Add a 2nd radiator in custom loop?

George Hudson

George Hudson

George Hudson

Associate
Joined
8 Jan 2020
Posts
67
I am running a custom loop consisting of the EK kit P360. https://www.ekwb.com/shop/ek-kit-p360 which has a single 360x38mm radiator. I have a Ryzen 3900X and EVGA RTX 2080 Ti in the loop in the sequence of pump/reservoir to GPU to CPU then to radiator and back to pump/resevoir. After an hour of overclocked GPU stress testing I am seeing GPU temps of 48-51c and CPU temps of 45-50c so both around the same. I am thinking of adding a second radiator (the same as that which is fitted) between the GPU and CPU. Will this be a waste of money? What kind of temperature drops can I expect to see? Thanks.
 
You might see the GPU drop below 50. That may lead to a small bump in boost clocks. But really it will be fairly marginal given the info you've stated.

That said, what happens under full CPU and GPU load? The water will probably heat up slightly faster and you will likely see higher CPU temperatures at least.

If you're happy with current temperatures, I would consider noise. If you're happy with current noise levels you could consider increasing fan speeds under load instead of a radiator. There's a tradeoff of noise and heat, and adding a radiator will probably only shift that up or down a little.
 
I appreciate the quick response. I think I got it in my head after watching quite a few Youtube guides that I needed radiators that provided a surface area for 4x120mm fans for my setup. One fan for each component in the loop and an additional fan for each component if overclocking, so admittedly based on that another 360 radiator would be overkill (unless I bought another GPU:) but spent enough on PC for a while). Noise wise the system is nice and quiet and that is the principal reason i went for this setup. I will leave the system as is and monitor temps as I play games. Thanks again.
 
TBF the "120 per component and 120 extra" advice is quite dated, some graphics cards chuck out 300W now! Your 2080 Ti is a top end card so consumes lots of power.

I'm actually a little surprised a single 480mm is offering temperatures quite so low but if it works, it works. Until you get the itch to tinker with it :)
 
LuckyBenski said:
TBF the "120 per component and 120 extra" advice is quite dated, some graphics cards chuck out 300W now! Your 2080 Ti is a top end card so consumes lots of power.

I'm actually a little surprised a single 480mm is offering temperatures quite so low but if it works, it works. Until you get the itch to tinker with it :)
Click to expand...
Yeah, the temps looked low for a single 360mm?

What case are you using?

I'm running 2 x 480mm rads in mine, but I'm using an AMD 3600X and a Vega 64. Just in the last sections of finishing the build so haven't got definite temps yet, though on my previous TR 1920X I saw similar temps to yours above, though my GPU was lower :)
 
I did more 3DMark benchmarking last night and I am consistently seeing 47-48 deg C on both CPU and GPU and hitting results above 98% of other benchmarks on 3DMark website. My inlet temps are 26 deg C and inside case is generally 5-6 deg C above. I was running a GPU overclock of +800 on memory and +130 on GPU Clock. Spec as follows :- Cooler Master H500M case, Gigabyte X570 Aorus Master Motherboard, AMD Ryzen 9 3900X, G.Skill Trident Z RGB 32GB DDR4-3600 Memory, Samsung 970 Evo Plus 500 GB NVME SSD, Samsung 860 Evo 500GB SSD, EVGA XC Ultra GeForce RTX 2080 TI in EK Waterblock, EVGA SuperNOVA P2 850W PSU, EK-KIT P360 Water Cooling loop, 4 x Cooler Master MF120R ARGB 59CFM fans, LG 34GK950F-B monitor
tH3DxjGpJH6hkhk47
 
Last edited:
If you do end up adding a RAD it does not have to go between GPU and CPU. You can add it to pretty much anywhere suitable for tube routing.
Temps will come to an equilibrium.
 
To be honest your temps are fine more so given you have overclocked. As someone mentioned, may see a drop in temperature which will allow the gpu to notch up a bin or two in terms of frequency with boost 3.0, but your talking sub 1% performance gain really.

I agree with the others, your temps are actually very nice for just a single 360mm, can see you have the H500M which I suppose helps with its large front mesh in getting air in.
 
LuckyBenski said:
You might see the GPU drop below 50. That may lead to a small bump in boost clocks. But really it will be fairly marginal given the info you've stated.

That said, what happens under full CPU and GPU load? The water will probably heat up slightly faster and you will likely see higher CPU temperatures at least.

If you're happy with current temperatures, I would consider noise. If you're happy with current noise levels you could consider increasing fan speeds under load instead of a radiator. There's a tradeoff of noise and heat, and adding a radiator will probably only shift that up or down a little.
Click to expand...

If you add 3 more rads you might see it at 35c... Do it do it. Jokes.

More rads less fan speed if you can, quiet life.
 
Avathar77 said:
If you do end up adding a RAD it does not have to go between GPU and CPU. You can add it to pretty much anywhere suitable for tube routing.
Temps will come to an equilibrium.
Click to expand...

This, so much this, it's a huge misconception in watercooling from what I've read after starting to get back into it recently. The loop order does NOT matter. Liquid is unlike air, it reaches an equilibrium, and the pump continuously moves that liquid around. Hence why it's always best to have a temp sensor in the loop so you know what your liquid temps are. CPU and GPU temps will always fluctuate, but you can't effectively know what form of cooling you have and how effective without being able to record the liquid temperature and comparing it with the ambient temp to discover the delta.

People create insane custom loops to simply have the radiators before the GPU, then to the CPU, back into the res. This is messy, adds head height for the pump to battle. Where as you should always take the shortest route, the less work the pump has to do, the faster the flow rate, the more effective it is at keeping the liquid temps down to a decent delta (with the exception of the heat dispersion from your radiators not being counted as a factor in this).
 
