Overclock.net banner

1 - 18 of 18 Posts

·
Registered
Joined
·
31 Posts
Discussion Starter #1
Hello,

I have custom loop composed composed of a 360 rad and a 240 rad, both with their corresponding number of fans (3 and 2).
The 360 rad is on the top of the case, exhausting air.
The 240 rad is on the bottom-front of the case, taking air in
I don't have any other fans right now.

I am cooling an i7 4770k running at 4.5GHz with 1.36v and a 980Ti on 1.27v.
Thing is, in these summer times, the CPU and GPU can go as high as 80°C, while the motherboard is reporting 42/45°C inside the case.
I think that's really high for a custom loop, and so I thinking about changing the way the air is flowing.

I think I'll be adding a 120mm exhaust fan on the rear of the case, and put all other fans as intake, to improve the radiator efficiency, and create a positive air pressure.
I still have room for 1x120 on the bottom, and 1x140 on the side, but I'm not sure it's would be of any use...

What do you think of this ?
 

·
Registered
Joined
·
946 Posts
IMO you are loosing RAD efficiency with the 360 as exhaust. I run all of my RADs as intake. The cooler the air entering the RAD the greater the efficiency. 80 degrees sounds unusually high for your setup. You could try taking the sides off of the case and see if the temp goes down. That could give you an idea that your air flow could be suspect. Do you have your fans and pump running at max speed?
 

·
Registered
Joined
·
2,302 Posts
Quote:
Originally Posted by apw63 View Post

IMO you are loosing RAD efficiency with the 360 as exhaust. I run all of my RADs as intake. The cooler the air entering the RAD the greater the efficiency. 80 degrees sounds unusually high for your setup. You could try taking the sides off of the case and see if the temp goes down. That could give you an idea that your air flow could be suspect. Do you have your fans and pump running at max speed?
Compared to water, air is a really lousy conductor for heat. Unless someone is running a high end GPU that's air cooled, the only heat not being dumped into the water in a loop is heat from MOSFETs, RAM, and whatever else is in the box. The ambient inside the box should only be 1-2 degrees above the ambient outside the box. This will equate to a fraction of a degree less for the loop itself if switching from exhaust to intake. As long as flow rate is decent in the loop, it should be difficult to measure the difference.

Switching from exhaust to intake does have the effect of dumping the heat a loop removes back into the case, and will cause every other component in the machine to run much hotter as it has the effect of raising ambient temps 5-10 degrees inside of the vase, which means all of the air cooled components run at least 5-10 degrees hotter, maybe more since air is a lousy conductor and you've now decreased the approach.

Unless someone is running a tiny little CLC with air cooled GPUs, total system efficiency should be much better with radiators as exhaust or something is probably not configured correctly.
 

·
Linux Noob
Joined
·
991 Posts
Quote:
Originally Posted by Oli3dfx View Post

Hello,

I have custom loop composed composed of a 360 rad and a 240 rad, both with their corresponding number of fans (3 and 2).
The 360 rad is on the top of the case, exhausting air.
The 240 rad is on the bottom-front of the case, taking air in
I don't have any other fans right now.

I am cooling an i7 4770k running at 4.5GHz with 1.36v and a 980Ti on 1.27v.
Thing is, in these summer times, the CPU and GPU can go as high as 80°C, while the motherboard is reporting 42/45°C inside the case.
I think that's really high for a custom loop, and so I thinking about changing the way the air is flowing.

I think I'll be adding a 120mm exhaust fan on the rear of the case, and put all other fans as intake, to improve the radiator efficiency, and create a positive air pressure.
I still have room for 1x120 on the bottom, and 1x140 on the side, but I'm not sure it's would be of any use...

What do you think of this ?
The quick solution is to leave all as-is and make your pump and CPU fan curve more aggressive. That is, have them spin at a higher rate @ CPU idle and ramp up at lower temps. You can even have some of the fans go to 100% when your CPU reaches 45C, if needed. You'll have more noise, but also more cooling. Low CPU / high GPU situations will still be your hottest.

Consider this: Additional non-rad intake fans will work against your front intake rad fans and any additional non-rad exhaust fans will work against your top rad exhaust fans. Therefore, additional fans are best added to the rads.

Unless you have incredible amounts of rad space, or an open-air case, it's best to equalize the air flow through rad-based intake fans and rad-based exhaust fans when possible. Push is more efficient than pull and push/pull is most efficient, except on ultra-thin rads where the gain is minimal (unless your fans are uncommonly weak static pressure-wise).

