Watercooling questions and thoughts (H220, radiators and additives)

[Timmyfox]

Okay, so I'm a bit of a watercooling noob. I've got a H220 which I've recently expanded with my 7970 using an EK copper-base block. I'm using distilled water mixed with a little bit of anti-corrosive fluid, I don't have any biocide in the loop as of yet but it's just a matter or ordering it.

 

Compared to cooling just the CPU alone I did notice some differences in temperatures:

 

• For idle temps, the GPU and CPU are idling at around the same temperatures (+/- 2C of each other), around 40-45C, this is roughly the same temps as the GPU was having with the stock cooler, although it is more consistent (the stock cooler slowly gained some temps over time and went up to about 50C). On the CPU side however, I feel like the temps are up by about 3-5C
• As for load temps, when running Prime95, the CPU ramps up to around 50-55C initially, this is roughly the same temps as I saw when I was cooling just the CPU, the GPU temps seem remain pretty unaffected as well. When putting load on the GPU however I see a different behavior; The CPU temps increase alongside the GPU temps (which I feel is especially strange as the GPU is connected after the CPU/pump in the loop).
  • The load temps for the GPU goes up to 55-60C in the most demanding games (with stock cooler it was anywhere between 70 and 80-85C) and the CPU temps increase with it. The CPU thus bumps up to as much as 5-10C hotter when playing a game than when running Prime95 and leaving the GPU idle. This up to 15-20C hotter CPU temps when in a game than I remember it getting when I was cooling just the CPU.

Note that these are all stock speeds so far.

 

I'm considering adding a second radiator to the loop, when I have a game open and put my hand over the top of my case, I can feel that the air exhausted from the H220 rad is quite very warm. I'm considering the EK CoolStream RAD XT 240. I'm using Noctua NF-F12's in pull for the H220 rad and will most likely use some F12's for the second rad as well.

 

Another question I have is regarding corrosion and algae buildup. As I've mentioned, I do not have any biocide in the loop yet as I have none to mix in. I've heard a lot on various forums about the "PT nuke" although I can't seem to find it anywhere where I live. I did however find a reasonably inexpensive one called "Mayhems biocide". Does anyone have experience with it and will it work fine? I've heard a silver coil will work just fine as well. However one of those seems to be fairly expensive compared to the biocide, I found only a single retailer that sells them as well. As for corrosion, I've got, as said, a small bottle of a no-name anti-corrosive fluid. I've read though on some forums that corrosion should be a non-issue in most cases unless you run a mixed-metal loop. The H220 as far as I know uses copper for the CPU block and copper/brass for the radiator and the GPU block is copper as well. What are you guys' standpoints on this? is the anti-corro fluid still a good safety precaution or would I be fine without it?

[airdeano]

Okay, so I'm a bit of a watercooling noob. I've got a H220 which I've recently expanded with my 7970 using an EK copper-base block. I'm using distilled water mixed with a little bit of anti-corrosive fluid, I don't have any biocide in the loop as of yet but it's just a matter or ordering it.

 

Compared to cooling just the CPU alone I did notice some differences in temperatures:

 

• For idle temps, the GPU and CPU are idling at around the same temperatures (+/- 2C of each other), around 40-45C, this is roughly the same temps as the GPU was having with the stock cooler, although it is more consistent (the stock cooler slowly gained some temps over time and went up to about 50C). On the CPU side however, I feel like the temps are up by about 3-5C
• As for load temps, when running Prime95, the CPU ramps up to around 50-55C initially, this is roughly the same temps as I saw when I was cooling just the CPU, the GPU temps seem remain pretty unaffected as well. When putting load on the GPU however I see a different behavior; The CPU temps increase alongside the GPU temps (which I feel is especially strange as the GPU is connected after the CPU/pump in the loop).
  • The load temps for the GPU goes up to 55-60C in the most demanding games (with stock cooler it was anywhere between 70 and 80-85C) and the CPU temps increase with it. The CPU thus bumps up to as much as 5-10C hotter when playing a game than when running Prime95 and leaving the GPU idle. This up to 15-20C hotter CPU temps when in a game than I remember it getting when I was cooling just the CPU.

Note that these are all stock speeds so far.

