SEEKING: Thermal Guru | Metalurgist | Extreme Custom Modder | Scientific Opinions on Heatsinks/Pipes

Custom heatsinking upgrades and thermal management solutions have long been an area I have been interested in for various reasons. I know how to MIG weld and solder copper pipe with Oxy/Ace as I used to work doing full-service appliance repair (including refrigeration system rebuilds) for a while.

I have seen some impressive laptop heatsink upgrades and mods on the internet but nearly all of them opt for thermal adhesive rather than a quality solder job.

SEE HERE: https://therandomlab.blogspot.com/2012/01/laptop-cooling-mod.html?m=1
SEE HERE: https://www.techpowerup.com/forums/threads/adding-heatpipes-to-a-richland-laptop.197346/page-2

Some range from bad/janky by just adding more copper mass which really only extends the length of time until heat soak occurs, while others at the more pro-end of the spectrum actually take air flow into consideration and don't just add a ton of mass, while still others at the more extreme/full-custom end of the spectrum properly integrate additional heat pipes and spreaders with thermal adhesive. (Nods to @iunlock and others here at NBR for being in this category.)

There are also questionable and head-scratching laptop mods such as adding liquid cooling loops, that has admittedly crossed my mind even before I sought out the pictures online, but would defeat the point in owning a laptop, at least for me. For now, my mods need to stay in the case and look stock on the outside for the mostpart. I even thought about how to add quick connect chucks and have a desktop stand-alone radiator with burping valve at the highest point and a small water reservoir that would rarely ever need to be refilled or topped up, due to only losing a drop or few each time you connect or disconnect. Am I the only crazy person who thinks of these sorts of things?

SEE HERE FOR QUESTIONABLE UNOFFICIAL LIQUID COOLING UPGRADE: https://sc02.alicdn.com/kf/HTB18Vj4...soft-red-brass-copper-tube-font-b-water-b.jpg

SEE HERE FOR PRODUCTION VERSION OF LIQUID COOLED LAPTOP OUTLINED ABOVE:

==================================================================

I don't know a whole lot about a lot of this stuff, but I can read and form some thoughts. That's where you all come in. I want your thoughts and opinions and corrections. Especially if you've got first hand experience or already tried any of this or know people who have. Links and resources are ideal.

==================================================================

NOOB STORY: https://forums.anandtech.com/threads/how-do-they-solder-heatpipes-to-heatsink.2256994/

So I started researching heatpipes and heatsinks and the low temperature (< 500*F) soldering methods used to join them which made me wonder... Does anybody know what the standard low temp solder alloy used to join the heat pipes to their finned heat sinks is in most laptops (composition / name of alloy / metalurgic properties)?

I ask because I am curious what its thermal conductivity is rated at. I can't imagine mass produced laptops are using the most thermally conductive alloy, but rather the cheapest and/or most durable alloy instead to keep their bottom line in check.

==================================================================

So I started researching what all it would take to either:
[A.] *properly* add additional heat pipes to a laptop

or

[B.] reflow / replace a stock heatpipe with the exact same (or slightly larger) heatpipe size but with a better solder job that is both more thermally conductive in the alloy of choice and surface finishes (lapping, etc) as well as more thoroughly joined to the finned heatsink in terms of solder coverage maximizing heatpipe joining area to the heatsink since a one-off custom modder job can spare the time and attention-to-detail as well as the cost to see that these specs are met.

==================================================================

Cheap heat pipes can be bought from Digi-Key among other vendors. I like the fact that Digi Key has all the relevant specs and data sheets in one place with an easy to use filtering system to sort through the tons of choices.

SEE HERE: https://www.digikey.com/products/en/fans-thermal-management/thermal-heat-sinks/219?FV=2dc1e7e,ffe000db&mnonly=0&ColumnSort=0&page=1&stock=0&pbfree=0&rohs=0&cad=0&datasheet=0&nstock=0&photo=0&nonrohs=0&newproducts=0&quantity=&ptm=0&fid=0&pageSize=25

==================================================================

Advanced Thermal Solutions Inc. has a very nice PDF general spec sheet for all of their heat pipes listed on DigiKey that mentions important details such as the working fluid being distilled water (I have seen others on the internet say it is acetone inside), the proper methods for joining, the suggested bend radii, and perhaps most importantly- the suggested low temp soldering range of temperatures. "For optimal results, heat pipes should be soldered using low temperature solder at temperatures above 139* C (282* F) but no greater than 250* C (482* F)"

SEE HERE: https://www.qats.com/DataSheet/Heatpipe_Datasheet_R4

==================================================================

INDIUM CORPORATION: http://www.indium.com/solder-paste-and-powders/low-high-temperature/

INDIUM CORP PRODUCT DATA SHEETS: http://www.indium.com/technical-documents/product-data-sheets/

SOLDER ALLOY SELECTOR GUIDE: http://www.indium.com/solder-alloy-guide/

SOLDER ALLOY COMPARISON TABLE: http://alasir.com/reference/solder_alloys/


Next I set out to find a source of low temperature solder that had the best thermal transfer properties and arrived at Indium Alloy 290 by means of analyzing raw specs, but not knowing much else about the metalurgic properties in terms of application, ease-of-use, and durability. After narrowing my alloy choices down to only a few, I tried locating information on actual usability for my intended application of joining heatpipes to heatsinks. I came across an article that laid out my choices perfectly, and in fact the Indium Alloy 290 was one of the two suggested alloys mentioned in a blog post for exactly the purpose I was intending to use it for.

