A Funny Thing Happened on the Way to the Vintage -- Part III

When we ended last week, Brian Ach’s recalcitrant stumbling stuttering ’73 2002tii that betrayed him and his wife Michelle in northern Virginia on the way to The Vintage had been towed to my house in Massachusetts. I’d offered to fix it. For free. So I could rewrite the ending to the story where I’d failed to fix it on the road. My plan was to throw equal parts logic, ego, and components stolen off my own tii at it until it or I cried uncle.

It seemed like a good idea at the time.

Now, there’s an expression in troubleshooting: Plan the work, work the plan. For a tough problem, take your emotions out of it. Don’t follow pet theories. Don’t take shortcuts. Sure, prioritize things that are more likely, and leave the ugliest least likely most expensive possibilities for last, but develop a systematic approach and execute it. If the approach is sound, you’re highly likely to solve the problem, sometimes in spite of yourself. So my plan for to finding and fixing the problem in Brian’s car was this:

·        Do a compression test to rule out anything mechanical and catastrophic.

·        Adjust the valves while the engine is cold… because adjusting valves makes me feel connected with a car.

·        Check that the TDC mark on the front of the crankshaft agrees with the one on the flywheel. (Paul Wegweiser’s crankshaft pulley was once installed backward, and the two marks were off from one another.)

·        Check that, at TDC, the camshaft mark lines up with the mark in the head.

·        Recheck the mark on the pulley of the Kugelfischer pump to verify it also lines up correctly.

·        Recheck the dynamic ignition timing and the basic advance functionality of the distributor

·        Verify that the voltage at both the battery and the coil is about 14 volts with the engine running and doesn’t vary much with engine RPM.

·        Drive my tii and verify that it runs correctly.

·        Drive Brian’s tii and verify that the problem still exists.

·        Swap the entire ignition system from my tii – dizzy, Pertronix, cap, rotor, coil, resistor, plugs, plug wires, everything.

·        If the problem goes away, begin swapping ignition parts back until the problem reappears to determine which component is bad.

·        If, instead, the problem persists, start to do something similar with the fuel system. Don’t swap everything right off the bat because pulling two Kugelfischer injection pumps and eight injectors and swapping them is a bit of a pain. Plus, the injection pumps are extremely reliable; when you hear hoof beats, think horses, not zebras and all that.

·        Start with the fuel pump – logical, since Brian’s had low fuel pressure (24 psi where it should have 29). Check fuel volume as well.

·        If necessary, use my portable exhaust gas analyzer to see what the thing is doing (rich or lean) when it’s running terribly.

·        If necessary, use an IR thermometer to help locate a weak performing cylinder.

·        If that doesn’t work, look at the gas, draining the tank and trying clean branded gas if necessary.

·        If that doesn’t work, systematically go through the entire fuel system.

·        If that doesn’t work, read up on how to cover up insurance fraud.

I was a man with a plan.

But if you asked me what I thought… There’s that old saying “your fuel problems are in your ignition.” Always suspect the ignition first. Plus, Brian’s fuel system was known good.

With that sly opening, let me offer a meditation on the concept of known good. Because if you don’t understand this, you can’t truly appreciate how weird the events that unfolded were.

About two years ago, during a general sort-out of my own tii, I was having terrible problems. I’d just had the distributor rebuilt by a highly reputable shop, and installed it and a new Bosch Red coil, the correct matching Bosch Red ballast resistor, new plugs and plug wires, and new points and condenser, the latter two procured from a BMW dealer in BMW logo’d boxes with BMW part numbers. But the car ran horribly. With all those new ignition parts, it clearly seemed to be a fuel problem. I posted my trials and tribs. My friend Lindsey Brown, a pro, texted me. “Can I come over and play?” he asked. He showed up with a distributor that looked, shall we say, experienced. “This,” he proclaimed, “is known good. I pulled it out of my own running 2002 this morning. Let’s put it into your car.”

“How can you say,” I retorted,” that that’s known good? MY distributor is known good. It was just professionally rebuilt. The points and condenser are brand spanking new. It’s certainly more known good than yours. It can’t be the problem.”

