Clevo M570TU Cooling Overhaul

Looking good, hyperbolic. Those are very strong temps for this notebook. I have to thank you for putting up those pics, hyperbolic, as they reminded me about the one thing that I forgot that has been itching in the back of my mind since I started revising my article: the picture of the GPU vent on the backpanel. I had to dig deep into my M570TU mods folder just to find the .raw, which took forever since my 1201N can't blaze through several thumbnails like my QX9300 or my current 920XM. I have updated the first post with the pic and a few more desciptions.



Anyway, I took a look at the blocks that you double mounted ontop of the GPU radiator. That is a very ineffective way to utilize the extra space. I say this because heat has to transfer through several mediums just to get to the top (or in this case bottom) blocks that you have stacked ontop of the middle ones. You're better off using blocks that are already tall enough to fit in that space. Also, the base of the top blocks will impede airflow as they are really thick. But more importantly, almost zero airflow is going to go through the fins on the top blocks since the base of the blocks already diverts CFM through the fins on the middle blocks. You can buy a pack of 8mm tall heatsinks at your favorite online retailer such as Newegg. Almost every computer parts store has them.

 

Yeah, I know it could have been done better, but I am well below the target temperatures I was looking for and I just want to use my notebook instead of taking it apart. Thanks for the informative thread.

 

@Soviet_Sunrise:
Very impressive !
I'm looking forward to mod my M17X cooling system just a bit.
I'm mostly concerned with the GPUs and the MCP. So far, using the MX-3 instead of the stock thermal pad on the MCP (lowered my max temp by 20C from 80+ to under 60c. Now with the uneven gaps between the chips and the heatsink on the GPUs have no choice but to use thermal pads.
Do you think there will be a huge difference in temps when swapping the stock pads with some high quality ones? And also what brand/type would you recommend for the GPUs (4870s)?
Also I have 1mm gap between the MCP crystal and its heatsink, so I thought maybe instead of using a rather thick layer of paste put a copper plate in between? Any suggestions?

Thanks in advance!

 

I would advocate to get some new thermal pads. Anything above thermal conductivity of 2.0 W/m. °C should be good and probably better than what you already have.

In my case, when I've replaced the MX-2 with Shin-Etsu (which is a torture to apply BTW) I got a small difference in temps, I think 1-2 C overall.

However, my video card will still reach 84-85C max after 2-3 hours of gaming, let's say MW2.

 

Thermal paste should never be used as a thermal bridge. http://forum.notebookreview.com/showpost.php?p=5771729&postcount=13

Swapping out the thermal pads with better one will yield some benefit. But the primary reason that we swap ours out is for durability reasons. I can recommend you Bergquist thermal pads, which you can find on eBay. They come in various gauges to fit your needs. The thing that makes them good is that only one side is adhesive while the other side has fiberglass gauze. So when you are removing the heatsink on your GPU, the pads will stay on the memory chips and not peel off somewhat onto the heatsink.

Recently I switched over to T-Global pads because I could not find Bergquist pads in 2mm thickness for a reasonable price. On my W870CU, I haven't made any comparison between the factory pads or the T-Global 3.2 W/m-K pads as I immediately swapped the new pads onto my GTX 285M, but the difference in performance should be marginal.

 

Thanks for the reply!
I wouldn't use a paste on the GPUs but had to try it on the MCP. I know it is not the best and probably dangerous option, but using the non-bleeding, non-conductive paste did reduce the temps by a huge 20C. After 2 months of testing the temps stay the same, no bleeding etc. And I seriously doubt it could be accomplished with a pad, though will give it a shot when the new pads arrive.
I guess the real question is whether or not better pads will help by more than 5c on the GPUs? My max temps are below 80c and only the memory can hit 82-84 in Crysis (ambient is 27c). So if there is a chance to lower it by 5-10c, I'd do my best.

 

Aikimox, did the factory thermal pad on your southbridge show any signs of wear or dirt? Replacing a thermal pad with thermal paste shouldn't have a difference as great as 20*C, but it's surprising to hear that it did. The initial thermal pad must have been a real dud.

One of the key properties that gives thermal pads their ability to transfer heat is moisture. Thermal pads tend to "dry out" over the course of it's service and ultimately lose it's ability to effectively conduct heat. Frequently exposing the pads during heatsink removal will speed up the degradation. This is why most users' GPU memory tends to fail fairly sooner than those that do replace them, and even moreso if the memory is overclocked. They are completely focused on replacing the thermal paste on the GPU die and not giving any attention to the thermal pads that transfer heat away from one of the most sensitive components on a GPU daughterboard, the memory chips. http://forum.notebookreview.com/showpost.php?p=5416851&postcount=17

However, if you can shove a properly gauged copper shim in there and sandwich it in thermal paste, that will be, by far, the best solution.

 

The pad is ok, it is in a sealed plastic bag now . It's clean and pretty wet. I think the problem with M17X-R1 is that both MCP and CPU heatsinks are connected and the gap between the MCP crystal and its heatsink is too big for the pad to fit tight. I think a thicker pad would suit better and will definitely check this out.
So 2mm thick pads for the GPUs should be enough?
I'd also make a shot with copper shims, but where can I get the 90%+ copper ones , besides Ebay?

 

I don't know the precise gap measurements for your southbridge and heatsink so you'll need to find that out on your own.

These guys should have everything you need for raw materials. C1100 copper is the standard for 99.95% purity. All of my RAMsinks, except for the Chinese ones, are made from the same copper. Make sure you lap both sides of the shim for the best possible contact. http://www.onlinemetals.com/merchant.cfm?id=1118&step=2&top_cat=87

 

Thanks for the great info!
Will keep you posted with the progress.

 

Hey SS, I guess I should replace my VRAM thermal pads as well. What size did you go with when you did your cooling overhaul? (I figure we have the same gap distance considering both cards have the same layout design and copper casing) I would rather not waste money using trial and error to figure out the right thickness...

 

The GPU memory uses 1.5mm thermal pads.

 

Hi Soviet, What do you think about a copper mod to replace the thermal pads on the memory and heatsink? Can i use 1.5 mm copper and apply it with Artic silver 5 on both sides? Or maybe apply the copper to the heatsink with thermal adhesive?

 

http://forum.notebookreview.com/showpost.php?p=5771729&postcount=13

The memory chips on notebook GPU's do not produce as much heat as you think. However, they are still sensitive to high heat, which is exclusively generated by the GPU die. All desktop GPU's compliment the GPU memory with small RAMsinks and are cooled by the CFM flowing through from the fan. The memory heatsinks are independent from the card's main heatsink that is dedicated to cooling the GPU die. All desktop GPU heatsinks either cool the memory with RAMsinks, or with a shared design just like notebooks where the memory and the GPU die are cooling by a single medium. The big difference is that desktop GPU heatsinks are far, far more effective and versatile than notebook heatsinks and the memory and the GPU core will both remain very cool compared to notebooks; 70*C is considered very hot in the desktop world. I think the only desktop GPU heatsinks that have direct contact with the memory are a few full GPU waterblocks in water cooling setups thats are tailored specifically for the card.

On notebooks it's more complicated. Notebook manufacturers cannot setup the design so that the memory can still have their own heatsinks apart from the die because of space and practicality issues. So on every single modern notebook, the memory shares heatsinks with the die. And since the GPU die produces more heat than the memory, the memory is going to be hotter when the GPU die exceeds a certain temperature where the heat that is shared throughout the cold plate will be higher than the relative heat output of the memory. http://forum.notebookreview.com/showpost.php?p=5186707&postcount=58. One has to take into account the thermal limitations of the memory chips. The GPU die, just like a CPU die, has an upper limit of around 100*C before it starts to incur heat damage, though it should realistically be around 95*C. Memory chips on the other hand have a lower tolerance of around 90*C. http://forum.notebookreview.com/showpost.php?p=5893584&postcount=18. This is why I am always telling users to avoid pushing their GPU temps over 90*C because it's ultimately the memory that is going to take the brunt of the heat.

My premise here is that you do not need to replace your thermal pads with copper sheets. It's really not going to benefit your overclock very much for the amount of time and effort that you will put in as the number one cause of your memory overheating is from your GPU die. Also, like I mentioned in my post in the first link of this post, retention pressure combined with heat will put physical stress on the solder joints that link the memory chips to the PCB. With thermal pads, all you need to worry about is heat instead of heat and pressure and a potential fitting issue that may endanger the memory chips or the GPU core depending on how offset in thickness the shims are. It's the primary heatsink itself that you want to improve since it has to carry the heat of both the memory and the core. Yes, the thermal pads still need to do their job and you need to make sure that they are maintained and replaced when needed as the medium still plays a part in keeping the chips cool. If the thermal pads were failing, or in the worst case scenario, there were no thermal pads at all, the chips would fry themselves as the ambient air in that small of an enclosed space with almost zero CFM would be heated up real fast by itself and the core.

