By John Anello
The Auto Tech on Wheels enters another dimension of troubleshooting when he encounters a case of profound ignorance and incompetence.
A shop called me in for a hard-start-when-cold complaint on a 1998 Mitsubishi Eclipse with the 2.4L engine (Figure 1). This shop had spent many hours trying to resolve this cold starting problem that two other shops had been unable to fix. The laundry list already included a new fuel pump, injectors, coil pack and ignitor assembly from the prior shops, but the current tech had decided to replace the fuel pump again, thinking the one just installed didn’t have enough volume. When the second pump didn’t cure the condition, the coolant temperature sensor was changed.
Why CTSs are always replaced for a cold-start problem is beyond me. I guess this tactic dates back to the early ‘80s when on-board computers were first introduced. The CTS was linked to this type of trouble because it seemed logical that the culprit would be the only component that responded to temperature changes. Well, there are far too many more variables involved in today’s cars to warrant this as a good practice anymore.
Too Many Cooks
This poor little car seemed like a soup kitchen with too many cooks lending a hand. It had been to so many shops who made so many attempts to solve the mystery that you have to just wonder about the patience of the vehicle’s owner, and marvel at his drive to find the root cause of the problem.
I was called in to see this Mitsu during a week when New Jersey saw some single digit temperatures. The tech could not get the vehicle started unless he sprayed carb cleaner into the intake. Once the car fired up and reached normal operating temperature, there was no hard-start problem until it got cold again. There were no codes in memory and the fuel pressure was right at spec. When I arrived at the shop, the car was still cold, so this was the perfect opportunity for diagnosis. I decided as a first step to place my emissions analyzer probe up the tailpipe to test the cranking emissions. As I cranked the engine for 15 seconds, I could see by the gas levels (Figure 2) that too little fuel was being delivered. The HC only peaked at about 1,675 PPM, while the CO climbed to about 0.61%. Notice how the CO level begins to drop off due to the lack of fuel delivery, thus creating the hard start. This engine definitely had spark occurring in the cylinders or there would not have been any CO at all. There was some combustion taking place, and this is even more evident with the O2 level dropping to 14.3% and the CO2 at 8.0%.
Data Looked Fine
I wanted to view some data parameters, so I hooked up the Escan to check some basic sensor inputs. It is always important to validate the air temperature and coolant temperature to make sure they closely reflect one another on an initial start-up. Any large deviation in readings could affect fuel delivery, especially if the CTS was giving false information. The readings in the scan data (Figure 3) while cranking the engine all seemed within spec and showed no signs of a lean fuel command. I wanted to see what the injectors were doing — were they all being triggered by the drivers properly? — so I hooked up my EScope.
I connected my scope to each injector and tagged another scope lead to the cam signal line. This is very important because you really want to check the frequency of your injectors while the engine is cranking or running. Most techs get so focused on the pulse width of an injector that they forget to take into consideration if a pulse or two are lost during one revolution of the camshaft. You need to make sure that the injectors all fire without any dropouts in the process. The injector patterns(Figure 4) seemed to be okay and they were all firing without missing a beat. Yet this car still wouldn’t start. The engine had good fuel pressure, rebuilt injectors, good sensor readings and even good fuel quality without any contamination. I was obviously overlooking something.
I decided to add another tool to the arsenal of equipment I was using, just to get some enhanced data to look at. I pulled out my factory Mitsubishi MUTT III scan tool and started scrolling through a list of data parameters not available in OBD II generic. I need to stress that OBD II generic information is very helpful for fixing many drivability problems, but sometimes you may need the added factory enhanced features to perform bidirectional controls, functional procedures, or just to view data parameters that are not available in the generic OBD II format. A scan tool provides the window to look inside the PCM to see what it sees, which will determine what it does, its strategy output. Having this enhanced data allows me to look a little deeper into what may be creating a lean fuel command during a cold engine start. The only thing that was not replaced on this vehicle yet was the PCM. Given the track record of these particular vehicles, there was a good chance that the PCM might indeed be bad. Then again, replacement could lead to an unwarranted cost for the customer if the problem still remained.
While I was sifting through the data there was one parameter that caught my eye. It was the starter crank input. I did not realize that this car even had one in its strategy to help start this vehicle. I immediately selected the rpm, fuel pulse command and the starter crank signal. As I cranked the engine, the scan data (Figure 5) revealed that the PCM never saw the starter input. How could this be? Was there a blown crank fuse like the ones used in some of the early Cadillac DFI vehicles? I now had to pull a diagram to find out how this circuit was laid out.
After viewing the diagram, I could see that the wire came off the ignition switch and fed both the PCM and the starter solenoid. I was looking around in the engine compartment when a heavy blue-coated wire running along the firewall caught my attention. Why didn’t I notice this during my initial visual inspection? Someone had cut the starter solenoid wire (Figure 6) and run it directly back to the ignition switch to splice into an aftermarket auto start alarm system. This cut wire bypassed the PCM wire needed to help start the vehicle. Once I saw this, I temporarily tied the PCM crank signal line into the starter solenoid wire and viewed the scan data while cranking the car again (Figure 7). The engine started up immediately once the PCM received the crank signal input from the starter, as seen on the MUTT III scan tool.
I went back to the scope to view the injector firing pattern and it was very easy to see why this vehicle had been hard to start cold. Notice how the PCM banks all the injectors together at the same time (Figure 8) in a non-sequential format to deliver four times the amount of fuel to get this engine started on an initial start-up. This is typical of a cold cranking enrichment strategy that I was aware of, but simply overlooked. This type of strategy was widely used on many OBD I systems, and relied on a starter input signal to the PCM from the ignition switch, starter terminal, or through a starter fuse. The starter fuse method usually became a problem when someone would borrow this fuse and not put it back, thus creating a hard cold start. The starter terminal method became another problem during starter installations where someone would forget to put the wire back in place, or, worse yet, mistake it for a fuse link and attach it to the battery stud, which would put the vehicle into a constant fuel enrichment mode. The third choice of running the starter input off the ignition switch left the odds of anything going wrong very slim to none. But, as you can see from this story, there are those special auto guys who can defy all logic and defeat all obstacles.
This was not a normal car to work on, but more of a product of the Twilight Zone. This car had entered another dimension and been taken to places that no car has been taken to before. Places where anything goes and diagnostics become open season with many different avenues open. Only the brave will survive these journeys and come back to tell the story. This crazy situation came about at the hands of some uneducated individual who decided that with a pair of diagonals, wire strippers and crimpers and electrical tape, no job was too big or too small. I personally know many aftermarket techs who are really good at what they do, but then there are others who just serve to give us that job security we need by cutting and splicing wiring harnesses without having any concern for or understanding of the kind of problems they may be creating along the way. It bothers me that the owner of the vehicle never noticed any problems after the alarm was installed. Perhaps it was installed during the heat of summer when the ambient temperature had no affect on starting the engine. Then, the cold weather moved in. I’m just glad the problem was found and hope that this story can open your mind to the possibilities of what may await you when a tampered-with car arrives at your shop.