3000gt MFI Problem

Sometimes I always feel like I am being tested. At work my official title is a "DBA" (Database Administrator), but in reality I do a lot of debugging and coding for a healthcare company. I'm on a special SWAT team in Development. Part of my job is to verify and fix the reproducible issues. That is the easy stuff. Other times we'll get the hard stuff. We'll have some intermittent performance problem, or some data condition that we know is happening, but we don't know how. We've had two other teams look at it, and our team is the last stop. I find this very exciting and probably why I've stayed with the company for so long. But you know what is not fun and exciting? When it's an electrical issue... An electrical issue on your favorite car... And you are not getting paid to debug it! Instead it's costing you time and money to figure it out.

Last summer I did extensive work on my 94 vr4. It had 66k miles on it and ran and drove fine. I had a lot of aftermarket parts stacking up and I wanted to have my transmission built. I figured it would be be best to just pull the motor and tackle everything at once. So that is what I did. While the trans was away I did new Valve Stem Seals, BC 264 cams, BC Cam gears, rebuilt turbos, new oil pump with billet gear, Fluidampr pulley, new timing belt components, new BOV, redid some IC piping, switched from factory coils to a Coil on Plug setup, removed my boost controller and went to an AEM controlled one, etc... The car was apart for 2-3 months and I took my time on it. Every motor bracket I cleaned, blasted, and painted. It was mid October when the transmission came back. I wasted no time getting the motor back in. The car fired right up, I topped off the fluids and all seemed fine...

The next day I went to drive it and the car shut off a few seconds after starting it. "Hmmmm. That's weird." I try again and this time it ran a little longer, but still died. This repeated. I thought it may be some safety feature in AEM EMS. Did I mention I upgraded the firmware? Maybe it would be easier to mention stuff I didn't change. The new firmware had some engine safety features, but they were all disabled. I kept messing with it before deciding to just revert the firmware. The car didn't stall after that. Since the car was warmed up I took it for a drive with no issues. I thought it was an AEM firmware issue and that all was well.

The next day I went to move it out of the garage to wash the dust off the car. And it died:/ It took me about 3 starts to move the car 100 feet so I could reach it with a hose. After the wash I did more troubleshooting. I thought it was the turbo timer. It would keep the car running for 30 seconds after shutting it off. If I didn't want to wait there was a button I could push to kill it early. Sometimes I'd have to press the button 2-3 times for it to recognize being pushed. I bypassed the turbo timer and the car didn't seem to stall. So I pulled it entirely from the car and figured that had to be it.

The following day I wanted to drive to the gas station to get some fresh fuel. While the car was warming up it died on me. I noticed I could hear the MFI relay click right before it shut off. This is a relay on the passenger side of the center console. I know other people have had problems with the relay. I had a few extra so I swapped it out to see if that helped. It didn't. To be safe I used a battery charger to test all my MFI relays. I then proceeded to test ground/power to the ECU and test continuity to the MFI relay. Everything checked out. I spoke with a friend and he suggested I try powering a light bulb through every connection. His reasoning was that even if a wire was 99% cut, a single strand would still read continuity. So I did exactly that. I unhooked my ECU to keep it safe then tested every power/ground wire to the ECU by powering a 35W foglight bulb. No dice.

Inside AEM I told it to log the outputs that control the MFI relay. There are two different outputs as the MFI Relay is really two relays put together side by side. In AEM the "MAIN RELAY" output controls one side, and then the "SWITCH1" output controls the other half. I logged these outputs using the internal logger so I could get the most accurate resolution. Here is a picture of the log when the car stalls. You can see these two outputs never toggle suggesting the ECU is not asking the either part of the MFI relay to power off.

Despite what I was seeing in the logs, perhaps AEM could still be dropping the connection somehow. I did some research online and other people leaned towards it being a faulty ECU. The AEM ECU was 10 years old. I contacted AEM, explained the problem and they suggested I send it in. They found no issues with it, but replaced the MAIN Relay inside it for peace of mind. By the time I got it back, we had snow and the roads were salted. So its not like I could drive the car if I wanted to. Regardless I wasted no time reinstalling and testing it.

It didn't take long for me to repeat the problem. Not only had I just wasted $150 having AEM look at my ECU, I still had the stalling problem to deal with. What proceeded next was me becoming an expert on the 3000gt MFI system. The first thing I did was download a copy of the 3000gt electrical manual. The MFI system spans many pages so I had to print and tape pages together.

I simplified the diagram some by drawing it out myself. Site Note: I knew I could ignore the factory Fuel Pump relay under the hood as I had already done the pin 5 to pin 2 bypass. This prevents it from using the factory resistor to lower voltage to the fuel pump under non load conditions. But seeing the diagram in person made me realize that I didn't even need to do that as AEM doesn't utilize this relay anyways. It never would activate the relay leaving the default connection to bypass the resistor.

The next breakthrough was testing the car with the MFI relay cover off. I knew I could hear the relay clicking before it would shut off, but I wasn't sure which side of the relay it was. With a suggestion from a friend I pulled the cover off and labelled each side relay with a marker. Then I started the car and had that same friend watch the relay. The Fuel Pump Side of the relay was clicking off. It's power source (PIN 7) comes from the ignition assembly. This shows where all of the connectors are.

I made a rough MS Paint sketch of it as well. You can ignore the red wire for now. When the key turns on, it sends power through the ignition harness to the back of the fuse block near the drivers feet. It goes through the fuse block (through the fuse) and out a connector in front of the box. From there it goes through a 12" long harness to another connector on the drivers side of the dash. It goes through the dash harness to the passenger side of the dash where it goes through another connector and to the MFI relay. In other words it's a long path with many connectors.

