First "custom" loop - have I done something wrong?

Associate
Joined
17 Feb 2021
Posts
29
I've just built my first loop as a fun project, with the primary intention of being a cost effective way of reducing my loud graphics card - but my temperatures seem a tad high. My secondary intention was for fun, so as I'm no expert I wanted to throw this to you guys for some guidance. All parts at stock speeds and ambient temperature was between 19-21C.

Issues/questions

I learned a lot on this project, most of it from searching these forums, so thank you all. I'll add what I think are my current issues.
  • Are these temperatures appropriate for the setup?
  • The pump can only run at 2600rpm and pushes 100l/h. Is this adequate? (Kit wasn't strictly designed for cooling two parts)
  • What is the solution to the fan RPM being linked to the CPU temperature? When I stress test the GPU I had to manually increase fan speed as the temperatures were only going up. Once I increased, they stabilised at around 77/78 hotspot. I've seen coolant temp being used - is this necessary/possible for my loop?
Main points aside, would you have done anything differently in my position? I was a total noob to watercooling when purchasing, so if I could go back I probably wouldn't buy a kit as I hate the CPU block mounting hardware.

PC specs

Ryzen 5 3600
Asus Prime B450m-a
2x8gb Corsair vengeance 3200mhz
Corsair RM750X
MSI mech OC RX5700XT
Corsair 4000D airflow

Cooling specs

Alphacool eissturm kit, containing:
  • Radiator: NexXxoS ST30 Full Copper 240 mm
  • CPU: NexXxoS XP³ Light
  • Reservoir: Eisstation
  • Pump: DC-LT 2600
  • 8x Eiszapfen 13/10mm straight
  • Tubing PVC 13/10 mm clear
2x120mm Thermaltake Pure A12 (pushing through radiator)
2x120mm Thermaltake Pure A12 (front intake)
Replaced 1000 ml Cape Kelvin Catcher with Mayhems eco X1 UV red
Bykski A-SP5700XT-X GPU block
Arctic MX-4 thermal paste on GPU/CPU

Fan curves for rad and intake

HeV3QZF.jpg.png

Setup

3K4PF6M.jpg.png

Temperatures

Prime95 blend test for 30 minutes. Minimum temperatures represent idle.

24CxQho.jpg.png

Furmark 1080p GPU stress test for 30 minutes. Again min temps = idle.
For comparison - on air cooling and ambient 26.2 (september) the hotspot max was 95 after only a minute.
DcvHe4E.jpg.png

Once again, thanks in advance.

Matt
 
Associate
OP
Joined
17 Feb 2021
Posts
29
The issue is two fold I think if I've read your post correctly

1. You are only using a single 240mm slim rad to cool both a gpu and cpu. That's simply not enough.
2. Both furmark and prime95 are high stress on temps so will push higher than typical operating temps but I think the above is fundamental.

This makes sense, watching the temperatures on the kind of games I play (GW2) it seems safe for now. Would you replace 240 with a 360 or add in a 240?

I'd agree with the above. You definitely need another rad in there for better temps or replace your current one with a bigger one. There's just not enough surface area to maximise heat dissipation.

But to answer whether the temperatures you're getting are appropriate, then yes they are given what you're using to cool your components.

That's a 4000d you're using right? So at most you can fit a 280mm rad up top and 360mm in the front. So there's room to improve.

It's your first time, you've done well. To be honest there's always things we look back on after our first time, either the components we wish we had bought instead, layout etc.

The next time round it'll be easier and you'll get even better results.

Thanks for your help. It is the 4000d - would you add in a 240 to front or swap 240 for 360?

I'm already looking forward to next time, and deciding all of my own fittings etc. Perhaps next year.


Not enough radiators.

Thank you!
 
Last edited:
Associate
OP
Joined
17 Feb 2021
Posts
29
Looks like your hardware is going to be pushing about 310-320W stock. A thick 360 would cover you but I'm not sure if you've got space in that case so maybe just chuck a mid size 360 in the front to give you a bit of headroom for upgrades.

Thanks for this. Can I ask what the general rule is with wattage and rad requirements? I was told 120mm per component as a rough guide, however, does this assume 60mm thickness?

Should be able to squeeze a 360 in the front if you turn your 240mm around. So go from the 5700XT to the rear conntecion on the 240mm, then out and down to the cpu, out to the front 360 and back to the pump.

That’s a great idea flipping the 240 - many thanks!
 
Associate
OP
Joined
17 Feb 2021
Posts
29
One bit of advise, I was also worried about my temps a bit on the CPU, turned out a really simple tweak in Ryzen master to fix the clock speed and set the score to 1.25 rather than auto which was running up to over 1.4v took about 15c off my max, if you haven't already, worth a look at that.

But ultimately it's down to what you want to achieve, for me it's aesthetics and silence but if you want the ultimate low temps for some reason then I'd start investing in more cooling power but that defeats the object as far as I'm concerned.

Thanks for the tip with Ryzen master! I’ll look into that this afternoon. I appreciate your alternate view point and candour, ultimately I do want a quieter system so adding in a rad will hopefully help. I’d also like some capacity to upgrade in the future I guess.

Extra rads will see the temps drop down lower and will also allow you to run lower fan speeds for a quieter system if thats what you are trying to achieve.

This sounds good to me. I’m not a huge fan of system noise!

