The Official Thermal Interface Material thread

[KLiKzg]

I know. Quotes because I know 100C isn't high enough, but I think it's more about the pressure and the repeated climb to 100C plus all the other little factors that contributes to a slow change in shape. Or...It isn't the CPU at all but, rather, the ILM that becomes deformed; I think that far more likely, especially given the reports that BCFs fix the problems.

It really is an odd thing, but computers are odd indeed.

I don't know if they still do it but, back in the 80's, some computer manuals stated the THEORY of how they believed that the computer would function.

Not really an odd thing, as CPUs need to e put in very precise place...measurement for motherboard placing of the components, including ILM & BCF are in 0,01mm regulated. That is high precision!

More likely is that Cu on IHS bends concave, because of its shape. Which is not so good for cooling after molding, but needs to be like that.
It could be that the top can be machined after molding, but that would increase the price of the IHS.
Then, there is also the tool (machine) accuracy...which can be better or worse, depending of the assembly of the machine & maintenance...etc.

Now that the CPUs are so delicate & needs to much power (& cooling also), there needs to be some other way of attaching & changing of CPUs...& update of complete package! But it is on people in IT industry, to lead with new standards...

 

[lexluthermiester]

But, wait, what CPUs were they? Mine is a 12700K that is not very hot.

The testing I did was a Xeon W3680 overclocked to 4ghz for a thread here at TPU. The rest of the systems with various CPUs all at stock or close to stock settings.

 

[TPCEA]

Peckers, aside. If you will!

Not really an odd thing

Sorry for not being clear enough. I was referring to the problem, not the ILM or BCF, since the temps aren't actually high enough to truly impact most of the CPU, let alone cause deformation of the copper.

The testing I did was a Xeon W3680 overclocked to 4ghz for a thread here at TPU. The rest of the systems with various CPUs all at stock or close to stock settings.

So it's not even that hot of a CPU, right? Interesting. Do you think that maybe your tube of H1 was out of tolerance, compromised or expired?

 

[lexluthermiester]

So it's not even that hot of a CPU, right? Interesting.

Overclocked at full load it can be. At 4GHZ the W3680 tops out at 185w of power draw and likely a similar amount in heat.

 

[Shrek]

This paste was used in for 1MW device, from my previous company...it has proven to be OK with heat dissipation in automotive industry!

Regenerative braking could hit the 1000 HP mark

 

[KLiKzg]

Sorry for not being clear enough. I was referring to the problem, not the ILM or BCF, since the temps aren't actually high enough to truly impact most of the CPU, let alone cause deformation of the copper.

Again, problem is not the edges on CPU IHS, problem is dipping of the top surface of the CPU IHS.

That is caused by other effect (mainly productivity & cost) & that is why we speak here about TIM (& why we need TIM).

 

[TPCEA]

Again, problem is not the edges on CPU IHS, problem is dipping of the top surface of the CPU IHS.

I understand that and did not specify the edges of the CPU. As I said before, maybe the actual problem is the ILM warping under the uneven pressure.

That is caused by other effect (mainly productivity & cost) & that is why we speak here about TIM (& why we need TIM).

If you believe that the IHS is deforming, I'd love to know why you think it happens.

Also, given the need for a BCF on some gen 12-14 that already use even top TIMs, I don't see how TIMs are going to resolve the problem once the gap exceeds the efficacious maximum depth of a TIM, which is why BCFs were made. Can you clarify?

 

[KLiKzg]

I understand that and did not specify the edges of the CPU. As I said before, maybe the actual problem is the ILM warping under the uneven pressure.

Not sure, as steel has so much more strength, then Si chip, plastics or Cu. It is deforming, as everything deforms with force. But its deformation is so low & well within 0,2% of elastic deformation.

If you believe that the IHS is deforming, I'd love to know why you think it happens.

I do not think IHS is deforming. But I do think that there are "manufacturing defects" present, mainly because of cost of manufacturing.

Also, given the need for a BCF on some gen 12-14 that already use even top TIMs, I don't see how TIMs are going to resolve the problem once the gap exceeds the efficacious maximum depth of a TIM, which is why BCFs were made. Can you clarify?

In order for TIM to be as effective as can be, it is better that it is thinner. So why do you ask for thicker TIM?
Of course, the thinnest possible also means that it would not separate between CPU IHS & cooler...