 

Thanks Soviet for this good information!! I like to upgrade my 570RU with a TU heatsink and fan but where can i buy this part and do they fit!? I can se in the pictures on the first page that the heatsink is different, but is it better than the RU heatsink?

 

The fans on the M570RU and M570TU are the same. The CPU heatsink on the M570TU is a four bolt mount as opposed to a three bolt mount on the M570RU.

 

Is it really the same gpu fan!? i read that the TU fan is running with less rpm,s than the RU fan!? The gpu heatsink look different than mine, the pipes are flatter and wider!

 

The GPU fan is the same too. As to how the ACPI handles the fan may be different on the M570RU though. The GPU heatsink only underwent a revision with the release of the GTX 280M. All cards before the GTX 280M have a heatsink with smaller heatpipes and radiator. The revised heatsink is compatible with the M570RU and the card it holds.

Where did you read that the M570TU's GPU fan is running slower than the M570RU? Lately a lot of people have been claiming to me that they have "read" something somewhere but can't recall the source.

 

SS

Where did you buy the Bergquist thermal pads? I see a few different models on their webiste but want to buy the right stuff

Thanks!

 

eBay. 10char

 

Ok...

This or this?

 

If you intend to use it for the GPU memory on the W870CU, both of those are too thin. The memory isn't going to make contact with the heatsink. You need to find 2mm pads.

 

That is all I found.

If you see some somewhere, shoot me a link. Much appreciated!

EDIT: I found some here and here, but they aren't the high-quality stuff you mentioned.

 

It's better than nothing, and by nothing I mean the ones that come with the GPU.

 

Wait, so Clevo doesn't use any thermal pads on the VRAM?

 

The second ones are fine.

 

I got the 2mm and 3mm from these per Blacky's recommedation in this thread:
http://www.newark.com/jsp/search/br...h_001&Ntt=gap+filler&Ntx=mode+matchallpartial

Europe based members should look at the corresponding Farnell site:
http://uk.farnell.com/jsp/search/br...ions=false&ref=globalsearch&_requestid=739499

 

What are you going to use those really thick pads for? Is the gap between the memory and the heatsink on the D901C really that big?

 

I was hoping to use the 3mm ones for the upper corner chips. There is a large gap there because the heatsink is getting away from the card in the upper part.

I used the 2mm for the memory chips, hopefully it is not keeping the card die away from the heatsink. I just couldn't find other proper thermal pads after i roughed up the original ones when swapping the cards.

 

Has anyone ever heard that high heat (soldering temperatures) will evaporate the liquid in heatpipes and make them stop transferring heat correctly?

I was thinking lead free solder (high percentage of silver) might transfer heat better for this kind of heatsink modification and expansion.

 

Computer heatpipes have an effective lifespan of around 20 years and over, about five times the service life of a typical notebook for most people nowadays. The only way the liquid in the pipes will degrade is if the copper or aluminium sealed tubing is compromised from physical damage or extremely heavy oxidation from temperatures far outside the range of what our CPU's and GPU's can produce. Specific tuning for the heatpipes regardless of what industry it is designed to serve will also not degrade or evaporate the liquid.

Solder is meant as a metal bond for long term permanence and is less effective than the heatpipes by as much as ten fold. If you mean to use the solder as a bond for mounting the additional copper onto the existing heatsinks, then yes it will be better than the thermal tape in terms of raw performance, though not by much.

Now that you mention soldering, I think I'll sound off again on the related topic. Here is the plain and simple reason: modifying the existing heatsinks by adding more mass to them will give them more thermal capacity. Increasing capacity means that it will take longer for the components to heat up on our temp monitors since the heatsinks can saturate and retain (and disperse) more heat. This is ideal because it is not only extreme temperatures that kills electrical components, it is also extreme change in temperatures. Lead free solder is brittle and fragile compared to lead based solder, which we stopped using years ago for health and safety reasons. When the solder is subjected from a room temperature immediately to the max temperatures that we like to take our notebooks up to when stressing them will crack the solder and reduce it's durability. Part of my philosophy is focused on mass to slow the change in temperature over time, as well as reduce max temperatures with proportionately increased surface area of course, to maximize efficiency and durability of the components. Though the practical drawback is the nominal weight of the notebook itself, it is still in absolute terms an advantage because, like I said in my dissertation, the notebook with the mod weighs less than an unmodded notebook with an external cooler such as the NC2000. The user will also need to carry less since the notebook cooler is literally inside of the notebook, meaning more space in the backpack for books and whatnot. However, if the user does not need ice cool temps and just wants to use his machine at the library for typing up work, then in this case would the latter be better. Though for me, my little 1201N serves that purpose. But for the big benchers out there, a full cooling overhaul and an external cooler will take your temps to a place they have never gone before.