What model fans are these? Pretty fans often don't move much air through radiators.
Photos would help.
 

·
Registered
Joined
·
946 Posts
After thinking about this and wondering if this is a newly built system? My gut feeling is the TIM on the CPU and GPU blocks could use a remount. Do you have a way of seeing your coolant temp? What is your ambient temp?
 

·
Registered
Joined
·
31 Posts
Discussion Starter #6
Thanks for your replies !

So, unordered :
- The loop has been mounted last november, and if it's my first loop, it's clearly not my first build. I used the TIM given with EK blocks. Anyway, could be that, but I'd be surprised. BTW, during the whole winter, the same settings were going in the low 60s°C on the GPU on more than 100 stable hours on TW3 with 1.26v on the 980 Ti running at 1493/2000, for somewhere around 20°C ambiant. Right now, above 75°C, the card cannot sustain that clock (you know, Maxwell and the heat...
wink.gif
). Don't remember precisely the CPU temps though, but I don't recall it going higher than the low 70s°C stability testing using cinebench. Should have noted back then. Still, above 80s on the CPU is really rare, it's the GPU that's worrying me the most.
- Ambient temp can go really really high here in summer. The 80°C were happening when ambiant was around 30°C. Whith the computer heating everything up, I think it was even higher (I was basically sweating in front of my screen). Old building, last floor, just under the roof. Right now, as I'm writing this comment, my CPU is idling quietly at 30°C (Haswell no delided, so 26-32 from one core to the other) for somewhere around 22/25°C ambient (wheather is cooler this days).
- The fans are EK-Vardar F4-120ER. I read reviews back then, and they should be fairly decent afaik.
- I quickly tried with the case opened, and it made a difference, around -5°C on GPU both idle and load (quickly ran Valley for about 5 minutes, so nothing scientifically proved right now).
- Front rad fans are pulling, but top fans are pushing the from the case through te rad to the exit. I can't do the other way, I don't have enough space on the top of the case (a Phantom 530 BTW), photo below.
- I don't have a coolant temp sensor. That's a mistake, and I'll amend it at the next drain.

Maybe the high ambiant is somehow defeating the dissipation. That's why, with probably around 45°C inside the case (at least according to MB sensors), I was looking for solution to put that air out as quickly as possible...So I'm really wondering how to exhaust that hot air in the more effective way. I already ordered a fan (slim 15mm noctua 120 fan) to put on the top rear, where the fittings are (had a 140 here when still air cooling, but now the fittings won't let me put a 25mm thick fan here). I understand the argument about pulling hot air inside the case if everything is taking in, but the air inside the case should already be about 10 to 15°C above ambiant. Wouldn't that be defeating the radiator cooling as well ? I'm wondering if the "less hot" ambiant air, even after going through the rads, would really be higher than that, especially if it is taken out right away by an exhaust fan.

And here the photo of the setup. The water is going out of the pump to the front rad, going up to the GPU, MB, 2d Rad, CPU and back to the pump.

 

·
Registered
Joined
·
2,290 Posts
Hi there

Yours temperatures are bit high for my liking and GPU temperatures should be in lot lower values

Have look I'm running 5960x OC to 4.5Ghz and my temperatures are in OCCT or in Aida64 are in 72°C as max in 24°C ambient

Regarding the GPU temperatures,have running 3*GPUs and with following OC on the cards GTX1080Ti with 2139MHz,GTX1080 with 2164MHz and other GTX1080 with 2100MHz and temperatures are in 36-38°c during the gaming on Ti, in mining or folding [email protected] temps on all GPUs are in 40-42°C as max that's with higher ambient temperature, during the rendering temperatures are in 36-38°C

Have run Titan X SC with EKWB waterblock and my temperatures never been higher than 34-36°C and that's with 1469MHz OC on the my Titan X SC

Did you check waterblocks if they're correctly mounted?

Other my suggestions or recommendations is,I would add extra fans on radiators, if you are running XE radiators then these radiators love push pull configuration and I would add extra fans for push pull, try to use same fans

I would personally go with different case as starting place as this case is not the best case for water cooling

Some cases have very restrictive top panel or front panel which will only cause lots issues

In my case usually ambient and inside temperature is within 0.5-2°C each other or if ambient is 21°C then inside the case my temperature is around 22-23°C on idle and during the load temperature inside is around 2-4°C as max and running similar setup to yours, top 360mm 60mm thick radiator and 240mm 60mm thick radiator on bottom, both radiators are running push pull fans, top fans are Phanteks PH-F120MP and bottom are EK Vardar F3 1850RPM

Add extra exhaust fans at top and front radiator and turn exhaust fan on case as intake as well this should bring colder air inside the case

Not sure how restrictive is mosfet or mobo waterblock but hard to say there

At what speed are you running fans and pump?

