Engine overheating sometimes

Question

Vehicle 2002 Mazda Protégé LX 2.0, 238,000 km

Symptoms Friend called, his girlfriend's car had been overheating. Mechanic told them the radiator was the problem, as it was out of coolant.

Repairs so far

1. When the car first came to me, it was running a temperature (temperature gauge about 2/3 of the way up), coolant was boiling over in the overflow tank, no CEL. We switched the car off and let it sit to cool off. Removing the rad cap revealed the radiator was empty (again). I filled it up with water, changed the rad cap for one form the junkyard, started the engine and revved it until it reached normal temperature and the fans came on. The temp seemed stable and the coolant wasn't boiling, although the rad fans were staying on seemingly full-time after reaching operating temperature. Friend was out of time, so I sent him on his way. He checked back in later that day to say the car ran a temperature again on his way home and coolant was boiling again.

2. Car came back to me 1 week later, we were going to replace the radiator. I got one from the junkyard (friend intends to junk the car in a few months and won't pay for a new radiator), cleaned it out with a household product that removes calcium, lime and rust, swapped it for his and filled it up with proper coolant. There doesn't seem to be a bleeder screw on this engine, so we ran the engine until it reached operating temperature. I noticed 2 things:

   1. There was a stream of bubbles coming out of the coolant in the overflow tank, I assume it's the system purging the air. This stopped right around the time the engine reached operating temperature.

   2. The rad fans came on as soon as the engine did. The car had been sitting for a couple of hours for the repair, so everything was cold. This is telling me the ECU knows something is wrong somewhere.

Friend still had time, so we took the car for a drive. Since the CEL was on, we stopped at a parts store to have the codes read, and there were 2: random misfire and something about the pre-catalytic converter temperature being inconsistent. We continued with our test drive, the rad fans were on the entire time (I could hear their hum). The temperature was fluctuating between normal (just under half) and high (over 80% of the way up). It seemed to reach high with hard acceleration, it seemed to cool off with mild engine load and it actually read the coolest (normal) while idling in traffic. When we got back home, coolant was boiling over.

Any tips as to the cause?

1. radiator not pressurizing: how do I check that easily?
2. bad water pump?
3. blocked radiator?
4. bad thermostat? (stuck closed?)

Update #1 Rented a rad pressure tester and went to the car (so it's not hot when it gets to me). The junkyard rad I put it was cracked. I will be returning it to the junkyard and getting another one. I never recommend junkyard radiators for this specific reason.

Update #2 Exchanged the junkyard radiator for another one, I will be putting it in later today. In the meantime, here's a relevant question on how to test for a bad head gasket, as that's a possibility here.

Update #3 Installed the second radiator form the junkyard last night, pressure tester wasn't holding pressure and there was a hiss, but couldn't find exactly where it was coming from. Heated the engine up, had some bubbles until the engine reached operating temp and then they stopped, I'm still assuming these are the system purging itself from air. Went for a short test drive, engine ran a bit of a temperature under hard acceleration and the coolant boiled (I heard it gurgling). Initially, the fans would come on for about 5 seconds at a time at rather short intervals, but once the engine got to temperature, the fans were on continuously. Got back home and popped the hood, there was a gurgling sound coming from the top of the engine, tranny-side (not pulley side). I'm suspecting the head gasket. Will poke around some more today.

Update #4 Rented a block tester, looks like the head gasket is good. While I was performing the block test, the coolant boiled when the fan turned on. Since, for the first time, the coolant boiled while I was poking around under the hood, I can state the following:

1. The cooling system pressurises, initially. Even with the block tester in place of the rad cap, the hoses got hard.

2. Something happens at some point to cause a pressure release from the cooling system, and this causes the coolant to boil and the hoses to go soft.

Any tips on what can be causing an intermittent seal? There are no visible external leaks, both rad hoses have new clamps on them at both ends. The overheating behaviour often (but not always) seems triggered by high rpms (2-3k).

Answer 1

Solved it. Removed the thermostat, dropped it into a pot of water and boiled it over the stove. The thermostat that came out of the car didn't open, the one from the junkyard did.

Reassembled, filled the system with water, purged the air and revved the engine at idle. It took longer to reach operating temperature, it didn't overheat and the fans didn't go crazy. The next day, I took it for a test drive, it took longer that on previous test drives to reach operating temperature, and once it did, I started doing jackrabbit starts and the temps stayed normal. Then I got onto a highway with a long steep uphill, I just put my foot into the floor all the way up the hill (3.6 km) and the temps stayed normal.

Yay for thermostats!

Answer 2

A bad head gasket can also generate bubbles in the overflow tank, and would certainly lead to overheating and coolant loss. You can get a relatively cheap kit to test for exhaust gases in the coolant (or any number of other methods to test the head gasket.) If it's a very minor leak, it's possible that it partially seals when it gets up to temp, so the bubbles aren't easily visible.

If the fans are kicking on when the car starts cold, the coolant temp sensor is either bad or disconnected. Or, there's a code (CEL) telling the ECU that the sensor isn't reliable, so it just turns on the fans to be safe. Oil in the coolant can foul the temp sensor, which would also be a sign of a bad head gasket.

Drain the oil and check it's color / consistency. If it has a two week-old chocolate milk look, the coolant is going into the oil, which means the head gasket(s) are bad.

Next option is leaking water pump. Many pumps leak internally, but have you found any signs of coolant leaking? Try a UV dye kit in the coolant system. (Sounds complicated, but Amazon has them for relatively little money - much cheaper than a head gasket job - and they work very well for coolant, oil and AC leaks.)

Basically, the coolant has to go somewhere. It can be:

1. Into the cylinder: Head gasket.
2. Out the exhaust: Head gasket, with white smoke out the exhaust. Only happens after (1) happens.
3. Onto the ground: Water pump, leaking hose, heater core, leaking radiator, REALLY bad head gasket (if it was that bad, though, I doubt the thing would run at all).

Best of luck.