"Pump-out Effect" Is it real or is it nonsense? Let's discuss...

[delshay]

@plastisch​

Can you point me to thermal paste that have 18W m/k?

 

[plastiscɧ]

@plastisch​

Can you point me to thermal paste that have 18W m/k?

i apologize i cant find it anymore in that short time window. the highest was 16 i could locate real quick. but i am pretty sure the was a 18 as well. i could swear...

BUT let me be clear: I NEVER tested them and i def. will not put my hand for these products into fire. Nore i will use paste ever again for my setup.

 

[Bones]

I guess I'll throw in and you guys know what I do. I'm always reTIM'ing my stuff anyway so I get to see what it does, both from things getting hot AND cold (Subzero) so if it has been seen before I certainly have.

The pump out effect is just that, it's an effect due to the thermal expansion and contraction of the components involved and depending on the conditions these pieces operate within that will determine what goes.
It's really nothing more than stating a natural effect of this repeated heating and cooling, when the pieces get warm they expand and want to squeeze to each other a little tighter displacing the TIM, when they cool they don't squeeze as hard so less TIM displacement happens - If they get cold thats when you see the opposite effect.
The extreme expression of it happens when things are subzero, in that event it's known as "Cracking" of the TIM. That's when the pieces get so cold they actually begin to pull away from each other and since the TIM is present (And frozen too) it becomes solid and can then be cracked by the force of this separation effect, leading to actual loss of cooling conduction.
XOC guys like me deal with this all the time BTW.

You must remember things like the CPU's lid/core and the cooler are solid, not liquid or even "Soft", the amount of movement from this thermal effect you can't perceieve yet it's there. As for the TIM, it's in a liquid or semi-liquid state so it "Flows" and that's what it's supposed to do by what it is and how it has to work.

Once TIM is present and the pieces have been installed with the expected mounting pressures the natural squeeze effect takes place, displacing some of the TIM to another spot. This means it moves along, either to just another spot or outside of the area where the cooler and CPU lid/core are together under pressure.

Since TIM is supposed to fill in the microscopic pits and valleys of these machined surfaces it will either find one to settle into or just keep moving along.

With each heating and cooling cycle it gets moved a little further, it can't move into a pit or valley that already has TIM settled in it largely because you can't compress a liquid/semi-liquid so it's forced to keep moving along, eventually finding a place it can settle into or winding up outside the surface area of the cooler and CPU lid/core.
It can displace what's there but you can never get more of it packed within the same spot period, something has to give or remain. If it's displaced that bit of it moves along too like the rest does.

What remains is just that, the TIM that did manage to settle into such a pit or valley and it will remain there, esp after the expected drying out takes place with that making it more of a solid than it was when first applied since when it's more of a solid so it doesn't get displaced readily if at all.

Obviously thicker TIMs will not move along as readily, thinner ones will in comparison.
The thinning that takes place from heat helps it to find and get into these places to fill in these gaps that exists on the surfaces of the cooler and CPU lid/core and it's supposed to do that. If there is no longer any spots it can settle into, it has no choice but to keep moving until it does wind up being outside the area the cooler and CPU lid/core "Meets" and you see it as TIM squeeze-out, hence the "Pump-Out" effect.

This can also be determined by just how flat these surfaces are to each other, the flatter/better they are to each other the more TIM you'll see getting outside the area. By the same token the less perfect or in this case, square to each other the less you'll have coming out.
Yes, it's possible for a cooler to not be perfectly square due to machining imperfections such as a wavy or slightly curved cooler surface for example - We've all seen it before to know it's a reality.

Another thing is in how flat/smooth these surfaces are when made in the first place.
Even with surfaces that are flat to each other, that in itself doesn't mean you don't have alot of places/spot it can settle into for holding it.
A more perfectly machined surface will have less spots (pit and valley's) so you'll have less TIM settling in since it's not needed in this case.

You also have to remember the thermal expansion and cooling of these pieces/surfaces doesn't stop, if it's being used you'll have that happening. Once enough has been "Pumped Out" with all the little pits and valley's filled, things will stabilize and this effect related to the TIM no longer happens.

Perhaps my explanation isn't perfect but that's what I've noticed over time and experience with it and believe me, I can go through some TIM in a very short amount of time with what I do.

