Temps High on Custom Loop

  • Thread starter Thread starter mk1_salami
  • Start date Start date
Didn't mean to scare you. I scared myself when the first time I tried to fill my system it overflowed! Fortunately I was only going to test the plumbing so there was no power at the time and luckily no water got to the important parts. Using a squeezy bottle and going slow I can fill mine and watch out for the liquid backing up. If I stop, it then usually drains down and I can continue. But if I have to fill the whole system, which is several reservoir fills, I use the inserted tube method so that I can fill a bit quicker.
 
Last edited:
My 2p would be not not bother with a permanent fill port/line at all. Just a temporary flexible tube just for when you are filling and bleeding which you will take off afterwards. Will help considerably with the case rocking bit mentioned previously.
 
Yeah I guessed you didn't go straight up
Because there's a brace there
Pictures too dark to really see but
Why use any Angled fittings?
Bulkhead fitting with a straight compression fitting
Come straight down a bit further before making a bend in the tube
Should work from what I can see
Yeah you could just use a bit of soft Tubing
Though I could never remember where I stored it when
I needed it lol
Plus I actually think the extra rigid tube into the reservoir
Adds to the look
Though just my opinion looks are of course subjective
 
Bubo said:
My 2p would be not not bother with a permanent fill port/line at all. Just a temporary flexible tube just for when you are filling and bleeding which you will take off afterwards. Will help considerably with the case rocking bit mentioned previously.
Click to expand...
Exactly. Despite always using soft tubing, which would make even easier to use one, never really used one.

About the angled fittings, I managed many times without any, and for hard tubing should be doable. From the little research I done when I was considering it, as long as you leave some space before the bend, should be fine. You need/want a straight piece of tubing which hasn't been "disturbed" to be the one connecting to your fittings to avoid any leaks.
For bleeding, only when I was using a so-so pump and 3 very restrictive radiator and very restrictive blocks that I had to shake the case a lot.
Using a D5 at full tilt, just tilting it a little and holding for few seconds, then opposite side, few times, will help. But just be aware that when at full tilt the pump will bleed air, some of the air bubbles will get to the intake of the pump and recirculate. Will take some time but it's normal. And after few hours or days, when you think most air is gone, you may notice the coolant level dropped a bit. You may try and tilt the case, gently to one side, then the other ever now and then for the first few days, as air will get stuck somewhere.
 
I see that you've removed the XD3 but for future reference having a 90 on the front outlet of an XD3 causes enough turbulance that at high pump speeds the air from the top of the res gets pushed into the loop. It's only about an inch and a half from the top of the port to the fill level. You can watch it happening. When I redid mine I came straight out the side outlet which cured it.

I drilled my case for a fill port but it didn't work well with the XD3, no other ports on the top to let air out as water goes in. Went back to the trusty old squeezy bottle.
 
So a bit of an update here. I threw it all back together with new Gelid Extreme pads on the GPU, Thermal Grizzly Kryonaut on the GPU and CPU and honestly it didn't make much difference! Was still seeing 68-70c on the GPU during Heaven or 3dmark stress tests.

I had my top radiator fans in a "pull" configuration, so cool air was being sucked in from the front rad, but hot air pulled through the top rad. At 2am last night i drained the loop - CONSIDERABLY easier with my little air inlet at the top of the rad and drain port on the reservoir. Like - really easy. Nearly a litre simply poured out without any tilting.

Anyway, I changed the top fans to be an intake too. So ALL intake, bar the 1x140mm rear exhaust fan. Well wouldn't you know, my temps have plummeted. Now a maximum of 61c on the GPU no matter what I do (with case completely closed and fans on max). In real world gaming it seems to hover around 57-59c. So from the previous setup before teardown, I've dropped the GPU temp from 70-72c while gaming to under 60c. Very happy with that result, and means a much quieter computer.

Also, the CPU temp at 90c was largely due to PBO2 settings in BIOS. With the bios reset to defaults, the CPU would not get hotter than 65c during Cinebench R23. With PBO2 turned on it was 72c but only boosting to ~4100mhz across all cores. With other PBO2 settings I can boost to ~4550mhz on all cores, but it basically boosts until the BIOS imposed thermal limit of 90c.

