Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Smoking hot hardware: Any pointers for DIY cooling?
View unanswered posts
View posts from last 24 hours

 
Reply to topic    Gentoo Forums Forum Index Kernel & Hardware
View previous topic :: View next topic  
Author Message
1clue
Advocate
Advocate


Joined: 05 Feb 2006
Posts: 2562

PostPosted: Thu May 22, 2014 4:38 am    Post subject: Smoking hot hardware: Any pointers for DIY cooling? Reply with quote

Hi.

I have a problem with heat. It's a laptop and a tower. I know the fans spin fine. This is at home, so no fancy server enclosures with built-in cooling.

The tower is able to keep cool enough because it has 5 big fans in it, but they are moving faster than I think they should have to, but it gets to the upper limit when I have extended compiling.

The laptop goes into thermal shutdown regularly. I've opened it up and vacuumed out the dust several times, and I know the fans spin fine. But you touch it when it's using video or skype desktop share or similar and it's literally too hot to touch, top and bottom. The fans sound like jet engines when this happens.

So I'm wondering what I can do. The laptop is a MacBook Pro. It has a metal case.

I'd almost go for making a box with a window air conditioner in it, but I need my laptop monitor. I don't need the keyboard or the pad.

Another thing that seems feasible but I have no experience is to go with liquid cooling. I was thinking a cold plate on the bottom of the laptop, but I don't know how effective that will be.

I'm really at a loss here, what works most effectively?

Thanks.
Back to top
View user's profile Send private message
lexflex
Guru
Guru


Joined: 05 Mar 2006
Posts: 359
Location: the Netherlands

PostPosted: Thu May 22, 2014 10:27 am    Post subject: Reply with quote

Hi 1clue,

Never used it myself, but this is a 'cooler' for laptops:

http://www.amazon.com/dp/B003ZMF27G/ref=asc_df_B003ZMF27G3146472?smid=ATVPDKIKX0DER&tag=pg-451-100-20&linkCode=df0&creative=395097&creativeASIN=B003ZMF27G

You might indeed be able to make something like this yourself using some old fans....

Alex.
Back to top
View user's profile Send private message
1clue
Advocate
Advocate


Joined: 05 Feb 2006
Posts: 2562

PostPosted: Thu May 22, 2014 4:39 pm    Post subject: Reply with quote

I've tried a couple of those coolers, they have marginal effect in my case. I literally wore them out, the motors got slower and slower over a few months, and it only drops the temperatures maybe 10 or 15 degrees at best.

I think I need some sort of chilled and forced air into the cooling ports of the laptop and tower. With the tower I suppose it's possible to get a cool plate and replace the CPU heat sink, but that only changes CPU temperatures and I think some of the other things like the video card are pretty hot too.

I have a window air conditioner I can donate, but I'm a bit at a loss as to how to proceed. Especially since there's both a tower and a laptop.

Still hoping for some sort of commercially manufactured device to use instead of the air conditioner.
Back to top
View user's profile Send private message
lexflex
Guru
Guru


Joined: 05 Mar 2006
Posts: 359
Location: the Netherlands

PostPosted: Thu May 22, 2014 5:29 pm    Post subject: Reply with quote

You might consider (for the tower) to connected the cooling elements to the case or to some outside (cooler) surface using heat conductors.
This will greatly improve the effective cooling surface and the total mass (as long as the conductors conduct enough, i.e. they need to have a large enough diameter and be copper.). Effectively you might be able to bypass the in-case airflow cooling.

Streacom uses a similar technique in some of their cases, but I am not sure if you can buy the heat-pipes and connectors separately; would be nice though if something like this would be available in various sizes as a 'DIY-kit'...
http://www.silentpcreview.com/Streacom_FC8_Evo


Alex.
Back to top
View user's profile Send private message
NeddySeagoon
Administrator
Administrator


Joined: 05 Jul 2003
Posts: 43753
Location: 56N 3W

PostPosted: Thu May 22, 2014 5:39 pm    Post subject: Reply with quote

1clue,

laptops are usually a lost cause. The best you can do is either prop the edges up on two thick books, so air can circulate underneath or stand them on edge, so the cooling path is vertical with the air intake lower than the exhaust.