When I revisited this mod and applied it to my W870CU months ago, it ended up weighing more than a fully loaded D900F. My temperature ceilings never exceed 70*C under 20 minute long stress tests in my typical room temperatures, and these were with moderately aggressive overclocks on both the CPU and GPU, and with nothing more than an Edova Xpad underneath. The goal here is to take advantage of the remaining space inside of the chassis. Though I consider my W870CU to be thoroughly complete in the mod, including the recently installed (and drilled out) W880CU backpanel and further mounting of copper onto the main radiators, I still plan to pick up a pair of W880CU fans. Why? Because I can.

Any Mr. Moneybags can buy a portable AC and shove it under his notebook, adjust a few numbers here and there for an overclock, and press the "bench" button. But it takes creativity and engineering ingenuity to make a notebook a real ice cold notebook that you can still take with you to your favorite places. This is the true art of modding.

 

very clear. good observations

+rep

 

I've had second thoughts of publishing my W870CU cooling overhaul, but I kicked myself again and decided not to. To put things in perspective, it's just a copy-paste of all my philosophies I've already talked about from my M570TU, but with a little bit more that I already covered in the past few posts. I'd rather just consolidate all of it into this thread.

Nonetheless, I had a few dollars sitting around as well as a pair of Notepal U2 fans that I picked up for someone but never claimed it, so I decided to splurge and buy a Notepal U3. NotePal U3 - Cooler Master

My decision to use the pair of U2 fans instead of the trio of U3 fans wasn't a hard one. I only need two fans: one to cool the CPU and the other cooling the GPU. If anyone wants to purchase the U3 fans from me, I'm open for offers.

Since then, I have only made one more mod, and that was to my 180W PSU. I have replaced the aluminium sinks with copper ones. http://forum.notebookreview.com/6709093-post13.html. But in regards to cooling, I ran my same stress tests under my same standards. Forgive me for not having pics or graphs like I usually do, but time is really pressing me lately. This time, my 940XM is under the magic of TS where I have it set for 27/26/20/20 on the multipliers with 100W/100A on the power limits, and HT enabled. My GPU is now slightly overclocked higher than before for 680/1700/1020 @ 1.05V. Ambient temp in my room was ~21*C. And instead of my old Edova Xpad, the Notepal U3 with U2 fans takes it's place. Running the fabled OCCT PSU test, I went for the suicide run. Past five minutes, 68*C on the CPU, 66*C on the GPU. Past 15 minutes, 71*C and 70*C. Past one hour, plus a bathroom break, 74*C and 73*C. The notebook's heat output and cooling capacity most likely reached equilibrium at around 10 minutes or so. The temps kept rising slowly because of the nearby ambient temperature being heated up by the exhaust over time. This is consistent with all of my previous runs on my W870CU and my M570TU, and any other long session stress test with any notebook. Surprisingly, I did not hit a temp ceiling; neither component throttled or spewed back errors.

Was this a dangerous run? Damn right it is. I am not doing it ever again. There is something about the cold winter season that makes people unleash the overclocking beast inside them, but mostly just to test the capacity of a new notebook cooler. But in the end, I'll keep the clocks listed above as my primary set because I know for a fact that I'm never going to surpass them on this U3 through my everyday usage. Though I even think without my U3 I won't hit those temps on realistic daily use.

The results are clear. An external notebook cooler combined with an internal cooling overhaul (and cold winter season) is better than s*x on ice, and yes that is a ice skating show. The U3 does it's job, but I am going to reserve it for home/desk use. On the road or in the computer lab or in the mall, the Xpad tags along simply because I don't want to take two more fans and a sheet of perferated aluminium with me just to lower my temps by, what I estimate compared to the Xpad, another ~8*C from a 70*C baseline. I'll just flash my lower clocks when I'm sporting the Xpad.

 

Impressive thread, I will be doing a similar overhaul/repaste to my G71 when I get my X9100 next month.


(Image from Ren7egade)

And PM me about the U3 fans, I may be interested in buying them off you next month.

 

I'm thinking of taking cooling mods like this to a different level...

I want to add heatpipes to the existing copper contact plates.
Does anyone know how this can be done well? I've heard some pretty bad things about arctic silver thermal epoxy vs everyday thermal tape.

 

I tried doing it with solder, but I wasn't careful and the heatsink fell apart (all of the stock solder melted).

So I would try epoxy if I was going to do that again.

 

Add more heat pipes is a straight forward thinking... but technically its not that simple. You need to integrate the heatpipe in the heatsink fins to have any palpable results, and that can only be made with a pressure cut machine when the fins are separated.

Even if the holes could be made(good drill bit on low speed with drill lubricant) with no harm to the fin struture, the hole will be perfect, and not with pressure borders(used for fin to heatpipe contact).

As I replied to you below, the best you can do is increase the contact between the heatpipe and the base of the cooler, wich in these systems is mediocre at best. That way you should increase the heat transfer between those two components.

 

That's a nice mod, is there a photo of the final result?

 

Yes, adding more heatpipes would be a more custom job that would require sacrificing surface area on the existing radiator to integrate the new heatpipe. It would also increase turbulance for the airflow gateway. Adding a solid block to encompass the heatpipe at the base would be nice, but it's not my cup of tea. The Alienware guys tried something similar and was able to shave off ~5*C from an 80*C baseline. Since the base of the CPU and GPU heatsinks on the W8x0CU are large enough, its better to just build up surface area on them to reduce the thermal load on the heatpipes since heatpipes (specifically the ones used in notebooks) become less efficient as the temperature gets higher and isn't being relieved by the small fan struggling to blow cool air to compensate for the high heat output from our overclocked components. Instead of relying solely on the process of the heatpipe, you would be dissipating heat directly off of the component itself at the base, and that heat would be sucked up by the fans and passed through the main radiators blowing away even more heat. This takes advantage of every sweep of the fan blade per revolution; a very efficient method. Combine it with an external notebook cooler like the U3, and you just made magic. High end desktop CPU heatsinks don't have this because the base is kept clear and flat for retention brackets, which is needed to keep the off center tower from wobbling during system operation and non-operation (transportation) to reduce risk of warping the motherboard.

http://forum.notebookreview.com/6759678-post4279.html
http://forum.notebookreview.com/6760659-post4282.html

I haven't taken a photo of the final product as over the last few months I've been making minor changes/additions to it. You can find most of my beta designs in the W870CU owners lounge. There are probably a few more I haven't refound yet, but they're there.

http://forum.notebookreview.com/5913413-post1189.html
http://forum.notebookreview.com/5916932-post1199.html

I'm still against posting the entirity of my project here for personal reasons. http://forum.notebookreview.com/6469725-post24.html

 

Thanks a lot, audigy! That was exactly what I wanted to know. My apologies, I hadnt realized you replied.

This is not what I was thinking. I agree it would be difficult. By copper plates I mean what you refitted the pipes to and refer to below as 'base of the cooler'.

What I want to add is entire heatpipe+fin assemblies, and now that I know how, also extra vents and possibly fans required to fully utilize 'active' cooling components (as opposed to ramsinks which are designed to be passive)

So what was the actual result of this mod for your temps? Big improvement?

 

I believe you can see it here http://forum.notebookreview.com/sager-clevo/532118-ati-x800-vga-heatsink-mod-fit-m570ru-chipset.html

I appreciate the reply soviet sunrise. You do have a lot of interesting insights especially into the subtleties and hidden types of things here... I think I will be adding heatpipes regardless, just because I can, and also because there simply Isn't a lot of room to add copper directly on top of the CPU or GPU, on all notebooks.
I do plan to develop a way to add heatpipes that can be used on many laptops, not necessarily overclocked but perhaps upgraded with components beyond their design limits (which technically my m860tu already is) possibly with extremely undersized cooling systems in the first place so I'll want to do this elegantly and fit active cooling in small empty corners of the chassis.
Small fans will be an issue i'm going to deal with.
One thing that I'm becoming increasingly interested in is the efficiency of different heatpipes themselves, what relationship is there to their size, the actual type of fluid they have and the speed they move heat.

Its a constant experimental process. My very first plan for this is not something I'm going to go into great detail about right away.
I'll post only my finished product with pictures and full details.

 

Other than the D900x/X7200 and a few other bigboy notebooks, I have not encountered any post 2005 notebook out of the hundreds that I have taken apart and serviced that did not have potential for mounting more heatsinking. Even my little 1201N can be outfitted with more heatsinking. Your M860TU has insane potential to become much cooler from mounting more sinks than adding a heatpipe. There have been projects on the M860TU by a handful of users that taken cooling down a similar route as my own and have reaped the rewards of deeply lower temps. What is needed is cooling capacity rather than transfer capacity. Computer heatpipes are generally 15-20 times more efficient at moving heat compared to just plain copper. But that's really all they can do, move heat. Unless there is surface area and some form of active airflow that isn't static, a naked heatpipe shoehorned into a radiator already occupied by heatpipes would just increase turbulance for air being blown into the radiator, and reduce surface area. Or even your other idea of mounting heatpipes to small empty corners of the chassis and mount some sinks on the ends of them will not be as effective as mounting directly on the heatplate.

The reason why heatpipes exist in the computer industry is because space is an issue. We can have direct mount heatsinks ontop of the CPU and GPU because they would be in the middle of the motherboard. Of course, it would then be much more difficult to remove that heat from the chassis unless more space is used to make an isolated tunnel out. The same goes for desktops; high end CPU tower heatsinks use heatpipes because the sheer size of the radiator ontop of the component would block so many capacitors, the RAM, and other components around the CPU socket. This is why heatpipes are used to raise the radiator to a space away from blocking motherboard components. Otherwise, the most effective way, disregarding cost, would be to use a vapor chamber, which is essentially a really flat heatpipe with a large footprint intended to be used directly at the source of heat. This is why I am always saying that the heatsink design of the D900x/X7200 is perfect because the radiators are right ontop of the component with short heatpipes inbetween. Obviously a vapor chamber wasn't used because of the tremendous manufacturing cost. Shorter heatpipes means quicker and more efficient transfer of heat from the component to the radiator per cycle. This is the philosophy that I have instilled in all of my mods, minus the vapor chamber. Heat being saturated directly from the component onto the mounted RAMsinks is almost as effective as a heatpipe because it has near zero distance to travel, atleast on our Clevo notebooks as the heatpipes are fairly long in proportion with a small radiator and fan, exluding the D900x/X7200 for the reasons previously mentioned above, compared to desktop counterparts; most desktop CPU tower heatsink heatpipes are exposed only for a short distance before they are covered with fins. Not only is my design modular for future changes or disassembly, they are low cost and easily obtainable. Customizing each RAMsink even further with dremels and belt sanders just multiplies the number of possibilities for an effective yet simple design for a cooling overhaul. Yes, the only downfall like I've mentioned before is the added weight on our backs, but if it's worth foregoing an external notebook cooler, then I don't see a reason for the willing to pass up such a fun and therapeutical opportunity. When we did experiments at the university on a TR U120 with an infrared thermograph, the efficiency of the heatsink was increased when we soldered a small 1U heatsink made to rely only on the airflow of the rack fans. We simulated the inside airflow that would be near the motherboard of a typical tower rig while keeping the single 120mm fan on the U120's radiator. If I can recall, we saw a drop of almost 4*C just by adding a passive heatsink made for space constrained server racks with very little CFM running by it. In short, the most economical way that yields cooling benefits greater than it's worth is to have a hybrid of both persay; dissipation directly at the source as well as where the original designated radiator is (again, because of space issue). Like I said in my dissertation, we are only restoring the compromised losses that were made when our notebooks were designed.

 

Heatpipes are more effective when they are longer (I think I read that somewhere, but I could be wrong), I don't understand why you even need them in the D900x/X7200.

 

Whevever you heard that from isn't true. The source of that information needs to go slap themselves over the head for failing thermodynamics. If that were true, all heatpipes would be folded up like the folds in our brain to elongate the distance. A heatpipe becomes less efficient the longer it is. The more distance a heatpipe has to transfer heat from the component to the radiator will result in loss of efficiency.

The D900x/X7200 needs heatpipes for the GPU daughterboard as the radiator cannot effectively pull heat from such a large surface area (the many components scattered across the card, but mainly the area above the GPU die). This is primarily because the radiator and the heatplate aren't one solid medium. The GPU fan shroud is open ported to allow incoming air from the fan to cool even more of the heatplate before it pushes air through the radiator. The vast majority of air is sucked in through the top of the fan. The bottom of a radial fan shroud is typically closed to minimize turbulance, but if opened like on some radial fans, some, but not a lot, of air can be sucked in from the open bottom as well. The heatpipe on the GPU heatsink in this design is used to help evenly distribute heat to all of the radiator fins rather than deliver heat to a farther away source of dissipation. In other words, it is acting as a poormans vapor chamber. The CPU heatsink is another example of a good design, albeit for a 95W+ TDP component. There is ample surface area for dissipation directly off of the CPU from the radiator sitting ontop of it as well as heatpipes delivering heat to another large radiator not in the immediate area of the CPU. This is why the X7200 was able to keep the CPU cool with less radiator fins compared to the D900F, which had two radiators (not including the third one over the CPU as that one did not have a fan for adaquate dissipation) both off the component.