So... a mixed result of the teardown. In hindsight I should have chaned the fan orientation first and it probably would have given me 80% of the gains I've seen in doing everything.

 
mk1_salami said:
So a bit of an update here. I threw it all back together with new Gelid Extreme pads on the GPU, Thermal Grizzly Kryonaut on the GPU and CPU and honestly it didn't make much difference! Was still seeing 68-70c on the GPU during Heaven or 3dmark stress tests.

I had my top radiator fans in a "pull" configuration, so cool air was being sucked in from the front rad, but hot air pulled through the top rad. At 2am last night i drained the loop - CONSIDERABLY easier with my little air inlet at the top of the rad and drain port on the reservoir. Like - really easy. Nearly a litre simply poured out without any tilting.

Anyway, I changed the top fans to be an intake too. So ALL intake, bar the 1x140mm rear exhaust fan. Well wouldn't you know, my temps have plummeted. Now a maximum of 61c on the GPU no matter what I do (with case completely closed and fans on max). In real world gaming it seems to hover around 57-59c. So from the previous setup before teardown, I've dropped the GPU temp from 70-72c while gaming to under 60c. Very happy with that result, and means a much quieter computer.

Also, the CPU temp at 90c was largely due to PBO2 settings in BIOS. With the bios reset to defaults, the CPU would not get hotter than 65c during Cinebench R23. With PBO2 turned on it was 72c but only boosting to ~4100mhz across all cores. With other PBO2 settings I can boost to ~4550mhz on all cores, but it basically boosts until the BIOS imposed thermal limit of 90c.

So... a mixed result of the teardown. In hindsight I should have chaned the fan orientation first and it probably would have given me 80% of the gains I've seen in doing everything.

Click to expand...
I think that looks really nice, for me it`s a lot nicer to look at than distro plate builds as you can appreciate the routing and bends that go with it.
 
mk1_salami said:
Anyway, I changed the top fans to be an intake too. So ALL intake, bar the 1x140mm rear exhaust fan. Well wouldn't you know, my temps have plummeted. Now a maximum of 61c on the GPU no matter what I do (with case completely closed and fans on max). In real world gaming it seems to hover around 57-59c. So from the previous setup before teardown, I've dropped the GPU temp from 70-72c while gaming to under 60c. Very happy with that result, and means a much quieter computer.
Click to expand...
Tidy looking build.

My setup is vaguely similar and I didn't see any change in temps by switching the top rad to intake. I'm using a thin rad on the top though.
I did see a big reduction in dust buildup because both intakes are filtered and the rest of the case is a sieve.
I'm at low to mid 50s for the gpu while gaming but it's an FE model and doesn't pull much more than 350w with the slider maxed.
 
I had big issues with temps on my set-up and simply had to move it around absolutely everywhere for all the air to work it's way out.
 
@mk1_salami - I have been interested in this thread because there are similarities between your system and mine. This is my first ever water cooled system so I must have made mistakes but I seem to be getting much better results than you.

Prompted by your thread I have been doing a few of my own tests, rerunning Cinebench and checking my PBO settings. However I don't have any need for CPU-intensive applications so I have focussed more on monitoring temps during gaming. I have my system balanced to be as quiet as possible by allowing temps to get near the operating maximum, which is dominated by the CPU. During gaming my 5900X peaks into the 80-85C range and my 3090 can reach up to 50C. That's with the CPU peaking at 4,900 MHz and around 175W. I managed to achieve -24 on the fastest 4 cores and -30 on the rest. The GPU is undervolted and limited at 0.875V and 1,980 MHz with a 360W peak.

As you have noticed, cooling certainly makes a difference. I have a more aggressive set of fan curves which will shave off 10C from the CPU peaks. This curve tends to increase the fastest fans from about 600-700rpm to 700-800rpm with a noticeable increase in noise, so I'm happy to keep with the quieter curves as long as my CPU never hits 90C. My front rad is a 420 HW labs Black Ice SR2 60mm with 3x140 as intake. I also have 2x140 in the base as intake (no rad). The top rad is a 420 Alphacool XT45 45mm crossflow with 3x140 as exhaust and the rear rad is a 140 Alphacool XT45 45mm crossflow with 1x140 as exhaust. I have dual D5 pwm pumps which run at about 65l/h at idle and peak at around 135l/h when the water rises to 37-39C.

Let me know if I can give you any more data to help compare to your system.

I should have noted that although it had good reviews, I mainly chose the HW labs rad as intake because it only has 9fpi as opposed to the 12fpi of the Alphacools, in the theory that it would be less restrictive in passing airflow. The two bottom fans are just to ensure that I am getting enough cold air into the case and try to keep a positive air pressure (to avoid extra dust - although I'm not sure how much effect that has). I think at load my case air hits nearly 40C although I have not paid much attention to that.
 
Last edited:
therealdippy said:
@mk1_salami - I have been interested in this thread because there are similarities between your system and mine. This is my first ever water cooled system so I must have made mistakes but I seem to be getting much better results than you.

Prompted by your thread I have been doing a few of my own tests, rerunning Cinebench and checking my PBO settings. However I don't have any need for CPU-intensive applications so I have focussed more on monitoring temps during gaming. I have my system balanced to be as quiet as possible by allowing temps to get near the operating maximum, which is dominated by the CPU. During gaming my 5900X peaks into the 80-85C range and my 3090 can reach up to 50C. That's with the CPU peaking at 4,900 MHz and around 175W. I managed to achieve -24 on the fastest 4 cores and -30 on the rest. The GPU is undervolted and limited at 0.875V and 1,980 MHz with a 360W peak.

As you have noticed, cooling certainly makes a difference. I have a more aggressive set of fan curves which will shave off 10C from the CPU peaks. This curve tends to increase the fastest fans from about 600-700rpm to 700-800rpm with a noticeable increase in noise, so I'm happy to keep with the quieter curves as long as my CPU never hits 90C. My front rad is a 420 HW labs Black Ice SR2 60mm with 3x140 as intake. I also have 2x140 in the base as intake (no rad). The top rad is a 420 Alphacool XT45 45mm crossflow with 3x140 as exhaust and the rear rad is a 140 Alphacool XT45 45mm crossflow with 1x140 as exhaust. I have dual D5 pwm pumps which run at about 65l/h at idle and peak at around 135l/h when the water rises to 37-39C.

Let me know if I can give you any more data to help compare to your system.

I should have noted that although it had good reviews, I mainly chose the HW labs rad as intake because it only has 9fpi as opposed to the 12fpi of the Alphacools, in the theory that it would be less restrictive in passing airflow. The two bottom fans are just to ensure that I am getting enough cold air into the case and try to keep a positive air pressure (to avoid extra dust - although I'm not sure how much effect that has). I think at load my case air hits nearly 40C although I have not paid much attention to that.
Click to expand...
Thanks for the input. Interesting to hear your GPU doesn't exceed 50c. Do you have the active backplate too? What model of 3090 is it? Seems you have 980mm of total radiators, vs my 760mm. I expect that'll made a decent impact on cooling.

Right now I'm still experimenting with PBO. My fastest cores are running -15 and the rest I incrementally spaced them down to -30 (-17, -19, -21, -23 etc).

Right now I'm not blown away by the performance of the watercooling and would like to run cooler, but I can't keep throwing money at it. I may look at adding a 140mm rear radiator.
 
therealdippy said:
@mk1_salami - I have been interested in this thread because there are similarities between your system and mine. This is my first ever water cooled system so I must have made mistakes but I seem to be getting much better results than you.

Prompted by your thread I have been doing a few of my own tests, rerunning Cinebench and checking my PBO settings. However I don't have any need for CPU-intensive applications so I have focussed more on monitoring temps during gaming. I have my system balanced to be as quiet as possible by allowing temps to get near the operating maximum, which is dominated by the CPU. During gaming my 5900X peaks into the 80-85C range and my 3090 can reach up to 50C. That's with the CPU peaking at 4,900 MHz and around 175W. I managed to achieve -24 on the fastest 4 cores and -30 on the rest. The GPU is undervolted and limited at 0.875V and 1,980 MHz with a 360W peak.

