New custom loop handles high GPU and water temperatures effectively
New custom loop handles high GPU and water temperatures effectively
Ooh... that's the chroma at 1200 RPM, it makes more sense now. Still high, but not too off. When the computer starts up, the liquid temperature should match the room temperature. At idle, it usually rises about 5 to 10 degrees depending on fan speed. The sensor looks fine too.
I think a mid-to-high 40s for the GPU would be expected. For troubleshooting, I don't have many ideas other than suggesting modifications that might need temporary or permanent changes, like adding a flow sensor. Just let me know and I'll begin typing.
"Going forward I can only suggest things that will require some temporary or even permanent (like adding flow sensor) modifications.
Just say yes and I'll start typing
"
I want this build to run flawless, so tell me what to do, and i will try it. Tho if it involves adding new stuff, like a flowmeter, to the loop, i wont do it in the next 14 days. Tho im kinda oldish, there is still a LAN and a new WoW exp comming up
😀
The process moves smoothly enough at this stage. Adding a flow meter proves beneficial no matter what current problem exists. It aids in spotting potential future issues such as partial blockages caused by debris accumulation or other factors. Noticing a drop in flow indicates it's time for thorough maintenance of the loop.
The situation becomes slightly more complex—you'll need a controller to link the sensor. I prefer models from aquacomputer. Aquaero works well but comes at a higher cost. Please let me know your budget limits, desired control features, and space constraints.
Now addressing the main challenge: let's begin with a simple solution.
1. As previously discussed, swap the side fans on the radiators to exhaust mode while keeping them active. This is useful for several reasons:
- The top radiator won't reuse hot air from the side radiator when the case is closed.
- It simplifies cleaning; in "push" mode dust accumulates between the fan and radiator, requiring disassembly. Switching to "pull" mode would need similar adjustments to the fittings connecting the tube from the radiator to the exhaust.
2. Remove the dust filters on panels where radiators are mounted—they're reducing performance without offering any practical benefit (once all radiators start exhausting air).
Test the adjustment with the case open, then close it and set fans to full speed. The result should show a temperature difference of 7-10°C above ambient and 10-15°C higher than the GPU, giving you a 17-25°C gap between ambient air and GPU temperature.
Another optional tweak:
Refer to https://noctua.at/en/nf-a12x25-pwm—they deliver significantly more airflow at lower noise levels. At 1500 RPM they'll match your current fans' noise at 1000 RPM while moving much more air through the radiators. However, they're costly and bulky—best if you can conceal them on the top radiator.
For testing flow, a flow meter or a flexible tube with 1.5-2 meters of length and barb fittings would be necessary.
I'm less likely to proceed with this test unless you clarify that the temperature sensor is located on the GPU inlet port, placing it in the coldest part of the loop after all radiators and before any heat-generating parts.
I just adjusted the fans on the side and took out the dust filters. Then I restarted the system again. Both tests used the panels installed. The fan speed affected the outcomes more than I thought, but the noise level is still excessive.
Fans 850 RPM
Ambient: 26
Water: 43.6
GPU: 53
Fans 1350 RPM
Water: 38
GPU: 47
I mostly prefer not to change the fans since I just got them for this build, and they’re really costly. This model seems adequate enough, I hope. I’ve been using the 140mm and 120mm versions on my machine and the GF’s AIO for several years. Currently, I’m running at about 750 RPM and 550 RPM when browsing.
If these settings are causing issues or being a bottleneck, I’ll switch them out. Otherwise, the budget is flexible—anything within reason works. I just want to avoid making the build too cluttered, so permanent changes should be subtle.
The difference between the coolant and GPU is quite noticeable.
Therefore, I’ll stick with another configuration – top and side panels removed – to determine whether the issue lies in airflow or radiator efficiency.
If it’s the former (a 5°C or greater rise in liquid temperature), you can either keep it unchanged (a perfectly acceptable choice) or think about replacing the fans, as they generate more airflow with lower noise levels, even at higher speeds. The expense is a factor, but their warranty is six years. Another justification for changing the fans is that your radiator is slightly thicker, which demands a bit more static pressure from them. Adding another set of fans in the same style (push/pull) should raise the temperature by 2-3°C while maintaining the same noise level.
In the other scenario, it’s likely due to a large volume of air trapped inside the radiator or some obstruction along the path.
I believe your loop performance could improve because I recently tested my system with these settings:
- Two 240mm 30mm radiators (thinner than yours, less critical) and 33% smaller surface area (critical)
- Strongly overclocked GTX 1070 plus a few Prime95 threads to balance power usage with your GTX 1080Ti
- 4xNF-F12 at 1350RPM
- Ambient temperature around 30-31°C
- Restricted airflow: rear intake via ~120mm fan opening and very limited bottom area
Despite this, the liquid never exceeded 41°C (removing the side panel would drop it below 38°C)
Considering your loop has 50% more radiator surface and cooler ambient conditions (~4-5°C lower), performance should significantly improve.
I’ll share the links to "gadgets" later.
I'm noticing some improvement. I'm beginning to think the low RPMs might be contributing to higher than expected temperatures. What are your radiator FPI values? Higher FPI radiators require stronger fans for optimal cooling performance because of their fin density, while lower FPI ones work better with slower speeds. Still, increasing RPM generally improves performance no matter the radiator type. I currently run my Gentles Typhoon AP-15s around 1200 RPM. They're audible but manageable. For running below 1000 RPM, you'll need significantly more cooling capacity to stay quiet and maintain low temperatures.