Found coolant leak/temp fluctuation I've been chasing... looking for thoughts

[sapote]

It's smooth. The needle pins 90-100C to 12 O'clock, but outside of the range it has some sort of sliding scale.

Also, I got the water pump out. It's a Geba (not OE).

1) So in order to see the needle jumps, the temp must had changed from 100 to 103 or 105, and not just 100 to 101. 3 or 5C is not small.
2) did you feel any radial or axial plays on the old pump?

[heinzboehmer]

Also, I got the water pump out. It's a Geba (not OE).

Was it the source of the leak?

[JustAWhisper]

Interesting. If M cars have buffering (quantization to be correct) as you said, say needle stays at 12 o'clock for actual temp in [90:100C] range, then it jumps to the next position, say 1 o'clock for [101:110] then we should see a jumpy needle and not a continuous moving needle. Do people see a smooth moving needle to jumpy needle?

It's smooth. The needle pins 90-100C to 12 O'clock, but outside of the range it has some sort of sliding scale.

Also, I got the water pump out. It's a Geba (not OE).

[sapote]

I think you are correct on the DME, but I did look into the buffer issue a while back and I learned the M3 is buffered, just in a much smaller range, 90-100C. Non-M's are 85-115C iirc.

Interesting. If M cars have buffering (quantization to be correct) as you said, say needle stays at 12 o'clock for actual temp in [90:100C] range, then it jumps to the next position, say 1 o'clock for [101:110] then we should see a jumpy needle and not a continuous moving needle. Do people see a smooth moving needle to jumpy needle?

[JustAWhisper]

I think the DME sampled the temperature and averaged out every 5 seconds or so, as there’s no benefit to do the average calculation faster than that. Obviously it doesn’t display the raw noisy sample data. But I thought the M3 don’t use buffering for the gauge as non-M cars.
As about the leak affected the pressure and temperature, I thought you had a very small leak that very hard to find, which should not change the pressure. A small leak as 10cc in 30 minutes should not change the operating pressure or temperature.

I think you are correct on the DME, but I did look into the buffer issue a while back and I learned the M3 is buffered, just in a much smaller range, 90-100C. Non-M's are 85-115C iirc. Without knowing the cooling system design and doing the math, it's hard to say for certain if a small leak is enough to cause the temperature difference I'm seeing in the gauge or not. I just know it's theoretically possible and it makes sense to me. You may be correct and it's just gauge issue. Once the cooling system is completely fixed I'll be able to find out for sure.

Coolant mist spray is very often the pump. I bet when you pull it, it’s not OE.

That's what I'm thinking, too.

[Obioban]

Coolant mist spray is very often the pump. I bet when you pull it, it’s not OE.

[heinzboehmer]

Go take a look at the inside of your water pump pulley with a mirror. My pump leaked through the bearing and there were a lot of specks of coolant on the inside of the pulley. Could also be the gasket between the pump and block.

[sapote]

I think the DME sampled the temperature and averaged out every 5 seconds or so, as there’s no benefit to do the average calculation faster than that. Obviously it doesn’t display the raw noisy sample data. But I thought the M3 don’t use buffering for the gauge as non-M cars.
As about the leak affected the pressure and temperature, I thought you had a very small leak that very hard to find, which should not change the pressure. A small leak as 10cc in 30 minutes should not change the operating pressure or temperature.

[JustAWhisper]

sapote, I don't think you're understanding what I'm trying to communicate. Maybe wiggle was a bad way to describe it, because it seems you're thinking the needle is bouncing around wildly.

What I've seen is moments where the coolant temp is hanging out right on the edge of the buffer, which is at 100C. The temperature is fluctuating slightly (it always fluctuates slightly, that's why BMW buffers the gauge), but it's averaging at a higher temperature than it should be (~100C instead of ~93C), and some of the fluctuation is spilling outside the buffer. These are slight needle movements. The reason the temperature is elevated is because there is a small pressure loss in the system from the leak. It's not a pressure jump, the entire system is running slightly outside it's design condition when a leak is introduced into it. This is why failing water pumps often cause small overheating problems before complete failure.

Speaking of water pumps, I've come to the conclusion that it is the likely culprit. I'm having trouble accepting it, as the water pump is less than 3 years old and only has about 14000 miles on it, but coolant is clearly getting on the pulleys and the thermostat housing is clean from what I've been able to see.
There are about 8 track days and 30+ autocross events on the pump, so maybe that has something to do with it, but I was really expecting more life out of it. It's also possible the shop that replaced the water pump lied when they said they put an OE BMW pump in. I never looked close enough as I trusted them at the time, but in retrospect I should have because I ended up having a bunch of issues with that shop.

[sapote]

If there is small leak then focus on fixing it.
Re to gauge needle jumping, sudden change in pressure still cannot cause the temperature to jump. We are not talking idea gas here with PV/T = nRT.

And what could cause sudden jump in pressure? Nothing unless tank ruptured or exhaust gas leaked into coolant, but you didn’t have that. So I say it jumped bc electrical or mechanical in the gauge.

[JustAWhisper]

I only use genuine OE BMW parts only for the cooling system. I agree, there should not be any excursions past dead center on the gauge with normal cooling system operation.

sapote, a pressure fluctuation can dork with the sensor. I guess "a couple" second could also be "a few." I would say it was definitely no more than 10 seconds, probably 5-7. You may be correct that there is an issue with the sensor/gauge. I've been chasing weird needle behavior for a while now before it became apparent there was a leak.

Anyways. There's clearly still "something" wrong. The question is what. I haven't had a chance to do any more work on it yet, so no new updates.

[sapote]

The "wiggle" I was referring to was a small needle excursion outside the upper end of the the band, which I believe is 100C. It was only a small wiggle, so I was only at 100.5-101C, and only for a couple seconds. My understanding is that it generally shouldn't happen in normal driving conditions, and indicates a pressure problem in the system.

Nothing could change the coolant big thermal mass that much within a couple seconds. Either the sensor or the gauge is jumpy, not the coolant temperature!

[8000RPM]

What brand water pump and radiator did you buy? Just for sanity sake, did you check the play on the new water pump. I’ve had OEM fail pretty quickly in the past whereas the BMW brand will go 100k miles safely.

A well maintained cooling system should not have temperature swings. Once I got mine squared away, it didn’t move even in 110F Las Vegas summer.

[JustAWhisper]

Temperature changes is normal.

if the white dried coolant spray appeared again, I would do this: connect the pressure tester to the tank but don’t pressurize it; run engine until reaching operating temperature, then pump the tester to increase the pressure slowly until the leak show up. But don’t pump higher than 35 psi.

In your original issue, using the pressure tester at operating temperature would have pinpointed the leak.

Testing at cold condition, the crack might not expand to rear its head.

The "wiggle" I was referring to was a small needle excursion outside the upper end of the the band, which I believe is 100C. It was only a small wiggle, so I was only at 100.5-101C, and only for a couple seconds. My understanding is that it generally shouldn't happen in normal driving conditions, and indicates a pressure problem in the system.

Anyways, after watching all weekend, there does appear to be some fresh coolant leaking out of somewhere. It may be the same or a different leak to before (perhaps there was more than one all along). I am quite frustrated, but there's nothing to do other than continue to investigate.

[sapote]

Fast forward to today and I saw a characteristic wiggle in the water temp gauge.

Temperature changes is normal.

if the white dried coolant spray appeared again, I would do this: connect the pressure tester to the tank but don’t pressurize it; run engine until reaching operating temperature, then pump the tester to increase the pressure slowly until the leak show up. But don’t pump higher than 35 psi.

In your original issue, using the pressure tester at operating temperature would have pinpointed the leak.

Testing at cold condition, the crack might not expand to rear its head.