There is no hyperbole at all the guy asked why his temps are higher than he anticipated being on water and the answer is he hasn't got sufficient rad surface area for heat dissipation. Nobody suggested he is running the risk of damaging his components. If he wants lower temps he needs more rad space. If he is fine with the temps then he can keep as is.

Thanks for the reassurance, it’s always a tetchy time with your first experience of something new. Which is why I appreciate everyone in this forum chipping in. I wasn’t happy with where Furmark was taking the GPU in my brief stress testing. I know AMD say “110°c hotspot is fine” but my temps weren’t stabilising at all until I raised fan speeds to unacceptable noise levels. Therefore, this indicates to me that I’m walking the line in terms of rad capacity.

Maybe hyperbole the wrong word and I get nobody suggested damage but considering this is his first time it's easy to get some excessive ideas in the head, folks come here for a bit of experience to be shared no?

I appreciate that Matt. :)

I did point out that IF he wanted LOWER TEMPS then yes more rad space is needed and at the end i DID say his system was fine as it is and is more than adequate. Don't take what i said out of context, it was only a suggestion :).

I got your drift, I’m aware I’m not damaging the parts but as said above, noise is important so it’s a key factor in adding a rad.

The old adage about 120mm per component +1 hasn't changed. So for cpu and gpu, 3x120mm would be safer but even that is such a generic statement.

to answer your original questions.
  • Are these temperatures appropriate for the setup? Probably about right considering the radiators and pump but very hard to be definitive sadly.
  • The pump can only run at 2600rpm and pushes 100l/h. Is this adequate? (Kit wasn't strictly designed for cooling two parts) 100lph will be the freeflow rate. With two blocks and a rad, that number will be significantly reduced. However, so long as there is flow, the pump flowrate is not a massive influence. Ultimately it's been shown numerous times to have relatively little effect on cooling so long as there is flow at a "reasonable" rate.
  • What is the solution to the fan RPM being linked to the CPU temperature? When I stress test the GPU I had to manually increase fan speed as the temperatures were only going up. Once I increased, they stabilised at around 77/78 hotspot. I've seen coolant temp being used - is this necessary/possible for my loop? This will depend on your motherboard a lot. Some BIOS allow setting fan profiles according to other heat sources than the cpu but having them work off the gpu is more tricky. I don't know that card; on some of my Geforce cards the fan header is reasonably accessible and adapters can be purchased that convert the pcb header to a standard fan connector allowing fan control through Afterburner or the like. You could therefore have one fan off the mobo cpu header controlled through the BIOS and one from the gpu controlled by the gpu.

Thanks for answering these! I’ll look in to the tips regarding the GPU.

As a brief update, I have one more day off work tomorrow before starting my new job on Monday - thus having less time to complete this project. ;) Therefore, this morning I purchased an EK 360mm rad (only slim will fit in front, as stated above by other users), EK CPU block (as stated above I despise the mounting hardware of the SP3 light) and a bunch of fittings to make it all fit better. Should arrive tomorrow - thanks OCUK! I’ve mixed a couple of brands but it’s all g1/4” and metals are the same - so from my research this is okay. Unfortunately, Alphacool are pretty much out of stock everywhere for the fittings/rad sizes I wanted.

I will update you all tomorrow afternoon. Once again, I’m astounded by the support offered to me in this forum and I appreciate all input.
 
Associate
OP
Joined
17 Feb 2021
Posts
29
Another thing to consider when going for lower temps is that (up to a point) GPU's will throttle clocks back as the chip heats up more. More thermal dissipation capacity means the card can run at higher power without reaching the bios level throttling thresholds. Once you've overcome that you will just run into the power limit of the GPU and beyond that you're trying to get closer to ambient or slowing fan speeds down.

As others have said, it depends what you're trying to achieve, but there's still benefit to be had by increasing your heat dissipation capacity beyond a 240.

If you want to look at it a different way, does your radiator look bigger than a traditional GPU and CPU cooling solution combined? There's only so much ability that a limited amount of rad surface area has to get rid of heat...

That's a good point. I definitely get some throttling, which I'm hoping to prove to myself by testing against historical cinebench scores.

Well - the first pic that came up when I googled my build with watercooling has a 360 :D Also, I like their drain valve. Can you think of a more optimal place for mine? Currently by the res, bottom right.

tt8wry369pj31.jpg
 
Associate
OP
Joined
17 Feb 2021
Posts
29
Just installed an EK SE 360 plus an EK quantum velocity CPU block. Note roughly similar ambient temps today. Overall I'm happy with the temperatures/noise levels and I also have capacity to upgrade components in the future. Performance wise, as expected I'm now seeing more from the GPU on average which is excellent. Appreciate all the help given in this thread :)

CPU temps in Prime95 blend test for one hour:
  • Fans at roughly the same audibly acceptable RPM as before
  • CPU on average 10C cooler
  • Max temps 5C cooler
  • Still need to try and reduce core voltage as suggested
3z59Tv3.jpg.png

GPU temps in Furmark 1080p:
  • Average GPU clock speed up by 17%, memory 14% (and thus naturally power to GPU increased)
  • GPU hot spot temp max down by 15C and very stable around 65C.
  • GPU temperature down by 10C average
oea7fRj.jpg.png



gbdK8hT.jpg.png
 
Back
Top Bottom