“Are you or are you not having a problem you can’t figure out?”

“Well… yes.”

“Then humor me,” he said. “Let’s put in my distributor and see what happens.”

My response was, essentially, “grumble… bitch… gripe… I’ll show you known good.”

We removed my essentially new dizzy and replaced it with Lindsey’s beat-up-looking one. I started my car. The problem was gone. I picked my jaw up off the floor.

“Now,” said the professional, “let’s transfer one component at a time from your distributor onto this one until we find out what is actually causing the problem.

It turned out that the culprit was the brand new condenser that I’d purchased from a dealer in a BMW logo’d box. It lasted about five miles before it died.

It was in this experience that I learned what was probably the most important diagnostic lesson in my Hack Mechanic life: There is no known good. There is only diagnosis.

Let that one seep into your bones for a few weeks until it’s part of your DNA. I’ve experienced this same thing in my engineering career, but it usually takes a slightly different form: Whenever you say “it can’t be X,” you’re admitting you have a blind spot surrounding X. Remove it. Because it may well be X and you need to be able to see or at least consider that.

Next, you must permit me a Star Trek moment. In the “Wink of an Eye” episode, Kirk and another crewman are kidnapped and taken into an accelerated timeframe. They move so quickly that they can’t be seen by the non-accelerated crew, only perceived as a faint buzzing. Kirk tries to communicate with the crew by recording a message and leaving it in what can only be described as the Enterprise’s 8 Track tape deck. Spock finds it and listens to it, but the tape contains only buzzing. At some later point, after the crew finds structures built around the Enterprise’s engines with no visible evidence of people, only buzzing, the light bulb goes on for Mr. Spock. He hears the buzzing again, and he says to McCoy “I know what it is.” He’s figured the acceleration thing out. He retrieves the tape, slows it down, and can now hear the captain’s message.

The point in all that geekiness is that, most of the time, we don’t get to say I know what it is, which appears to be the Vulcan equivalent of eureka. It’s really rather rare, at least for me, to look at and think about a car problem and engage it so rationally that it leads me inexorably to a path where I do one test, go “a-HA!” know that I’ve found the answer, and pat myself on the back for my diagnostic skills. Most of the time, you just need to pound it out. And, by the time you do, you’re often robbed of that “eureka” moment because you’ve spent a boatload of time and money, you’re tired, and instead of eureka, your reaction is often “THAT was it? You have got to be shitting me.”

I began to execute the plan. I did the timing and voltage checks. It made no difference. I then drove my tii and verified that it ran fine. In one of those big circle of life things, my distributor that had been thought good but actually bad was now the one that was known good, or as close as you can get. I put the two cars nose-to-nose in the garage like a rotated scene out of Frankenstein. I pulled the ignition out of Brian’s car and put it in a box. I then yanked the entire ignition – everything – out of my tii and installed it into Brian’s. As I was doing so, I had the thought “you know, this is really too big of a chunk to do at once. Plugs and plug wires, coil and resistor… probably not necessary. You’re going to have to swap them back when you start this puppy up, drive it, and the problem is gone. You probably should’ve swapped only the dizzy and its components.”

But, hey, plan the work, work the plan, right? Everything was swapped, I started Brian’s tii up, pulled it out of the garage, drove it around the block, and…

No difference.

Well, then. Wasn’t that cagey of me to misdirect you by floating three times how much I regretted not having had a distributor with me when I looked at the car in the parking lot? In truth I had regretted it, right up until this moment when I realized it would’ve made no difference.

I came inside and told all this to Maire Anne. She’d heard me explaining “the plan.”

“So,” she said, “that means it can’t be an ignition problem, right?”

I was careful.

“No,” I said, “it means I don’t see how it could be an ignition problem.” Remember—try not to have blind spots.

The path was clear. If it’s unlikely to be ignition, look at fuel. This meant engaging The Mystery:

·        The entire fuel system was supposedly “known good,” with the injection pump, injectors, fuel pump, and gas tank all serviced 700 miles ago.

·        The Kugelfischer injection pump had been rebuilt and the injectors cleaned and tested by one of the five experts in the country who specialize in this work. Unless some unspeakable fate befalls an injection pump, it should be the last thing you should when a tii goes from running to not running or running poorly.

·        The gas tank had supposedly been cleaned, yet when Brian pulled out the fuel filter in the parking lot and tapped it out on a paper towel, “black stuff” came out of it.

·        The small brass screen at inlet of the injection pump had a big tear in it. So either the pump rebuilder had installed it that way, or something had muscled its way through the fuel filter and ripped the screen. Both possibilities were unsettling. Brian’s nightmare scenario was that the small brass threads of the screen had gotten into the injection system and were mucking up the pump and/or injectors.

·        The E28 fuel pump—also installed only 700 miles ago—was putting out low fuel pressure, 24 psi instead of 29.

Since the fuel pump pressure was low, addressing it was not only the obvious next step, it seemed the next most likely source of the problem. I realized that I didn’t even need to pull the fuel pump out of my tii—I’d prophylactically replaced mine with an E28 fuel pump last year and had my original fuel pump sitting on a shelf. I undid the fuel lines and the three 10mm nuts holding the fuel pump assembly (the bracket, pump, and its expansion tank about the size of a small pickle jar) in Brian’s car and pulled it out as a unit. I removed his newly-installed E28 pump and installed my old tii pump. I tightened everything up and turned the key to fire up the pump so I could check for leaks.

Gas everywhere.

I shut it off and rechecked the hoses. They were all tight. I looked at the fuel pump carefully, and fuel seemed to be leaking from around the electrical plug itself, rather a terrifying sight. I pulled it all out, took the pump out, cleaned it and dried it, and blew into the inlet. Sure enough, bubbles came out around the connector. My old fuel pump was cracked. Sheesh.

Try this again. I crawled under my tii, pulled out my pump assembly, and swapped it en masse into Brian’s car. I spliced a fuel pressure gauge in the engine compartment with a tee fitting. I cracked the key and checked for leaks. Tight. I checked the pressure. 28 psi—much better. I fired up the car, backed it out of the garage, and took it down the street.

No difference.

This was the first of many anti-eurekas. (e.g., “That wasn’t it? You have got to be shitting me.”)

Okay then. It was time to listen to my own there is no known good advice, as well as all the folks who had chimed in saying “go back to basics and check the gas tank.” I undid the eight bolts holding the filler flange onto the tank and pulled out the pickup assembly. There’s a screen at the bottom of the pickup tube. The screen had a small amount of particulate matter in it, but nothing alarming. I also noticed that the screen wasn’t sitting completely flush with the bottom of the tube, as if it had been pried out, cleaned, and put back in slightly askance. This wasn’t a big deal. If anything, it said to me that the tank and the screen had likely been cleaned.

I then looked into the gas tank with a flashlight. In the recess at the bottom where the pickup tube sits, I saw a small amount of particulate matter. I’d done this same thing with my own tii two years ago, and what I found in it was horrifying; it looked like pot roast. What I saw in Brian’s tii struck me as minor. Just a little bit of rusty scale. And some odd black bits. I fished a few out. They were soft, unlike the rust, but I couldn’t tell if they were paint that had fallen off the outside of the tank and had found their way in, or pieces of the inside of a fuel hose that had been degraded by ethanol, or what. I thought about Brian’s observation that “black stuff” came out of the fuel filter in the parking lot. Was it the same stuff? There was no way to tell.

I thought that, given the persistent nature of the problem, I’d be wise to do a front-to-back component-by-component check of the fuel system, skipping no steps and leaving nothing to chance.

It was the best diagnostic decision I’d ever made.

I undid the fat rubber fuel hose that connects the pickup tube to the inlet of the fuel pump and let the gas run out into a clean bottle. I then blew out the hose with compressed air into the same bottle. When I looked at the bottle, there was a surprising amount of rust, scale, and particulate matter in it. What the…? If that came out the hose that was between the gas tank and the pump, then…

I grabbed the E28 fuel pump I’d removed from Brian’s car. Original tii fuel pumps have a small screen at the pump inlet that keeps rust and other gas tank crud out of the pump, but E28 pumps apparently don’t. I looked inside the inlet, and saw some brown stuff. I swabbed it with a few Q Tips, and immediately pulled out some sizeable rust flakes.