Oh, I just remembered right now that some M860TU ruined their fan shroud by boring it out. This kills static pressure and airflow direction. Boring out the intake shroud past it's effective clearance for the fan blade length (usually 3/5 the length of the blade) will ruin the vacuum effect. http://img4.imageshack.us/img4/5361/heatsinks.jpg

 

I agree that more mass is always helpfull, but you need to consider something. You can not have just the ramsinks and the base, so the heatpipe is the major responsible for taking the heat from you component. If you can improve it, even by just a small percentage, the gains should surpass added masses like 8/9 ramsinks. Remember that the aluminum block that I've considered before will act as mass too, with even better contact area and better contact heat transfer(AS Epoxy VS thermal tape).

About the notebook heatpipes being less efficient as any others when the temperature rises, I think is the contrary. Notebook heatpipes use a higher temperature phase change coolant than heatpipes used for general/desktop applications. Thats simply because the higher Tj temperatures involved in the notebook platform. The most CPU desktop counterparts have Tj temperatures of around 70/80ºC, before the thermal management kicks in. On mobile CPUs the Tj temps can go as high as 100/105ºC, even higher than most GPUs.

You need to consider another aspect, one that I call inevitable design flaw. As you know our motherboards are upside down, wich means the heatpipes too.

Considering that an heatpipe is not fully filled with colant, and that during the heat up of it most of the liquid vaporises and goes up to condensate in the radiator, the direct contact with coolant in liquid form on the side of the heatpipe that touches the base will be highly unlikely, even considering an higher angle of case inclination. I know, it's an unknown subject, no one knows wich type of coolant and how much filling %. So is just a theory...

Just to help understand:

At the time I had taken the X800 thread photos, I was still making the aluminum block(~6 months ago). But as you can see in one of the photos at that time, I was already making some kind of refitting, just for strengthening. Now with the block, I've lowered about 5ºC on the CPU.



For permanent refitting purposes I recommend that you use silver solder. It will be difficult and expensive, but the final work will not get any better. You can use AS Epoxy instead, the performance should be near silver solder. It's safer to apply and cheaper. It's very hard and once it dries becomes a permanent solution.

You have part A and part B, same as epoxy glue. Just mix and use.

 

Yes, I agree that the block would yield the improvement, but the fit and finish needs to be tight and precise. If the aluminium block isn't making contact with all of the pipe, whether it be the block itself or assisted with paste, the efficiency gains won't roll in to your expectations. And this case, 5*C (assuming at a 80*C baseline) for a large aluminium block, which could have been surpassed by adding RAMsinks. Some users that I have helped in the Asus forum last year saw results as high as 10*C from an 85*C baseline in their G51's just by adding RAMsinks and drilling out their backpanels. And they don't have as much room to put several RAMsinks like on my W870CU and other Clevo notebooks. This is also something that isn't readily doable as the tools and machinery needed to mill out an aluminium or copper block tailored just for that heatsink base isn't a luxury that most users have, and is what I wanted to avoid as it alienates users with only basic resources.

Almost all high end desktop CPU heatsinks use either acetone or methanol, both of which have a boiling temperature of ~56*C and ~65*C depending on the purity and the pressure inside the pipe. However, notebook heatpipes aren't too far off as they all use ethanol (basically the stuff we drink during Oktoberfest). They boil at around ~78*C. I've cut open both types of heatpipes for computer and workshop purposes and I'm pretty sure that most notebook heatpipes are ethanol based. I've never opened up a heatpipe from a Clevo notebook, but I'm sure it is the same as well. The reason why I say that they lose efficiency as they get hotter is because ethanol is a heavy molecule. It is not going to move as fast as acetone or methanol as they are heavy, and they need to be kept at a higher pressure gradient in the pipe in order to remain effective (fast), rendering it even less efficient compared to the lower vacuum pressure heatpipes in desktops. The hottest point in the process is at the component. If a CPU is reporting 70*C, the heatpipe's temperature at the base directly above the component is slightly less, usually around 2-3*C lower. A heatpipe will operate just fine as heated liquid will still be moving very rapidly through the pipe to the radiator and cooled liquid will still return to the hot area. But once that boiling temperature is attained, heatpipes have reached their thermal load and having an operating temperature higher than that at the base would slow it down as there would be more vapor in the pipe than there would be liquid, and they cannot carry the same thermal load. So I probably should have worded my statement better in your quote.