 

[plastiscɧ]

I guess I'll throw in and you guys know what I do. I'm always reTIM'ing my stuff anyway so I get to see what it does, both from things getting hot AND cold (Subzero) so if it has been seen before I certainly have.

The pump out effect is just that, it's an effect due to the thermal expansion and contraction of the components involved and depending on the conditions these pieces operate within that will determine what goes.
It's really nothing more than stating a natural effect of this repeated heating and cooling, when the pieces get warm they expand and want to squeeze to each other a little tighter displacing the TIM, when they cool they don't squeeze as hard so less TIM displacement happens - If they get cold thats when you see the opposite effect.
The extreme expression of it happens when things are subzero, in that event it's known as "Cracking" of the TIM. That's when the pieces get so cold they actually begin to pull away from each other and since the TIM is present (And frozen too) it becomes solid and can then be cracked by the force of this separation effect, leading to actual loss of cooling conduction.
XOC guys like me deal with this all the time BTW.

You must remember things like the CPU's lid/core and the cooler are solid, not liquid or even "Soft", the amount of movement from this thermal effect you can't perceieve yet it's there. As for the TIM, it's in a liquid or semi-liquid state so it "Flows" and that's what it's supposed to do by what it is and how it has to work.

Once TIM is present and the pieces have been installed with the expected mounting pressures the natural squeeze effect takes place, displacing some of the TIM to another spot. This means it moves along, either to just another spot or outside of the area where the cooler and CPU lid/core are together under pressure.

Since TIM is supposed to fill in the microscopic pits and valleys of these machined surfaces it will either find one to settle into or just keep moving along.

With each heating and cooling cycle it gets moved a little further, it can't move into a pit or valley that already has TIM settled in it largely because you can't compress a liquid/semi-liquid so it's forced to keep moving along, eventually finding a place it can settle into or winding up outside the surface area of the cooler and CPU lid/core.
It can displace what's there but you can never get more of it packed within the same spot period, something has to give or remain. If it's displaced that bit of it moves along too like the rest does.

What remains is just that, the TIM that did manage to settle into such a pit or valley and it will remain there, esp after the expected drying out takes place with that making it more of a solid than it was when first applied since when it's more of a solid so it doesn't get displaced readily if at all.

Obviously thicker TIMs will not move along as readily, thinner ones will in comparison.
The thinning that takes place from heat helps it to find and get into these places to fill in these gaps that exists on the surfaces of the cooler and CPU lid/core and it's supposed to do that. If there is no longer any spots it can settle into, it has no choice but to keep moving until it does wind up being outside the area the cooler and CPU lid/core "Meets" and you see it as TIM squeeze-out, hence the "Pump-Out" effect.

This can also be determined by just how flat these surfaces are to each other, the flatter/better they are to each other the more TIM you'll see getting outside the area. By the same token the less perfect or in this case, square to each other the less you'll have coming out.
Yes, it's possible for a cooler to not be perfectly square due to machining imperfections such as a wavy or slightly curved cooler surface for example - We've all seen it before to know it's a reality.

Another thing is in how flat/smooth these surfaces are when made in the first place.
Even with surfaces that are flat to each other, that in itself doesn't mean you don't have alot of places/spot it can settle into for holding it.
A more perfectly machined surface will have less spots (pit and valley's) so you'll have less TIM settling in since it's not needed in this case.

You also have to remember the thermal expansion and cooling of these pieces/surfaces doesn't stop, if it's being used you'll have that happening. Once enough has been "Pumped Out" with all the little pits and valley's filled, things will stabilize and this effect related to the TIM no longer happens.

Perhaps my explanation isn't perfect but that's what I've noticed over time and experience with it and believe me, I can go through some TIM in a very short amount of time with what I do.