I wondered if that was the tip of the rustberg. I began tapping the inlet of the pump on a paper towel. A horrifying rust mound quickly formed.

I let it rest for a while, then repeated. More rust continued to come out.

So, rust had gotten past the pickup screen and into the E28 fuel pump. I would’ve screamed eureka, but it hardly seemed appropriate. I’d already changed the fuel pump and it made no difference. And the questions. Was the rust from the pre-cleaned gas tank? From a single bad tank of gas after all the work had been done? I didn’t know. I asked Brian about the exact order in which all the known good work had occurred. In fact, it wasn’t all at the same time. In one thrust, the fuel tank was cleaned, and the rebuilt injection pump and cleaned and tested injectors were installed. But the E28 fuel pump was apparently installed earlier. This made some sense. It meant that rust from the pre-cleaned tank could’ve gotten into the screen-less E28 fuel pump and hung out in there, still present even after the tank and injection components were cleaned.

The question was how far forward in the car the rust had gotten. Had the E28 fuel pump acted like a food processor, grinding up the rust and spitting it into the injection system? I removed the main fuel filter by the battery—the one Brian had installed in the parking lot in Virginia, then driven perhaps 40 miles on while nursing the stumbling car to where he could rent a U-Haul trailer and truck. I tapped the filter on a paper towel, and only clear gas came out. Again, this was the replacement filter. The previous filter was the one where Brian reported that “black stuff” came out when he tapped it on a paper towel. Well, whatever had happened had happened, but it was clean now, and there was no evidence now that rust was getting past it into the engine.

Still, I wanted to be thorough, so I used compressed air to blow out the line from the inlet of the filter all the way to the back of the car. Note that this line consists of a hard metal line that runs underneath the car, with a short rubber hose on each end. I captured the blow-out in a bottle.

I was greeted with a horrifying combination of rusty scale, sediment, and powdered rust.

Known good my ass.

I blew the line out probably ten times trying to get it clean, spraying carb cleaner in one end, and using a rag held on the other end with a hose clamp, and still kept getting a faint halo of rust on the rag.

A new mystery presented itself: How could all this crap come out the line, yet none of it be present in the fuel filter? I theorized that, since the tii’s fuel filter is mounted vertically next to the radiator, it was possible that lose crud could come into the filter and get blocked, but when the fuel pump was shut off, gravity could cause the gas and crud to simply drain out the bottom of the filter into the fuel line beneath it, and accumulate there.

I also theorized that there were probably at least three separate issues. The big rust was likely from the original unclean tank or a load of bad gas, whereas this fine powdery stuff I couldn’t get rid of was likely because there was rust forming on the inside of the long metal line itself, as well as inside the metal expansion tank. I blew the metal line and the tank out, individually and repeatedly, and always saw a faint rust halo on the rag.

There was one more fuel line—the return line from the pressure valve on the back of the Kugelfischer pump to the gas tank. I did the same test, putting a bottle on one end and blowing compressed air into the other. This line showed up clean as a whistle. This was good—there was no evidence of rust making the round trip back to the tank. And with no rust coming out of the main fuel filter, that meant there was no direct evidence anything had gotten into the Kugelfischer pump or the injectors. At least not yet.

Finding all this rust may not have deserved a eureka, but how could it not affect the car’s running? How could the car not run better with it all cleaned out? I put everything back together and drove the car.

No difference.

Okay. Okay. There was one more thing to try before zeroing in on the challenging and expensive issues of the Kugelfischer injection pump and the injectors. That was the gas itself. One school of thought is that an additive containing alcohol can break up the water, allow it to mix with the fuel and be burned, but the other school of thought is that, if you have water in the gas, there’s little alternative but to drain the tank.

Rather than wonder about it, I just did it.