I had 3 successful warmups with no stalling if I ran a jumper wire from the back of the fuse block to the MFI relay (Now you can look at the red wire in the above photo). I had assumed it was a bad connection at a connector or maybe my fuse block had some corrosion. So that is what I tackled next. I removed the fuse block and disassembled it. I cleaned all the metal connectors with electrical cleaner. All the fuse crimps I soaked in alcohol and squeezed gently with pliers so they'd make a tight connection. The red shows how power comes into Fuse 12 via the C-82 connector. The green shows where power leaves the fuse to the front C-71 connector. I also de-pinned the wire at every connector and cleaned the pins with electrical cleaner.

This didn't help. I even tried a spare fuse block I had just to completely rule it out. Someone online had mentioned the ignition switches can go bad over time. There are copper points inside the switch that can wear and flatten out over time. I pulled mine apart and it had some wear but didn't look awful. To be safe I swapped it with a low mileage spare I happened to have. Here is a pic showing the inside.

At this point I knew running the jumper wire seemed to take care of the problem but I couldn't explain why. I measured resistance of the entire circuit and it was perfect. I tried bypassing one connector at a time to see if I could fix it, but I couldn't. The only time it seemed to fix it was when I went directly from the ignition harness to the MFI relay, or directly behind the fuse box to the MFI relay. So it was puzzling. I decided to keep investigating and check some other stuff.

The next thought was maybe the fuel pump side of the circuit had something to do with it. Maybe it was intermittently grounding out and causing the relay to fail. So at the MFI relay I de-pinned pin 5. This is the pin the AEM ECU would ground (via SWITCH1 output) to turn on the fuel pump. Then I jumped my hotwire fuel pump relay in the hatch so that it would be on all the time. I started the car and it still died. Only now I wouldn't hear the relay click before the stall as it was bypassed and half of the MFI relay was disabled. This told me that the relay clicking BEFORE the stall was a symptom and not a cause.

This meant the only logical thing would be the power wire going to the ECU from the ignition switch was intermittent. I had noticed that if I was logging via my laptop and the car died I would lose the log. The "stop and save" menu item would be disabled. But if I put the car in 6th gear and purposely tried to take off stalling it; the car would die but the log would continue. I had been doing my previous datalogging via the internal ECU logging so I could have he fastest resolution into the parameters I was monitoring. This also helped support my new theory that pin 82 of the AEM ECU was losing power.

After some discussion online, people were questioning my alternator and battery terminals. I had already cleaned my terminals and even cut and freshly stripped the wire going to the terminals. My alternator was newish, but I had painted all my brackets. Maybe it grounds through the block. So despite being a pain to do as the radiator needs to come out; I pulled the alternator. I did have some paint and even anti-seize on the mounting points. Even the engine bracket it bolts to had paint on the mounting points, so it may not have been making a great connection to the block. I cleaned all the mounting points and then reassembled.

Unfortunately that did not make a difference. I wanted to test my theory of the pin losing voltage so I ordered an oscilloscope. I had wanted one for a while and they have really came down in price. I did a ton of research and went with a Hantek DSO5202P from CircuitSpecialists. It was on sale for around $300 with free shipping.

The scope came and I began monitoring pin 82 at the ECU. This is also the same wire that feeds the Fuel Pump side at the MFI relay. What I saw was very strange. I'd see the voltage actually spike and dip at times. At some points it would actually dip low enough (around 8v) which I assumed was enough for the AEM ECU to lose power. I tested the same wire on my red vr4 and saw the same spikes, only the voltage dips were not nearly as bad.


So what was it? It turns out it is related to the Coils. This should have been more obvious to me as the Coil on Plug setup was new. But I found it strange that a spark event could somehow cause my ECU to shutoff. But that is exactly what was happening. The wiring diagram of the MFI circuit above is not complete. I don't know why the 3000gt manual leaves items off. I found this other page later on in the service menu which talks about the PTU and Coils.

I also added the newly discovered wires to my hand drawn diagram. The same power wire that feeds the MFI relay and feeds pin 82 of the ECU, also feeds the factory PTU and Coils. Now I no longer have a PTU and I'm running 6 coils with no wasted spark. So you would think that cutting the spark events in half and having brand new coils (coils that are newer in technology than the 25+ year old ones I had in the car) would be less stress on the system. But that is not what happened.

Notice the "Capacitor" on the left side of the factory diagram above. It's normal for a coil to introduce some voltage feedback back into the system when it fires. The factory capacitor (also called a "noise condenser") is the little cylinder looking thing that screws onto the factory coil pack. It's job is to absorb that noise. Now I still had this installed and I tested the capacitor using my multimeter and it checked out. The issue was that the new coils are just more chatty than the factory ones. My car must be right on the edge too as its only when the car is cold and warming up. The aftermarket coils cause more of a spike/dip than the factory ones did. The dip will sometimes be enough to cut power to the ECU.

I solved my problem by hotwiring my coils power source to the battery via relay. Technically I tied them to the ABS fuse under the hood which ties to the battery. I did the ABS delete a while ago and my battery already has enough wires going off the positive terminal. I plan on replacing the fuse with a lower 25A one. Turns out a battery is like a giant capacitor and is the best thing you could have to absorb the noise from the coils. This keeps the feedback out of my MFI ignition wire. When I monitor the wire inside the car now it's a constant 14v. Most importantly is that my car no longer stalls.

As for the red jumper wire I had tested up above. My theory regarding why that also worked is that it gave feedback a more direct path back to the battery. But I feel better about hotwiring the coils under the hood. I gave Rvenge this feedback and he's going to recommend it now for anyone installing his COP kit. Hopefully I'm the first/last person to deal with this. I'd like to thank my friend Tommy for letting me bounce some ideas off him and also the 3si community for over 100 comments on this matter. If anyone wants to walk through the thread it can be found HERE.