You need to tell us more about the tower. Lots of fans does not translate into lots of useful airflow. Fans also need to move air in the right direction in the right place.
A picture would be useful.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.
Back to top
View user's profile Send private message
khayyam
Watchman
Watchman


Joined: 07 Jun 2012
Posts: 6228
Location: Room 101

PostPosted: Thu May 22, 2014 8:10 pm    Post subject: Re: Smoking hot hardware: Any pointers for DIY cooling? Reply with quote

1clue wrote:
The laptop goes into thermal shutdown regularly. I've opened it up and vacuumed out the dust several times, and I know the fans spin fine. But you touch it when it's using video or skype desktop share or similar and it's literally too hot to touch, top and bottom. The fans sound like jet engines when this happens.

1clue ... when Apple switched from powerpc to intel CPU's they suddenly had to deal with the additional heat the processor created, with their towers they solved this with the insertion of a clear plastic internal frame (or 'mold') that allowed the fans to funnel air right over the CPU and pci slots and with a fairly huge heatsink this just about worked (though these machines were often noisy, particularly when doing any kind of graphics processing). In the case of the macbook's they pretty much failed to take this excess heat into account as the first generations of macbook's (one of which I own, so I speak from experience) had all manner of heat issues (of which Apple's response was to provide a firmware update that basically caused the fans to spin up constantly). This switch also coincided with Apple being less a computer company and more a consumer electronics company (iPod, iSight, etc) so, like other items in their catalogue they are designed with the initial gleam in mind but not to last (that white macbook will look like a car mechanics hands within the first year of purchase, and your ethernet port may just break off the motherboard due to snap together case not providing any support for the peripherals enclosed, etc, etc). The later "alu" and "air" books have a similar build quality, there is *a lot* put into the look, but little thought and attention behind the function ... and if it gets too hot then this simply equates to a shortened lifespan and so the possibility of further sales.

Their problem is that these machines (unlike other netbooks) need to support Apple's line of products (particularly Final Cut Pro) and so they can't opt for a cooler (but slower) cpu such as the Atom, this means they are caught trying to keep them slick looking dinky portables but able to run graphics intensive programs ... and somehow run without bursting into flames or sounding like a jet engine. Enter integrated graphics/discrete GPU and all the defects these brought with them but none of this has been able to solve the key problem ... they are simply too hot.

My solution has been to use every mechanism possible to under-perform ... so, cpufreqency scaling, scheduling (using BFS), ionice/nice, -j1, load-average=1.00 for software builds, and various other attempts to tune behaviour and use less resources. I also use a "coolpad" (which is just the equivalent of the bookstop that Neddy suggested) to keep some air circulation on the underside. With these in place most of the time the fan is off. I avoid the use of a lot of apps (like flash, and skype) as these inevitably cause the temperature to rise and the fans to kick in. I'm sure that's not what you want to hear, but my wanting the machine to last and be usable (which for me equates to being somewhat quiet and yet cool) has prompted me to go in the above direction ...

best ... khay
Back to top
View user's profile Send private message
1clue
Advocate
Advocate


Joined: 05 Feb 2006
Posts: 2562

PostPosted: Thu May 22, 2014 8:55 pm    Post subject: Reply with quote

I'm not happy with underperforming.

I have the machined aluminum MacBookPro5,2 -- mid 2009, 17 inch. It's maxed out and I need it to be faster.

I write software, and this thing is owned by my employers. So it runs Mac OS as a main OS and several Linux and Windows VMs. It only takes 8g and I'm using all of it every day.

The newer faster machines are useless to a software developer because of the lack of screens that you can read with 48-year-old eyes.

I'm running dual monitors, and I have a frame I welded up which holds the external monitor at the right height for me, and the laptop monitor open just under it. The chassis is at an angle such that I can barely see the keyboard from my sitting position, with the hinge about 4" above the table and the wrist rest end of it about 10". I've experimented with the angle and it doesn't seem to make much difference. It's a really handy setup, I think the next one will be a bit different. Maybe I can get the whole works into a case that can be an airplane carry-on.

Anyway, the way I understand it the aluminum Mac laptops were supposed to use the chassis as a heat sink. I know mine gets pretty hot, but blowing air across it doesn't seem to help much.

I've been pretty rude to this thing, and it's held up well except for the part about heat. There are dents here and there but that's to be expected.

Realistically speaking I need another computer. I'm trying to limp along until I can get some new hardware, but until such time as that happens I need to make the best of what I have.

I was thinking of a strongly refrigerated stream of air forced into the hinge from the bottom (where the fan is at) and maybe some leaked so it flows under the bottom. Or, maybe get/make a cold plate and run refrigerant through it, and maybe circulate refrigerated air around the bottom part? I don't know.

I'd love to get dried air in an enclosed space, and get that temperature down near freezing.


The tower is a Gentoo VM host running various VMs, there are 2 in-facing fans on the front panel which blow past the drives (filter's dirty again, gotta clean it) and 1 in-facing fan on the side which blows straight at the video card, which has its own heat sink and fan. One fan is outbound on the top, and one outbound at the top on the back. CPU is 1st generation i7 920. Also slow, but faster than the mac. 12g RAM at the moment, needs more but I'd rather get a newer box.

I don't get overheating on the tower that I've noticed, and it's never done thermal shutdown.

Anyway, I have a welder and enough tools to be a construction worker for a number of professions, so I can build something but I'm just not sure what to do here. It makes sense to me that if I could get these boxes into a colder environment they would fare better. Most important is the laptop since that's my workstation and it has thermal shutdowns.

So what about this:
Let's say I get a container of something like water. I disassemble the air conditioner to separate the inside loop. The outside loop and fan stay in the window. Inside loop goes into the water as a thermal energy store, sort of a thermal flywheel. A second heat exchanger runs a circuit to the laptop, let's say I make an insulated box fully enclosing the bottom part and have a couple "cold plates" by pumping coolant through those plates. Or maybe pump the water (or whatever) from the reservoir through there?

The air conditioner brings the reservoir temp down, shuts off. The individual cooling systems attached to it can pump coolant through at whatever rate is appropriate.

This has to be useful no matter what the laptop or other system right? Especially if the internals are set up for liquid cooling.

And I think any future purchases will be set up with liquid cooling, far more capable than the system could possibly overwhelm. I wonder if I can get a case that's essentially an external heat sink, and bond the CPU and other hot bits thermally to that case?

Sorry this probably makes no sense, but I'm posting it anyway. I'm desperate for ideas here.
Back to top
View user's profile Send private message
1clue
Advocate
Advocate


Joined: 05 Feb 2006
Posts: 2562

PostPosted: Fri May 23, 2014 2:03 am    Post subject: Reply with quote

Nobody? Is this so out in the ether nobody wants to touch it?
Back to top
View user's profile Send private message
khayyam
Watchman
Watchman


Joined: 07 Jun 2012
Posts: 6228
Location: Room 101

PostPosted: Fri May 23, 2014 3:09 am    Post subject: Reply with quote

1clue wrote:
Nobody? Is this so out in the ether nobody wants to touch it?

1clue ... I doubt there is much in the way of refrigeration expertise in these parts, those AC units I've seen (and they are not common in my neck of the woods) would probably quickly destroy a macbook , but I'm not really sure how such a system might work. I should note that computers (and batteries) not only have an upper limit but a lower limit too, so the idea of "down near freezing" is probably not a good idea. The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) states their recommended temperature range for a server room to be from 18°C to 27°C (64.4°F to 80.6°F).

best ... khay
Back to top
View user's profile Send private message
NeddySeagoon
Administrator
Administrator


Joined: 05 Jul 2003
Posts: 43753
Location: 56N 3W

PostPosted: Sat May 24, 2014 4:53 pm    Post subject: Reply with quote

1clue,

The thermal gradient through the laptop is fixed ... unless you open it up and fix Apples design.
It follows that to drop the CPU tempreture by 10 Deg C, the inlet temp has to fall the same amount all other things being equal.
You also have to stay above the dew point, since you really don't want condensation forming in the laptop.
In an office environment you would be lucky to be able to get 10C drop.

The other alternative is to increase the mass of air flowing through the laptop. by forcing more air through it.
This has the downside that it may destroy the internal fan as the fan will windmill in the forced airstream.
It may even windmill well above its design speed which will quickly kill the bearings.

You can mix the above options too.


For the tower, make yourself a water manometer from a drop of ink and a few feet of clear plastic tube.
Put the ink in the tube, bend it into a U shape and add an inch or two of water in the bottom of the U.
Use this manometer to check that the case of the tower is at a positive pressure.
If not, rearrange your cooling. You need need air being blown in, not sucked out.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.


Last edited by NeddySeagoon on Sat May 24, 2014 8:15 pm; edited 2 times in total
Back to top
View user's profile Send private message
djdunn
l33t
l33t


Joined: 26 Dec 2004
Posts: 780
Location: Arrakis

PostPosted: Sat May 24, 2014 7:11 pm    Post subject: Reply with quote

best you can do with the laptop is underclock it, besides that, possibly a rig up a cooling plate to rest it on, and make sure its intake is cool, but not cold air,

desktop, its always about airflow, I have 0 cords on top of the motherboard, they all lay behind it, i think thats the best thing i ever did to combat heating in the tower.
_________________
A process cannot be understood by stopping it. Understanding must move with the flow of the process, must join it and flow with it.

-The First Law of Mentat
Back to top
View user's profile Send private message
pa1983
Tux's lil' helper
Tux's lil' helper


Joined: 09 Jan 2004
Posts: 100

PostPosted: Sun May 25, 2014 5:02 am    Post subject: Reply with quote

Laptops is harder. Replacing the Thermal Past with better one could do some difference but realy unless its old and dry that wont do enough.

Best option I could think of is this.
http://www.liontech.se/PartDetail.aspx?q=p:4610247

basically metal liquid at room temperature. It corrodes aluminum so only use it with copper and nickel plated coolers.

People use it to replace the intel thermal past on IVY and Haswell processors by removing the IHS and relapsing the thermal interface material with that. Difference in Delta T for a desktop CPU at 84W is less then 0.5C compared to over 10C with original TIM so theres 10C to save stock and 20C during Overclocking.

For a laptop one might save 4-5C on a quad core i7 at 35-45W TDP.

You have to check some guides on how to use it. Have not tried it my self but I have done the numbers on it VS normal past and spoken with people that used it and the real world figures added up so it works. Basically will give you the same results as if the IHS or cooler was soldered to the silicon.

If I recall one must apply it with a cotton swab. Also make sure to clean of any old thermal past properly with some alcohol.

Im actually better at hardware and cooling then linux and gentoo. Studied Thermodynamics many years ago. Never practiced it but the basics principals I still remember. I my self cools a 110-130W idle system and 400-450W load system with just 120mm fan at 700-100rpm and a 92mm at 950 rpm at the most.

Most important thing is to exhaust any heated and used air asap.
Also using good cooler is important. I use a NH-D14 with a modded TY-140 fan. Runs at 320-370rpm idle and 700-800rpm load and thats with a i7 3930K @ 4Ghz = 180W TDP.
System is basically dead quiet at idle (less nosy then a usb 3.0 2.5" external drive) and that was with a GTX570 with reference cooler. Recently replaced it with a GTX780 with reference cooler and its just ridiculously quiet and the CPU never hits much over 70C during the hardest stress test I can trow at it, two sessions of intel linpack.
Sure I dont have the overclocking margins on the GFX as the hotted up aftermarket models but the advantage is that I dont need chassis cooling to keep the card cool and the card wont heat up the air for the CPU and It allows me to soundproof the howl PC and in the end its dead quiet.
Ducted cooler sucks air in to the chassis them self since they expel it in the rear and logic suggest that if air goes out air comes in to replace it so its a win win.

So one can build a high end powerful system that is quiet and where cooling dont cost a fortune.
I was in to water cooling back in 1999-2003, built my own systems bought system etc but I gave it up because modern aircooling is just so much cheaper and quieter basically due to heatpipe technology. No noisy pumps etc.
There are cheaper cooler out there that are good like the CM Hyper 212 cpu cooler. But good fans usually has to be bought separately if silence is a concern for both CPU cooler and chassis.

I usually hand pick all man fans from experience (I have tried a few over the years in my chassis) and properly made reviews with scientific data to back up any claims by the reviewers.
Also having to many fans is not good and there placement is important. More is not always better. I usually avoid intake fans and use exhaust only.
Fans should never have airflows that interfere like opposing one and another thats worst case scenario. One wants the straitest possible path for the airflow from intake to exhaust point.

A fan creates a turbulent flow in its exhaust path, its good for picking up heat from coolers. But turbulent flow creates lower pressure and can create hot pockets of air in a chassis if its dominating so it can actually harm cooling if fans are badly placed.
If you remember the old trick where you take a vacuum cleaner and run it backwards and then you can balance a ball on top of the exhaust stream. If the same thing happens with intake fans in you PC and the components needing cooling are not in the path of the airflow cooling can actually be compromised. I have seen test showing like 10C increase in temperature on components using the same fan as intake instead of exhaust.

Plus having redundant intake and exhaust fans realy dont speed up the volume of air going trough the chassis unless its severely restrictive. Better to have two fans either as intake or exhaust. Only real good reason to have intake fans is if you use dust filters so you can dictate where the air enters.

Also with modern hardware temperatures is not that big of a problem and people tend to be a bit over cautious on whats safe and not. if your CPU hits 80C during stress test like intel linpack then its fine. No need to improve cooling more. It wont die from heat then, if the hardware survives for decades electro migration will kill it because thats what kills electronics. Basically current is the killer not heat since atoms will move with the electrons.

But in the end how good you cooling is depends on how well you match up your components. All components plays a part and selecting the wrong one on any of them can seriously effect cooling and noise. Picking Chassis, CPU cooler, GFX card, PSU and Fans with a thought out flow pattern is important.

I prefer to exhaust all hot air in the rear so I prefer if all heat sources are there and the coolers are aimed backwards.
Smaller risk for hot pockets to be created since air can enter where ever there are holes in the chassis.
For me that means a tower cooler for the CPU, ducted cooler for the GPU, also means using a good PSU that can operate at an ambient temperature of 50C that way it can handle even the hottest exhaust air for the CPU cooler and aid in chassis cooling for a lower total noise level, therefor I avoid Chassis with separate PSU vent.
Also make sure that the chassis has proper intakes in the front (front of the sides is fine to) or bottom to make sure I get a strait flow of cold air from the front that then can get to the components that need cooling and after that immediately being exhausted. Very little airflow is actually needed to cool away 400W+ that way.

If one selects the more common open air GPU coolers I would suggest a closed loop water cooling kit for the CPU. Then take in air from the outside, back for example for the radiator. Then make sure the chassis has opening vents in the bottom for fresh air and 1-2 12-140mm fans in the top for exhaust of hot air from the CPU and GPU cooler.


Efficient cooling is about using the air once then exhaust it.
You dont want CPU or GPU coolers that has a tendency to mix fresh air with the hot air just used. Basically old style CPU coolers are bad for cooling efficiency.

To keep noise down if important one dont want to use more air then needed. This is easily done by measuring the air flowing it to the GPU or CPU cooler and measuring the ambient air at the same time. Reduce chassis fans RPM, manually or as I do use BIOS built in fan profiles and if needed fine tune the fans RPM manually (Resistor, fan controller etc) until one is just on the edge on raising the chassis air temperature coming in to the coolers above ambient air during idle and synthetic loads.
Because if you chassis is pulling trough more air then your CPU and GPU coolers consume its just a waste of airflow and a sources of noise and unnecessary build up of dust.

Added advantage of optimizing chassis ventilation to match CPU and GPU cooler airflow (only plays a role on open air GPU coolers not ducted cooler since they exhaust hot air them selfs air) is that you minimize build up of dust.

People have all kinds of ideas that using more intake fans then exhaust fans reduces dust. Well it alters where you collect dust but logic suggest that there are X amount of dust particles for every cubic meter of air so reducing the amount of air flowing trough the chassis per hour should reduce the build up of dust.
Plus I have never found any actually scientific claims about overpressure in chassis reducing dust build up.
Places I have worked we used overpressure simply to keep dust air out but that still requires that over pressurized air to be dust free it self for that to work.
Also remember chassis fans has a max static pressure of 1-2mm water pillar (that also means zero airflow) and one atmosphere is 10000mm and chassis fans work against little to no pressure difference since a good chassis or modded chassis is optimized to have little to no restrictions (badly designed fan grills etc are removed) so that means there realy is no pressure difference to talk about any way since fans and pumps at there peak flow has virtually no pressure.

We have two parrots and the dust here is 10x what a normal family would have at home per day and still I just need to use a can of air a few times a year and my system is on 24/7 with no dust filters simply due to the fact that I only use the air that I need for cooling and no more.

With out knowing more about your components and layout ( pictures would be nice) I cant help you more then this. Also dont forget to experiment with fan placement, fan models, fan profiles in bios etc.
You migth also need a better CPU cooler. After market once are recommended.
One might be surprised to know that removing a fan, halving the RPM might not make things worse at all.
But in my world a high end PC should get away with 1-2x 120-140mm fans at the most.

If you have 5 fans its either a very power hungry system or they are badly placed or you just dont have the CPU/GPU coolers to remove the heat to the air in the first place.

Post pics and specs and I might be able to give some tips on cooling.
_________________
WS: i7 3930K@4Ghz, 32Gb ram, 256Gb NVME & 128Gb sata SSD, GTX780 3Gb & RX 460 2Gb
NAS: i3 4360 3.7Ghz, 20Gb ram, 256Gb SSD, 42Tb HDD, NIC: Intel 2x1Gbit
ROUTER: J1900 2Ghz, 8Gb ram, 128Gb SSD, NIC: 2x1Gbit, WIFI: Atheros AR9462 and AR5005G
Back to top
View user's profile Send private message
eccerr0r
Watchman
Watchman


Joined: 01 Jul 2004
Posts: 7230
Location: almost Mile High in the USA

PostPosted: Mon May 26, 2014 7:52 pm    Post subject: Reply with quote

I'll have to second it, a lot of the times laptops are just not cut out for high performance computing. Ensure the heatsink is attached properly, and if that doesn't help you'll have to run it with CPU throttling turned on and put your high performance tasks on a full sized machine.

I think apple is another problem, they are focused on aesthetics over functionality. Aesthetics unfortunately sells very well but you can't really make a high surface area heat exchanger with just a flat surface...

Unfortunately for now, high performance computing will generate heat. Not much you can do about it other than deal with it. Maybe someday the capacitance beast in on-die interconnect can be slain (the leakage monster seems to be fairly dead now, but it may just be due to dynamic switching now) but until then...
_________________
Intel Core i7 2700K@ 4.1GHz/HD3000 graphics/8GB DDR3/180GB SSD
What am I supposed watching?
Back to top
View user's profile Send private message
NeddySeagoon
Administrator
Administrator


Joined: 05 Jul 2003
Posts: 43753
Location: 56N 3W

PostPosted: Tue May 27, 2014 5:04 pm    Post subject: Reply with quote

eccerr0r,

Even the 1/2CV^2 is getting better with smaller feature sizes and the reduction in V (Vcore really)
Unfortunately, clock speeds and transistor counts are still being pushed up to keep the power disapation fairly constant.
Well, in the AMD/Intel world anyway.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Kernel & Hardware All times are GMT
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum