"It doesn't do it all the time"! Do you still cringe when a customer says that? I know I still do. For many of us intermittent problems are bittersweet. It's a love hate relationship. In one way you hate the large amount of time you must spend testing or even test driving the vehicle. On the other hand, once you have figured out the problem, the solution can bring an overwhelming sense of satisfaction. It is that sense of satisfaction that most of us technicians live for. 

Qualify the vehicle. 

In our industry an "intermittent" is generally a problem that occurs at least once per day. That being said, you need to begin to qualify your customer. "Just how intermittent is it"? This should be one of your initial questions to a customer who requests service for an intermittent problem. I warn you now: If the customer describes a vehicle that is symptomatic only once or twice a week, don't bother wasting your time. A car like this actually has what many like to call a "random problem" an not an intermittent. If you dare to diagnose a random problem you will most likely spend a lot of "unbillable" time test driving the car just to see if the problem rears its ugly head. In a polite way advise your customer that their vehicle has to be symptomatic at least once a day in order to efficiently and economically diagnose. This can be difficult for a customer to comprehend. We have developed a hand out at our shop to help explain this to customers. You may wish to do the same. 

Qualified! 

Once you actually have a legitimate intermittent problem, ask the customer as many questions as you can about the nature of the problem. The answers to these questions will start your mind painting a picture of the problem and in the wink of an eye your mind will already be busy at plotting a solution. 

Most intermittent problems fall into one of two categories. The first is a vehicle that will not start intermittently. If this is the case try to question the customer to determine if the engine is actually turning over when the problem occurs. This way you will know weather or not you can eliminate the starting circuit of the car. I'd like to have a dollar for every repair order in the country that has been written up as: "Customer reports car would not start" when actually a little more questioning would have led to the complaint being: "Customer reports car cranked over but would not run". Two entirely different problems that, more often than not, do not get explained correctly at customer write up. 

The second type of problem is the intermittent running problem. These are by far the most interesting to diagnose. They can also be the ones that send you home at night with an upset stomach. The intermittent running problem can manifest itself in many different forms like: Hesitations, stalling, bucking, pinging, racing, etc. Question the customer further to find out important details that will effect diagnosis like: Does the problem occur more when the engine is hot or cold, after a hot soak, during inclement weather, when a certain accessory is turned on or off, during acceleration or deceleration, during turns, uphill or downhill? 

Once you have noted the answers to these questions advise the customer of the initial tests you will perform and the cost of those tests. Make sure the customer is very clear in understanding that your initial tests may not find the problem but will most likely point you to a certain system on the vehicle that needs specific testing. We ask our customers to authorize a "Level One Diagnostic, (LID)" Our LID is a package that includes a visible, (no tools), underhood inspection for obvious problems, scan-tool connect fee, and one hours labor to cover the technicians time reading the datastream, road testing, etc.  After your initial tests are complete you can now begin to formulate a differential diagnosis. In a differential diagnosis you are basically taking your test results and the vehicles symptoms and listing all the possible causes. At this point you may have actually discovered the cause of the problem. If so, you can advise the customer as to the cost of repair. Generally the vehicle still requires further testing to pinpoint.  If so, you can now contact your customer to ask if they will authorize additional testing time and/or other specialized tool costs to aid in further diagnosis. If the diagnosis has gone on longer than you like, a call to a tec-line may be helpful. If the customer does give you authorization to continue, be sure to give them status and current cost updates after each actual billable hour you spend on their car. That way there will be no question as to the amount owed should the customer decide at some point to halt the testing. 

Digging In. 

In this example we will demonstrate the intermittent starting problem. Specifically a "cranks but won't run problem". By far the most interesting yet difficult problems will be those with the smallest symptom window. The symptom window is the actual time frame in which the vehicle is symptomatic. In some of the more difficult cases the symptom window can be as short as 1-2 minutes long. This means any testing you hope to accomplish must be done during that small time frame. In the worst cases the symptom window is short and the time between symptom windows is long. An example of this would be a vehicle that has a problem for only 1 minute and the problem only occurs once per day. Lets take a 1994 Subaru Justy 3 cylinder that we repaired as an example. This customer reported that each morning after being parked all night, the vehicle would crank but would not start. The customer went on to say that if the vehicle were cranked a second or third time the car would start and be fine for the rest of the day. We had to keep this vehicle for a total of six days because the symptom window was so short and the time between windows was so long. What made matters worse was the fact that on three of the days the car started fine with no problem. Nothing to test on those three days! To make a long story short, we were able to compile some interesting test data from several of the mornings when the car would not start for that first minute. When a car has a starting problem like this we generally install a spark tester on one of the plug wires and we disconnect one of the fuel injectors and plug in a noid light. Generally speaking, if the car will neither spark nor run the noid light we start leaning towards a crank sensor system problem. Hence you begin your differential diagnosis. (Don't forget that some cars with a snapped timing belt will show up as no spark and no illuminated noid light). We found that while cranking their was no spark and no flashing noid light. So now your thinking...timing belt or crank sensor right? Well, not exactly. The solution became a lot closer when I was sitting in the vehicle staring at the dash with the key on one morning after the car would not start.  While sitting their I heard a relay "ping" and at that moment I noticed the check engine lamp popped on and I heard the fuel pump run its two second prime. The lamp should have been on as soon as the key was turned on for the bulb test but it came on about one minute after the key had been on. Referring to Figs. 1 and 2 we could see in the schematic that the power for the ECM, check engine light, coil, igniter,  and the fuel pump came from what Subaru calls the "ignition relay". The relay is a "dual contact relay" in which one contact sends power to the coil and the igniter while the other contact sends power to the ECM, check engine light and fuel pump relay. We decided to hook up a volt meter to monitor the that relay. When the car would not start the next morning our volt meter showed that the one of the contacts in the relay, (wire WB), was fine as the ignition coil had its 12 volt supply. The other contact in the relay, (wire R) did not have power. This side of the relay supplies power to the ECM and the fuel pump relay. A simple tap on the relay and the power was their.  A new relay resolved the problem. Most of us don't think to bother looking at the check engine light on a car that "cranks and won't run" but I have forced myself to make it a habit. We recently had a late model Ford Explorer where the customer had an intermittent "cranks won't run" complaint. Upon initial inspection of the vehicle, one of the first things I noticed was that the check engine light was not on at key on. A simple tap on the PCM relay in the underhood relay center put the check engine light on and the fuel pump was heard making its two second prime. I guess now I could say I solved this with a little luck and past experience.  Another faulty relay. 

Then their are the cars with the intermittent running problems. These cars run well most of the time but act strange intermittently. First, try to duplicate the problem based on the answers the customer gave to the questions. If the problem was a bucking in inclement weather then you might road test the car after wetting down the secondary ignition components. Sometimes when a customer reports the symptom of bucking, the first thing we think of is ignition. If the vehicle is a late model Ford the torque converter may be suspect as this has been shown to be a weakness in some of their models. Once you duplicate the condition you will develop a hunch or "gut feeling" as to weather the problem seems like a lack of fuel, ignition problem, etc. At this point you may want to check the service bulletins on the vehicle for the system you think is suspect.  If no relevant bulletins are found you must push on with further testing. Not to pick on Subaru again but we recently had a 1996 Subaru Legacy 4 cylinder that developed a very nasty intermittent hesitation. Fortunately this car exhibited several symptom windows per day in which to analyze the problem. In fact, parking the car for about a half hour or more and restarting it almost always forced the problem to surface. The difficult part was that the symptom window was short. When the vehicle hesitated it did so for all of 1-2 minutes then the hesitation was gone until the car was parked a half hour and driven again. This car has an OBDII datastream but the data updates on the scanner were so slow that I did not believe the scanner would have caught the problem. Beware that this is becoming an increasing problem on OBDII cars with generic scan tools. (The government said we as independents had to communicate easily with the car but did anyone say at what speed?) Of course the sensitive OBDII computer did not store any fault codes! Anyway, a recorded movie of the datastream looked good during the hesitation so we decided to connect the Snap-On Vantage to the TPS signal wire. When the vehicle hesitated the TPS showed a normal voltage reading that increased with throttle angle. Eventually we monitored fuel injector pulse width and the MAF signal wire. When the car hesitated and I was applying throttle, the fuel injector pulse width did not get any greater. That would indicate that someone was lying to the ECM telling it that more fuel was not needed, but who? The MAF sensor showed no more air entering the engine. That must be the problem right?  Cleaning the sensor's sampling wire and even substituting another MAF sensor did not solve the problem. The sensor still showed a decrease in air entering the engine. What was actually happening was that the MAF sensor was correctly showing a decrease in air entering the engine since the engine hesitated for another reason. What I thought was a cause was actually an effect. Be careful not to get fooled with strategy scenarios like this. It's like saying: "What came first the chicken or the egg"? Its no different than monitoring a crank sensor signal and noticing the frequency change when the engine slows down. You have to ask yourself: Did the signal I just saw on my scope change due to a problem with the sensor and that caused the engine to slow down? Or did the engine slow down for another reason and the crank sensor is just reporting the change? Makes you think harder doesn't it? In the end the problem with this vehicle was the O2 sensor. In Fig. 3 the Snap-On Vantage is shown in dual graph mode. The top graph is the O2 sensor signal wire to the ECM. The bottom wire is the O2 sensor heater voltage. Normally .2 to .8 volts is observed on the O2 sensor signal wire as the O2 sensor switches back and fourth from lean to rich. What actually happened here is that some of the O2 heater voltage bleed across inside the O2 sensor to the signal wire.  As you start out on the test drive at first their was no hesitation. This is due to the fact that the car is still in open loop. As you can see in Fig.4, the hesitation occurs the moment the car goes into closed loop. During closed loop the O2 sensor signal is believed by the ECM. At that point the voltage is still way too high. The ECM believes this reading and views it as a rich exhaust. Therefore, fuel delivery is reduced and the hesitation occurs. After about 30 seconds the voltage comes back down into normal range and the car is fine for the rest of that trip. Once the car is parked for about ½ hour the cycle repeats itself and the hesitation is back again. 

Their is no doubt that intermittent problems are getting more frequent as vehicles are built with more electronics. Hopefully you can apply the ideas presented here to some of those future problems.