DeepL Translate: The world's most accurate translator

Translate texts & full document files instantly. Accurate translations for individuals and Teams. Millions translate with DeepL every day.

www.deepl.com

deepL! my best friend with those complex textes! perfect written text.

and this is just the perfect describtion how it works!
like the sand and stones in a river. they move not themselves. they get moved by external motion

 

[Bones]

DeepL Translate: The world's most accurate translator

Translate texts & full document files instantly. Accurate translations for individuals and Teams. Millions translate with DeepL every day.

www.deepl.com

deepL! my best friend with those complex textes! perfect written text.

and this is just the perfect describtion how it works!
like the sand and stones in a river. they move not themselves. they get moved by motion

Thanks!
For what I do, understanding how TIM works is important like all else related.
To make the point I took a screenie and it shows I'm right there with Der8auer himself, maybe I had to run more MHz (Due to lack of dual channel with the chip) but I got it. If you can't get everything right when on the edge you won't get anything in return for your efforts.
He's #1 all-time for FX in 1M Super PI, I'm #2 for the same but know the MHz I was running IS currently the WR for the clockspeed ran in any benchmark period regardless of chip/socket to get an actual result, not just for a CPU-Z validation. There are validations that are higher in MHz but not for benchmark results with the clockspeed I was running.

To catch him I have more work to do and ATM I'm working on a few things for that.
Maybe I will, maybe not but should be fun anyway.

 

[plastiscɧ]

Thanks!
For what I do, understanding how TIM works is important like all else related.
To make the point I took a screenie and it shows I'm right there with Der8auer himself, maybe I had to run more MHz (Due to lack of dual channel with the chip) but I got it. If you can't get everything right when on the edge you won't get anything in return for your efforts.
He's #1 all-time for FX in 1M Super PI, I'm #2 for the same but know the MHz I was running IS currently the WR for the clockspeed ran in any benchmark period to get an actual result, not just for a CPU-Z validation. There are validations that are higher in MHz but not for benchmark results with the clockspeed I was running.

To catch him I have more work to do and ATM I'm working on a few things for that.
Maybe I will, maybe not but should be fun anyway.
View attachment 204707

haha. der Bauer! well he has BEEF with another very smart older Guy/man which is famous for his PC-Tech blogs as well on Youtube. they are so salty nearly SULFUR to eachother. Igor said about the Bauer that he is a true Marketing-Whore.

let me find out what he says.

the screenshot is impressive.

this is DER Bauer's biggest Opponent _IGOR_
gimmie some time to make some research. but i wanted to show u the nightmare for this young guy in person.

 

[Bones]

I woudn't doubt it.
Some of what happens is sheer luck, some is skill and knowing how to tweak.
I've always known for each of us that does what we do there is another out there that could be even better at it, so I bear in mind what one has done another can do too.

I'm not the best by any means, I just know what I know and hope it's enough whenever I go for it.

 

[plastiscɧ]

I woudn't doubt it.
Some of what happens is sheer luck, some is skill and knowing how to tweak.
I've always known for each of us that does what we do there is another out there that could be even better at it, so I bear in mind what one has done another can do too.

I'm not the best by any means, I just know what I know and hope it's enough whenever I go for it.

i think we all are enthusiasts. and if there is something not precise enough there is always space to correct it in a fair and positive way. no master has ever fallen from the sky by just random..

 

[lexluthermiester]

However, it does solve the pump-out effect.

Well, there IS that!

Folks we're getting a bit off topic here. Let's rope it in..

 

[freeagent]

I may have had some pump out on my 980 with the TIM I had in there before.. was used to a 62c load that turned into 72c.. I've got some thick stuff in there now, load is now mid 50s with a hotspot in the mid 60s I think? I haven't spent a ton of time on it..

 

[Falkentyne]

I woudn't doubt it.
Some of what happens is sheer luck, some is skill and knowing how to tweak.
I've always known for each of us that does what we do there is another out there that could be even better at it, so I bear in mind what one has done another can do too.

I'm not the best by any means, I just know what I know and hope it's enough whenever I go for it.

One thing I should say is that Conductonaut is nowhere near 73 w/mk. Just like Thermalright Silver King's 79 w/mk is even more in outer space.
I made my own Galinstan (liquid metal) from raw materials, and Galinstan is 16.5 w/mk, and I was only 1C behind Conductonaut when tested on the same laptop.
When you have an Eutectic alloy, the w/mk of the entire mixture is always going to be lower than the lowest w/mk of the individual components.
In some deep buried post on notebookreview forums, that I have no hope of finding, someone said that Conductonaut was tested in a lab to around 30 w/mk.

Another property of LM is its thermal conductivity increases as temp rises, while with normal TIM, it decreases.