SEE HERE: http://www.indium.com/blog/low-temperature-indalloy-solder-alloys-for-heat-pipe-attach.php

BUY INDIUM #290 ALLOY (wire): https://buy.solder.com/Indium-Silver-Wire-97In-3Ag-3-FT/P177_61/

BUY INDIUM #290 ALLOY (ribbon): https://buy.solder.com/Indium-Silver-Ribbon-97In-3Ag/P158_52/

BISMUTH SOLDER DATA SHEET: http://www.indium.com/technical-documents/product-data-sheets/download.php?docid=713


==================================================================

I wonder if the correct solder alloy could be used to permanently attach a heatsink to CPU & GPU by either reflow oven or heat gun so as to avoid the hassle of having to worry about repasting and/or liquid metal leakage or pump out / bake out... Can anybody say with a degree of certainty what the maximum temperature is that a motherboard can safely be taken up to while removed from a system for maintenance/modding without damaging any of the other chips or components?

Here is an interesting case study I found on thermal cycling of various low temp solders.

HP CASE STUDY ON LOW TEMP SOLDER ALLOYS: http://www.hpl.hp.com/hpjournal/96aug/aug96a10.pdf

==================================================================

Non-traditional superior-performance (compared to pastes / greases / and possibly even liquid metals) TIMs can also be found in the form of thermal pads as well as what Indium Corp. calls their " heat spring " AKA metallic TIM that does not suffer from the breakdown and degradation in performance that traditional TIM greases, pastes, and liquid metals do. It yields 86 W/mK of thermal dissipation using 35-100 psi clamping force. It is also safer in terms of melting and spilling when comparing it to liquid metals. Does anybody happen to know what the average amount of clamping force used on most processors and traditional modern heat sinking systems is?

SEE HERE: http://www.indium.com/blog/solder-redefined-part-4-baseplate-to-heat-sink.php

BUY HEAT SPRINGS HERE (warning: expensive! group purchase anyone?): https://buy.solder.com/Thermal-Interface-Materials/C1013_1/

TIM PROCESS ENHANCEMENTS: http://www.indium.com/thermal-management/tim/

BROWSE INDIUM CORP.'s OTHER PRODUCTS: https://buy.solder.com/

==================================================================

That's all for tonight I guess. Let me know what you think. Cheers.

 

Before you go and work on more heatpipes, finding the right solder etc. You need to understand how to diagnose where the problem is coming from. More than anything the main cause for bad cooling are contacts, not enough mass and bad airflow. These are far more likely your problem than anything else, attaching more heatpipes will very likely result in no difference or very meh difference in temps.

In the case of the person who modded his sandy bridge, having a single heatpipe (although bigger than average) is not enough to tranfser the heat generated, so in his case it was effective. I assume you try to improve temps on your Dell inspirion, which has 2 large heatpipes which are shared with the GPU and CPU. The heatpipes are not the problem, they are more than plenty to cool down your system. Your issue is more than likely bad fit and terrible tim. Getting Conductonaut and doing a proper lapping job on your GPU and CPU heatsink plates, will give you the results you're looking for.

 

@Mr. Fox and @bloodhawk have experience in custom soldering job for heatsinks.

@Donald@HIDevolution (or the company) has experience in soldering shims for laptops they sell, last I called you can get the soldering job if you request.

I forgot there's also one more tech here that understands the chemistry behind using liquid metal on heatsinks but I can't remember name. @D2 Ultima ?

 

Naw, it'll likely be years before I go that far with my Dell. I understand the primary concepts and reasons why heat transfer is poor from the factory. This is mostly just a theoretical discussion to present various data and info and get opinions from others because I have a genuine interest in this sort of stuff. But I do have a 3rd Gen i7 3632QM system w/single fan that is on its deathbed I may end up experimenting on for fun. It currently idles around 45-55 *C. It's been a while since I repasted it.

I'm somewhere between a computer science, mechatronics, and materials science/mechanical engineer. Who knows if I will ever actually finish school for any of the above, but I've at least have credits towards CS and been accepted for a couple of the other programs if I ever arrive.

My main points with presenting my findings in this thread are:

1.) Using Indium Alloy #290 (73 W/mK) in place of whatever the factory uses which is likely NOT 97% Indium and 3% Silver like alloy #290 would likely yield better thermal transfer from die to heat pipes. #290 is pretty much the highest thermal conductor besides pure tin (73 W/mK) [EDIT: or pure Indium (86 W/mK)] which matches it at 73 W/mK. Pretty much all other alloys are worse conductors. I am mostly interested in what alloy the factories use to do the job and if it is pure tin (73 W/mK) or some other worse-conducting alloy. Even the worst conducting soldering alloy should still yield better results than the best thermal epoxy adhesives though.

2.) As I mentioned above on the other thread related to thermal pastes and compounds last night, I am curious why people who use NON-conductive TIMs aren't just going with Fujipoliy XR-m pads (17 W/mK) which are superior to the two highest regarded NON-conductive TIMs GELID Extreme (8.5 W/mK) and TG Kryonaut (12.5 W/mK).

3.) Same as my second point, why aren't people using CONDUCTIVE TIMs like TG Conductonaut (73 W/mK) experimenting with Indium Corporation metallic "Heat Spring" TIMs (rated at 86 W/mK) ? This just seems like a gain in thermal throughput with even less risk and more longevity imo.

 

I think @Khenglish has us all beat. And, he lives not very far from you, bro.

 

Ah yes, how could I forget about him.



Khenglish even has experience soldering BGA components, definitely would vouch.

 

When soldering heat pipes I end up using 63/37 solder paste since it's cheap. Heat pipes will start to expand when heated over 200C, but usually this is not a problem as I'll have areas that need to be thin clamped down, so the expansion just improves heat transfer by widening parts that can be widened. When going over 250C like the guide says to avoid they explode.

I've attached an image of my current custom built GPU heatsink. I built my own denser radiator, added a 3rd 6mm heat pipe to the core, and added 5mm heat pipes to a copper fan shroud.

This is one of my uglier mods as I modified a previous custom heatsink I had, but despite how they look I managed not to pop or kink any heat pipes.

I have to say the hardest part in these mods is bending heat pipes without kinking them. I'll clamp the heat pipe in a vice, bend it a little, reflatten, then bend a little more, reflatten, etc until I have the bend I want (or something good enough).

Internal by Khenglish posted Nov 24, 2017 at 1:10 AM

 

Jesus that is one heavy mod. Why the copper plate on the CPU tho?

 

THANK YOU @Khenglish for sharing what essentially amounts to trade secrets with myself and others. This is exactly the kind of response I was interested in hearing. I really appreciate the info coming from someone with first hand experience. You are my heatsink hero and I really admire your amazing work there.

This is my kind of pr0n, considering I love to weld and do custom metal work (primarily in relation to bicycles, but I do other fabrication projects as well). Do you use torch, heat gun, or oven to heat your work prior to soldering?

 

Ok I'm a noob but I have to ask.

Can these "heat springs" be used on our own laptops, like the BGA turdbooks like the MSI GT73VR?

If someone like me were crazy to buy something like this, how exactly is it installed?
Do you cut it to shape to match the BGA silicon slug? Or a larger cut (taking care to use proper SMD component insulation as always)?

And still the question no one has answered:
What is the clamping force used in BGA turdbooks like the MSI GT73VR? And on LGA superbooks like the MSI 16L13 and Clevo P870 DM3/TM1?

I have a GT73 so I'd like to know, and I am not interested in spending $300 on a piece of fancy technology and wasting it unless I am 100% sure it willi work.

tl;dr; need MSI clamping force pl0x

 

Just to clarify, that $300 gets you 20 heat springs. So by the piece they are $15/each which is why I suggested a group purchase, haha. If you check out all the links I have inserted as well as the links available from Indium Corp's heat spring webpage, you will see it is essentially like a square of foil (NOT SURE if it has an actual adhesive or not) that is very soft and compressible. They can be bought in different metallurgical compositions and thicknesses which dictate their melting points and necessary clamping forces respectively.

It is basically like a non-liquid liquid metal application, but thinner and solid rather than liquid, which amounts to better heat transfer properties. If you read through the numerous webpages and technical details/documents on their Heat Spring webpage, it actually mentions that heat is primarily transferred through crystal lattice vibrations, hence the reason a square of (special metallurgic composition) soft solid foil can actually outperform a liquid metal (which should technically have better gap-filling coverage and ability than a solid).

The square heat springs are very soft, there is a video of them cutting them with scissors and using razors to precisely chop them to size.

 

Thermal pads have thickness many times greater than a light coating of TIM that we use on CPU / GPU's, and it's likely the designs wouldn't allow for a thick pad between CPU / GPU and heatplate.

Same goes for the IHS, a thermal pad would be too thick.

It's something to try, there may be mechanical designs that wouldn't pop from the bending stress when fastening it down.