 

[TPCEA]

Not sure, as steel has so much more strength, then Si chip, plastics or Cu. It is deforming, as everything deforms with force. But its deformation is so low & well within 0,2% of elastic deformation.

If that is the case, then how is it that a BCF can fix the problem?

I do not think IHS is deforming. But I do think that there are "manufacturing defects" present, mainly because of cost of manufacturing.

problem is dipping of the top surface of the CPU IHS.

By dipping and defects, it seems that you are stating that this is a problem caused prior to use. If that is the case, and considering your statements against the bending of the IHS and ILM, that again raises my above question about why a BCF has been proven to help. Please explain.

In order for TIM to be as effective as can be, it is better that it is thinner. So why do you ask for thicker TIM?
Of course, the thinnest possible also means that it would not separate between CPU IHS & cooler...

Of course thinner is better; that is already well-established! My statement was based on the idea of the bending that we have been discussing. As the surface deforms, the gap between it and the cold plate increases, thus the NEED for a thicker layer of TIM. That was my point. We are having communication issues; I'm sorry.

In reference to your last sentence, what do you mean by separate, since the subject of your sentence is TIM? Do you mean pump out?

 

[ShrimpBrime]

If that is the case, then how is it that a BCF can fix the problem?


By dipping and defects, it seems that you are stating that this is a problem caused prior to use. If that is the case, and considering your statements against the bending of the IHS and ILM, that again raises my above question about why a BCF has been proven to help. Please explain.


Of course thinner is better; that is already well-established! My statement was based on the idea of the bending that we have been discussing. As the surface deforms, the gap between it and the cold plate increases, thus the NEED for a thicker layer of TIM. That was my point. We are having communication issues; I'm sorry.

In reference to your last sentence, what do you mean by separate, since the subject of your sentence is TIM? Do you mean pump out?

Super interesting that the idea an IHS plate bends. I doubt this honestly.
Can't say much about the coolers however, I'd think heat pipes would be prone to bend before the IHS plate.
Anything easier to bend than the IHS plate will bend first.

The BCF plate is to prevent the motherboard from warping, not the IHS plate.
Which has been a known thing since LGA #1 that the board can warp enough the pins loose contact and can result in issues even no post.

Using TEC formula for mounting pressures which range 75 to 200psi of clamping force.

TEC Installation | Install | Tech Info | Thermoelectric Cooling | Peltier Coolers | TEG Water Blocks TEC

Shop Custom Thermoelectric - We sell peltier cooler, seebeck effect and thermoelectric cooling that generate electricity from waste heat. TEG water blocks, & TEC.

customthermoelectric.com

A single 40x40mm TEC with two # 6-32 machine screws.
Screw diameter = 0.138 inches (from the table above)
Pressure = 125 pounds per square inch (PSI)
TEC surface area » 40mm = 1.575" so 1.575 x 1.575 = 2.481 square inches
So, (0.2) X (0.138) X (125) X (2.481) / (2) = 4.28 in-lbs.

This formula should be adequate for processor cooler mounting while a TEC and a CPU can both generate similar amounts of thermal energy.

 

[Noci]

Super interesting that the idea an IHS plate bends. I doubt this honestly.
Can't say much about the coolers however, I'd think heat pipes would be prone to bend before the IHS plate.
Anything easier to bend than the IHS plate will bend first.

Exactly that .

For example, when the first reviews of the Alphacool Core 1 came out showing it's outstanding performance but at the same time also the concerns of some regarding the very thin cold plate, my first thought was there must be some interaction/relation between these factors.

So my thoughts from a metallurgic point of view went something like this and untill proven otherwise still stand;

Like an IHS either being convec or concave being the rigid part, the ductility of the copper coldplate combined with the mounting pressure, forces it to form to a near perfect fit provisioning conditions for a very good heat transfer. (But only in this case, regarding the Alphacool Core 1 with a very thin coldplate, thicker coldplates my act different)

Anyone who wants to share his/hers ideas on this? Let yourself go... .

 

[KLiKzg]

If that is the case, then how is it that a BCF can fix the problem?

Because of 2 things:

• stacking tolerances
• manufacturing deficiencies

Both of which, give out some loose ends...so the thing is not firmly in place.

But even that, is not the issue, for any of your theories...even without those, we would need TIM...otherwise the surfaces would not touch perfectly between CPU IHS & cooler...so there you not be heat transmission, only emission & receiving of heat...so TIM is here to fill out the gap, as a medium for heat transfer.