If opening case window or side panel lower temperatures then you need run faster fans to bring colder air or exhaust hot air

Hope this helps

Thanks, Jura
 

·
Linux Noob
Joined
·
991 Posts
So your top fans are pull exhaust and your front fans are push intake.
Vardar fans are usually good.
Best scenario is that in which opening the case has little or no effect on normal operating temps.

Get a small air compressor or 8-pack of compressed air and blow out the case, rads and screens. Clean the fans.
Check for air bubbles in both rads.

Is your case front solid or screen? A screen/grilled case front can sometimes fit fans between the case and outer cover.
 

·
Linux Noob
Joined
·
991 Posts
Jura has a great point. I didn't realize those rads were extra-thick. Double rads need fans on both sides or at least higher single fan speeds to make proper use of their extra cooling capacity.

Here's a review comparing single fan performance vs push/pull. The Black Ice Nemesis GTX in the review is also double thickness.
 

·
Registered
Joined
·
1,166 Posts
There are lots of threads regarding air flow in multi rad systems. It can be something of a controversial subject with some insisting its best to always have cool air to the rads and others insisting intaking rads heats the components in the case.

The only testing I have seen supports intaking rads.

http://www.overclockers.com/forums/showthread.php/700744-AirFlow-direction-for-rads?s=79be81e91b35fbc19e34426a42ab2ae6&p=7111920#post7111920
http://www.overclock.net/t/1430077/corsair-750d-owners-club/400#post_21061809

The results seem to contradict the idea that radiators heat the internal components in any more than negligible ways and actually cools better in some cases. Although it seems logical that rads exhausting warmer air into the case could be a bad idea I think thats over simplifying the reality. A couple of rads both intaking air creates a lot of air flow. That higher air flow has a cooling effect even if the air is a bit warmer than ambient.

If a system has one front rad intaking and one top rad exhausting then the MB section is sitting in the exhaust of the front rad. The air is warmed slightly more by the air cooled components and exits through the top rad. The MB section gets two or three fans volume of air flow to cool it.
With both rads intaking the MB section is still in the exhaust path of the rads, but, the coolant temp is lower so the exhaust air temp is lower. The MB section gets 4 to 6 fans volume of air flow.
More air flow volume at a lower temp.

So its maybe not quite so straight forward. Cooling is more complicated than simple temperatures. The flow rate of coolant mediums plays a large role too.

Other fan placements can of course further complicate the issue. Naturally the ideal case is to have rads with dedicated air flow sections seperated from the rest of the system like pedestals and dual chambered cases like the corsair air series
 

·
Registered
Joined
·
3,467 Posts
Before assuming that this problem is caused by airflow through the radiator you need to know your water temperatures. That's the only way to gauge how efficiently your radiators are dispersing the heat load.

Ambient temp
Idle water temp
Load water temp

These will help in figuring out what's going on much better than just assuming you need to add a fan, which you probably don't.
 

·
Banned
Joined
·
3,726 Posts
Quote:
Originally Posted by Oli3dfx View Post

Thanks for your replies !

So, unordered :
- The loop has been mounted last november, and if it's my first loop, it's clearly not my first build. I used the TIM given with EK blocks. Anyway, could be that, but I'd be surprised. BTW, during the whole winter, the same settings were going in the low 60s°C on the GPU on more than 100 stable hours on TW3 with 1.26v on the 980 Ti running at 1493/2000, for somewhere around 20°C ambiant. Right now, above 75°C, the card cannot sustain that clock (you know, Maxwell and the heat...
wink.gif
). Don't remember precisely the CPU temps though, but I don't recall it going higher than the low 70s°C stability testing using cinebench. Should have noted back then. Still, above 80s on the CPU is really rare, it's the GPU that's worrying me the most.
- Ambient temp can go really really high here in summer. The 80°C were happening when ambiant was around 30°C. Whith the computer heating everything up, I think it was even higher (I was basically sweating in front of my screen). Old building, last floor, just under the roof. Right now, as I'm writing this comment, my CPU is idling quietly at 30°C (Haswell no delided, so 26-32 from one core to the other) for somewhere around 22/25°C ambient (wheather is cooler this days).
- The fans are EK-Vardar F4-120ER. I read reviews back then, and they should be fairly decent afaik.
- I quickly tried with the case opened, and it made a difference, around -5°C on GPU both idle and load (quickly ran Valley for about 5 minutes, so nothing scientifically proved right now).
- Front rad fans are pulling, but top fans are pushing the from the case through te rad to the exit. I can't do the other way, I don't have enough space on the top of the case (a Phantom 530 BTW), photo below.
- I don't have a coolant temp sensor. That's a mistake, and I'll amend it at the next drain.

