Getting it wrong: Misdiagnosing a cooling system problem
I recently helped a friend with a cooling system issue in his 1974 BMW 2002. It was one of those experiences where everything seemed straightforward, yet we came to exactly the wrong conclusion. It’s worth understanding why.
We were on a road trip to a BMW event in Greenville, South Carolina. My friend’s car and its recently-rebuilt engine had performed flawlessly on the thousand-mile drive. But after we arrived in Greenville and settled in at our separate hotels, he texted me reporting that the car was overheating, with the temperature gauge rising into the red while warming up, standing still, or driving. He felt the lower radiator hose and found it cold, indicating that the thermostat wasn’t opening. He sourced another thermostat, drove the car carefully to a nearby NAPA store, bought coolant and a catch basin, and changed the thermostat in the parking lot, but it didn’t fix the problem.
First, let’s nail down some terminology. The word “overheating” implies one of two things. It can mean that the temperature gauge is running in the red but the system is still closed and pressurized, or that the owner has ignored the warning signs and continued to drive the car, and the coolant has blown out from under the radiator cap, creating the classic steaming engine compartment. On any car with an aluminum head, you really want to avoid the latter situation, as it’s likely to cause the head to warp or crack, so by stopping when the temperature kissed the red zone, my friend did exactly what he should’ve done to forestall disaster.
Next, let me explain the distinction between overheating as described above, and hot running. If the cooling system is basically functioning (not leaking antifreeze, thermostat opening, water pump turning, fan blowing), and the temperature runs about in the middle of the gauge when the car is driven at speed on the highway but heads for the red in hot weather when the car is in traffic, the problem is almost always caused by either a partially-clogged radiator, a cooling fan that’s too small or too slow, or both.
But if a car is overheating in all situations, it’s usually caused by one of six things. The thermostat can be stuck closed. The radiator can be badly plugged up, which often masquerades as the thermostat being stuck. The impeller on the back of the water pump may not actually be moving any coolant because it’s broken off or is massively corroded. There may be air that needs to be bled out of the cooling system. There could be a restriction in one of the coolant passageways in the engine. Or the head gasket could be blown.
Of course, there is another possibility—the car isn’t actually running hot and the problem is in the gauge or sensor. You’ll often hear “Check it with an infrared temperature gun,” but this isn’t the slam-dunk many people think it is. Most thermostats open up somewhere in the 160–200°F range, and most engines are warmed up at about 200°F, but that doesn’t mean you’ll see those numbers on the IR gun. There are a few reasons why. First, the car’s temperature sensor is immersed in the coolant of the closed system, whereas you can only point the infrared gun at the outside of a hose or metal component, so the IR gun is never going to tell you the actual coolant temperature. And you may get substantially-varying readings on the rubber hoses, metal radiator necks, and thermostat housings. Second, the temperature gauge on the dashboard isn’t calibrated in degrees, so unless you’ve already poked around with the IR gun when your car was running normally and learned that halfway up your gauge is, say, 200°F at the radiator, the reading on the gun isn’t as meaningful as you’d think. Once you’ve been around cars for a while, the “if it feels hot, it is hot, and if it doesn’t, it’s not” seat-of-the-pants metric can be more meaningful than the gun.
But even if you use an IR gun and determine that nothing’s really wrong and that the problem is in the gauge, have you ever tried driving a car a long distance with a pegged temperature gauge? It’s incredibly stressful. All that you can think about is “What happens if something really does go wrong while it’s pegged?”
With that, let’s get back to the story. My friend limped his hot-running car to my hotel, stopping each time the temperature gauge approached the red and waiting for it to cool down. When I met him at the curb, I verified his observations. I watched with my own eyes as the car was started and warmed up, and saw the temperature gauge steadily climb from the blue to the center and then to the red. I felt the lower radiator hose and it was indeed cold. I agreed that this appeared to be demonstrating the textbook symptoms of a car whose thermostat wasn’t opening. The fact that he’d replaced the thermostat and it didn’t solve the problem was odd, but in this age of globalization and junk parts, I know people who have had several bad thermostats in a row.
I checked to make sure that the water pump was actually moving coolant. If the thermostat is open, you can usually see coolant moving through the tank at the top of the radiator, but with the thermostat closed, that won’t work. The 2002’s engine has a “water divider” bolted to the side of the head. The coolant temperature sensor is screwed into it. It was an easy matter to unscrew the sensor and verify that coolant was actually moving through it. And both the water divider and the thermostat were hot to the touch. I reached inside and turned on the heat, and warm air poured from the dash vents. It sure seemed to me that coolant was moving.
I tested the gross functionality of the sensor and gauge by unplugging the sensor and verifying that the gauge dropped to zero. I then grounded the sensor wire, and the gauge immediately pegged high.
So, what else? This was a newly-rebuilt engine. Looking in the radiator, there were no bubbles or oil in the coolant indicating an obviously blown head gasket or a cracked head. I couldn’t figure it out, but as they say in the medical world, “When you hear hoof beats, think horses, not zebras.” “Horses” would be a stuck thermostat or a plugged radiator. “Zebras” would be some odd restriction somewhere.
It was getting late, so we pulled my friend’s car into the parking garage of my hotel and parked it next to my own BMW 2002. I took him back to his hotel for the night.
In the morning, I picked up my friend and we got more coolant. I verified that my own car ran halfway up the gauge and that its lower radiator hose was warm, so clearly the thermostat on my car was opening and my gauge and sensor were working correctly. After we bled his system—which made no difference—I then began swapping cooling system parts with my car.
I did the easiest one first—the coolant temperature sensor. I forgot to loosen my radiator cap first, so when I unscrewed the sensor, coolant shot out, but fortunately, it wasn’t scalding. I swapped sensors and found no change; my car still read in the middle of the gauge, but his still started low and floated into the red as it warmed up.
Next, I swapped thermostats. The externally-mounted ’stat of the 2002 made this easy, though messy. But my known-good thermostat had no effect on his car; the lower hose still didn’t open, and the gauge still headed for the red.
I asked my friend about his radiator. He said it had recently been re-cored. Still, it seemed likely that the radiator was the problem.
I was about to swap radiators (it’s easy enough; they’re only held in by four 10mm bolts), but I paused and thought that I should hold off because the next-easiest thing to do would be to swap instrument clusters. Because the failure mode of the temperature gauge on a 2002 is usually for it to be jumpy or immediately peg high (I’ve never seen one start cold, warm up, and then settle on an incorrect high value), I had no reason to suspect the temperature gauge, but it was something to cross off. And the longer I’ve been at this, the more I’m aware that whenever I think “I don’t think it’s that,” I’m admitting I have a blind spot in front of “that.”
At this moment, four friends walked through the garage, folks I’ve known from the Boston BMW Car Club of America chapter for 30 years. They asked what we were doing, and I gave them the rundown. There’s often a certain annoyance that can come from well-meaning people chiming in with advice on something you’ve already spent hours on, but as I didn’t know what the cause of the problem was, I tried to be open-minded.
One friend, the oldest and most-experienced of the group, laid his hands on the radiator and immediately said, “You said it’s overheating? This thing’s barely warm.”
I paused and said carefully, “It’s not overheating in the sense of blowing coolant, but it’s running in the red, and I don’t have reason to doubt the gauge.”
“If this engine was really running in the red,” my senior friend said, “it’d be scalding hot. You wouldn’t be able to get near it. And when you undid that radiator cap, it would blow coolant everywhere and burn you.”
I realized that he was right. After all, when I foolishly undid the temperature sensor from my car earlier, it released pressure and coolant shot out. Suddenly, swapping the gauge clusters was no longer an academic exercise. My friend removed the cluster from his car, and I yanked the one from mine. We put mine in his, and to my surprise, the temperature read only about a quarter of the way up the gauge. So maybe it was the gauge.
“OK, let’s put your cluster back in,” I said. “We should see the high temperature again. If we do, then we’ve nailed it.” My friend balanced the cluster on the steering column and plugged the cable into the back. It read fine. I scratched my head.
“Now, install the cluster for real,” I said. “Hook up all the cables and tighten everything down.”
And that’s when we found the root cause of the problem.
Because vintage BMWs often have jumpy gas and temperature gauges, it’s common to run an extra ground wire from the back of the cluster to the body of the car. Some folks ground the metal plate on the back of the cluster, but some directly ground the nuts on the backs of the posts securing the temperature and fuel gauges. This is what my friend had done.
But the grounding wires weren’t the problem. It was when he went to connect the ring terminal to the back of the temperature gauge that he noticed that simply moving around the nut on the back of the post changed the reading on the gauge.
I fiddled with the back of the gauge. It seemed that it wasn’t an issue of the nut being loose but was some electromechanical connection between the post and the gauge itself. I found that simply by leaning the post to one side or the other, I could make the gauge read three different things: The low (and apparently correct) temperature reading, an artificially high but reasonable reading, or pegged high. Through trial and error, I managed to get things tightened down while still generating the correct reading.
And with that, we not only fixed the car but reconstructed the incorrect forensics. It’s a fascinating study in getting it wrong. To summarize:
1) The owner of a car, who had recently paid a stiff bill for installation of a rebuilt engine, sees the temperature gauge heading into the red. The owner doesn’t want to crack the head of the newly-rebuilt engine, so he stops driving.
2) The owner feels the lower radiator hose and finds it cold. He concludes that thermostat isn’t opening, sources another thermostat and installs it. No change.
3) The owner drives the car to my hotel, stopping whenever the gauge approaches the red.
4) I verify all this at curbside.
5) The next day, I swap the thermostat again for a known-good one from my own car, still no change.
So how did we get it so wrong? Because the car was never driven long enough to get warmed up. Every time the temperature gauge touched the red, we shut it off. That’s why the thermostat wasn’t opening.
We followed decades of experience that hot gauge + cold lower radiator hose = stuck thermostat. The cool October temperature in Greenville and the fact that the radiator was recently re-cored made for a non-problem masquerading as a problem.
The outlay was cheap—several hours of time and the cost of a thermostat, a few gallons of antifreeze, and a catch basin. My friend was a little embarrassed, but I would’ve done the same thing.
Live and learn.
Rob Siegel has been writing the column The Hack Mechanic™ for BMW CCA Roundel magazine for 33 years and is the author of five automotive books. His new book, Resurrecting Bertha: Buying back our wedding car after 26 years in storage, is available on Amazon, as are his other books. You can order personally inscribed copies here.