 

[plastiscɧ]

One thing I should say is that Conductonaut is nowhere near 73 w/mk. Just like Thermalright Silver King's 79 w/mk is even more in outer space.
I made my own Galinstan (liquid metal) from raw materials, and Galinstan is 16.5 w/mk, and I was only 1C behind Conductonaut when tested on the same laptop.
When you have an Eutectic alloy, the w/mk of the entire mixture is always going to be lower than the lowest w/mk of the individual components.
In some deep buried post on notebookreview forums, that I have no hope of finding, someone said that Conductonaut was tested in a lab to around 30 w/mk.

Another property of LM is its thermal conductivity increases as temp rises, while with normal TIM, it decreases.

and u put this liquid-metal fed mixture on the head of a CPU?
help me alittle bit. i am not that familiar with extreme OC "sports"

 

[Shrek]

Here is an AMD report on Thermal Pads vs. Thermal Grease
Thermal Interface Material Comparison: Thermal Pads vs. Thermal Grease (chipdb.org)
and it mentions the pump-out effect, so I guess we can finally take it as real.

It also recognizes the need for the "Long-Term Stability and Reliability of the Material"

 

[Falkentyne]

and u put this liquid-metal fed mixture on the head of a CPU?
help me alittle bit. i am not that familiar with extreme OC "sports"

Yes you apply it to a BGA Direct die CPU and you insulate the area around the CPU so the LM won't spill off onto something and destroy the laptop. As long as the heatsink is a copper base and is reasonably flat, with somewhat decent mounting pressure, it works.

 

[plastiscɧ]

Here is an AMD report on Thermal Pads vs. Thermal Grease
Thermal Interface Material Comparison: Thermal Pads vs. Thermal Grease (chipdb.org)
and it mentions the pump-out effect, so I guess we can finally take it as real.

It also recognizes the need for the "Long-Term Stability and Reliability of the Material"

this article can be def. considered pro-pad (carbon sleeve).

at least it is written objectively and benevolently towards the "genre" and does not treat these materials as stepmotherly as die-hard paste fans would possibly do because they suspect that the "age of oil" is coming to an end and the mass with which some like to "lubricate" the CPU will soon be a thing of the past...

yes well now the objectivity is already lost to me again.
to be fair. some numbers would have been nice as well

 

[Shrek]

Here are some numbers for thermal grease

• Power 100 W
• Area 3 cm^2
• Thickness 0.05 mm
• Thermal conductivity 5.6 W/K m

resulting temperature drop ~1°C; so why sweat going much above 5 W/K m for thermal grease

A graphite pad is thicker, perhaps 0.2mm uncompressed but has a thermal conductivity of 35 W/K m which brings us back to a little under 1°C (better if it compressed).

Had I used 4cm x 4cm, things would be a good bit better compared to the 3cm x 3cm I actually used.

 

[plastiscɧ]

Here are some numbers

• Power 100 W
• Area 3 cm^2
• Thickness 0.05 mm
• Thermal conductivity 5.6 W/K m

resulting temperature drop ~1°C; so why sweat going much above 5 W/K m for thermal grease
View attachment 204739
A graphite pad is thicker, perhaps 0.2mm uncompressed but has a thermal conductivity of 35 W/K m which brings us back to a little under 1°C (better if it compresses).

Interesting that it states 'single use'

Carbon fibers are electrically and thermally very good conductors, the electronegativity EN has a very high value of 2,50. The difference to iron (EN=1.64) is very high at 0.86, which is already considerably corrosive in the presence of an electrolyte. By comparison, the difference between iron and aluminum (EN=1.47) is only 0.17. Carbon fibers also have a negative coefficient of thermal expansion in the longitudinal direction at lower temperatures. When heated, they therefore initially become shorter and thicker.**

The two aforementioned properties result in the compelling need to insulate carbon fiber-based components from other metallic components both mechanically and electrically

** ergo they DO NOT move and stay between the materials and "block" their position

 

[lexluthermiester]

Here is an AMD report on Thermal Pads vs. Thermal Grease
Thermal Interface Material Comparison: Thermal Pads vs. Thermal Grease (chipdb.org)
and it mentions the pump-out effect, so I guess we can finally take it as real.

Agreed.