Fortunately, 2002s have small (10 gallon) tanks anyway, the tank was less than half full. I took the old E28 fuel pump—the one that had all the rust in it—wired it to a battery, and used it to pump out most of the gas into a gas can. I then manually cleaned the bottom of the tank with rags until it was visibly free of scale and particulate matter. Again, it didn’t really look all that bad to me. It seemed to me likely that the tank had been cleaned, but some load of particulate matter came in, perhaps with a fill-up. But wherever it had originated, I cleaned it up.

For a moment, I thought “you know, now that it’s drained, if I remove a few bolts, I could pull the gas tank completely out of the car and have it professionally cleaned.” But I didn’t know that the tank, or bad gas, was in fact a problem, much less the problem, and I didn’t want to wait a week for the tank to be cleaned and reinstalled to find out. Instead, I drove to AutoZone, bought a brand new clean 5 gallon gas can and a generic high-pressure fuel filter, drove to a branded gas station, bought a few gallons of branded 93 octant gas, dumped it into the cleaned tank, installed the generic fuel filter between the tank and the fuel pump (which was now my fuel pump out of my tii) to reduce the possibility of any stray particulate matter from the tank getting into the fuel pump, and drove the car.

No difference.

My future seemed to ineluctably involve a Kugelfischer injection pump and injectors.

Then I got a call on an unrelated matter from my friend Tom Samuelson. Tom has eclectic car tastes. He has a killer E9 and an E24, but also a beautiful E Type convertible, a Volvo P1800, and a hot rod. Sometimes cooking in a bunch of different kitchens gives you better perspective on problems. I explained to Tom what I’d found.

“Hmmn,” he said, “Kind of sounds like an ignition problem to me.”

“That’s what I thought,” I said, “but the diagnosis hasn’t borne that out.”

“Was anti-seize paste ever used on the plugs?”

“I don’t know. Why do you ask?”

“Well,” he said, “I’ve seen it cause intermittent grounding of plugs, translating into weak spark.

It fit the symptoms. I texted the question to Brian. “You know,” he said, “in fact, yes, I did use anti-seize on the last set of plugs.”

Bingo. And thus we see the value of knowing the difference between “it can’t be an ignition problem” and “I don’t see how it could be an ignition problem.” The latter leaves room for anti-seize paste; the former does not.

I called back Tom. “What do you use to get it off?”

“Acetone and a toothbrush.”

This was perfect. There’s a Rite-Aid next to Bentley Publishers. At the end of the day, I walked next door and bought some acetone nail polish remover and a toothbrush. I could barely contain my enthusiasm as I drove home. I didn’t even head inside the house; I went straight into the garage. I pulled the plugs and cleaned them and the plug holes with acetone. There wasn’t much residue, but, hell, who knows. I cleaned it all up, put it together, drove the car, and…

No difference.

Crap.

Sigh.

I’d always thought that Brian’s worst fear about there being brass screen fragments inside the injection was highly unlikely. Now, even though I still had no direct evidence that any contamination had gotten past the main fuel filter, I didn’t know what to think. There really didn’t seem to be much else it could be other than the injection pump and/or injectors. And whether they’d gotten contaminated by rusty fuel or by screen fragments didn’t seem to affect the course of action. Either way, I needed to look at the pump and injectors.

The Kugelfischer injection on the 2002tii has the reputation of being arcane and impenetrable, but in fact it’s quite well documented. The injection pump itself is often referred to as a Swiss watch of engineering. Back in the day, BMW of North America published a slim volume “The BMW 2002 tii Fuel Injection System” intended for service technicians. It’s available for download on Bob Murphy’s 2002tii site at www.2002tii.org/pump/pump_guide_v1.pdf. The manual largely treats the injection pump as “no user serviceable parts inside,” but, to continue the Swiss watch analogy, there are four small Allen key covers on the top of the pump, and beneath them are four little suction valves that look remarkably like little watch batteries (e.g., sure, pop the back off the watch to change the battery, but don’t even think about touching anything else inside). There are also four delivery valves screwed to the top of the pump. The injection lines are attached to the delivery valves. The manual does a nice job leading you through a rough-running troubleshooting exercise to isolate  the problem to a cylinder, and from there to the suction valve, delivery valve, injector, or, by default fall-through, the pump. It essentially says:

·        First, make sure it’s a fuel problem and not an ignition problem.

