Cause and effect – it's all about processing and feedback.
Your brain gathers inputs, processes them and produces outputs by the millions every day. Most of these transactions are handled on full automatic without you even having to think. There are a few hundred that you do think about, and any time you try to make something happen without sufficient input, you're apt to make a mistake. For one example, setting a cup of liquid on a table with your eyes focused on this magazine can cause a spill. Carrying a piece of merchandise to the cash register without having checked the price can bring a surprise.
Just about everything we do requires gathering data, processing that data and crafting an output action that brings the desired goal or outcome.When men and women started building computers, the model on which they worked was pretty much the same. But in the beginning, computers could only process the data the operator provided. Like a plain old wristwatch, that was an open loop system.
Then came more complicated machines. For a simple example, consider the thermostat on your wall that controls your heating and cooling by processing feedback from the air it is trying to condition. The data you provide is the temperature you want. We'll call this the target temperature. The thermostat compares the target temperature to the prevailing temperature of the air, and if the two don't match, the thermostat activates the air handler and the heat pump. The input data came from two sources: You entered data (the target temperature) and data the thermostat gathered (ambient temperature). When the target temperature matches the ambient temperature, the thermostat shuts the air handler and heat pump off until the ambient drifts out of its programmed range, which is yet another piece of data.In its simplicity, this system works beautifully, but until recently, a climate control thermostat couldn't diagnose its own problems and it certainly couldn't adapt. Automotive computers have had this capability for years.
Automotive Computers
This Venture van got its owner's attention just a couple of weeks after we replaced the harmonic balancer. The original balancer was coming apart and out of line to the point that it was wobbling and ruining the belt, so we replaced the balancer and the belt.
The next time this lady called, she was describing a rough running problem and a flashing MIL. Had we done something wrong? It had started right after we replaced the balancer and belt.
Well, the flashing MIL meant she obviously had a misfire, but when I interviewed her in person, she told me that the MIL would either flash, illuminate without flashing or go completely out for a day or two. And it had repeated this cycle more than once or twice. That's kind of weird for a misfire. What that meant was that one or more of the cylinders was misfiring very badly sometimes, not so badly or not at all at other times, and with no misfires at all for enough trips that the PCM was clearing the code itself.Misfire codes don't ordinarily behave that way.
The GM 3.1L and 3.4L V6 engines have a dreadfully weak intake gasket, and busy shops that do engine work have replaced a few dozen at least.
Well, I pulled the dipstick and didn't see any milky oil, so the intake gasket wasn't at fault here. And besides, the engine was running just fine with no MIL at this point.
Snatching the history codes produced a P0300 – that's just peachy – misfiring on more than one cylinder.Water in the gas?
Uncommanded EGR flow sometimes?
No on both counts!
Historical misfire data was showing that cylinders 1 and 2 had 12,343 and 16,535 misfires, respectively. Cylinder 4 had 7,365, and Cylinder 5 had 21,348. Cylinders 3 and 6 were the best performers, with only 141 and 206, respectively. I took a digital photo of the screen, then cleared everything for a fresh start and test drove the van. As the engine warmed, I began to see a misfire or two and feel a little roughness in the idle when I stopped. No misfires cold, only hot. Interesting, but not conclusive.OK, I've gathered data, now what?
Gather MORE Data
One thing I really hammer home is that the best source of data you can have is the data you've gathered from normal vehicles, like the one in question. The more you scroll through the PIDs on a GM 3.4L of this vintage, the more likely you are to spot bad data. Like a treasury agent looking for funny money (they study the real ones), you'll recognize out-of-range PID data very quickly if you're really familiar with the datastream on a given vehicle.
In this case, the fuel trim data was consistently running from -15 percent to-30 percent, and that's a problem on any vehicle. This PCM was correcting for a rich condition by subtracting fuel.Double-digit long term fuel trim corrections (10 percent or greater, either positive or negative) point to an out-of-balance situation that needs to be corrected. In my experience, positive fuel trim readings are a lot more common than negative ones, because vacuum leaks are fairly common.
Somebody that packs the fuel tank might saturate the carbon canister and cause a rich condition, so with that idea in mind, I went to the EVAP Canister Purge PID and found 17 percent, which is the normal purge rate for one of these when it's idling with a warm engine.I punched her number into my cell phone. Had she been packing the tank? Her answer, in so many words:
"Heck no, who can afford to do that any more? It never has more than half a tank."
Well, just in case the Canister Purge valve was allowing more flow than it was supposed to, I took the valve (it's on the rear valve cover) out of the system and saw only slight movement toward the zero fuel trim mark. It was an expected but less than significant shift. Leaving it disconnected, I removed the PCV valve and let it suck on a piece of paper instead of the crankcase – fuel contaminated oil can cause fuel trim numbers to go negative, but once again, I only saw a slight fuel trim correction, and we were still really high in the triple digits. I reconnected the EVAP purge and the PCV.
The fuel pressure was between 30 and 40 pounds as expected, and there was no fuel at the regulator vacuum hose, so a pinhole in the regulator diaphragm wasn't the cause of this concern.Houston, we still have a problem.
Back to the Blue Screen
With the Nemisys in hand again, I rooted my way through the rest of the data, but nothing really got my attention.
Remembering that the computer science instructor had a Pontiac Montana that was near this same year model, I garnered her permission to bring her van over to the shop and make some comparisons.
With a second Nemisys connected to the Montana, I began to compare pertinent PIDs. I brought the Montana to operating temperature and found fuel trim readings in the +5.5 range, which is near enough to zero to be comfortable – product variability in the field makes adaptive learning necessary. The EVAP on the Montana was also running at 17 percent idling, which matched that PID on the other van.
What I did see was something I suspected but absolutely couldn't verify with anything I had on hand. The MAF on the Venture was reading 4 grams per second, nearly a full gram per second higher than the MAF PID on the Montana, which was reading 3.05. The Montana was newer and its PCM was reporting hundredths of a gram – GM's friends at Delphi had obviously learned something between 1998 and 2001. With MAF as critical as it is, the guy looking at the blue screen needs tighter measurements than I was getting on the Venture.
Sorting The Data
OK, so let's backtrack and debrief. The MAF was over-reporting the mass of the incoming air. The PCM was factoring for 4 grams per second of air coming in, which was truly a lie. I found that out later that week while troubleshooting a 2001 Buick that a 3.8L normally reads 4 grams per second, but that's a bigger engine.
Since the Venture's MAF was over-reporting, the PCM was over-fueling, but the O2 sensors were providing after-the-fact data and they were telling the truth. Furthermore, they weren't happy.
"Hey, back off on the fuel pulse up there, we're reading low oxygen and that means you're too rich!"
The PCM's program subtracted fuel to make the O2 sensors happy, and in so doing, it ratcheted short fuel trims negative until they maxed out. It then began to make the coarser long term fuel trim adjustment necessary to bring the short term fuel trims back to zero, but there were times when the MAF drifted far enough out of reality that even the long term fuel trim couldn't do enough to compensate. That's what brought about the misfires – it was like having a carburetor out of adjustment.
A replacement MAF saved the day. Thus endeth this adventure.