Like I said in my previous posts, because notebooks are contrained for space and manufacturers cannot easily add a large radiator and fan because of the demands for lighter and slimmer notebooks nowadays. It's all a matter of making compromises. Because most notebooks are left with small radiators and fans (and other inefficient ways of cooling), CPU and GPU temperatures commonly exceed the near boiling/boiling point of the heatpipe as much as 15*C. This is also why I always say to others to never push their CPU and GPU past the 85-88*C range.

In your diagram, liquid is pooled up on the bottom of the heatpipe. This is not how computer heatpipes are built. Liquid resides around the outside layer or the pipe (not the copper casing) in a wick. In computer heatpipes, these wicks are typically made of sintered powder, which allows for maximal capillary efficiency regardless of where the heat is coming from, just as long as the pipe is touching the base of the component. In other words, they keep liquid in the outer layer like a sponge (they are metal foam anyway) regardless of position or gravity. The liquid isn't going to be at the bottom of the pipe as they are not entirely in a freeflow liquid state due to the wicking. The wick holds the liquid so that the empty inner core of the pipe is used for vapor when the liquid evaporates. Other than the empty core, the wicks, at room temperature, are almost fully saturated with liquid to allow the most heat transfer.

http://forum.notebookreview.com/6759678-post4279.html

I just remembered something a bit more off-topic. I'm going to hate on Gophn's "focused airflow" mod that was procured two years ago. http://forum.notebookreview.com/accessories/302094-zalman-coolers-easy-mod-focused-airflow.html The design concept is effective, but the holes were made so that airflow from the notebook cooler would be aimed at the four fan intakes of the notebook, which does nothing other than increase turbulance. The only reason why the resulting drop in temps was evident is because the mod wasn't "focused" enough. http://forum.notebookreview.com/sager-clevo/519315-laptop-cooler-3.html

In my linked post above, the push/pull design isn't effective because so much static pressure is lost due to the NC2000 not having any static pressure to begin with, and becuse of the large gap between the top of the notebook cooler and the notebook's fan intakes as the notebook is raised by it's own feet. Essentially, there is no isolated tunnel to have >90% retention of airflow going into the fan intakes. Airflow is not fully focused into the fan intakes and is instead dispersed throughout the bottom of the notebook, thus dispersing heat from all building up underneath the notebook. The notebook's fans are sucking in the cooled ambient air that the Zalman cooled, somewhat, which yielded the mediocre temp drop. Had the holes on the Zalman mod been made to focus air over the CPU and GPU, and other hot components of his choice, the results would have been more desireable, and even more so if there were holes made into the bottom of the notebook chassis above the hot components.

 

Heard from a clevo employee that the liquid used in clevo NBs is destilliert water.

 

If it is distilled water, then that is good. It is light and fast and environmentally friendly. The boiling point is near 100*C, and a lot more liquid can be condensed inside the pipe.

 

OK, I just cant reply to everything in regards to what I was going to do, obviously it was not all directed at me anyway.

Just want to say 2 things that it think will make it better than only ramsinks:

Mostly what I am thinking is replacing the mass of ramsinks with heatpipes. It will hardly be less copper at all on the CPU. Im not even talking about the copper of the pipes leading away from the cpu, Im saying that there will be as much copper directly on top of the CPU as you could fit with ramsinks.

And secondly in response to the turbulence.
Once again I am not attempting to add to the existing radiator. It wont even be near it. And let me just say, the fan I have is ridiculously powerful. Im almost thinking its overkill.
I'm going to make a custom fan shroud so the air will be going in more than 1 direction.

 

Lolwut? If the RAMsinks were replaced with an equal volume of heatpiping, the heat still has nowhere to go. You might see some low single digit changes, but it doesn't warrant the effectiveness of using a heatpipe right on the base. As I've said, heatpipes are only meant to move heat. What I am imagining here is that you would have heatpipes on the component bases with some array of surface area for heat to dissipate from (either on the base and/or piped away to small corners of the chassis), and with your radial Delta it will blow the heat all around the interior of the chassis. This is a good idea, but it's still not getting heat out of the notebook, rather it is just spreading it around. A less powerful fan would be better suited so that the main fan isn't interrupted by the various jet streams. The M860TU's fan is powerful, but not that powerful. It can only suck in and blow out as much heat as it can spin.

With that said, I think what you can do is mount a fan that is sucking air from inside the chassis and blowing it outside aimed at your desk. For your main fan, you can isolate it with rubber or foam so that airflow won't be terribly intefered by the second fan, and remove the factory fan grill and replace it with a standard metal finger guard for maximum CFM intake. This can turn out to have nice results once you get the fan, the heatpipes, and some sinking installed.