 | |||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||
|
Within Acceptable Parameters...Posted 6/15/2000By Brian Manley
I'm a Star Trek freak; I'll admit it. I've followed the series, all of them, and I've seen all of the movies - at least twice! My favorite character is Data. Why? It's his ability to diagnose. With no more than the circuits and program in his Positronic Brain, he can collect data, formulate a foolproof diagnostic routine, correct the fault and save the day. Some days I feel just like Data. I grab the ticket that says "Car Died - No Start." We push the car in, I apply the appropriate visual checks, spark and fuel tests, then funnel my diagnosis until I find the root cause. I still get a charge out of hearing these cars fire up after a successful diagnosis. Just like Data, I systematically collected information, located the cause, and performed the correction. The only difference between us is that my emotion chip is installed, and I still yell "Yippee!" when I'm victorious. But what about those tickets you grab that state "Vehicle dies - intermittently," or "Car stalls once-per-week, restarts immediately"? We still need to approach these in the same manner - Complaint, Cause and Correction. We just know that we'll spend a significantly longer time on the Cause portion of our diagnosis! I've collected, in my files, dozens of case studies where I used the movie mode function of my scan tool to identify a system problem, then I funneled the diagnosis with TSBs, lab scope, logic probe, DVOM, wiggle mode, and so on. I offer the following repairs, all OBD-I, GM vehicles, with intermittent faults, that made me wish that I could have my emotion chip removed!
I drove this car initially with no equipment hooked up. The car never acted up for me. Back at the shop, I plugged in my MT-2500 scanner, which revealed the following codes: 23 Manifold Air Temp Too Low; 43 ESC Module Or Sensor Problem; 53 Defective ECM (PIA Error). Now I know that the PCM isn't happy with itself, so I proceed to the "tap test." Tapping the PCM revealed no falters on the scan tool, or in the way the engine idled. I grabbed my hair dryer to warm things up, but the PCM still never hiccuped, even when tapping while warm. I looked for rubbing wires or tight harnesses that could cause an intermittent fault, but that wouldn't explain the code 53. I then proceed to check the overall condition of the car, looking for previously replaced parts, tuneup condition, general maintenance items - all are good. This car is well maintained, and I see no signs of abuse, accident damage or replaced parts. At this point, I'm fighting the urge to apply the "Corsica Theory," or that little voice that says, "Hey, last time we saw this problem on a Corsica, it needed a new PCM." I don't know about you, but I don't like guessing. I always have to find a reason for the complaint, even if I have to drive the car home for multiple days with equipment cables noodling their way from under the hood. That's exactly what I did for this customer. I called him to let him know my preliminary diagnosis, and he agreed to let me drive it until I found some tangible cause for his self-described "possessed car." I knew that the movie mode on my scanner was going to be my best ally for catching this intermittent fault, so I cleared codes and drove the car home, then back to work. No problems. I performed a charging system diagnosis to test for defective diodes, and voltage dropped multiple circuits. All within normal parameters. I added a DVOM to my arsenal set to read AC voltage and hooked it to the alternator output wire. This sat on the seat next to me, just in case I caught the alternator exceeding 300 millivolts AC. I also hooked my lab scope to the EST wire leaving the ICM and going into the PCM, in case I saw a fault with ignition signal processing. Finally, on the way home that evening, I felt a stumble on acceleration that lasted a split second. I enabled the movie mode on my scanner and saw no changes in the AC voltage or on my DSO. Reviewing the movie revealed the data in Figure 1. Notice the frame number is -26. Now, compare this to the data in Figure 2. This is frame number -25. The time from start is 7:16 in both figures, but the data recorded in 200 millisecond intervals due to the baud rate of the PCM. Which values differ? What is the next course of action? I replaced the PCM and my customer has had no more trouble. You GM techs may have done this two days prior, then put it behind you, and perhaps you are wiser than I am! Ever wonder why the late '80s PCMs fail in this manner? I've gathered information explaining that back in 1988, GM issued many in-house bulletins and some shared bulletins addressing the harsh means by which we techs perform the "tap test." We were causing the boards to crack and the surface-mounted components to lift from the board, causing a "no contact" situation. The official GM stance: Technicians were causing PCM failures by abusing the "tap test" and when changing PROMs, were pushing down too hard on the board. Furthermore, we were told that most of the PCMs returned were not defective when put through GM's "rigorous" testing procedure. Nothing was done to "rebuild" them. They were repackaged and shipped out to dealers. Skip forward to 1996. It turns out that the sheer number of returned units caused GM to look hard at their "test and rebuild" process. The biggest fault found was that the "goo" that covers the PCM main board was expanding more than the surface-mounted components could take. When heated by high engine temperatures, the goo would actually lift the component from the board, breaking contact. This was a special goo requested specifically by the GM engineers. The high PCM return rate was exacerbated by the fact that GM switched from components that had their legs pushed through the board, then soldered, to a process that placed the component on the face of the board before soldering (surface soldering). GM engineers have since refined their testing procedures by including thermal cycling and tap testing for all returned PCMs.
OK, time to apply the Corsica Theory! Let's just order a PCM, put it in, and go home early! No, wait; Data would never shoot from the hip like that, so I'd better find a cause for our concern. I approached this car just as I did the Corsica, but this PCM revealed no trouble codes to me. Based on past experience, I heated and tapped the PCM, ignition module and crankshaft sensor. The vehicle never skipped in the stall. It was time to pull two more diagnostic tools from by box: the logic probe and the DSO. I have found that more is better in respect to catching glitches in electronic components. I hooked my logic probe to the crankshaft sensor's input into the ignition module, and probed the EST wire between the ignition control module and PCM with my lab scope. Then, went for a drive (see Figure 3). Once this car traveled at operating temperature for 20 miles, I felt sputtering; then the car died. When this occurred, the logic probe continued to give me a strong audible beeping noise, which indicated that the crank sensor still had a signal going into the ignition control module. Conversely, the lab scope caught a pattern from the ICM showing stunted amplitude, which was below the PCM's acceptable threshold voltage. The data stream did reveal, with the movie mode, that the rpm signal dropped abruptly during the stumble. After replacing the ignition module, the car was cured. How many of you use a logic probe for this type of diagnosis? I bought mine about 10 years ago for 20 bucks from Radio Shack; it sure is a valuable tool.
No, I'm not going to apply the Corsica Theory or the Toronado Theory here. This ticket came with a paper clip and a stack of repair orders showing that the wires, plugs, PCM, fuel pump (twice) and fuel filter had all been replaced in an attempt to cure his complaint. After finding no trouble codes in memory, and no data parameters out of spec in movie mode, I began with probing the crankshaft sensor and EST wire, because of past experience. Ultimately, I only caught the car faltering once. On a cold start-up, the engine didn't fire for a few seconds, and there was no 18x signal reaching the ICM! The engine will not start unless the ICM sees both the 18x and the 3x signals during cranking. Both signals must be present or the ICM will not deliver the fuel injector pulse to the PCM. Additionally, this system has no provisions for codes if the ICM loses the 3x or 18x input signals. This explains that while the customer had many stalling events, there was not a code in memory. The ICM uses the 3x input from the CKP sensor strictly as a sync signal for crankshaft position. The 18x signal, in addition to verifying the 3x pulse for the ICM, is buffered by the ICM and sent to the PCM to serve as an engine spark timing reference (18x high resolution) during cranking. The only other sensor used on the Fast Start system is the CMP sensor and it is identical to the CMP sensor used on the C3I systems on engines with SFI; however, the CMP sensor signal on the Fast Start system is used by the PCM solely for the purpose of sequencing the fuel injection timing. The engine will still run and start if the CMP sensor signal is lost, but the CMP fail-safe operation of this system is different from the C3I system used on SFI engines. If the CMP sensor signal is lost, the PCM will not revert to pulsing the injectors like a multi-port engine. Instead, the PCM logic will attempt to sequence the injectors based on the 3x signal after two engine revolutions (six pulses). Therefore, there is a one-in-six chance that the fuel injection pulse will coincide correctly with the valve timing. The estimated fuel injection sequence will allow the vehicle to start and run, but its performance will be reduced.
This is yet another component failure that can be revealed by taking a movie, or simply by freezing stored data, and trouble codes, then reviewing them. Scrutinizing every line of data can point directly to the cause of the complaint, even without hooking up another piece of equipment, or even opening the hood. Look at the data in Figure 4. Which parameters are out of spec? Block learn has climbed to 155, indicating more than just a slight compensation for a lean exhaust. This movie was taken under a hard acceleration with a marginal loss of engine performance. When block learn climbs above 150, a code 44 will set. All of you will look at the scan data in Figure 4 and conclude that we have a fuel delivery problem. See how good you are? At this point, you can either hook a current probe around the fuel pump wires, or hook up a gauge to catch it in the act. With my fuel pressure gauge taped to the windshield, I drove hard onto the highway. The gauge dropped to 20 psi, well below the specified 40-44 psi for this fuel system. I had also hooked my DVOM to the pump wires with my extra-long bungee leads. With the car acting up, and the pressure low, the meter read 13.8 volts; time to drop the tank and replace the pump and filter. What is intermittent driveability data collection in the year 2000? It's the ability to use multiple techniques and varied equipment to catch the "glitch." If you have data stream available, use it to your advantage. If you don't, then you may have to hook your DSO to every critical input, one at a time, until you find a faulty pattern. Good luck, and may the Force be with you! Oops, wrong movie.
|
||||||||||||||||||||||||||||||||||||
Copyright (c) 1996-2009. Automotive Service Association®. All rights reserved.
 | |||||||||||||||||||||||||||||||||||||