·        One line at a time, use a 14mm wrench to slightly loosen the connection of each fuel line to each delivery valve. Fuel should immediately come out the loosened line, and the engine should run noticeably rougher. If it does, that cylinder is probably not the problem.

·        If fuel leaked out but the engine doesn’t run rougher, the problem is likely either the delivery valve or the injector. Swap delivery valves between cylinders to see if the problem stays with the delivery valve or the cylinder. If it stays with the cylinder, it’s likely the injector. Swap injectors and retest.

·        If no fuel leaked out, the problem is likely either the suction valve or the injection pump. Swap suction valves between cylinders to see if the problem stays with the suction valve or the cylinder. If it stays with the cylinder, it is likely an internal problem with the pump.

I went through this exercise and found that, when I cracked open line #4, no fuel leaked out.

A-HA! Finally! Got you, you little sucker!

Following the troubleshooting procedure, I undid #4 Allen key cap and pulled out its suction valve. The underside had a red powdery coating on it. I immediately removed the other three suction valves and checked them. They all looked similar. This appeared to be the first direct evidence that rusty fuel had in fact made it into the injection system. This was big.

(I’m not sure what happened with this photo, but you can still make out the slight rusty cast on the valves.)

Almost giddy with the discovery, but biting my tongue to not shout “EUREKA” and anger The Hack Mechanic Powers That Be, I thought “to hell with swapping #4 suction valve with #3. I’m swapping all four suction valves with the, ahem, known good ones from my tii and driving the car.” I had a moment where I thought “no, don’t do this. Don’t skip a step. And don’t risk contaminating your own injection system with rust.” But I could smell victory. And I reassured myself that I’d thoroughly blown out every fuel line on Brian’s car, and the risk of cross-contamination was negligible.

I swapped all four suction valves with those from my tii and drove Brian’s car.

It made no difference.

Unbelievable. Unfreakingbelievable.

I repeated the “crack the line” test. #4 still had no fuel coming out. If I had a “known good” suction valve in there, by the troubleshooting procedure in the manual, this meant the problem was likely in the injection pump itself.

I felt like I was at the scene of a train wreck. I tried to gather my wits and soak it all in. I started the car, and looked, and listened.

I said that, back when I was checking out the car in the parking lot in Virginia, “I put a finger on each of the plastic injector lines and verified that I thought I could feel fuel pulsing into them.” With this new knowledge about #4 pump line apparently not outputting fuel, the validity of that observation was clearly in doubt. I again put my finger on #4 line. Could I feel pulsing, or was it just sympathetic vibrations from the other lines? I couldn’t tell.

But, then, I saw something. The fuel lines are translucent plastic. And #4 line looked different than the other three. With the car running, I could see nothing in the other three lines, but #4 looked like, well, like when you slurp liquid and air up a straw from the bottom of a glass. That is, I could see what I assumed to be both fuel and air in #4 line but not in any other. It wasn’t right. Plus, how was it that I could crack the nut holding #4 line and not have fuel come out, but see fuel in the line?

It didn’t make any sense.

The fuel pressure in the lines coming out of the Kugelfischer pump is supposed to be very high, about 500 psi. In a “kids don’t try this at home” moment, with the car running, I completely undid #4 line from its delivery valve and watched. There were small upward spurts of fuel out the delivery valve, but it didn’t look like the 500 psi Vesuvius I’d imagined.

Something appeared clearly wrong with the fuel delivery from #4 port of the injection pump.

So much for horses not zebras.

Did I mention that I was doing this for free, in an attempt to change the ending of a story—this story—from “I failed” to “yeah, I kick ass?”

Okay. I’m going in. In the immortal words of Groundskeeper Willie, “Lunch lady Doris, have you got any grease? THEN GREASE ME UP WOMAN!”

(Next week: The Hack Mechanic goes in, and lengthens his list of life’s regrets.)

(copyright BMW CCA 2014. All rights reserved.)