 

I'm considering adding a second radiator to the loop, when I have a game open and put my hand over the top of my case, I can feel that the air exhausted from the H220 rad is quite very warm. I'm considering the EK CoolStream RAD XT 240. I'm using Noctua NF-F12's in pull for the H220 rad and will most likely use some F12's for the second rad as well.

 

Another question I have is regarding corrosion and algae buildup. As I've mentioned, I do not have any biocide in the loop yet as I have none to mix in. I've heard a lot on various forums about the "PT nuke" although I can't seem to find it anywhere where I live. I did however find a reasonably inexpensive one called "Mayhems biocide". Does anyone have experience with it and will it work fine? I've heard a silver coil will work just fine as well. However one of those seems to be fairly expensive compared to the biocide, I found only a single retailer that sells them as well. As for corrosion, I've got, as said, a small bottle of a no-name anti-corrosive fluid. I've read though on some forums that corrosion should be a non-issue in most cases unless you run a mixed-metal loop. The H220 as far as I know uses copper for the CPU block and copper/brass for the radiator and the GPU block is copper as well. What are you guys' standpoints on this? is the anti-corro fluid still a good safety precaution or would I be fine without it?

 

thank you for posting your findings on a straight-forward GPU addition. seems

about par as the 240 radiator just isn't large enough to support CPU/GPU. nothing

was noted if the CPU/GPU were overclocked or not.. or fan speeds.

 

basically a 360x 60 radiator or 2x 240x 60 will allow great cooling performance along

with low fan speed noise as the rpm doesn't have to be WOT.

 

as to coolant, yes, they are straight copper components so any off-the-shelf PC

coolant (might already have biocide already included in coolant)or distilled water

will work, just fine. for a biocide agent a silver kill-coil or additive will be just fine.

corrosive agents are mainly for mixed metal systems other than copper

(aluminum mainly).

 

airdeano

[Timmyfox]

Thanks for your reply. I wouldn't say it's all bad as the GPU temps certainly did improve. Although it did come with some slight negatives for the CPU. Hopefully a second rad should improve this.

As for clock speeds, everything is stock from what it was out of the box; CPU is an i7 2600k and GPU is a 7970 GHz edition.

Fan speeds, both the NF-F12's are running at about 700 rpm when idle and 1000rpm under load.

[Timmyfox]

Anyone else have any comments or advise for me?

[Ghost]

Anyone else have any comments or advise for me?

If you asked before expanding the H220 I would have recommended not doing it. Now that you have your only choice is to add a radiator.

 

In terms of other things you can add, you may want to add a reservoir to your loop. There is a flaw in the H220 that causes the pump to become loud because air remains in the loop. Adding a reservoir can fix it. Even if you don't have it now you may get it in the future and a reservoir isn't the most expensive thing in the world. Plus it can look good in a loop too. In this case the reservoir doesn't have to be above the pump as the radiator reservoir will still be feeding the pump directly. The second reservoir will only be used for bleeding. (You will have to fill your loop through the radiator reservoir though.)

[Timmyfox]

If you asked before expanding the H220 I would have recommended not doing it. Now that you have your only choice is to add a radiator.

 

In terms of other things you can add, you may want to add a reservoir to your loop. There is a flaw in the H220 that causes the pump to become loud because air remains in the loop. Adding a reservoir can fix it. Even if you don't have it now you may get it in the future and a reservoir isn't the most expensive thing in the world. Plus it can look good in a loop too. In this case the reservoir doesn't have to be above the pump as the radiator reservoir will still be feeding the pump directly. The second reservoir will only be used for bleeding. (You will have to fill your loop through the radiator reservoir though.)

How come you say that? My plan was to get into watercooling in a reasonably inexpensive and more safe (in terms of not having to pay as much attention to every single component) way. I had intended to see how this would work out and then go from there in a way to gain some experience without diving right into it.

 

I might look into adding another res, thanks. The cheapest in stock that seems reasonable would be the Alphacool Cape Corp PRO 10LT which is priced at around $25 followed by the Swiftech MCRES-Micro Rev.2 which is about $5 more. Other ones around that price range are the Phobya 5.25 bay res and Phobya Balancer 250 nickel.

 

Also as for radiators, any brands/models to avoid? I'm considering an EK CoolStream RAD XT 240 as said in the OP.

[Ghost]

How come you say that? My plan was to get into watercooling in a reasonably inexpensive and more safe (in terms of not having to pay as much attention to every single component) way. I had intended to see how this would work out and then go from there in a way to gain some experience without diving right into it.