Maybe the high ambiant is somehow defeating the dissipation. That's why, with probably around 45°C inside the case (at least according to MB sensors), I was looking for solution to put that air out as quickly as possible...So I'm really wondering how to exhaust that hot air in the more effective way. I already ordered a fan (slim 15mm noctua 120 fan) to put on the top rear, where the fittings are (had a 140 here when still air cooling, but now the fittings won't let me put a 25mm thick fan here). I understand the argument about pulling hot air inside the case if everything is taking in, but the air inside the case should already be about 10 to 15°C above ambiant. Wouldn't that be defeating the radiator cooling as well ? I'm wondering if the "less hot" ambiant air, even after going through the rads, would really be higher than that, especially if it is taken out right away by an exhaust fan.

And here the photo of the setup. The water is going out of the pump to the front rad, going up to the GPU, MB, 2d Rad, CPU and back to the pump.

here is the issue that I got from reading your two posts.

it was running cooler, now it's not.

this really jumped out to me, as that is a indication that something is going on with the loop itself.

so what could have happened to alter the functionality of the loop?

one, is that the pump has slowed down thus lowering loop efficiency. if this is the issue replace the pump

two, the tubing is breaking down on the inside, this is clogging up the water blocks, if this is the issue replace the tubing and clean out the water blocks.

three, the TIM has broken down, and needs to be replaced.

I cant think of any other reason for lowered performance after start up. hope this helps.
thumb.gif
 

·
Registered
Joined
·
31 Posts
Discussion Starter #15
Quote:
Originally Posted by RX7-2nr View Post

Before assuming that this problem is caused by airflow through the radiator you need to know your water temperatures. That's the only way to gauge how efficiently your radiators are dispersing the heat load.

Ambient temp
Idle water temp
Load water temp

These will help in figuring out what's going on much better than just assuming you need to add a fan, which you probably don't.
You're right. I have to add water temperature sensor, as well as flow sensor I guess. It's on my todo list on the next drain.
Without data, I can only guess and try. Anyway, I bought that fan before it occurs to me to discuss it here, so... I'll be doing my little tests.

But still, the ambient temp is way cooler now, as is the weather, and I lost 5 to 10 degrees on the GPU temps, maxing out at 70-72°, with the both CPU and GPU idling around 27°C.
That is really looking like some kind of threshold in ambient, that's why I was wondering if the setup was wrong, or if taking air in could improve the efficiency.
I also changed the fan response so that they go up faster, we'll see.

Anyway, it's quite clear the case is not ideal, thing is I was miles away of building a custom loop when I bought it.
The top and the front are "semi-open", mid plastic mid ventilation grid. And I don't have a lot of space, for disks especially. I had to put them in 5.25" trails just below the top radiator.
I plan to have a Case Labs case for the next build (Mercury S8 probably). Maybe I could get it right now, the current build could use the move...
smile.gif


But still, I may have to check out the wb on the GPU, cause I'm miles and miles away from 36° at load. I put it on quite carefully, but who knows. Maybe I screwed it up. :s
I have PE rads, the "mid" offer by EK. The fan speed and pump are both connected to the CPU PWM.

And lastly (and I think I've answered everything), all the fans already are between the case and the cover. No more space sadly. :s
I did a complete radiator clean up some weeks ago.

I think I'll try taking air in anyway when I get the new fan putting it at exhaust, and see how it reacts in stress testing.
There is also room to place a side fan just at the same position as the front rad is taking air in. Could exhaust it right away to.

Thanks for links on some tests about taking in or exhausting, I'll read that more carefully.

EDIT : never mind, opening the case doesn't seem to change a thing after more tests. It does improve the time to idle, but not max temps.
 

·
Premium Member
Joined
·
2,307 Posts
Im rge on overclock.com..I did the testing in first link in Ashcroft's post (and same testing in most builds I have done for past 20 yrs), just wanted to add clarification of that testing.

