case orientation leads to drastic cpu temps change with tower air cooler

[Patrick-x13]

hi everyone, 

I purchased an Arctic Freezer 33 esport air cooler to cool my old i7-3930k cpu, installed it and tested. 

I was pretty disappointed tho, cause the temps under load were not so good at all, the cpu was reaching 80/90 degrees on cinebench and intel burn test, and even throttling a bit towards the end. This temps are worse than the ones with the intel stock cooler, with  witch I could at least fiinish both tests without throttling.

Then I decided to try another time and did it with the case in a horizontal position... much better results. Cpu temp not going over 70 degrees, and both test ended with no throttling.

 

Could it be that the cooler is not doing enough pressure on the cpu spreader with the case in vertical position?.. are tower cooler supposed to be mounted only in horizontal builds? 
If this is the case, what possible solution could I adopt... Its very disappointing that a new aftermarket tower cooler (even if cheap) gives me worse temps than the stock one!!   

 

thanks ! 

 

patrick   

[Nocte]

15 minutes ago, Patrick-x13 said:

Could it be that the cooler is not doing enough pressure on the cpu spreader with the case in vertical position?

Maybe? Have you tightened the screws properly?

16 minutes ago, Patrick-x13 said:

are tower cooler supposed to be mounted only in horizontal builds?

No.

[DragonTamer1]

Sounds like it is not making full contact, are some of your mounting screws loose?

[191x7]

Did you use a good thermal paste? Removed the stickers/protection?

[Patrick-x13]

3 minutes ago, Nocte said:

Maybe? Have you tightened the screws properly?

 

2 minutes ago, DragonTamer1 said:

Sounds like it is not making full contact, are some of your mounting screws loose?

 

1 minute ago, 191x7 said:

Did you use a good thermal paste? Removed the stickers/protection?

yes of course... screws thigtened properly ( I even tried to add some metal washers to increase the tension) and the cooler seems pretty firm even in vertical position... but the difference is evident... 

I used Arctic Mx-4 thermal paste and yes of course removed the stickers.. 

[DragonTamer1]

Just now, Patrick-x13 said:

 

 

yes of course... screws thigtened properly ( I even tried to add some metal washers to increase the tension) and the cooler seems pretty firm even in vertical position... but the difference is evident... 

I used Arctic Mx-4 thermal paste and yes of course removed the stickers.. 

What happens when you re-run the tests with the system vertical after laying it down?

[.Apex.]

26 minutes ago, Patrick-x13 said:

Then I decided to try another time and did it with the case in a horizontal position... much better results. Cpu temp not going over 70 degrees, and both test ended with no throttling.

CPU cooler is most likely not installed properly

[Patrick-x13]

Just now, DragonTamer1 said:

What happens when you re-run the tests with the system vertical after laying it down?

same results as before in vertical position, the temps returns to higher values and cpu is slightly throttling toward the end of both test

[DragonTamer1]

1 minute ago, Patrick-x13 said:

same results as before in vertical position, the temps returns to higher values and cpu is slightly throttling toward the end of both test

Do the cross braces for your cooler that connect to the socket go over top of your coolers main cross member? I can't find any pictures of it mounted to a 2011 socket.

 

If you undo the screws a little bit (barely at all) does the cooler freely wobble around? Just loosen them until they are not tight, you still want it to be attached.

 

I have a Dark Rock Pro 4 on Xeon 1650 (basically the xeon version of what you have) and it doesn't make full contact out of the box. I had to put tape between the cross member and the back of the cold plate. It had noticeable play in it.

[Patrick-x13]

10 minutes ago, DragonTamer1 said:

Do the cross braces for your cooler that connect to the socket go over top of your coolers main cross member?

yes its mounted correctly I think... i followed the manual.. here's  a pic of the cross braces screwed to the socket

 

18 minutes ago, DragonTamer1 said:

If you undo the screws a little bit (barely at all) does the cooler freely wobble around?

yes if I undo the screws the cooler can freely move a bit, but I'm seeing that is making contact with the cpu lid... 

[DragonTamer1]

4 minutes ago, Patrick-x13 said:

yes its mounted correctly I think... i followed the manual.. here's  a pic of the cross braces screwed to the socket

 

yes if I undo the screws the cooler can freely move a bit, but I'm seeing that is making contact with the cpu lid...

You mentioned that you added washers, where did you add them and how many?

[Patrick-x13]

3 minutes ago, DragonTamer1 said:

You mentioned that you added washers, where did you add them and how many?

yes I added 4 washers betwenn the head of the screws and the hole in the cross brace, but now I've removed them since they made almost no difference... 

[DragonTamer1]

Just now, Patrick-x13 said:

yes I added 4 washers betwenn the head of the screws and the hole in the cross brace, but now I've removed them since they made almost no difference... 