 

I might look into adding another res, thanks. The cheapest in stock that seems reasonable would be the Alphacool Cape Corp PRO 10LT which is priced at around $25 followed by the Swiftech MCRES-Micro Rev.2 which is about $5 more. Other ones around that price range are the Phobya 5.25 bay res and Phobya Balancer 250 nickel.

 

Also as for radiators, any brands/models to avoid? I'm considering an EK CoolStream RAD XT 240 as said in the OP.

• The pump in the H220 is weak compared to custom loop pumps and impossible to upgrade unless you change the whole CPU block. Also as it is in the CPU block and the res have to be above it this makes for somewhat awkward tubing runs if you ever want to get rid of the rad/res combo at the top.
• I would get a res that looks good. That would be one of the points of having a res. If you cheap out on the res then you may make your build look worse. I would look for a standard tube res.
• Avoid aluminum radiators as they tend to perform worse than copper radiators. Don't both with full copper rad though. Look for Cu/Al rads. Otherwise bigger is better.
• Don't forget you ware going to have to invest into 6 fittings at this rate. The cost of fittings can ramp up drastically. Look at EK fittings for reliable and cheap fittings.

[R4STABAN]

(which I feel is especially strange as the GPU is connected after the CPU/pump in the loop).

 

Just going to jump on this

 

I assume your setup is cpu-gpu-radres-cpu? It is a common misconception that components after a rad in loop order will be significantly cooler than if they were before the rad.

 

The coolant moves too fast through the radiator to be cooled substantially. After a certain period of time of having a heat load applied the coolant will reach a near equilibrium temperature throughout the loop (small differences).

 

The radiator acts to dissipate heat from the entire loop.

 

So if your ambient temp in the room is the same, heat load from blocks is the same, but you turn up your fans significantly on your rad, the equilibrium temperature will drop throughout the loop.

[Timmyfox]

Just going to jump on this

 

I assume your setup is cpu-gpu-radres-cpu? It is a common misconception that components after a rad in loop order will be significantly cooler than if they were before the rad.

 

The coolant moves too fast through the radiator to be cooled substantially. After a certain period of time of having a heat load applied the coolant will reach a near equilibrium temperature throughout the loop (small differences).

 

The radiator acts to dissipate heat from the entire loop.

I'm aware of this, yes as Linus explained it quite well in the watercooling guide.

What I mean though is that I find it strange that putting load on the CPU doesn't affect the temps of the GPU (CPU running 10C hotter than GPU in prime95) whereas putting load on the GPU causes the CPU temps to rise with it (they both run at the same temperatures). I suppose this could be because the GPU outputs a lot more heat compared to the CPU due to a much higher TDP but it's a bit interesting since physically the liquid does in fact pass the radiator before reaching the CPU (which, as you note, indeed shouldn't matter).

[Ghost]

I'm aware of this, yes as Linus explained it quite well in the watercooling guide.

What I mean though is that I find it strange that putting load on the CPU doesn't affect the temps of the GPU (CPU running 10C hotter than GPU in prime95) whereas putting load on the GPU causes the CPU temps to rise with it (they both run at the same temperatures). I suppose this could be because the GPU outputs a lot more heat compared to the CPU due to a much higher TDP but it's a bit interesting since physically the liquid does in fact pass the radiator before reaching the CPU (which, as you note, indeed shouldn't matter).

You forget that your CPU is also used in games. ;)

 

Try stressing your GPU only (Only way i can thing of is by decreasing CPU affinity to one core and running a GPU burn in test, even though this will still use one CPU core) and I doubt you will see the same increase.

[Timmyfox]

You forget that your CPU is also used in games. ;)

 

Try stressing your GPU ( I don't actually think you can) only and I doubt you will see the same increase.

I've monitored the CPU load, even in game menus with minimal load I get these increases. I also ran Furmark at one point and still got a pretty significant increase in CPU temps with only 2% load. The CPU temps does match or even go over that of what I get when running Prime95 when in a game.

 

I do know what you mean though, the CPU temps where indeed at their lowest (compared to the game menus) when running Furmark which intends to put load only on the GPU as it seems. They did still rise a reasonable amount though, as said in my previous post, probably due to the liquid heating up a bit for obvious reasons.