Once burned by milk you will blow on cold water

TheMawn · 2014-04-12, 01:04

Well, with a bit of a lull in my exams, it's time to take the next step with my baby. The hardware has been sitting on my desk for two weeks.

I'm going liquid, baby.

My GTX 660 Ti has the ridiculously powerful ASUS cooler that is inaudible next to my EVGA GTX 670 and 10C cooler, so the GTX 660 Ti will remain under air. However, it is occupying the top slot meaning all its air goes over the back of the GTX 670 PCB on its way in. Now I'll be able to bring it lower and give it better and cooler airflow.

My i5-3570k is currently under an H80i and I'll be a bit sad to put that guy away. I'll have to find it a home. It's using two Noctua 120mm (also sad I won't have a place for those) that are very quiet but probably not suited for radiator cooling, so I'm expecting CPU improvements as well.

I'll be cooling the GTX 670, the i5-3570k and my CPU power circuitry with the built-in waterblock.

Before I start, I'll log today's 24-hour max temps:

GTX 660 Ti: 67C (a bit higher than usual actually)
GTX 670: 69C
i5-3570k: 78C, 83C, 82C, 81C (don't worry, current are 5C cooler across the board)

I leave as just a wee boy. I shall return still as just a wee boy.

TheMawn · 2014-04-14, 06:22

I'm back. I was a massive idiot for the first chunk of the process. I started way too late at night for my own good. Anyhow, I am back and using a computer that is liquid cooled.

I am going to need to reseat the CPU waterblock, which is going to be an absolute nightmare with how crowded the loop is in that area. I am getting wild temperature fluctuations, spiking from 30C to 50C and back down in two refresh cycles while completely idle, and the CPU temperatures don't even stabilize before reaching the 80's so I stop the torture test in about 15 seconds.

In the last fifteen minutes, the GTX 660 Ti has stabilized at its max of 56C, so that's an improvement there. I would expect the max to probably reach 60C over the course of a day, but this is a consistent temperature 10C cooler already.

The GTX 670 is completely unbelievable. 34C maximum! WTF??

I had to give up on cooling the VRM. I got a bit greedy and went with a 360mm radiator on the top side of my case and a 120mm on the back, but things got really cramped back there with just the CPU, GPU and Radiator, so trying to put in two more points was not going to happen.

I also had to abandon the push-pull on the 360mm radiator. That monster is just too big. Still able to get push pull on the 120mm though.

TheMawn · 2014-04-14, 06:53

Here is a question directed at anyone who knows a bit more about this stuff than me:

My pump has a cable that plugs into the CPU header so I can read off the pump RPM (instead of the CPU fan RPM). It is missing the first two pins, obviously, since I'm not sending power, but it has the wire for RPM and the one for PWM. The PWM appears to work between maybe 20% and 90%. There's also a bit of a grinding noise that is at its weakest around 50%-60% but gets really annoying below 40% and above 70%. It's not like a metal on metal grinding, but sounds kind of like an old hard drive during a defrag, but with more humming.

Has anyone experienced this before? I am lead to believe that this is the sound of air bubbles stuck in the loop, going through my pump. I've had the pump going for a few hours now and the level HAS been very slowly dropping but the noise hasn't really improved. Some people say to tilt the case every which way, some say to take the radiators out and give them a good shake (somehow) to clear out the air, and others will say that air will naturally bleed out over the course of a few days.

kracker · 2014-04-14, 15:26

Is the CPU delidded? For a custom loop, it's almost a must as that crappy paste will really hold you back a lot(so I've heard).

TheMawn · 2014-04-14, 17:39

No, I didn't de-lid. I may attempt such a thing in a generation or two, when I really feel I can afford to risk $300 like that. My H80i was applied to a CPU with a lid, too. There's no reason the CPU should be hotter and fluctuate more with a properly applied custom block versus an all-in-one from Corsair.

TheMawn · 2014-04-14, 19:02

Got it. There is a recommended flow direction, and I have the inlet and outlet backward.

Reseating the CPU waterblock didn't change anything, so I'm going to assume it was and still is properly set. Still, swapping the inlet and outlet is going to be major pain in the ass.

chalsall · 2014-04-14, 19:40

Quote:

Originally Posted by TheMawn

Got it. There is a recommended flow direction, and I have the inlet and outlet backward.

LOL... Good thing the pumps didn't overpower the seals, even though they likely spent a great deal of energy trying to do so (as they were instructed to do).

TheMawn · 2014-04-15, 06:54

I switched the inlet and outlet. It took a bit of wrestling. Sadly I'm still having trouble. Stressing all four cores slowly creeps temperatures up to 90C and probably beyond, but that's where I stop the tests.

I may try a third re-seat of the block, but I'm running out of ideas. The second re-seat didn't go so well, the block still being attached to the loop and whatnot.

Batalov · 2014-04-15, 07:43

Sorry, couldn't resist (it is on HBO right now + the bloody moon is in the skies)

kracker · 2014-04-15, 19:04

Maybe you could post a diagram of your water setup perhaps?

chalsall · 2014-04-15, 19:29

Quote:

Originally Posted by TheMawn

I may try a third re-seat of the block, but I'm running out of ideas. The second re-seat didn't go so well, the block still being attached to the loop and whatnot.

This should be a simple engineering problem.

Where are you able to take temperature measurements?

Where are you able to take measurements of the speed of pump(s) and/or fan(s)?

Are you taking measurements regularly, and plotting all of these over time? Cacti et al may be very good friends.

2014-04-12, 01:04	#1
TheMawn May 2013 East. Always East. 11×157 Posts	Once burned by milk you will blow on cold water Well, with a bit of a lull in my exams, it's time to take the next step with my baby. The hardware has been sitting on my desk for two weeks. I'm going liquid, baby. My GTX 660 Ti has the ridiculously powerful ASUS cooler that is inaudible next to my EVGA GTX 670 and 10C cooler, so the GTX 660 Ti will remain under air. However, it is occupying the top slot meaning all its air goes over the back of the GTX 670 PCB on its way in. Now I'll be able to bring it lower and give it better and cooler airflow. My i5-3570k is currently under an H80i and I'll be a bit sad to put that guy away. I'll have to find it a home. It's using two Noctua 120mm (also sad I won't have a place for those) that are very quiet but probably not suited for radiator cooling, so I'm expecting CPU improvements as well. I'll be cooling the GTX 670, the i5-3570k and my CPU power circuitry with the built-in waterblock. Before I start, I'll log today's 24-hour max temps: GTX 660 Ti: 67C (a bit higher than usual actually) GTX 670: 69C i5-3570k: 78C, 83C, 82C, 81C (don't worry, current are 5C cooler across the board) I leave as just a wee boy. I shall return still as just a wee boy. Last fiddled with by TheMawn on 2014-04-12 at 01:04

2014-04-15, 07:43	#9
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 41×229 Posts	Swithing to custom ...khmm... liquid Sorry, couldn't resist (it is on HBO right now + the bloody moon is in the skies)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Firefox burned me again	jasong	jasong	3	2017-04-09 03:50
How cold is it?	petrw1	Lounge	14	2015-01-19 03:22
Hot and cold running crud	xilman	Science & Technology	2	2013-04-12 00:49
Warming cold ∞	xilman	Lounge	7	2013-01-21 20:38
Cold Fusion? Is it possible?	Fusion_power	Lounge	3	2003-08-19 01:13

2014-04-14, 06:22	#2
TheMawn May 2013 East. Always East. 11·157 Posts	I'm back. I was a massive idiot for the first chunk of the process. I started way too late at night for my own good. Anyhow, I am back and using a computer that is liquid cooled. I am going to need to reseat the CPU waterblock, which is going to be an absolute nightmare with how crowded the loop is in that area. I am getting wild temperature fluctuations, spiking from 30C to 50C and back down in two refresh cycles while completely idle, and the CPU temperatures don't even stabilize before reaching the 80's so I stop the torture test in about 15 seconds. In the last fifteen minutes, the GTX 660 Ti has stabilized at its max of 56C, so that's an improvement there. I would expect the max to probably reach 60C over the course of a day, but this is a consistent temperature 10C cooler already. The GTX 670 is completely unbelievable. 34C maximum! WTF?? I had to give up on cooling the VRM. I got a bit greedy and went with a 360mm radiator on the top side of my case and a 120mm on the back, but things got really cramped back there with just the CPU, GPU and Radiator, so trying to put in two more points was not going to happen. I also had to abandon the push-pull on the 360mm radiator. That monster is just too big. Still able to get push pull on the 120mm though.

2014-04-14, 06:53	#3
TheMawn May 2013 East. Always East. 6BF₁₆ Posts	Here is a question directed at anyone who knows a bit more about this stuff than me: My pump has a cable that plugs into the CPU header so I can read off the pump RPM (instead of the CPU fan RPM). It is missing the first two pins, obviously, since I'm not sending power, but it has the wire for RPM and the one for PWM. The PWM appears to work between maybe 20% and 90%. There's also a bit of a grinding noise that is at its weakest around 50%-60% but gets really annoying below 40% and above 70%. It's not like a metal on metal grinding, but sounds kind of like an old hard drive during a defrag, but with more humming. Has anyone experienced this before? I am lead to believe that this is the sound of air bubbles stuck in the loop, going through my pump. I've had the pump going for a few hours now and the level HAS been very slowly dropping but the noise hasn't really improved. Some people say to tilt the case every which way, some say to take the radiators out and give them a good shake (somehow) to clear out the air, and others will say that air will naturally bleed out over the course of a few days.

2014-04-14, 15:26	#4
kracker "Mr. Meeseeks" Jan 2012 California, USA 3²×241 Posts	Is the CPU delidded? For a custom loop, it's almost a must as that crappy paste will really hold you back a lot(so I've heard).

2014-04-14, 17:39	#5
TheMawn May 2013 East. Always East. 11×157 Posts	No, I didn't de-lid. I may attempt such a thing in a generation or two, when I really feel I can afford to risk $300 like that. My H80i was applied to a CPU with a lid, too. There's no reason the CPU should be hotter and fluctuate more with a properly applied custom block versus an all-in-one from Corsair.

2014-04-14, 19:02	#6
TheMawn May 2013 East. Always East. 11010111111₂ Posts	Got it. There is a recommended flow direction, and I have the inlet and outlet backward. Reseating the CPU waterblock didn't change anything, so I'm going to assume it was and still is properly set. Still, swapping the inlet and outlet is going to be major pain in the ass.

2014-04-15, 06:54	#8
TheMawn May 2013 East. Always East. 11·157 Posts	I switched the inlet and outlet. It took a bit of wrestling. Sadly I'm still having trouble. Stressing all four cores slowly creeps temperatures up to 90C and probably beyond, but that's where I stop the tests. I may try a third re-seat of the block, but I'm running out of ideas. The second re-seat didn't go so well, the block still being attached to the loop and whatnot.

2014-04-15, 19:04	#10
kracker "Mr. Meeseeks" Jan 2012 California, USA 3²·241 Posts	Maybe you could post a diagram of your water setup perhaps?