I always prevent exhaust air from heating my intake air..., ie currently all exhaust out rear of case through a hole in back of my desk, so exhausted air cant heat my intake air without heating most of the room. Air conditioner prevents heating the room. Ie if your summer ambients are 30C with poor air conditioning and/or not directing exhausted air, and you end up with 35C intake air....that requires a different fix.

That being said your case from reviewers testing has bad air flow, ie top/front some air restriction. Plus you have no exhaust fan, and no fan inside case moving air only through rads, ie poor airflow in a case that is air restrictive to begin with.

If you are going to use rads as both intake, you need good air flow and you need a pull fan at top rad which gives air flow over ram/mobo...push fan through rad doesnt do same. And since pull is 1C worse than push for rad, you really need push/pull at top if your going to run top/front rads as intake with rear exhaust (like I always do).

I think your on right track with 1) case with better air flow (ie vents/grills in front/top), 2) use rear/dedicated exhaust fan regardless of what rad setup in all testing scenarios it lowered internal temps, 3) push/pull on rads improves cpu/gpu temps, internal case temps, ram/nb temps (via pull top fan blowing across ram/nb heatsinks) compared to push or pull alone at same speed or noise level. (in single rad testing, push/pull is ~2C better temps than push in 30mm rad, 3C better in 45 mm rad, and 4C better in 60mm rad, and pull only is ~1C worse than push. 4) Try to think of a way to exhaust air, without it easily recirculating back to intake. But like suggested, definitely get a water sensor and at least a cheap thermometer for measuring intake and water temps so you can see what is happening. Then just left with controlling your house temp.
 
  • Rep+
Reactions: Oli3dfx

·
Linux Noob
Joined
·
991 Posts
Quote:
Originally Posted by Oli3dfx View Post

You're right. I have to add water temperature sensor, as well as flow sensor I guess. It's on my todo list on the next drain.
Without data, I can only guess and try. Anyway, I bought that fan before it occurs to me to discuss it here, so... I'll be doing my little tests.

But still, the ambient temp is way cooler now, as is the weather, and I lost 5 to 10 degrees on the GPU temps,
maxing out at 70-72°, with the both CPU and GPU idling around 27°C.
That is really looking like some kind of threshold in ambient, that's why I was wondering if the setup was wrong, or if taking air in could improve the efficiency.
I also changed the fan response so that they go up faster, we'll see.

Anyway, it's quite clear the case is not ideal, thing is I was miles away of building a custom loop when I bought it.
The top and the front are "semi-open", mid plastic mid ventilation grid. And I don't have a lot of space, for disks especially. I had to put them in 5.25" trails just below the top radiator.
I plan to have a Case Labs case for the next build (Mercury S8 probably). Maybe I could get it right now, the current build could use the move...
smile.gif


But still, I may have to check out the wb on the GPU, cause
I'm miles and miles away from 36° at load.
I put it on quite carefully, but who knows. Maybe I screwed it up. :s
I have PE rads, the "mid" offer by EK.
The fan speed and pump are both connected to the CPU PWM.

And lastly (and I think I've answered everything), all the fans already are between the case and the cover. No more space sadly. :s
I did a complete radiator clean up some weeks ago.

I think I'll try taking air in anyway when I get the new fan putting it at exhaust, and see how it reacts in stress testing.
There is also room to place a side fan just at the same position as the front rad is taking air in. Could exhaust it right away to.

Thanks for links on some tests about taking in or exhausting, I'll read that more carefully.
Yes, 72C is way too high, but 36C for extended full load is ultra-optimistic. My set up is similar to yours with smaller rads and my GPU high when benching and overclocked is 50-53C which should probably be your peak target as well.

How are your pump and fans run by your CPU temp? Are they separate curves or together?

If needed/desired, it looks like you may have room enough to add fans between your pump and front rad.
 

·
Registered
Joined
·
31 Posts
Discussion Starter #18
Thanks for your answers.

Yes, the pump and fans are all reacting to CPU temp, all responding to the same PWM signal.

I'll definitely get that water temp sensor. I planed to do my loop maintenance in the coming weeks anyway.

I'll definitely see for a better case. I'd love to have a Case Labs case, but it cost a hefty price to make them come to Europe... I don't know if it's worth it.
On the other hand, I don't find anything like it by other vendors, and it would pretty much be the last case I'll ever buy.
tongue.gif


I'll also buy a knew thermal paste for the blocks. I was planning to go for the Thermal Grizzly Kryonaut, which seems to be quite good. Any thoughts ?
 
1 - 18 of 18 Posts
Top