Put them on the two screws on your cooler that connect the braces to the cooler. Place them under the braces but above the cooler so it pushes down it there. As long as your screws are long enough you could keep adding as necessary.

 

If that doesn't work you can try rotating the whole cooler 90* clockwise so that the tension is put on the brackets in a different way.

[Patrick-x13]

Nice solution, I'll give it a try on monday since I finished my thermal compound and waiting for a refill... I'll post the results... thanks 

 

 

[aezakmi]

This is basic physics, not about pressure, washers or whatever. in horizontal position the heat spreader of the CPU (lid) will have an even temperature with a hotter point in the center where the cores are, this way the heat transfers better and faster to the heatsink and into the air, thermal paste helps mitigate this effect by filling all possible microscopic holes and cracks on both metal surfaces.

 

As heat and hot air tends to rise, in a vertical position one part (top) of the lid and heatsink will always be hotter than the opposite resulting in a slower dissipation. Do try to tighten it a bit but not too much since you could damage the CPU.

[Patrick-x13]

thanks @aezakmi, this makes a lot of sense, still I can't explain to myself why this cooler performs worse than the stock one... 

 

based on your explanation, do you think there is something I can do to make this cooler work better?

 

..thinking about orientation of the cooler, case fans ecc... I mean 20 degrees between vertical and horizontal position of the case is quite a big difference to me

[WoodenMarker]

Did you gradually tighten the cooler in an X-pattern instead of one side before the other and are the screws as tight as you can make them? An even or non-tightened mount can easily cause high temps. 

 

[DragonTamer1]

1 hour ago, aezakmi said:

This is basic physics, not about pressure, washers or whatever. in horizontal position the heat spreader of the CPU (lid) will have an even temperature with a hotter point in the center where the cores are, this way the heat transfers better and faster to the heatsink and into the air, thermal paste helps mitigate this effect by filling all possible microscopic holes and cracks on both metal surfaces.

 

As heat and hot air tends to rise, in a vertical position one part (top) of the lid and heatsink will always be hotter than the opposite resulting in a slower dissipation. Do try to tighten it a bit but not too much since you could damage the CPU.

What?

 

Heat rising is irrelevant in this situation because there is not a large amount of heat being generated and the IHS of a processor is so small that it quickly becomes saturated. If you are talking about the cooler, the heat pipes in the cooler would work to move the heat away regardless of angle.

[MaxPower4U]

Sounds similar to a problem I run into several years ago with a  Noctua NH-C12P & NH-C14, on an overclocked Phenom 1100T (4GHz - 1.425v). Would run Prime95 perfectly fine all day, with the case laying on it's side, but as soon as I stood the tower case up into the correct orientation, would overheat and blue screen pretty fast.

At stock speeds/voltages, it didn't matter what the case orientation was, but OC'ed the heat pipes just couldn't remove the heat, unless the case was laid on it's side.

 

You can read the whole story here if interested

https://forums.overclockers.com.au/threads/amd-phenom-ii-1100t-4ghz-high-temps.973767/

 

 

 

[Patrick-x13]

very interesting... it seems to me like its a common problem with heatsink tower coolers, maybe they don't perform very well with vertical builds... or at least some of them
Altough it sound strange, I think it could be possible... 
I'm trying @DragonTamer1 advices tomorrow, but I'm starting to think about getting a liquid cooler like a corsair H60 or something like that... 

[Patrick-x13]

UPDATE: I tried @DragonTamer1 solution, and it did slightly help but not to the point that I think the contact of the heatsink was the actual problem.. I realized that I was running the system lightly overclocked so the higher temps could be related to that in the first place, sencondly the heat dissipation was better in horizontal position for sure.

 

So I guess that is a combination of elements making this particular cooler not good for this machine, I got a masterliquid lite 240 and the temps are now perfect even with oc. 

Curious fact is that I mounted the same cooler in an x299 system (i9-7900x) at stock speed, and its actually quite good in that. thermals are balanced and I could bench without throttling and with safe temps. 

 

still digging

 

thanks 

bye

[DragonTamer1]

11 hours ago, Patrick-x13 said:

UPDATE: I tried @DragonTamer1 solution, and it did slightly help but not to the point that I think the contact of the heatsink was the actual problem.. I realized that I was running the system lightly overclocked so the higher temps could be related to that in the first place, sencondly the heat dissipation was better in horizontal position for sure.

 

So I guess that is a combination of elements making this particular cooler not good for this machine, I got a masterliquid lite 240 and the temps are now perfect even with oc. 

Curious fact is that I mounted the same cooler in an x299 system (i9-7900x) at stock speed, and its actually quite good in that. thermals are balanced and I could bench without throttling and with safe temps. 

 

still digging

 

thanks 

bye

That is really weird.

 

I tried laying my system on it's side for shits n' giggles and I did see about a 3-4*C drop on it's side on average but nothing substantial like yours. I wonder if some heat-pipes are more efficient at moving heat than others.