By dipping and defects, it seems that you are stating that this is a problem caused prior to use. If that is the case, and considering your statements against the bending of the IHS and ILM, that again raises my above question about why a BCF has been proven to help. Please explain.

BCF helps....but the Si motherboard is not of same thickness, it varies a little...so does any manufacturing part we make.
Get on top of that force between BCF & ILS, which deforms the CPU (with its IHS) & motherboard a little.
Expecting to have everything as in place & to lean one surface to the other - impossible!

CPU IHS will always have defects & if you measure it on a fine scale...it would show how smooth & straight is the surface...you would be amazed!

Anyhow, to cope with all that deficiencies in manufacturing & assembly (or better or worse sort of reading audiences here), there is TIM....a medium to transfer heat & not to conduct electricity. A perfect solution for imperfections in the world (on a micro-scale).

Of course thinner is better; that is already well-established! My statement was based on the idea of the bending that we have been discussing. As the surface deforms, the gap between it and the cold plate increases, thus the NEED for a thicker layer of TIM. That was my point. We are having communication issues; I'm sorry.

In reference to your last sentence, what do you mean by separate, since the subject of your sentence is TIM? Do you mean pump out?

If you have some irregularities on BCF & ILS, as well se some incompatibility with you CPU IHS - yes, there might be a need to put more TIM...

But, as previously mentioned, TIM as a paste & thermal pad, is good transmitter. But not perfect. & compared to Cu-Ni CPU IHS & Cu cooler, TIM acts as a barrier!

If there is separation of a layer of TIM between CPU IHS & cooler, then you do not have heat transmission anymore...you will get heat emission & receiving, which is on a magnitude lower then transmission.

 

[TPCEA]

Thank you all for your explanations! Very helpful, indeed!

Funny how none of the YTers I've seen who have promoted the BCF have mentioned that it's for the mobo. I wonder why?

 

[KLiKzg]

Thank you all for your explanations! Very helpful, indeed!

Funny how none of the YTers I've seen who have promoted the BCF have mentioned that it's for the mobo. I wonder why?

They either:

• don't know about it, as they lack the knowledge (or degree or both) of a subject that they are talking about...
• or they are too young to remember companies staring to put BCF on P4 & later processors, to cope the load from cooler on motherboard

 

[ShrimpBrime]

Thank you all for your explanations! Very helpful, indeed!

Funny how none of the YTers I've seen who have promoted the BCF have mentioned that it's for the mobo. I wonder why?

It's a "contact" bracket, or contact frame. And it only comes in contact with the motherboard.

According to a quick search asking "what does a contact frame do"

The search engine AI generative response is.
"A contact frame is a mounting aid that improves CPU cooler performance by optimizing contact pressure. It replaces the original Integrated Loading Mechanism (ILM) on the motherboard."

____

So if I'm not mistaken, this bracket magically makes cooler cool better than ever!! It doesn't even touch the cooler.

Here's a nifty video about installation. Notice the reported temps by the cooler. 1 minute and at 6 min 10 sec.ish
This poor guy went through all that trouble, basically for nothing.

 

[maxfly]

It's a "contact" bracket, or contact frame. And it only comes in contact with the motherboard.

According to a quick search asking "what does a contact frame do"

The search engine AI generative response is.
"A contact frame is a mounting aid that improves CPU cooler performance by optimizing contact pressure. It replaces the original Integrated Loading Mechanism (ILM) on the motherboard."

____

So if I'm not mistaken, this bracket magically makes cooler cool better than ever!! It doesn't even touch the cooler.

Here's a nifty video about installation. Notice the reported temps by the cooler. 1 minute and at 6 min 10 sec.ish
This poor guy went through all that trouble, basically for nothing.

Intel contact frames are largely used to keep z690/790 MBs from warping and twisting. Which is still a common problem with z690/790 ILM. It also helps with temps when mounted properly.

The AMD contact frame didn't/doesn't have any appreciable benefit other than adding rigidity.

 

[ShrimpBrime]

Intel contact frames are largely used to keep z690/790 MBs from warping and twisting. Which is still a common problem with z690/790 ILM. It also helps with temps when mounted properly.

The AMD contact frame didn't/doesn't have any appreciable benefit other than adding rigidity.

Most of this I had already said above.

""Helps temps when properly mounted""
You are talking about the cooler? I don't see any way to mess up installing an aftermarket contact bracket. It's 4 screws.