Coollaboratory already has a non-liquid pad, which I've seen some report using successfully, and IDK why more don't use it as it's safer to apply in solid form:

Coollaboratory Liquid MetalPad
http://www.coollaboratory.com/product/coollaboratory-liquid-metalpad/

I don't see a thickness measurement given, but it's got to be very thin to use for CPU / GPU TIM replacement. IDK how it would be for IHS, you'd probably need to cut it down to fit.

It's also inexpensive, less than $10 for 1 CPU application:
https://www.newegg.com/Product/Prod...Model_COOLLAB1xCPU-_-9SIA1K65259564-_-Product

Or, a 3 pack:
https://www.newegg.com/Product/Prod...oollaboratory+Liquid+MetalPad&N=-1&isNodeId=1

Coollaboratory Liquid MetalPad Thermal Grease Pad for GPU & CPU 3-Pack - NEW
https://www.amazon.com/Coollaboratory-Liquid-MetalPad-Thermal-Grease/dp/B001CXNSU8

 

The Fujipoly pads are offered as thin as 0.5mm. If anything it seems like it would even double as a shim and increase contact pressure imo. Thanks for sharing the CLMP link. The Heat Springs are offered down to precise micron thicknesses (not sure what the inch or millimeter conversion is off the top of my head rn) so they should definitely be usable. Would love to see a comparison of Heat Spring vs CLMP in terms of performance, metallurgic composition, thermal properties, physical properties, etc. Thanks for sharing your input.

 

The pads are terrible because you have to burn them in at around 80c. Which is way more risky than a simply liquid metal application. Also due to the fact that there is mixed other metals in it etc. makes them only around 10WK

 

Fujipoly XR-m pads = 17 W/mK . Specs are specs. Burning in at 80C isn't a big deal on laptop hardware, especially if it is a high performance or gaming laptop like they would likely be used on.

 

Was refering to the metal pads from coolaboratory.

 

You'd do better to research your comments before posting them.

Installation of the Liquid MetalPads
Version 1.10 English
http://www.coollaboratory.com/pdf/manual_liquid_metalpad_englisch.pdf

"The Coollaboratory Liquid MetalPad consists of only metal and no non-metallic additives. The situation when the Liquid MetalPad changes’ (melting) is at approximately 58°C from solid to liquid. When you are using the revolutionary Liquid MetalPad, it has to melt only once with a “BurnIn” process to achieve full performance.
...
You should have now better temperatures (averaged 4-7°C) with the Liquid MetalPad compared to other heat-conductive pastes, as example silicone pastes. Please note that you will get no or only marginal changes compared to the “Coollaboratory Liquid Pro”."

So you were wrong on both guesses.

On top of that they are made of thin foil metal material, MetalPad is just the name:

A little tricky to use since it looks like foil, keeping it from blowing away while I ...
By Anonymiss on May 1, 2015
Verified Purchase
"A little tricky to use since it looks like foil, keeping it from blowing away while I try to put it on the heatsink was an issue. I had to hold my breath...lol. Would recommend the thermal liquid for a standard desktop processor but these are ideal for systems that are considered to be locked down and un-upgradeable. Such as a laptop, video card, and video game console. The MetalPad lowered the temperature on my laptop quite a bit. I have an A10 socket for a laptop and I do some minor games on it but it was getting up into the 90 celsius which scared me quite a bit. Now it's down to about 62 celsius on a full load. It saved my laptop! It could probably save yours too, the heat transfer is amazing. Just remember you have to raise the temperature to FUSE the MetalPad to the heatsink to around 58 celsius."

 

I was curious why @Danishblunt was suggesting thermal pads had to be burned in. I was assuming he knew something I didn't. My rebuttal still stands that 17W/mK > 10W/mK. The confusion is coming from calling CLMP "pads". I would call it a MTIM (metallic TIM) or FTIM (foil TIM) just like IC's Heat Springs are MTIMs/FTIMs.

The CLMP cannot be compared 1:1 to IC's MTIMs because their melting points are widely dissimilar. The CLMP is basically just an easier and more foolproof way of applying liquid metal. The IC MTIM (AKA: Heat Springs) operate on a completely different metallic composition and theory and are meant to stay solid.

The burn-in that is referred to is literally just heating your system up enough to melt the solder into a liquid so that it flows and disperses between die and heat spreader. You could achieve the "burn-in" with a heat gun all the same.

 

The CLP is 80W / mK, and although I can't find a claim for the CL Metalpad (foil), the text indicates it's "the same" or close enough to CLP:

Additional Information:
http://www.coollaboratory.com/product/coollaboratory-liquid-pro/

Thermal conductivity (CLP)
80 W/mK

And, in the CL Metalpad Installation manual:
http://www.coollaboratory.com/pdf/manual_liquid_metalpad_englisch.pdf

"You should have now better temperatures (averaged 4-7°C) with the Liquid MetalPad compared to other heat-conductive pastes, as example silicone pastes. Please note that you will get no or only marginal changes compared to the “Coollaboratory Liquid Pro”."

So, it looks like the CL MetalPad is better than the best Fujipoly thermal pads?

 

The Coollaboratory Liquid Metal Pad thing never worked well for me. Even using Kryonaut or Phobya NanoGrease Extreme was as much or maybe even a little more effective. I tried several times and it just wasn't anywhere close to as good as CLU or Conductonaut. The metal pad never fully melted no matter how long I did the burn in. Maybe the heat springs would be better. Hard to say without trying it, and way too expensive to even consider it without a group purchase.

I also tried Indigo Extreme four times and it never worked right. But, that was on a laptop. Maybe not enough contact pressure for it to flow correctly. I am going to try the Indigo again on my desktop.

 

A couple of things.

Why go to the trouble of a solder when you can use conductonaut that's a similarly high thermal conductivity (73W/mK) and it's far easier to use than thermal paste - I have reused it time and again without needing to clean and reapply.

I have been looking into the clamping force issue recently after Gamers Nexus did an article on comparing it across a variety of RX Vegas but can't find a source of non carbon pressure sensitive film that's not a huge roll of the stuff for $100s - any help here from anyone would be much appreciated

Nobody round here uses thermal pads on cores (other than experimenting) since if you have a 0.5mm gap between heatsink and core the usual approach is to either get rid of the gap or put a shim of 400+W/mK copper in there. The pad would also be sensitive to the evenness of the gap over the contact area (sorry I'm not an engineer so don't know the technical term) since the fujipoly extreme is not very compressible and paste doesn't have this issue. Many ppl use the 11 and 14W/mK pad options because they are more durable and less sensitive to fit issues. Specs aren't always the only consideration - getting the m bit of W/mK as small as possible also improves TC

Lastly I have been thinking about fiddling with a diy hybrid WC setup since I first saw the Asetek M18x years ago, and will soon get started on it after accumulating parts and learning about open loop watercooling over the past year. Starting with a test of how much heat a diy hybrid watercooling setup can remove from a GPU heatsink when restricted by a 1/8" airgun quick disconnect and a small loop of 6mm copper tube for contact. I was going to use low temp solder paste and a heat gun because that will be easier to reverse than thermal epoxy.

 

Indium Corporation's Heat Springs are rated at 86 W/mK so they would technically edge out CL FTIMs a bit further. The difference here is in the transparency, marketing, and target buyers mostly. CL is focused on end-users where IC is focused on other corporations, researchers, and big manufacturers mostly I think. I admire that IC is so transparent with all their data and still willing to sell to end users though.

 

We still need to know the clamping force on the laptops. These pads require 40 PSI.

 

Similar reports to your's in the customer comments section, along with a larger number of success reports. It's likely down to the fitment of the heatplate on the CPU / GPU, as it is for the liquid CLP - it needs a good fit to make contact.

The nice thing about the Metalpad instead of the liquid, it's not going to flick globs of doom all over the interior of your laptop, the foil is a safer application method.

For cases where CLP works, the Metalpad is worth trying too.

 

Many users have reported for it to properly melt and perform you need to burn in at 80c. It's the typical "read the fine print". It doesn't state that it will melt completely at 58c, it says when the Liquid MetalPad changes’ (melting) is at approximately 58°C. Also Never said you have to repeat the burnin, I mean you only burn something in once...Duh.

here the performance of the metal pad:
http://pclab.pl/art25661-3.html

There are countless users who have tested the metal pads and only saw a slight improvement over AS5, which is being beaten by almost all modern tims, so yeah, the metal pad are not even close to liquid metal. There is a reason why metal pads aren't a thing.

What are you talking about? I already stated that I was talking about the metal pads, I was not even talking about your other tim, I don't know it and neither did I find any useful reviews or experiences from users, so I wouldn't know if they need a burn in like coolaboratories metal pads or not.

 

The Coollaboratory Liquid MetalPad Installation instructions caution against using a heatgun for water-cooling applications, and instead to disconnect the pump for a brief "heat up to temperature", if it's possible. Otherwise it's going to not work in that application.

"Beside the presented varieties of installation there are many other possibilities, to melt the Liquid MetalPad between for example the processor and the cooler. Very effective is the use of hot-air gun/ hair-dryer, which heat the cooler accurately. Please note that you regulate this according to the capacity of the unit and that you only heat precise the cooling element. This method works especially at air cooler with fins / heat-pipe function. A short-term heating of the cooler up to approx. 60°C leads to the wanted result. Please note that it is advised against using hot-air guns with water cooling. Alternative methods such as these are at your own risk."
...There are more detailed installation description in the instructions:
http://www.coollaboratory.com/pdf/manual_liquid_metalpad_englisch.pdf

 

A small square isnt too hard to machine 0.5mm square in the middle with the right tooling. I know this is far beyond the scope of most extreme modders even, but it's the areas I like to dabble in myself. Sounds like you have an idea similar to my detachable laptop water cooling concept I mentioned in the first post. I am really glad to see this discussion back and forth happening.

If you look at the Solder Alloy Comparison Table (and other sources) you will see the Solidous/Liquidous temperatures for the various alloys. There is indeed a plastic state in between solid and liquid where not all material will be completely solid or completely liquid. I find your brushing off this topic without any actual experimentation or first hand experience contemptuous. I'm not here to hear naysayyers nay. That happens all the time and they have been proven wrong more than once. If that was the case, I don't think Indium Corporation would be in business right now.

 

The reason they say that is because they don't want to be responsible for an idiot user overheating his waterblock and cooking the orings or other materials it interfaces with that could cause leaks and/or damage to the PCBs and other components nearby.

 

Oh, I don't disagree. Almost everything is worth trying in my book. That's the only way to figure out what works best. If you don't try something, you'll never know how good (or crappy) it might be.

Laptops have always had an issue with heat sink contact being poor as well as low pressure. Even the best fitting heat sinks on the best built laptops have relatively low contact pressure. I can tighten my desktop heat sink/water block down tight enough to break something if I wanted to, so contact pressure should not be an issue. It might even need to be tightened a bit further as the Indigo flows, because the end of the package where the liquid metal is stored is pretty darned thick.

Every time I tried the Indigo on a laptop, it only flowed 1/3 to 1/2 way across the IHS and stopped flowing. Where it stopped was still a thick layer, so that seems to suggest not high enough pressure. On a laptop, you tighten down the heat sink until the screws stop turning, and the only way to get it tighter is to add washers to the screws to bottom out the coils on the springs, but that is also good way to break the CPU or socket, or fracture the solder on a BGA CPU.

 

"Countless users" sounds like unwarranted hyperbole, given the user reports I am seeing that show a high % of success.

It's not just Metalpads, or CLP, or Thermal Grizzly with installation issues, there are always reports of people not getting any particular product to work well for them, and these reports need to be heeded in that you'd better figure out how to do it right, or you'll have the same problems getting results.

Failure reports are offset by reports from those that had successful installations. Learn from them what not to do, and what to do.

It's unsafe to use liquid metal, or any liquid electrically conductive material, so the solid at installation Metalpad is the only safe choice.

I've had lots of very successful low temperature AS5 installs, but I've done a lot of them, and have a technique learned over time that gets very good results.

AS5 is still viable today as an inexpensive TIM for great results. I prefer using it over using conductive pastes, as I don't see the need to risk hardware for a few degree's of improvement. The results I get with non-conductive pastes put temperatures way below thermal throttling, which is all I need. Some installations I am still in contact with are still running after 7+ years with no rise in temperature.

If the Metalpad is better than AS5, it's worth a try, if you want to spend more money for a few degree's better results.

Kryonaut is also a good choice, a bit more expensive than AS5, and is also non-conductive:
http://forum.notebookreview.com/threads/thermal-grizzly-kryonaut.790919/page-16#post-10642351

Both AS5 and Kryonaut need about 24 hours of at temperature use to attain best results, so don't panic if the brand new application of either is running hotter than expected, game for a few days and keep the temperatures up, and give the curing process time.

 

Contact pressure aside, the reason for the flow not happening all the way to the edges is likely related to the point of thermal equilibrium being reached for your given hardware's heat dissipation requirements. What you are explaining actually makes sense to me if you have a well-executed water cooled system. I'm not sure how to explain this the best, but as the metal pad melts and spreads out and increases its surface area, its cooling ability for your system would go up until it is pulling enough heat off your system to lower the total sink surface heat level to a point below the foil pad's melting point, if that makes sense. At least that's my take on it. I'm not sure if once it melts once then re-solidifies how that would affect its melting point in the future or if it would be as easy to remelt in the future. I anticipate if it was used on a non-water cooled system you would likely get a better spread when it melts because of the additional heat stress the system would maintain as the metal pad begins to flow.

 

Not following directions is seldom a good approach, and good results are seldom achieved flying by the seat of the pants. Following directions is really tough for some people. Experimenting is OK to see if something works better if following the directions does not work, but following them first is the best place to start. If one does not follow the directions, then they have no business being critical of a product that does not work for them.

With the Coollaboratory Liquid Metal Pads and Indigo Extreme stuff the directions were followed meticulously, more than once. So, it has to be dependent on proper contact, sufficient pressure, and probably both of those conditions need to be correct to achieve the desired outcome.

I have never tried it on a water cooled desktop. Where I tried it and the Indigo flow never completed (4 times on two systems) was an air-cooled laptop with a desktop CPU. It got hot enough (and then some) for much, much, longer than necessary. I never saw the peak and rapid drop of the temps as the instructions explained would happen just because it never flowed far enough. The only thing I can identify is it ran out of pressure because the contact was too weak to facilitate it, and gravity alone clearly wasn't enough.

 

I do have experience in this, that's why I know that until a certain point it really doesn't matter how good your tim is. If it was as simple as applying new high end tim and your temps would magically drop, then people would not struggle having high end hardware with high temps. There are way more factors to cooling than tim.

When you look at kenglishe's picture, you will notice that he drastically increased the size of the heatsink. He made the GPU plate way bigger, connecting it with his own copper Fan (yes he made the fan conduct heat) and then added more heatpipes to transfer heat that way. I tried Collaboratory liquid ultra and even conductonaut, only to see that they perform about the same as IC Diamond 24 and Kryonaut on a 4940MX @ 4.4ghz on a P157SM. However, when adding another fan and put a very small desktop heatsink for experiment on top of the heatpipes on the CPU, the temps would drop massively, because I increased airflow and mass.

If it truly was so much better to put proper and good tim on your notebook and if it would truly be so much better, then it would be a thing. It would be offered as an expensive service, but it is not a thing. What I find weird is that you are sitting here, claiming I don't have any first hand experience when modding my cooling, after writing in your first post admitting that you don't have experience yourself.

All I'm doing is showcasing why some solutions like the metal pad aren't as great as the sheet @hmscott posted. I can't explain why the metal pads are only a little bit better than AS5 for instance. If you want to know why it isn't as effective, maybe see if you can find a physicist. All I do is look up for facts, see reviews and user experience and sum it up.

I am not interested in what I want or what I don't want, I am only interested in facts.

 

I can easily post forum topics etc. but how about we take the benchmark for now, where is your benchmark showcasing it's much better? I mean I provided some links showing how it doesn't perform evven remotely on liquid metal levels alongside with my statement and mr.fox multiple experience with them himself, I've yet to see a single link from you proving me wrong.

 

Yes, there is no doubt about it, the contact pressure is definitely important as it is directly related to the heat being put through the interface materials of the cooling system. I think of the contact pressure as a constant scalar/multiplier myself. It has its limitations of adjustment though as mentioned prior, about cracking solder joints or CPU dies.

 

Ok guys I'm sorry for being the dumb person here, seriously, but should I even try to buy these pads for a GT73VR?

@Maleko48 @Mr. Fox @Volcano_God_In_The_Sky ? anyone?

 

Your test example was against AS5 and some unnamed original paste, and not in the same application against CLP, you can't draw conclusions from other tests with CLP against this comparison.

I don't have enough time available or interest in continuing another pointless long never ending exchange with you, so be happy with what you have, and do what you want.

 

You're not dumb. You're pretty smart, dude. If you want validation... Sure you should. There is no legitimate reason not to try them if you can acquire some of the material for a reasonable price. They could work extremely well, but if you don't try them you will never know.

 

@Mr. Fox I have the money, but I wanted to save it for the P870 TM-G power system with a prema mod. I really am impressed by the cooling in it and it SEEMS Like they learned their lesson from the KM1 radiator gap debacle. If they infact did, and all that is needed is a Prema Bios to get what you truly want, then that would finally be a decent DTR that didn't need all of those difficult mods you had to do yourself, to get it usable. I was just considering if these indium pads would actually work on a BGA system, since I was sort of impressed I was even able to run SuperPi 1M on my 7820HK trashbook @ 5 ghz.

And then there's having to decide if, or when I should upgrade my desktop 2600k to either a 7920X or wait for the 9700K. Moar Corez, right?

 

I won't deny some of the points you make here to be valid, but I also won't take them as an end-all either. Some of your responses are misleading or not detailed and thorough enough which leads to confusion and mis-speaking about irrelevant facts that aren't on the same level in terms of comparison. I believe there are always ways to improve things in areas people might not necessarily think there is that possibility, that's just the way I am and I enjoy tinkering in these areas. All of this theory aside, it is obvious that less extreme mods can accomplish more than enough reduction in temps and there are also hard limits involved in all of this modding, due to sheer physics- I don't deny that either.

Unfortunately most of this conversation has gone the way of Coollabratory's Metal Pads which are essentially supposed to be easy-to-apply solder. It's the same as Indium Corporation's solder ribbon basically. You could buy that and cut it to size and fire up furmark to burn it in and it would likely flow and act the same as Coollabratory's Metal Pads.

TBH, I am more interested in the Heat Springs which are intended to stay solid rather than melt. There is no legitimate comparison available on this front right now due to lack of competition. Unfortunately they are difficult-to-acquire at a non-cost-prohibitive price point right now. I believe in the new-age technology Indium Corp is developing though and can't wait to see more results and examples of it being used in the wild. There is something to be said of a company that is transparent with all of their research and data and still moving forward with new developments despite that others may think it is a moot point to develop in those areas due to the status quo having a stranglehold on the market.

 

Sounds like a good plan to me, bro. Save the money for that instead.

 

I try to mostly use an oven, but that GPU heatsink was too complicated to have everything clamped in place to heat the whole unit. I did the GPU core section in the oven, then when I added the memory and FET cooling sections I used both a heat gun and soldering iron simultaneously. I'd have just the pipe I was working on, and maybe a nearby one clamped down, and the rest of the heatsink would not get hot enough to require being secured.

I switched out one of the 6mm CPU heatpipes for a 7mm pipe I found. I had the pipes and contact plate pressed together in a vice and I was worried that the solder paste would flow up to the vice and solder the vice onto the heat sink, so I added the copper flashing to block the solder. Later on I just used wood. Wood smoulders a little at those temps, but it doesn't flash burn (a heat gun will make it flash burn though. wood is only OK in an oven).

 

Yeah, I have always thought that was kind of an interesting product as well. It was just too expensive to purchase at full cost as just one person on faith that it would be worth such a gigantic wad of cash. A group buy to make the cost minimal is the only way to do it. Now, if it worked better than anything else it would be worth buying it and having enough to do a lot of systems.

 

I have found liquid metal to improve temps over conventional pastes by around 4C. What Coolaboratory provides to spread the metal is trash, which I think is why many users do not get good results. I put a dot in the middle of the die, then wrap my finger in plastic wrap and spread the metal with my finger. This works drastically better than their brush and q-tip. There is definitely a point when thermal contact gets as good as it can though, and the rest of the cooling system will need to be improved.

 

My original thoughts were that most laptop manufacturers likely use the minimum size possible of heat pipes to save as much cost as they can, but that in most cases there is the ability to step the size or thickness up a mil or two and still fit in the same space constraints, but with better performance. Obviously that would require a resolder job, but the idea there would be for it to remain in an identical configuration to the original installation. There could be additional gains as well from using a solder with superior thermal pass-through conductivity compared to the stuff the factory uses as well.

There are also the more extreme variations which reconfigure or add to the existing system which I am just as interested in as well.

@Khenglish have you experimented with an arduino electric heating circuit? Basically setup a short circuit by using two leads clamped or attached to the specific section of heat pipe you need heated and use a relay to switch on and off the pulses while simultaneously monitoring the surface temp with a thermocouple / thermistor which feeds back to the Arduino (or CLICK PLC) to know when to stop based on a set temperature target? I know in San Francisco, refrigerator repairs that involve soldering/brazing aren't supposed to use torches but actually use electric heating systems based on shorting a circuit until the metals are cherry red hot, so the arduino thing is basically a scaled down concept of that. A converted Microwave Oven Transformer also makes a good spot welder which is another method or option depending how you want to use it or implement it.

Also, do you heat the pieces then feed the solder or do you clamp the solder between the surfaces and let it flow as the whole work heats up, then adjust and tweak the angles of everything before allowing it to cool? Sounds like there needs to be a purpose-built heat-pipe roller or jig setup. Could use a couple different sized bearings and some fabricated steel to make one up... hmmmmm.

 

Yes you usually can remove and replace the manufacturer heat pipes with larger ones and get a temp improvement. CPUs usually are very adequate on high end systems, but GPUs can use better heat transfer.

I've never done anything with a microcontroller and heating.

Whenever possible I have solder paste or a thin layer of hardened solder + flux between the pipe and the contact plate prior to heating. This is necessary for high heat components like the core, or else there will be no solder between the plate and pipe. For lower heat components like memory though just having solder on the side of the pipe is fine.

 

Heat-Spring?

 

Excellent addition to this thread @yrekabakery THANK YOU! I know Indium Corp has been around for a while now, but I totally missed this thread when searching for thermal threads. Perfect addition and cool to see how far back it goes.

EDIT: It looks like the thread has since been closed and idk if I have seen any final consensus from my skimming of it. There are definitely good bits and pieces of info throughout though.

 

I won't deny that I am sometimes not detailed enough, fair point.
Also it's fine that you want to "think of a new way" but if you want to go that route, then don't stay stuck on something that has been chewed over and over again. If you want to find a solution that will give you good results, then you need to think out of the box, try to find variables nobody has thought about. Don't think about mass of the heatsink, tim or airflow, but other factors that can be helpful. You need to realize that until a certain point, tim won't do you any good anymore.

Sadly it isn't as simple as just look at the K value of the tim to find out which one is best, there are way more factors such as thermal resistance between the die and tim, bondline thickness and bonding area dimensions. This is also the reason why IC Diamond performs better than Promalitech TK-3 despite having lower K value.

If you find something new and interesting, which could finally fix this dilemma with desktop CPU's in notebook finally running at acceptable temps, then I'm ofc all ears. The thing that Heat Springs are advertising is also that they have much less heat resistance btw.

 

Yeah I have a P157SM which is basicially the same as you have cooling solution wise. I tried a fairly thick conduconaut layer but it was exacly the same as IC Diamong 24 tempwise. It seems that for you LM does improve your temps a bit, did you mod your CPU cooling in any way other than the thicker heatpipe?