To be honest the only reason i can see for adding more rads is for quieter performance. Like someone has pointed out, the more fans and rads you have the lower the rpm you could technically run them at to achieve the same level of performance. In retrospect, adding more rads will just give diminishing returns. For instance, i ran 2 x 240's which gave a decent temp of around 45-46c on the gpu and 65c average on the cpu, i then swapped the 240 for a 360 and saw a further drop of 2c on the gpu and 5c on the cpu. Finally i added another 360 to the bottom which gave me a further but very small drop to 41c max on the gpu and 55-58c on the cpu.

That's a total of 960 worth of rads with some in push pull config to get just a couple of more degrees in temp drops.
 
Can you recommend suitable in-line water temp sensors? I have two temp sensor headers on mobo which I currently use for ambient and case internal air temps but would use those headers for water temps instead. Probably monitor pre and post radiator water temps.
 
I fitted the temperature sensor recommended by LuckyBenski a couple of weeks ago. No problems with fitting it immediately on the exit of the radiator. Readings appear to be fine. I have been logging a lot of data while playing around with the fan curves and playing games/running Heaven benchmark. I am using the temperature sensor as the controlling sensor for the radiator fans speed. The most I have seen the ambient/water Delta-T at is 13 degrees C but I have now got it to a pretty constant 10 degrees C. My curiosity has got the better of me and I have just ordered a 2nd 360mm radiator although it is thinner than my front mounted one being only 28mm thick due to the amount of room at the top of the case. I have also ordered a BarrowCH OLED display flow meter which can be hooked up to a fan header for monitoring. I am not expecting to get Delta-T any lower. What I am hoping to achieve is lowering the fan speeds even though the system is pretty quiet as it is. At present I am running pump at 30% constant, inlet fans on radiator between 40% and 50% depending on Delta-T and the 4 exhaust fans at 40% constant. The two constant fan speed values are while I have been determining optimal radiator fan settings. I will post this thread with my test results regarding 2nd radiator.
 
If you have the space, it's worth it IMO. Particularly as you can run fans and pump at lower speed with more radiator area.
 
Just got my system back together with second radiator in and flow meter. I was shocked to learn that running my water pump at 30% was circulating less than 0.3 litres per min. Quick test to establish all flow rates from 100% down to 40% gave a range of 6.8 to 1.3 liters, so I have settled for now running 50% which equates to 2.55 litres min and its maintaining a Delta-T of 9 degrees C. Lots of testing to do. Keep you posted.
 
Update.


I am using the Cooler Master H500M case with a EK-XRES 140 Revo D5 PWM combo pump and reservoir. Also fitted is a Aqua Computer water temperature sensor and Barrow flow meter. An ambient temperature sensor is connected to the motherboard and located just to the side of the front 2x200mm fans.


I am running a 360mm radiator at 38mm thick positioned at the front of the case, with 3x120mm Vardar fans inside the case pulling the air through the radiator and 2x200mm fans in front of the radiator pushing air through it. Possibly not the best configuration. I may try removing the 2x200mm fans in case they are reducing the air flow through the radiator.


In the top of the case I have a 360mm radiator at 28mm thick with 3x120mm Cooler Master MasterFan SF120R ARGB in push configuration, expelling air from inside the case through the radiator. I also have 1 of these fans on the rear of the case pushing internal case air out. So 9 fans in total.

The custom loop is setup from reservoir and pump -> flow meter -> GPU -> top radiator -> CPU -> front radiator -> temperature sensor -> back to reservoir and pump.


I am running my EVGA RTX 2080 Ti with a stable overclock of +800 on memory and +125 on the GPU seeing 2070Mhz in the EVGA Precision X1 software.


I have data logged over 30 hours of testing.


The best Delta-T I achieved was during a 200 minute run of Heaven benchmark with the pump and all fans at 60%. The 2x200mm fans always run at full speed. This produced an average Delta-T of 6 degrees C with ambient at 26 and coolant at 32 degC. The fans were audible but in no way loud. The mobo temp sensor averaged 39 degC with the CPU average at 42 degC ( and a peak of 51), the GPU average at 37 degC (with a peak of 48).


The fan and pump speeds I am currently using, after a total of 390 minutes of testing with Heaven benchmark are 40% pump (moving 1.27 litres/min), 50% front fans, 40% top and rear fans. Over 4 runs, this produced a repeatable average Delta-T of 8 degrees C over ambient. The fans are quiet. The mobo temp sensor averaged around 40 degC with the CPU around 46 degC ( and a peak of 77), the GPU 45 degC (with a peak of 52).


I used HWinfo64 to record these sensors.

My pump has developed a fault in that the sense signal to the mobo header is not producing a signal. I can still control the speed of the pump on the PWM speed control line but not getting the RPM data from the pump. I have contacted EK and their customer support is brilliant. Communication has been very rapid over 24 hours since reporting the problem via emails. A replacement pump is on its way and collection of faulty unit at a time suitable for me when I have swapped them over. I have no regrets in purchasing the EK kit and very happy with the results I am achieving.
 
Last edited:
In my experience rad space is more about fan speeds than it is anything else.

More rad space will see lower temps, maybe 5-10C odd the GPU and a few off the CPU.

The real beauty is you'll be able to drop your fans speeds; I've got a 360 and 240, and I can't hear my system. Temps are a bit better than yours, but not by much. If I ramped up my fan speeds under load, that's where I'd see real improvement.
 
The noisiest fans in my system are the 2x200mm sitting in front of the front radiator. These are connected directly to the 12v and speed can not be adjusted. I am going to remove them and see what effect it has on temps. I like the argb lighting on them so may replace them with a set of the speed controllable 3x120mm I have in the top. Fun times.
 
You must log in or register to reply here.
Back
Top Bottom