway if I didn't have a spare ecu.

Unhook the power wire at the pump and run a test wire from the battery through a switch of some kind like a toggle etc. to the pump If the pump continues to run and maintain pressure after your 20 second window at least you know it's not the pump. One less item to worry about.

It's a start.

That's a very good point and something I have been meaning to do. It's the first thing on my list, as it is an easy test.

A 42 fault code means that the ECU thinks it's activated the fuel pump but it doesn't see battery voltage on the line to the pump. Pin 13 on the ECU is the sense line for the pump power. Pin 56 on the ECU is the output line to activate the fuel pump relay.

You should see pin 56 (which connected to the MPI relay pin 7, white w/red stripe) have battery voltage on it when the pump isn't running and go to 0v when the ECU is activating the pump. Pin 2 on the MPI relay is the output (black w/green stripe) to the pump and is connected to a black w/white wire that runs to Pin 13 on the ECU and to the fuel pump check connector under the hood.

While your cranking the fuel pump relay is activated by getting battery voltage on pin 9 of the MPI relay (black w/yellow stripe) from the starter circuit. Once the ECU starts receiving CAS pulses it pulls pin 56 to ground and holds it there while the engine is spinning. Every time it gets a pulse it resets a timer, if the timer expires (the ECU has seems any CAS pulses in awhile) it lets pin 56 float back up to battery voltage deactivating the fuel pump relay.

The simplest test is to ground pin 7 on the MPI relay and see if the engine stays running. If it does either the ECU or the wiring is damaged. Measuring the voltage on that circuit will also show if this is the problem. If you don't see the ECU pull the pin to ground or if you don't see battery voltage at the ECU when it's not activated the pump are both problems.

I wouldn't be surprised of it's the ECU. People changing the caps don't usually know how to fix the damage that takes awhile to show itself. The electrolyte is corrosive and unless it's all cleaned and neutralized it continue to eat at the circuit traces until they fail.

WOW, exactly the information I need to troubleshoot the ECU/wiring. I will get to work armed with this info and let you know my diagnostic progress.

Thanks again guys, I really appreciate it.

aether · Jan 16, 2012

A quick update.

The fuel pump is good. When I hotwire it directly to the battery, the engine runs fine.

Then, after hooking the fuel pump back up normally, I did what Steve suggested and grounded pin 7 on the MPI relay. With pin 7 grounded the engine runs fine. Therefore, it must be the ECU or the wiring...

I took the ECU out again and opened it up. Something that I forgot to mention last time was a white iridescent residue all over the entire board. I had asked the guy I bought the car form about this a few days ago, and he said it was just the cleaner used when the new caps were soldered on. However, from my experience with PCBs, a white residue is usually a BAD thing, and a cleaner that leaves a white residue is usually because of a precipitation reaction with either the PCB substrate or the solder. Has anyone ever seen this residue before? Is it possible that it is a harmless cleaning residue as I have been told? Do you think if I cleaned it in an isopropanol bath in an ultrasonic cleaner it could fix the problem, or should I just pony up the cash for a new ECU?

I also noticed some small burns on the back side of PCB. The two marked C106 and C108 are obviously solder burns from when the caps were replaced, but there is another small burn towards the corner of the board. I am guessing this is also from the iron... maybe the guy replacing the caps accidentally brushed the iron against the PCB or dropped it for a second when he was working.

I'll probably check the circuit voltages next. I would have done it today, but it was freezing outside and the car is in the street, since I don't even have a driveway at my current house, and contorting myself under the dash with my feet dangling in the street grew tiresome very quickly in 30 degree weather.

One last thing. I have found many sources for ECUs online, but if anyone on here has an idea where I could find one (maybe even locally...) for a good deal that would be awesome. I would like to get an EPROM if possible, but that is definitely not a deal breaker.