I tried looking for some videos showing end user experience. I found one where the temp difference was negligeable, within margin of error and or simply lowering room ambient temps by 5f or 2 to 3c.
Another person commented a drop of -5c. This seems like quite a bit of difference, but this is simply a statement.

Another little search, I found Neutronman, post 3, had tested a bunch and claims no real difference. Also states these plates are made from aluminum. The original bracket is stamped steel that's nickel plated. Which would be a stronger more rigid material. So the aluminum contact frame is more likely to bend than the original ILM.

Thermal Grizzly CPU Contact Frame(s)

www.overclock.net

 

[maxfly]

Most of this I had already said above.

""Helps temps when properly mounted""
You are talking about the cooler? I don't see any way to mess up installing an aftermarket contact bracket. It's 4 screws.

I tried looking for some videos showing end user experience. I found one where the temp difference was negligeable, within margin of error and or simply lowering room ambient temps by 5f or 2 to 3c.
Another person commented a drop of -5c. This seems like quite a bit of difference, but this is simply a statement.

Another little search, I found Neutronman, post 3, had tested a bunch and claims no real difference. Also states these plates are made from aluminum. The original bracket is stamped steel that's nickel plated. Which would be a stronger more rigid material. So the aluminum contact frame is more likely to bend than the original ILM.

Thermal Grizzly CPU Contact Frame(s)

www.overclock.net

Not true. The original ILM, PCB and backplate weren't designed correctly. The PCB itself on many MBs is simply too thin around the socket. They don't keep many MBs from warping. Which obviously effects bread board testing. My first z790 MB folded up like a taco when I mounted my peerless assassin on it with the stock backplate and ILM. I never had an opportunity to test temps as the board wouldn't lay flat on my test bench. Bought a $10 Thermalright frame and the problem was rectified. Point being, temps aren't the only reason the contact frames are used. Temps are what the toobers like to talk about, because clicks but that's not the original reason the contact frames were created.

 

[KLiKzg]

Not true. The original ILM, PCB and backplate weren't designed correctly. The PCB itself on many MBs is simply too thin around the socket. They don't keep many MBs from warping. Which obviously effects bread board testing. My first z790 MB folded up like a taco when I mounted my peerless assassin on it with the stock backplate and ILM. I never had an opportunity to test temps as the board wouldn't lay flat on my test bench. Bought a $10 Thermalright frame and the problem was rectified. Point being, temps aren't the only reason the contact frames are used. Temps are what the toobers like to talk about, because clicks but that's not the original reason the contact frames were created.

I would not trust Al frame so much...1/3 of tensile strength, ~1/3 the melting point, ~1/4 of the plastic deformation temperature (we in Croatia call it a flowing point) & has 2x the amount of temperature expansion on heat compared to a Steel.
Only thing better from Al is the thermal & electric conductivity. Which is why people see the drop in temperature (as some heat is dissipated that way).

All that says: do not trust Al parts, all over!

Reasons why YouTubers say that, is because they are paid to say positive review on the merchandise!

 

[ShrimpBrime]

Not true. The original ILM, PCB and backplate weren't designed correctly. The PCB itself on many MBs is simply too thin around the socket. They don't keep many MBs from warping. Which obviously effects bread board testing. My first z790 MB folded up like a taco when I mounted my peerless assassin on it with the stock backplate and ILM. I never had an opportunity to test temps as the board wouldn't lay flat on my test bench. Bought a $10 Thermalright frame and the problem was rectified. Point being, temps aren't the only reason the contact frames are used. Temps are what the toobers like to talk about, because clicks but that's not the original Board the contact frames were created.

Temps are not the reason. Board warps. Yes, I believe we are on the same page

 

[maxfly]

I would not trust Al frame so much...1/3 of tensile strength, ~1/3 the melting point, ~1/4 of the plastic deformation temperature (we in Croatia call it a flowing point) & has 2x the amount of temperature expansion on heat compared to a Steel.
Only thing better from Al is the thermal & electric conductivity. Which is why people see the drop in temperature (as some heat is dissipated that way).

All that says: do not trust Al parts, all over!

Reasons why YouTubers say that, is because they are paid to say positive review on the merchandise!

Between the steel back plate my block uses and the Thermalright frame, I ain't scurred I trust the frame a whole lot more than Intels janky ILM solution. I get the distinct impression you've never used one?

If a persons heatsink is allowing 2-12c (the reported averages)waste heat to seep down into the frame? Dey got prollems lol.