As you have noticed, cooling certainly makes a difference. I have a more aggressive set of fan curves which will shave off 10C from the CPU peaks. This curve tends to increase the fastest fans from about 600-700rpm to 700-800rpm with a noticeable increase in noise, so I'm happy to keep with the quieter curves as long as my CPU never hits 90C. My front rad is a 420 HW labs Black Ice SR2 60mm with 3x140 as intake. I also have 2x140 in the base as intake (no rad). The top rad is a 420 Alphacool XT45 45mm crossflow with 3x140 as exhaust and the rear rad is a 140 Alphacool XT45 45mm crossflow with 1x140 as exhaust. I have dual D5 pwm pumps which run at about 65l/h at idle and peak at around 135l/h when the water rises to 37-39C.

Let me know if I can give you any more data to help compare to your system.

I should have noted that although it had good reviews, I mainly chose the HW labs rad as intake because it only has 9fpi as opposed to the 12fpi of the Alphacools, in the theory that it would be less restrictive in passing airflow. The two bottom fans are just to ensure that I am getting enough cold air into the case and try to keep a positive air pressure (to avoid extra dust - although I'm not sure how much effect that has). I think at load my case air hits nearly 40C although I have not paid much attention to that.
Click to expand...
Do you have pictures of your setup?
 
Because this was my first water-cooled build, I was nervous about stripping a GPU to add my own waterblock, so I opted for a pre-built. Mine is the Gigabyte 3090 xtreme waterforce WB (which was ridiculously expensive :mad:). It has a passive backplate and the memory seems to reach 80-85C under load, which is a healthy margin below the maximum operating temperature.

I never took any photos of my PC but I guess I could do one.
 
Bit of a mini update. Some may have seen in the other thread that the second Xd5 pump I received made the same grinding noise as the first.

So … time to drain the loop again and replace it. While I’m there, I relocated the 480 to the roof, the 280 to the front and have ordered an additional 240 rad for the base of the case.

Top and front will be in push pull configuration. It’s a bit of an excuse to have a play again and add a third radiator. Will report back when I receive the new radiator and fans with the results. Here’s a small work on progress - mock up (not final runs) pic while I wait.


online image hosting
 
Last edited by a moderator:
back again, rotten luck with the pump's, good to see the 480 in the roof and hopfully its taking the heat from the gpu straight out of the case, the 280 will be more than enough to cool your chip, not much heat will vent back into the case and the additional 240mm rad should work wonders to cool the fluid down before returning to the pump (i'm guessing thats how your going to go with loop order?)
i've got everything crossed that the replacement pump works and you dont have any more problems with excess sound, nothing worse than a bad pump as a loop has to be drained and taken apart, which is frustrating at the best of times, let alone when something goes wrong.

keep us all updated and good luck!
 
Last edited:
thats looking a really solid setup now, the air will move around in time thats trapped in the loop, try making the pump go faster then slow it down a few more times, probally take a couple of days by which point you'll notice your coolant level will drop in the res, when that happens top it back up and happy days, out of intrest how are temps now under load and what rpm's are you running at, also what kind of pump rpm's at idle and load are you hitting?

it's important not to run the pump too fast at idle as the coolant will be moving too fast through the loop to take heat away when you play a game
 
Last edited:
I'm surprised the change of fan configuration made any notable difference to your absolute GPU/CPU temps. Especially so given that your coolant temperature was absolutely fine

You REALLY need to record and analyse your Coolant and CPU/GPU/Ambient DeltaT values. Its the difference between the coolant temp and ambient temp which will indicate the performance of your loop.

You should also analyse your temperatures when the loop reaches temperature equilibrium.

The CPU and GPU blocks can only dissipate so much wattage. Beyond that it's simply a case of your radiators being able to manage the coolant temperature. The only true metric of that is as above - The DeltaT of your coolant over ambient. As an example my loop when handling full synthetic load from a 2080 and 3900X has a deltaT (Coolant over Ambient) of between 1 and 3 degrees C. Thats a very good value, but I have 960mm of rad space, 2x120mm intake and 1 x 140mm exhaust.

That being said I would suspect your fan change may have contributed a small amount to the absolute temperature inprovement but I suspect different factors have played a part. I.E What was the recorded ambient between each observation?

I'd be interested to know your typical Ambient and Coolant temperatures at idle and under load.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom