Okay, old pump out, and can't find anything wrong. Passages are clear as far as I can see in. So now I'm looking for some small sign of an issue. Here are some pictures and questions:

1. The blades on the old pump appear to have the corners rounded off. It looks likely they are rounded off by erosion and not by a stamping machine during manufacture. The overall diameter of the impeller is the same on old and new when I put a caliper on them to measure:


2. The shaft is purple. It looks like from heat. Since there are a couple of bare-metal chips on there, I'm guessing that the purple is discoloring from during the manufacturing process, but I don't know that for sure:


3. Side shot of a rounded blade on the old impeller. That discoloring appears to be from a spot weld during manufacture. All blades have it and it is consistent:


4. Side shot of a square blade on the new impeller:


5. Gap between the back of the old impeller and the old case. This measurement shows 5/16":


6. Gap between the back of the new impeller and the new case. This measurement looks like 3/16" but that is a bad angle to cause the camera to focus. It is really more like 1/8":


The rounded blades don't appear to be significant enough to cause the severity of problem that I am having. Thoughts?

The 3/16" extra gap in back gives me more concern, but still would this cause a chronic severe overheat? Related to this, I did notice the same difference between the amount of shaft sticking out of each impeller. The new pump has about 3/16" more shaft sticking out than the old one. Thoughts?

Doesn't seem to be off by that much. The 3/16 may give it a tighter fit and move more coolant. Let'* see how the new one does.

Okay, it will be going on in the morning!

Well . . . that didn't happen.

I decided to beat my head on the overheat issue a bunch more by doing lots of research, then decided to take it all apart and flush everything. All of it came out clean. I put it back together, including the new water pump, and still same issue. I got out my IR thermometer and checked around. Lower hose halfway between radiator and thermostat housing, upper hose at radiator, and thermostat housing, all have nothing like the temperatures that the gauge and the PCM (through OBD2) are reporting. Watching the thermostat housing, I see it get up to about 185 degrees then cool itself off. Watching the hoses I see the same, and about 25-35 degrees difference between upper and lower hoses when it is flowing. It can start cold and idle forever and not have the issue. If it is driven, the issue follows engine effort, and when parked and idles it eventually recovers, then can idle forever without issue.

Earlier in the project, I noticed I can have a nearly-immediate direct effect on the temperature reading by squirting a little water on the coolant temperature sensor. Way back then, I thought I was just tricking the sensor by cooling it off, and I'd watch it rise right back up again fairly quickly. Now I'm thinking that the coolant sensor tricked me instead, by really reading head temperature or somesuch, instead of coolant temperature.

So now I'm wondering if the coolant temperature sensor is just reading hot because it happens to be located next to the exhaust manifold. The PCM and dash gauge closely follow the IR-thermometer temperature of the head next to the coolant temperature sensor. If everything else is reading okay with the IR thermometer, and this is closely following what the IR thermometer says is happening right next to the coolant temperature sensor, then I probably don't have any overheat issue in the first place, just a measurement issue.

I love having these revelations after putting bunches of parts and skinned knuckles into a project.

So here'* the part that doesn't compute. It worked fine for over 150,000 miles, then suddenly started doing this. Change to new sensor and does same exact thing. What changed to cause this behavior to start?

You have a genuine head scratcher there. Maybe the PCM is getting finicky.

Good to hear. I don't like asking the easy questions.

I was thinking some possibility of PCM delivering bad voltage to the thermistor. Checked a bunch of circuits between chassis ground and various fuses and all show 14V. I'll backprobe the temp sender when I get a chance and see what it shows.

I also watched my IR thermometer, aimed at the metal next to the temperature sender, tell me almost the same temperatures as the PCM. This makes me think it is reading the temperature of the head, not the coolant. Almost like I need to isolate the temp sensor from the head while allowing it to read the coolant. That doesn't seem possible.

Okay, stayed up later than I should last night devising next steps. I'm going to install a coolant temperature sensor hose adapter in the upper radiator hose and relocate the sensor there:



I am concerned that the PCM won't see the coolant temperature rise quickly enough from a cold start as compared to being on the exhaust side of a head . . . especially if it considers ambient temperature in its equations and predictions. At the same time, where it is currently installed cooks it really well, and the little sleeve heat shield seems inadequate to the task of defending it from all of that heat:



We'll see what happens.

Related: I know of a 2017 of the same platform has already cooked the same sensor to death, and had it replaced. Looking around online it appears that it is common at around 50,000 miles, like on the one I'm talking about. If this works on the 2007 without setting codes etc. then I'll be doing the same on that one in a few months. In the olden days, I would just wait for it to fail again, but this one refuses to allow the air conditioning to work as a punishment for the sensor going bad. I understand the thinking, but it still sucks when it happens 500+ miles from home in triple-digit weather.

It will take a little longer to respond from a cold start as it will be cold until the thermostat opens,. I'm just wondering why it'* doing this now.

Me too. Baffles me. I thought about jury-rigging airflow directly to the coolant temp sensor and shield, but then decided if I am going to change stuff, I may as well change it away from that exhaust manifold and hopefully improve it overall. I don't like deviating from factory on stuff like this, but I'm done fiddling around with this issue if this is a good fix for it.

I forgot to post an update. Thanks to Mgiledi who asked about this.

It has about 8,000 miles on the "repair" and is doing fine.

I installed one of these in the upper radiator hose:
Don't forget to buy two more hose clamps.

I tapped it to the 12mm threads of this car'* temperature sender, and installed it in approximately the middle of the upper radiator hose, on the straight part.

Remember, don't just cut the hose in half. It needs to have the same length cut out of it as the main body of the adapter.

The temp sender'* wire harness can be pulled back out of the bundle that it comes from such that it can connect to the sender in its new location. Be sure to wrap the now-exposed pair of wires in something protective. This was a brand new sensor that I used, and the original sensor is in the head where it is supposed to be, but only serves to plug the hole.

The coolant temperature on the dash gauge takes longer to come up on a cold start, and then comes up more quickly than otherwise when the thermostat opens for the first time. After that, it is no different than before. The only downside that I can see here is that if the thermostat fails closed, I wouldn't see the temperature rise on the gauge as quickly as in the factory location.

I am 99.99979% convinced that the little insulating heat shield that is supposed to be around the sender is the problem. It must let heat in that otherwise would have been reflected away. Or something. I like to keep things factory, but this was too big of an issue with too easy of a fix, to fiddle with it any longer.

Someday, I'll take the thermostat out and drill a couple of small holes in it that will allow a bit of coolant to bypass when it is closed. That will help the temperature cause to climb more easily and gradually, and will also solve the issue of a stuck-closed thermostat not reading on the coolant temperature gauge. I'll probably do this when I replace all of the old hoses.

Oh, and by the way: This is on a 2007 Saturn Outlook. The Saturn forum is so seldom visited that I knew I'd get better results here . . . and help the Acadia folks as well.