 

[KLiKzg]

Between the steel back plate my block uses and the Thermalright frame, I ain't scurred I trust the frame a whole lot more than Intels janky ILM solution. I get the distinct impression you've never used one?

If a persons heatsink is allowing 2-12c (the reported averages)waste heat to seep down into the frame? Dey got prollems lol.

You are right, you trust what you trust.

But don't "assume" too much, as it is with the word...without checking the facts (even right here on forum), you make "ASS of yoU & not ME"!

& well, again...you need to check how the cooler is positioned on the CPU IHS & ILM, to make that statement...maybe they use the cooler which is cooling (but also heating up) the ILM...& steel has less thermal conductivity hen Al, so over time - it gets heated up. But as you game along, it will cool down also...
My suggestion, as some experience as BOINC cruncher...stick to the facts, not YouTubers...talk to the mech.eng. about cooling. & from my experience, even though Al has a better thermal conductivity...my choice is Steel! Problem is & will be: putting too much weight on ILM, as you do not check facts about weight & cooling capacity of the cooler itself.

Now, even though this is topic about TIMs, lets get back to them. TIMs of all sizes, shapes, forms & conductivities.
& just to mention, those 2~4°C (od 2~12°C mentioned by you) change can be because of: cleaning of cooler & replacing TIM with a newer one! But I am open to hear what are the measurements...maybe Igor's lab would make another one of those measurement...

 

[maxfly]

You are right, you trust what you trust.

But don't "assume" too much, as it is with the word...without checking the facts (even right here on forum), you make "ASS of yoU & not ME"!

& well, again...you need to check how the cooler is positioned on the CPU IHS & ILM, to make that statement...maybe they use the cooler which is cooling (but also heating up) the ILM...& steel has less thermal conductivity hen Al, so over time - it gets heated up. But as you game along, it will cool down also...
My suggestion, as some experience as BOINC cruncher...stick to the facts, not YouTubers...talk to the mech.eng. about cooling. & from my experience, even though Al has a better thermal conductivity...my choice is Steel! Problem is & will be: putting too much weight on ILM, as you do not check facts about weight & cooling capacity of the cooler itself.

Now, even though this is topic about TIMs, lets get back to them. TIMs of all sizes, shapes, forms & conductivities.
& just to mention, those 2~4°C (od 2~12°C mentioned by you) change can be because of: cleaning of cooler & replacing TIM with a newer one! But I am open to hear what are the measurements...maybe Igor's lab would make another one of those measurement...

Nvm on all counts. Your not hearing what I'm saying.

 

[ShrimpBrime]

folded up like a taco

I wanted to share an experience from 15 years ago when I had first gotten into LN2 and extreme methods of cooling.
Had this Biostop board that was AM2 socket. Was horrible overclocker, thought to try LN2, I got the board real cheap like. Well even with LN2 the board was crap for OC, so after the disappointment, I poured the rest of the thermo cup (held about 16oz) straight onto the board after disassembly. It literally curled up like a taco shape. Like, no joking here. It was quite impressive! After a couple of days drying out, I decided to try the board. There was a lot noise and crackling sounds happening..... that Fkn thing actually posted right up! I couldn't believe it. I did recycle the board. It was no better than a paper weight anyways. Was Biostar TForce 550SE. I have no evidence of this board, the cpu-z says it was an asus M3A (Nov 10 2008). So I never did get a worthy submission on any cooling with it.

 

[maxfly]

I wanted to share an experience from 15 years ago when I had first gotten into LN2 and extreme methods of cooling.
Had this Biostop board that was AM2 socket. Was horrible overclocker, thought to try LN2, I got the board real cheap like. Well even with LN2 the board was crap for OC, so after the disappointment, I poured the rest of the thermo cup (held about 16oz) straight onto the board after disassembly. It literally curled up like a taco shape. Like, no joking here. It was quite impressive! After a couple of days drying out, I decided to try the board. There was a lot noise and crackling sounds happening..... that Fkn thing actually posted right up! I couldn't believe it. I did recycle the board. It was no better than a paper weight anyways. Was Biostar TForce 550SE. I have no evidence of this board, the cpu-z says it was an asus M3A (Nov 10 2008). So I never did get a worthy submission on any cooling with it.

Nice! Mine wasn't nearly that bad, it only rolled the ends of the MB up off of the test bench about an inch and a half. It was quite the sight.