So now the question is: How serious of a problem(if any) is it for PC components and DIY/Boutique/Enthusiast PC builders?

this article can be def. considered pro-pad (carbon sleeve).

No, this thread is about "Pump-out" effect of fluidic/paste based TIM. Discussion of Thermal Pads needs it's own thread of discussion if users want to flesh out more information about them.

 

[Shrek]

Now things get complicated... then again some people here say they have experienced elevating temperatures with age, and I can only guess that is due to pump-out, but it is only a guess. I'm still wondering about surfactant additions to have the grease fight its way back in.

Well graphite pads might be argued to be a solution to pump-out, but at about $3 a pad compared to the thermal grease I use (about 1¢ an application) I also need to be convinced.

 

[lexluthermiester]

but at $3 a pad

WOW! That's cheap! The least expensive quality pad I can find is $8 for 40x40mm.

 

[robot zombie]

Here is an AMD report on Thermal Pads vs. Thermal Grease
Thermal Interface Material Comparison: Thermal Pads vs. Thermal Grease (chipdb.org)
and it mentions the pump-out effect, so I guess we can finally take it as real.

It also recognizes the need for the "Long-Term Stability and Reliability of the Material"

I kind of wonder if this is part of the reason for their 'concave' IHS.

Now things get complicated... then again some people here say they have experienced elevating temperatures with age, and I can only guess that is due to pump out, but it is only a guess.

I always just chalked that up to the gradual deterioration of the paste.

260°C will do for me

I like that Super Lube for many things. It's a good general dielectric grease. Gotta love that teflon stuff. Good at keeping moisture out. Really great on bikes and really any similar mechanisms. Maybe even the more rugged low voltage contacts outdoors. Never thought to plop it around a cpu socket though

 

[plastiscɧ]

Agreed.

So now the question is: How serious of a problem(if any) is it for PC components and DIY/Boutique/Enthusiast PC builders?


No, this thread is about "Pump-out" effect of fluidic/paste based TIM. Discussion of Thermal Pads needs it's own thread of discussion if users want to flesh out more information about them.

your question is relatively easy to answer. i hope

big components = paste ofc. the good/proper one
thin, slim and narrow build components = carbon

it should make sense in some manner and serve the ppls needs (andy has budged e.g.).


maybe i can make a technical comparison with a different example:

normal cars run with accu-batteries in the future. fair!
trucks, trains, and busses just can't - too heavy. they need the stronger hydrogenium H1 BUT it's only 50% as effective and quite expensive to create like nowadays engines.

taking advantage of costs is a must here. then this will be the right patch to success.

squeezed out pasted and ruined components are expensive to restore.
present GPU for notebooks ~850€

 

[Shrek]

WOW! That's cheap! The least expensive quality pad I can find is $8 for 40x40mm.

Amazon.com: IC Graphite High Compression Thermal Pad (40X40mm 5-Pack) …: Computers & Accessories

I like that Super Lube for many things. It's a good general dielectric grease. Gotta love that teflon stuff. Good at keeping moisture out. Really great on bikes and really any similar mechanisms. Maybe even the more rugged low voltage contacts outdoors. Never thought to plop it around a cpu socket though

I meant the SuperLube thermal grease, not the straight silicone grease or the one with PTFE.

 

[plastiscɧ]

I kind of wonder if this is part of the reason for their 'concave' IHS.


I always just chalked that up to the gradual deterioration of the paste.

I like that Super Lube for many things. It's a good general dielectric grease. Gotta love that teflon stuff. Good at keeping moisture out. Really great on bikes and really any similar mechanisms. Maybe even the more rugged low voltage contacts outdoors. Never thought to plop it around a cpu socket though

we/I use this for anything whats electric to clean and lube. even the bycicles chain. keep out liquids and delete rust as well

 

[robot zombie]

Amazon.com: IC Graphite High Compression Thermal Pad (40X40mm 5-Pack) …: Computers & Accessories


I meant the SuperLube thermal grease, not the straight silicone grease or the one with PTFE.

Ahhhh, haha. Gotcha. Why do the tubes all look the same?

we/I use this for anything whats electric to clean and lube. even the bycicles chain. keep out liquids

Good stuff. I have so much CRC around though. Nasty stuff. Probably canned cancer. But then so is brake cleaner and that gets a lot done. Completely different products from anything in this thread though